Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Ragnar Nurkse's balanced growth theory wikipedia , lookup
Fear of floating wikipedia , lookup
Business cycle wikipedia , lookup
Globalization and Its Discontents wikipedia , lookup
Economic growth wikipedia , lookup
Monetary policy wikipedia , lookup
Early 1980s recession wikipedia , lookup
Phillips curve wikipedia , lookup
MACRO POLICIES TARGETING LOW INFLATION: REVISITING THE CONVENTIONAL WISDOM IN SELECTED DEVELOPING COUNTRIES OF ASIA A thesis submitted in fulfilment for the degree of Doctor of Philosophy Ahmed Taneem Muzaffar School of Business 2013 DEDICATION To Ishrat and Talal ii Acknowledgements All praise be to All Mighty for His infinite generosity without which I could not accomplish this task. Faith in Him sustained me when I was in doubt and guided me when I felt lost. My profound gratitude is to my supervisors – Professor P. N. (Raja) Junankar, Professor Anis Chowdhury, and Professor Ronald A. Ratti – for providing me with important academic support. Both Raja Junankar and Anis Chowdhury were instrumental in encouraging me to think about broader developmental issues rather than focussing on a narrow technical work. Their intellectual support instilled confidence in me, which helped me think outside the box and to provide a sharp critique of some of the conventional ideas of mainstream economics. Three of them continuously pressed me to improve the clarity of my writings. This helped me understand the aspects of theoretical and applied economics that were most difficult to master. They also provided me moral and psychological support when I was struggling at the initial stage. The privilege of receiving guidance from such intellects shall remain an inspirational experience for my career in the days ahead. Outside the panel of supervisors, I would like to thank Dr. Girijasankar Mallik, Senior Lecturer of School of Business for his enthusiastic support, especially when I needed to understand issues relating to applied econometrics. His door was always open even though he was not on my supervisory panel. I am, indeed, grateful to him. I am also extremely grateful to Dr. Zulfan Tadjoeddin for his constant material, intellectual, and moral supports. Quite often he was the bridge between Professor Anis Chowdhury and myself since Professor Chowdhury’s move to the United Nations in October 2008. My foremost gratitude is to the University of Western Sydney for awarding me with the International Postgraduate Research Scholarship and UWS International Award. Without this financial support I could not have embarked upon my PhD study. I also express my gratitude to Professor Andrew Cheetham, Pro Vice-Chancellor (Research) for allowing me to submit the thesis with a tuition waiver. I should also iii thank the School of Business for approving release of funds to attend conferences at the University of California at Berkeley, USA and the University of New South Wales, Australia. The feedback I received from the conference participants was important in strengthening the arguments of this work. I am also indebted to Professor John Lodewijks, Head of School of Economics for his support and giving me an opportunity to teach in a number of units as a tutor. Besides helping me financially, this has been intellectually very rewarding for me. I also gained experience in teaching at a higher education institution. My thanks also go to Craig Berry, Carolyn Love, Bec Campisi, and Barbara Pinning for their support on administrative issues. Finally, I should like to thank my family members, my wife, son, parents and parents-in-law, particularly for their tremendous moral support. My parents shared their own gruelling experiences of pursuing postgraduate studies and their words helped me keep going. I am specially indebted to my wife Ishrat Khan who has sacrificed her own higher studies and potential career in order to help me achieve my own. Last but not least, I thank my little son Ahmed Talal Muzaffar for his tolerance of my endless excuses for why I could not play with him. iv Statement of Authentication The work presented in this thesis is, to the best of my knowledge and belief, original except as acknowledged in the text. I hereby declare that I have not submitted this material, either in full or in part, for a degree at this or any other institution. ……………………………………………… Ahmed Taneem Muzaffar University of Western Sydney v Table of Contents Acknowledgements iii Statement of Authentication v Table of Contents vi List of Tables viii List of Figures x Abstract xi Introduction: The Wisdom of Low Inflation in Developing Countries 1 1.1 An Overview 1 1.2 The Policy of Low Inflation in the Developing Countries and the IMF 4 1.3 Growth under Moderate Inflation: Experience from Selected Developing 7 Chapter 1 Countries 1.4 The Nature of Inflation Growth Relationship 11 1.5 The Research Objectives of the Thesis 19 The IMF and the Policy of Low Inflation: A Review of Article IV 21 Chapter 2 Consultation for Selected Developing Countries of Asia 2.1 Introduction 21 2.2 A Review of Related Literature 23 2.3 A Review of the IMF Article IV Staff Consultation Reports in Selected 28 Developing Countries of Asia 2.4 Chapter 3 2.3.1 Bangladesh 29 2.3.2 Bhutan 31 2.3.3 Cambodia 32 2.3.4 China 34 2.3.5 India 35 2.3.6 Indonesia 36 2.3.7 Lao PDR 37 2.3.8 Malaysia 38 2.3.9 Maldives 39 2.3.10 Nepal 40 2.3.11 Pakistan 41 2.3.12 The Philippines 42 Conclusion 44 Cross-Country Evidence on Inflation-Impact on Growth: A Review 46 of Related Literature 3.1 Introduction 46 3.2 Cross-Country Evidence of Growth-Inflation Nexus 46 3.3 Conclusion 55 vi Chapter 4 An Enquiry into the Relationship between Inflation and Growth in 57 Selected Developing Countries of Asia 4.1 Introduction 57 4.2 Trends in Inflation and Growth in Some Selected Developing Countries 58 of Asia 4.3 The Nature of Relationship between Inflation and Growth 64 4.4 Identifying a Threshold Level of Inflation 76 4.5 Experiences with Policies Targeting Low Inflation 87 4.6 Determinants of Inflation: Demand Driven versus Supply Side 96 4.7 Conclusion 107 A Panel Study on Growth-Inflation in Selected Developing Countries 109 Chapter 5 of Asia 5.1 Introduction 109 5.2 Estimation Methods 111 5.2.1 Selection of variables and summary statistics 113 5.2.2 Empirical model 128 Empirical Results 131 5.3.1 Static panel estimation results 134 5.3.2 Dynamic panel estimation results 139 5.3.3 Summary of the results from static and dynamic panel estimations 145 5.3.4 Robustness checks and comparing the differences in the threshold 147 5.3 levels of inflation 5.4 Chapter 6 Conclusion 152 Appendix to Chapter 5 153 The Relationship between Inflation and Growth: The Bangladesh 156 Experience 6.1 Introduction 156 6.2 An Overview of the Bangladesh Perspective 157 6.3 Trends in Growth and Inflation in Bangladesh 163 6.4 A Time Series Analysis of the Growth-Inflation Relationship 166 6.5 Identifying the Threshold Level of Inflation in Bangladesh 170 6.6 Conclusion 181 Chapter 7 Conclusion 183 References 196 vii List of Tables Table 1 Average Annual Real GDP Growth and Inflation Rates 1950-2010 9 Table 2.1 Inflation Targets and IMF Policy, Selected Recent Studies 24 Table 2.2 Inflation Targeting Countries (Emerging and Developing Countries) 25 2011 Table 2.3.1 Average Growth-Inflation Figures in Bangladesh 29 Table 2.3.2 Average Growth-Inflation Figures in Bhutan 31 Table 2.3.3 Average Growth-Inflation Figures in Cambodia 32 Table 2.3.4 Average Growth-Inflation Figures in China 34 Table 2.3.5 Average Growth-Inflation Figures in India 35 Table 2.3.6 Average Growth-Inflation Figures in Indonesia 36 Table 2.3.7 Average Growth-Inflation Figures in Lao PDR 37 Table 2.3.8 Average Growth-Inflation Figures in Malaysia 38 Table 2.3.9 Average Growth-Inflation Figures in Maldives 39 Table 2.3.10 Average Growth-Inflation Figures in Nepal 40 Table 2.3.11 Average Growth-Inflation Figures in Pakistan 41 Table 2.3.12 Average Growth-Inflation Figures in The Philippines 42 Table 2.4 Median CPI Inflation and Growth in Countries in South Asia, Pre and 44 Post 1980 Table 3 Selected Cross-Country Studies on the Relationship between Growth 47 and Inflation Table 4.1 Summary Statistics of CPI Inflation in 15 Developing Countries of 58 Asia Table 4.2 Summary Statistics of Real GDP Growth in 15 Developing Countries 59 of Asia Table 4.3 Studies on Inflation-Growth Causality 65 Table 4.4 Studies on Threshold Level of Inflation 77 Table 4.5 Studies on Low Inflation Targeting 87 Table 4.6 Inflation and Growth Statistics in Indonesia and Thailand, Pre and 94 Post IT Implementation Table 4.7 Determinants of Inflation Table 5.1 Summary Statistics of Growth Rates of Selected Variables, 1961- 97 125 2010 Table 5.2 Correlation Matrix of Growth Variables 126 Table 5.3 Impact of Inflation on GDP Growth, OLS Estimations 131 Table 5.4.1 Fixed and Random Effects Panel Regressions (Case 1) 135 Table 5.4.2 Fixed and Random Effects Panel Regressions (Case 2) 137 Table 5.4.3 Fixed and Random Effects Panel Regressions (Case 3) 138 viii Table 5.5.1 Impact of Inflation on Growth, Dynamic Panel Estimations 140 Table 5.5.2 System GMM Estimations (Case 1) 142 Table 5.5.3 System GMM Estimations (Case 2) 144 Table 5.6.1 Summary of Results from Static and Dynamic Panel Estimations 146 (Case 1) Table 5.6.2 Summary of Results from Static and Dynamic Panel Estimations 147 (Case 2) Table 5.7.1 Inflation Turning Points in Different Time Frames 148 Table 5.7.2 Impact of Inflation on Growth, the Regional Effect 149 Table 5.7.3 Averages of Selected Variables in the Sample Countries, 1990-2010 150 Table 5.7.4 The Impact of Inflation on Growth Based on Structural 151 Characteristics of the Economy Table A.5.1 List of Variables and Their Sources 153 Table 6.1 Key Demographic and Geographic Features of Bangladesh 158 Table 6.2 Conditions of IMF’s US $ 1 billion Loan to Bangladesh, 2010 161 Table 6.3 Decadal Averages of Growth and Inflation in Bangladesh, 1971-2009 164 Table 6.4 Unit Root Tests with Real GDP, CPI, Growth, and Inflation, 1977- 167 2009 Table 6.5 Structural Breakpoint Test for Real GDP Growth in 1992 169 Table 6.6 Cointegration Test Results 170 Table 6.7 Impact of Inflation on Growth, Conditioned upon Other Covariates, 172 1977-2009 ix List of Figures Figure 1 Scatter Plot of Inflation and Growth in 150 Emerging and 10 Developing Countries, 1961-2011 Figure 4.1.a Inflation and Growth in South Asia – Bangladesh, India, Pakistan, 61 and Sri Lanka Figure 4.1.b Inflation and Growth in South Asia – Bhutan, Maldives, and Nepal 62 Figure 4.1.c Inflation and Growth in East Asia – Indonesia, Malaysia, 63 Philippines, and Thailand Figure 4.1.d Inflation and Growth in China and East Asia – Cambodia, Lao 64 PDR, and Vietnam Figure 4.2.a Scatter Plot and Linear Fitted Line in Four Countries of South Asia 74 Figure 4.2.b Scatter Plot and Polynomial Kernel Fit Curve in Four Countries of 75 South Asia Figure 4.2.c Scatter Plot and Linear Fitted Line in Four Countries of East Asia 75 Figure 4.2.d Scatter Plot and Polynomial Kernel Fit Curve in Four Countries of 76 East Asia Figure 4.3 Inflation and Growth in Selected Countries (Long Run Median in 82 per cent) Figure 4.4 Scatter Plot between Average Real GDP Per Capita and Long Run 83 Inflation Figure 5.1 Inflation and Growth in Selected Asian Countries, 1961-2010 111 Figure 5.2 Distribution of Real GDP Growth and Real GDP Per Capita 114 Growth in Selected Developing Countries of Asia, 1961-2010 Figure 5.3 Distribution of Inflation in Selected Developing Countries of Asia, 116 1961-2010 Figure 5.4 Relationship between Real GDP Growth and Inflation 117 Figure 5.5 Deviation from Trend of Growth Rates in Oil and Commodity Price 124 Indexes, 1961-2010 Figure 5.6 Mean of Cross-Country Data for East Year, 1960-2010 128 Figure 5.7 Conditional Correlation between Inflation and Growth, 1961-2010 134 Figure 6.1 Real GDP Growth and CPI Inflation in Bangladesh, 1971-2009 164 Figure 6.2 Growth Rate, Actual and Fitted, 1977-2009 177 Figure 6.3 Scatter Plot between Unexplained Growth and Inflation, 1977-2009 177 Figure 6.4 Parameter Stability Test using CUSUM Test 178 Figure 6.5 Parameter Stability Test using CUSUMSQ Test 178 Figure 6.6 Impulse Response Analysis for Growth and Inflation 179 x Abstract This thesis questions the validity of macroeconomic policies that emphasise targeting low inflation rather than economic development. It provides systematic research in the context of Asian developing countries to raise doubts about this conventional wisdom of keeping inflation at a low single digit level. In this connection, the study addresses three questions relating to the growth-inflation nexus. The first question is whether there exists an inflation threshold at 5 per cent; in other words, whether inflation becomes harmful to economic growth when it exceeds 5 per cent. This is because the International Monetary Fund (IMF) advises developing countries to restrict the inflation rate to not more than 5 per cent per annum. It is regarded as safe and conducive to long term economic growth. The second question is whether the threshold level varies according to country-specific circumstances and levels of development. The purpose is to examine the validity of IMF’s policy advice of keeping inflation at 5 per cent irrespective of country-specific circumstances. The final question is on the nature of policy suggestions relating to growth and inflation in country reports by the IMF. This is to determine whether the advice in the IMF reports have any reflection of the need for a change suggested by leading economists, including the Chief Economist of the IMF, in the aftermath of the global economic crisis of 2008-2009. The study examines the research questions, broadly, in two ways. First, we resort to a critical review of existing literature and an in-depth content analysis of the IMF Article IV country consultation reports. The second method provides an empirical analysis using panel and time series estimation techniques. For both types of analyses, our sample includes selected developing countries from Asia. Although a number of studies in recent years have argued against policies targeting low inflation, both empirically and analytically, a systematic empirical research on Asian developing countries is limited. Besides, our interest in Asian economies, particularly the Asia-Pacific region, is due to their good performance in terms of both growth and macroeconomic stability. The selection of countries, however, is influenced by the availability of data on growth, inflation, and other control variables; existence of literature pertaining to the research issues; and release of country reports by the IMF xi during 2009-2010, around the time of the global economic crisis. For instance, country-specific literature review and content analysis of the IMF country reports include 15 and 12 South and Southeast Asian countries (including China), respectively. On the other hand, the cross-country panel study includes 14 countries covering economies from South Asia, Southeast Asia, Asian countries within the Commonwealth of Independent States (Asian countries of the former Soviet Union), and Small Island States (SIDS). The empirical analysis using panel estimation models uses annual observations between 1961 and 2010. The data are gathered from standard sources such as World Bank’s World Development Indicators and the IMF’s International Financial Statistics and World Economic Outlook. Both static models such as Fixed Effects and Random Effects models, and dynamic panel estimation model (System Generalised Method of Moments, SGMM) are used. Our methods have a number of strengths over the methods used in the existing literature. First is the use of annual observations instead of five or ten year averages. The relationship between inflation and growth becomes stronger with greater frequency of observations. Second is the use of SGMM since the technique is the most advanced so far, in order to take care of the endogeneity problem. The empirical investigation also includes time series regression analyses in a country-specific context. We examine the case of Bangladesh between 1977 and 2009. Bangladesh is selected because the country is widely regarded as an interesting example in the development discourse. The findings of the research provide strong empirical evidence that an inflation rate beyond 5 per cent is not necessarily detrimental to economic growth. The crosscountry panel study shows that the threshold varies between 8 and 15 per cent. Similarly, results from country-specific study on Bangladesh reveal that the threshold is around 15 per cent. Empirical evidence also suggests that the threshold level varies according to the levels of development. Poorer and less developed countries demonstrate a higher inflation tolerance. Analyses from the IMF country reports show that policy advice of achieving low inflation at around 5 per cent or less continues to receive priority in general. There is little evidence that IMF’s policy prescription reflects the need for a change in the wake of the recent global economic xii crisis. Nor is it consistent with country-specific historical experience on inflation and growth. The results, in short, help to conclude that the empirical basis for maintaining inflation at 5 per cent or below in developing countries of Asia is weak. We also conclude that targeting inflation should be done after a careful consideration of country-specific circumstances and levels of development. Taking all these into account, the IMF should rethink its macroeconomic policy advice to developing countries and policy makers should highlight their country-specific circumstances during Article IV consultations with the IMF officials or when they negotiate with the IMF to access its support facilities. Thus the analyses and findings of this thesis reveal its two important contributions. First, it provides a unique contribution to the literature pertinent to the growth-inflation nexus. Secondly, it helps rethinking macroeconomic policy making in the developing countries that still follow a rule of low inflation targeting policies. Macroeconomic policies require a broader perspective, creating a balance between the need for stabilisation and development. The study suggests that the developing countries should give priority to poverty reduction and employment creation instead of pursuing a restrictive policy of low inflation. JEL Classification: E31, O40. Keywords: Inflation threshold; growth; Asian developing countries; panel estimation; IMF macroeconomic policies. xiii Chapter 1 Introduction: The Wisdom of Low Inflation in Developing Countries 1.1 An Overview Attaining price stability by keeping inflation low receives priority in macroeconomic policies of the developing countries in recent decades. This follows episodes of high and accelerating inflation accompanied by growth stagnation during the 1970s and the 1980s, especially in Latin American countries. There seems to be a strongly held belief by the mainstream economics that targeting low inflation is good for achieving macroeconomic stability and therefore beneficial for long term economic growth. The experiences of the “Great Inflation” of the 1970s and the “Great Moderation” during 1993-2007 have led to a growing consensus on this.1 Founded on this belief, policies targeting low inflation have been spearheaded by international financial institutions such as the International Monetary Fund (IMF). The IMF strongly advocates the developing countries to pursue macroeconomic policies targeted at keeping inflation rates within 5 per cent. As a result, central to the economic management of some of these countries, compelled to embrace IMF suggested economic reforms, has been to restrain inflation at 5 per cent. Typically, these policies include tight monetary policy and reining in fiscal deficits because of their supposed link to cause inflationary pressure.2 This policy of fighting inflation first, keeping it as low as 5 per cent, is suggested even at the expense of immediate adverse impacts on growth on the ground that this would foster sustained growth in the future. That is, the argument is couched in terms “short-run pain for long-run gain”. 1 The 1970s was a period of high inflation, caused mainly by commodity price shocks and the breakdown of the Bretton Woods system as a result of high US inflation. The Great Moderation was characterised by an unusually high degree of macroeconomic stability, with steady growth and low and stable inflation in most of the advanced economies since 1993 until the Great Recession hit in 2008. 2 Chowdhury (2006, p. 407) notes that “[o]ne major component of conditionality [by the IMF] has been macroeconomic stabilisation…[which is] solely meant to be nominal stabilisation…of a low inflation (targeted at less than 5 % in the medium term) and a low budget deficit (targeted at near zero primary deficit).” 1 The IMF strongly believes that inflation above 5 per cent is harmful for sustained economic growth and long term prospects for the developing countries. The concern is that inflation might accelerate beyond 5 per cent and runaway inflation may lead to a crisis. The cost of this stabilisation process, then, may turn out to be painful. This sense of apprehension appears to have led the IMF to advise the developing countries to become extremely cautious about inflation. An inflation rate restricted at 5 per cent is regarded as safe. Macroeconomic policies, irrespective of country specific circumstances, are therefore aimed at achieving this target. This is clearly not what IMF’s Article of Agreement states that “each member shall endeavor to direct its economic and financial policies toward the objective of fostering orderly economic growth with reasonable price stability, with due regard to its circumstances” (Article IV, I(i)).3 It seems that now the IMF believes this one policy fits all works the best. The nature of relationship between inflation and growth has been studied extensively in the literature, with mixed findings – outlining both reasons for inflation being beneficial and detrimental to economic growth. The literature also provides reasons why both low and high inflation may be growth-inhibiting. On the other hand, in the wake of the crisis that led to the Great Recession of 2008-2009, the efficacy and implication of IMF policies have also been questioned.4 Although a number of studies (for example Chowdhury 2006; Epstein & Yeldan 2008) have provided both empirical and analytical arguments questioning the wisdom of keeping inflation essentially at 5 per cent or below, no systematic empirical research is available for developing countries in Asia particularly Asia-Pacific region which have generally done well in terms of both growth and macroeconomic stability. Incidentally, low inflation and high growth experience of East and Southeast Asia vis-â-vis Latin American countries’ hyperinflation and growth stagnation provided a strong rationale for low inflation macroeconomic policy paradigm. The broad generalisation of East and Southeast Asian experience, however, ignores the fact that both South Korea and Indonesia grew rapidly since the mid-1960s until about mid-1980s when inflation rates ranged between 13 and 17 per cent. Poverty also declined dramatically in both Indonesia and South Korea during this period of rapid growth. 3 Source: http://www.imf.org/External/Pubs/FT/AA/index.htm#art4 In fact pointing out that economic crisis sometimes bring about enormous changes, Palley et al. (2012) comment that the “economic crisis that these events have generated, combined with the failure of the mainstream economics profession, has again put the question of change on the table.” 4 2 Therefore, there is a need for systematic investigation of the Asian developing countries. This research aims to do that by thoroughly investigating the nature of inflation-growth relationship in selected developing countries of Asia. The findings of this thesis raise doubt about the validity of the widely held belief that low inflation is always beneficial for growth, particularly in the context of developing countries. The study is unique as it specifically seeks to answer whether inflation beyond 5 per cent is always harmful for growth in some selected developing countries of Asia which are pursuing low inflation targeting policies. Such empirical investigation with particular emphasis on developing countries of Asia is absent to the best of our knowledge. Thus, the findings of this research will have significant implications for macroeconomic policy making in these countries. More broadly, it is a significant contribution to the literature pertaining to growth-inflation nexus. It is also important because despite the policy application suggested by the IMF based on this conventional wisdom, economic performance failed to improve and poverty remained high in many countries, even though inflation fell (see Wilkinson 2000, p. 643). Even the influential people at the Bretton Woods Institutions seem to have acknowledged it. A former president of the World Bank (WB) admits, “…if we take a closer look, we see something else – something alarming. In developing countries, excluding China, at least 100 million more people are living in poverty today than a decade ago. And the gap between rich and poor yawns wider” (Foreword to Thomas et al. 2000). The findings of the thesis reveal that despite suggestions to revise its position by the leading economists, both inside and outside the IMF, in the aftermath of recent global crisis, the IMF seems to continue with its policy prescription of achieving low inflationary environment (usually below 5 per cent) in developing countries. This is evident from the country consultation reports released by the IMF, still emphasising strongly on macroeconomic stabilisation by attaining price stability. The policy advice in the reports shows little reflection of flexibility being suggested even by its own senior economists. The advice, irrespective of country-specific circumstances, seems to focus on achieving certain nominal targets, for example inflation within 5 per cent in particular. It is clearly against the mandate of the IMF to fostering economic growth and attaining reasonable price stability, taking the country specific 3 circumstances into account. This notion of one policy fits all does not find support in the country-specific studies as well as historical evidence on inflation and growth. Likewise, cross-country evidence, as well as empirical analysis of this thesis, finds no justification for claiming that inflation above 5 per cent necessarily deteriorates growth in the developing countries of Asia. Poorer countries appear to show higher tolerance for inflation. In other words, whether inflation would cause an adverse effect on growth depends critically on country-specific circumstances and historical conditions. In short, evidence from the analysis of this research reveals that the conventional wisdom of low inflation has some serious flaws and macroeconomic policies in developing countries should not be concerned only about price stability. Macroeconomic policies need to balance between the need for stabilisation and development. The rest of the chapter is organised as follows. Section 1.2 briefly discusses the rationale behind the IMF policies geared towards keeping inflation within 5 per cent benchmark and the problems relating to such policies. Section 1.3 takes a look at the empirical evidence on growth performance during moderate inflation regimes. Section 1.4 provides a brief literature review pertinent to the nature of the growthinflation relationship. Section 1.5 sets the research agenda of this study, outlining the key research questions it attempts to investigate. 1.2 The Policy of Low Inflation in the Developing Countries and the IMF In response to why inflation should be kept within 5 per cent the IMF notes: “Inflation is the most pernicious tax on low-income households that lack the means to protect their salaries and scant savings against inflation. In addition to the negative impact on the poor, a large body of empirical evidence has established that when (annual) inflation passes the 5 per cent mark investment and economic activity also suffer. It is against this background that 4 the Fund supports policies aimed at achieving or maintaining low inflation.”5 In line with the mainstream economic view, as mentioned in the previous section, the IMF does not seem to acknowledge that ensuring price stability by keeping inflation at 5 per cent may conflict with the objective of fostering economic growth. Antiinflationary policies are believed to ensure both the objectives of low inflation and long term growth – a divine coincidence as termed by Blanchard and Gali (2007). One can, however, put forward strong theoretical and empirical evidence against such a policy stance. Nobel Laureate in Economics, Paul Krugman in his opinion page of The New York Times in 2011 calls this “the low inflation trap.”6 When inflation falls it creates a deflationary expectation and thus even if nominal interest rate is kept at a very low level7 real interest rate continues to rise. This leads to a higher real cost of borrowing and eventually depresses the economy.8 Therefore, pursuing a very low single digit inflation rate may create a stabilisation trap, a situation attributable to low inflation and insufficient growth for poverty reduction (Chowdhury 2006, p. 409). A couple of decades ago, another influential economist James Tobin (see Tobin 1987) also pointed out the danger of paying too much attention to inflation control. On empirical side, Akerlof and Shiller (2009) bring in the example of Canada to highlight the danger of placing too much emphasis on attaining a low inflation target at any cost. In the early 1990s, the Bank of Canada vigorously pursued restrictive monetary policies to bring down inflation from 4.8 per cent to 1.8 per cent. The authors note that this had a terrible cost as unemployment reached 11.3 per cent in 1992, a level unprecedented since the Great Depression. Yet the then Governor of the Central Bank of Canada commented that the costs of lowering inflation were 5 Source: http://www.imf.org/external/np/exr/faq/progfaqs.htm#q4 Paul Krugman, “The Low Inflation Trap (Slightly Wonkish),” The New York Times (23 September 2011); available at http://krugman.blogs.nytimes.com/2011/09/23/the-low-inflation-trap-slightlywonkish/ 7 It cannot go below zero. 8 The Japanese economy is a classic example in this regard. This idea is also captured in the Keynesian Liquidity Trap. There is however an opposing view which considers higher inflation rate to cause higher cost of borrowing. Rising inflation leads to higher inflation expectation. Fisher’s equation explains that this would lower the real return for lending and hurt lenders. As a result lenders tend to charge higher nominal interest rate. 6 5 temporary, but there is a permanent gain in terms of changed expectations on inflation. However, as the authors note that the recession as a result of this was “deeper and longer than the Great Depression” (ibid., p. 115). In the context of developing countries, a classic example of an economy trapped in a stabilisation trap could be Argentina (see Chowdhury 2006, Note 5, pp. 426-427). In the 1980s, hyperinflation in the country (average annual inflation rate of around 391 per cent) had led to a crisis and the IMF stabilisation program brought down inflation to a single digit level, 1.5 per cent annually. However, as Chowdhury (2006, p. 427) notes that the continuation of the tight macroeconomic policies created a “deflationary-spiral” and the economy plunged into recession, causing unemployment rate to rise from 6.5 per cent in 1991 to 17.5 per cent in 1996. This had dire consequences for the poverty rate (head-count ratio) as well – about 13 percentage point rise in a decade, from 21.8 per cent in 1993 to 34.3 per cent in 2002. Referring to the experience of Argentina, Chowdhury (2006) alerts that while inflation needs to be kept under control, too much emphasis on a very low inflation rate may lead to an inadvertent consequence that exacerbates poverty. Epstein and Yeldan (2008) also observe the sacrifice made in terms of growth and employment in developing countries in order to pursue a low inflationary environment. Besides supply shocks are common in developing countries and restrictive policies create challenges for them in such circumstances. The authors note that there are instances where stabilisation policy of price stability, in the hope of achieving higher growth, has not been materialised (ibid. 2008). Ensuring growth by only combating inflation first, and by achieving a single digit rate, may not be as simple as it appears to be. As Rodrik (1999b) explains that growth performance in developing economies may depend on domestic social conditions and its interaction with supply shocks. In the event where social divisions are deep and conflict management is weak, external shocks may magnify the effects and reduce productivity and growth. Against the backdrop of the Great Recession of 2008-2009, the issue of revisiting the traditional macroeconomic tenets has come at the forefront. Even the heads of the IMF have recognised the need for it. For instance, Strauss-Kahn a former managing 6 director of the IMF felt the need for a “wholesale re-examination of macroeconomic policy principles” in the face of the crisis.9 The Chief Economist of the IMF Olivier Blanchard also criticised the one policy one instrument, consensus of the mainstream economists.10 He comments that “beauty is…not…synonymous with truth…[and] there are many targets and there are many instruments.”11 In a different paper, raising doubt about the conventional view of low inflation strategy, Blanchard et al. (2010) make an important observation that higher average inflation and higher nominal interest rates to start with, would have helped to lower interest rates more, thereby most likely reducing the decline in output and deterioration of fiscal positions in the aftermath of global crisis. This criticism against the low inflation strategy is not just shared within the IMF. As The Economist notes that “[e]conomists of highly divergent stripes…[such as] Kenneth Rogoff, Greg Mankiw, Scott Sumner, Paul Krugman, Brad DeLong – all have indicated that higher inflation would be a boon to the economy.”12 1.3 Growth under Moderate Inflation: Experiences from Selected Developing Countries There is much historical evidence from some of the high growth performing countries such as Brazil, Israel, South Korea, and Indonesia that both growth and inflation can move together, during their early phase of development period. For instance in Brazil, throughout the period between the 1940s and early 1980s high growth occurred while there were several bouts of high inflation (Armijo 2005, Table 1, p. 2014). Brazil is often cited as a high inflation high growth counter-example to the common notion that inflation was bad for the economy (Bruno & Easterly 1996). In fact, Latin America in general had double digit inflation rates in the 1950s and 1960s, but economic growth was respectable (Pazos 1972). Growth in Brazil started to decline when the country experienced hyperinflation in the early 1980s. In this 9 Remarks by Dominique Strauss-Kahn at the conference “Macro and Growth Policies in the Wake of the Crisis” at the IMF’s Headquarters in Washington, DC on March 7-8, 2011. 10 The target was inflation and the instrument was policy rate. 11 Source: The speech “Monetary policy in the wake of the crisis” delivered by Blanchard at the IMF Macro Conference, March, 2011; available at http://www.imf.org/external/np/seminars/eng/2011/res/pdf/ob2presentation.pdf 12 The Economist (15 February 2010), “Monetary policy: a healthy dose of inflation.” Available at http://www.economist.com/blogs/freeexchange/2010/02/monetary_policy_1 (this is because of the problem of the zero bound to nominal interest rate.) 7 connection, Baer (1987) explains how orthodox austerity programmes, for which growth had to be sacrificed, failed to control inflation because of its inability to identify the causes of inflation. Citing some development theories suggesting inflation is a good way to mobilise resources for capital accumulation, Bruno and Easterly (1996) argue that there was little in the early experience of developing countries to contradict this view.13 Israel’s economy, the authors state, grew at around 10 per cent per annum between 1948 and 1973, with inflation rate of around 6 per cent to 7 per cent per annum. Both of these figures were double compared to the figures for members of the Organisation for Economic Cooperation and Development (OECD) around the same time. The study argues that “the higher largely anticipated inflation was a price considered well worth paying, especially as widespread indexation of wages, exchange rates, and savings minimised the distortionary costs of inflation” (ibid., p. 139). Bruno and Easterly (1996, Figure 1, p. 140) also show that the per capita growth rate actually rose as one went from single to double digit inflation. Only when the annual inflation rate exceeded 20 per cent did the relationship seem to turn negative. In the Asian region, a good example is South Korea and Indonesia both of which grew rapidly amidst high inflation rate. In explaining the policy attitude of South Korea towards growth and inflation, Chowdhury (1996, p. 150) comments that during the 1970s “Korea’s inflation rate remained above 20 per cent and it maintained its growth rate at about 8 per cent, showing that Korea was more interested in growth than price stability.” South Korea grew by 8 per cent in the 1960s and 1970s when the inflation rates were at double digit levels. The same was true for Indonesia in the 1970s, when the country grew by 7.7 per cent while the inflation rate was over 17 per cent (Table 1). 13 It is interesting to note that both Michael Bruno and William Easterly worked for the World Bank in different capacities. So there is much evidence that policies suggested at country-level reports by the international financial institutions often have little reflection of what the senior members of these institutions think. 8 Table 1: Average Annual Real GDP Growth and Inflation Rates 1950-2010 Decade Republic of Indonesia Republic of Korea Growth Inflation Growth Inflation 1950-59 4.1 22.4 4.7 25.6 1960-69 3.5 196.0 7.7 13.5 1970-79 7.7 17.3 8.2 15.0 1980-89 4.8 9.4 7.8 8.3 1990-97 6.9 8.7 7.5 6.0 2000-2010 5.2 8.2 4.5 3.3 Sources: Frank et al. (1975) and Woo et al. (1994, Tables A1 and A2). Note: Output growth and inflation rates for the Republic of Korea in the 1950s are average annual rates for 1954-1959, excluding the Korean War period, and are based on real GNP and wholesale price index respectively. Contrary, to the widely held views, high inflation did not affect poverty reduction. For example, the poverty rate declined from over 60 per cent in the early 1970s to about 11 per cent in Indonesia just before the Asian crisis hit it in 1997. The poverty rate shot up only after the crisis when inflation rose to 60 per cent and the economy contracted by close to 14 per cent. As noted in Chowdhury (2006), the impact of inflation on poverty depends on a number of factors. For example, when inflation reduces real wages, that should encourage firms to expand employment. Therefore, the net effect of inflation on poverty would depend on the relative elasticities of real wage and employment with respect to inflation. One IMF study by Cukierman et al. (1992) shows that the inflation elasticity of the income (real wage) was only 0.03, while the output (employment) elasticity was 0.94. These estimates are similar to those found by Romer and Romer (1999), implying that, moderate inflation may actually reduce poverty.14 There are also other factors at work. For instance, most of the poor are net debtors who benefit from inflation as it lowers the real value of debt. What matters for the poor is not the aggregate price level or overall inflation, but the prices essential commodities that dominate their consumption basket. Monetary policy is a blunt tool as cannot deal with sector-specific price rises. It can hurt the poor if such policy causes job losses. In general, it is the unskilled workers at the lower end of the labour market who lose their jobs first. 14 There is a caveat here. The argument assumes that real wages decrease with moderate inflation or wages are partially indexed to inflation. 9 In the light of the above discussion, Figure 1 takes a look at the broad experience between inflation and growth from all the 150 emerging and developing countries, as listed by the IMF. The scatter plot between inflation and growth sheds some light on the nature of relationship between these two variables. First, the relationship is nonlinear in nature as evident by the quadratic functional relationship. The signs of the linear and the squared terms of inflation, positive and negative respectively, confirm an inverted U shaped relationship between the two. What is also evident is that in the range between 5 and 15 per cent of inflation there does not seem to be any relationship. Growth does not appear to decline within this range of inflation. This means, the upper part of the inverted-U is flat, displaying a plateau in the relationship rather than a sharp cliff edge. Therefore, the fear of accelerating inflation and sliding down of growth beyond 5 per cent inflation is unfounded. Thus, policy makers can have a good deal of flexibility in dealing with inflation while pursuing strategies for sustained growth. Figure 1: Scatter plot of Inflation and Growth in 150 Emerging and Developing Countries, 1961-2011 y = -0.0032x2 + 0.0598x + 4.3199 20 Real GDP Growth 15 -15 10 5 0 -10 -5 0 5 10 15 20 25 30 35 40 45 -5 -10 -15 CPI Inflation Source: Author’s calculations based on the data from World Bank, World Development Indicators. Notes: Growth is between greater than -10 per cent and less than 15 per cent to avoid the unusual circumstances. Inflation is less than 40 per cent to avoid any outlier effect. 10 This finding motivates us to take a further look into the existing literature to see what it states about the nature of the relationship. This review is carried out in the following section. 1.4 The Nature of Inflation Growth Relationship Understanding the inflation-growth relationship is perhaps one of the most extensively studied macroeconomic topics in the literature. In determining how inflation affects growth, macroeconomic theories offer both conclusions – positive contribution as well as negative impact of inflation on growth. The positive contribution of inflation on growth comes mainly from a Keynesian perspective (for a summary see, for instance, Chowdhury (2002, p. 22)). In the Keynes-Kaldor effect (see Kaldor 1955-56), inflation raises profits of entrepreneurs and therefore redistributes income in favour of profit earners who, compared to workers, have a relatively higher propensity to save.15 The higher profit and savings can be channelled into investment and thus leading to higher economic growth. In the Tobin effect, also known as Mundell-Tobin effect, (see Tobin 1965) inflation helps the economy to become more capital intensive. Higher inflation may raise the share of savings devoted to capital accumulation rather than to hoarding – the accumulation of real money balances – since the real rate of return from financial investment goes down due to inflation (Montiel 2011, p. 300). Thus physical capital intensity is raised. This helps transformation of the economy from a traditional agrobased to a more modern industrial one. What this therefore implies is that inflation is an integral part of growth. The third proposition comes from the Kalecki effect which explains that during inflation, increased seignorage, in other words inflation tax revenue, leads to forced savings by people in an attempt to retain the real value of savings.16 This transfer of resources from households to government allows government to take up necessary 15 The redistribution of income, according to Keynesian perspective, takes place because nominal wages are assumed to be sticky and do not change quickly in response to changes in prices. A rise in the price level therefore reduces real wages making entrepreneurs better off and workers worse off. 16 Seignorage is the amount of revenue government raises by printing money. Against the Keynesian view, Cagan (1956) suggests that attempts to push the collection of seignorage above its maximum may lead to hyperinflation. 11 investment (see Johnson 1969, Chapter 9 cited in Paul et al. 1997, p. 1388).17 The writings of Kalecki also suggest that an upsurge of inflation would arise from some initiating sector(s) which seek to increase their share of national income (Sawyer 1985, p. 286). This will lead to a competition amongst different competing sectors trying to offset the gains from each other through further price rises. Inflation would not proceed at a uniform rate either through time or across sectors under such circumstances (Kalecki cited in Sawyer 1985, p. 286). In short, the above propositions suggest that inflation is inevitable in a growing economy and inflation may even be needed to support or induce faster structural changes in developing countries. On the other hand, proponents stating inflation having a negative impact on growth derive their arguments from neoclassical theories. For example, according to the Pigou effect, which considers real money balances as part of households’ wealth, rising price level would cause real money balance to fall making consumers feel less wealthy. This leads to lower consumer spending and ultimately becomes harmful for economic growth. Inflation is also believed to be detrimental to investment and, therefore, harmful for economic growth. Higher inflation may raise the cost of maintaining working capital, in the form of real money balances. If money and capital are complements – higher physical capital requires greater need for working capital – then higher inflation would make capital accumulation more costly. Thus, it would impede investment and growth. The economists of neoclassical origin also argue that inflation leads to higher volatility in inflation thus raising uncertainty in terms of potential income flow and giving misleading signals to the markets. Friedman (1977) argues that increased uncertainty reduces the information function of price movements and 17 According to Keynesian mode such intervention by government might be essential in the absence of a developed capital market ensuring a mechanism whereby saving is automatically transforming into investment. According to Keynes, for example cited in Wilkinson (2000, p. 645), higher income may lead to higher proportion of it to be saved by the workers and thus this increased propensity to save may cause a decline in consumption and aggregate demand. This ultimately leads to unemployment as workers will not be able to find jobs because of shortage of demand for products or services they could produce and squeezes out excess savings. This proposition is sometimes known as paradox of thrift and proposed to refute the orthodox view – Say’s Law – supply creates its own demand (analogously: savings create investment). 12 impedes long term contracting. The differences in sectoral prices, moving at different rates, create distortions in investment decisions and misallocation of resources. All this would lead to reducing investment and growth. In an open economy, inflation causes domestic goods and services becoming relatively more expensive leading to a fall in exports. That is, inflation may lead to real appreciation of domestic currency and therefore adversely affecting export and economic growth.18 Investment and growth can be negatively affected through another way in an open economy. If the households have a choice between holding deposits in domestic banks and holding foreign financial assets, a rise in domestic price level may reduce the real return from domestic deposits. This may cause capital flight from the domestic economy. Because of the reduced flow of credit to domestic financial institutions for potential investment, economic growth, ultimately, is hampered (see Montiel 2011, p. 302). The above theoretical discussion on whether there is a positive or negative relationship between inflation and growth also stems from the debate on how steep the aggregate supply curve is. This is regarded as a key controversial issue in macroeconomics (Dornbusch et al. 2008, p. 10; Mankiw & Ball 2011, p. 355). The Keynesian perspective on this is that in the short run and medium run the aggregate supply curve is flat or upward sloping and therefore inflation is less likely to be harmful for growth.19 The neoclassical perception is founded on the long run aggregate supply curve which is believed to be vertical and therefore output does not depend on the price level.20 18 Real exchange rate, e, is defined as follows: e = ne × (P/P*); where ne is the nominal exchange rate, P is the domestic price level, P* is the foreign price level. So an increase in the domestic price level (rising inflation) causes appreciation of real exchange rate and therefore makes domestic export more expensive (see for instance Mankiw & Ball 2011, p. 138). 19 In the short run prices are assumed to be fixed and therefore aggregate supply function is horizontal. Changes in output, therefore, should not have any impact on price level. In the medium run, aggregate supply is upward sloping. Since wages are assumed to be sticky, increases in prices would cause real profit to increase and therefore firms would have an incentive to increase output in response to price increases. Inflation and growth may move together. 20 In the long run prices and wages are flexible and output is determined by real variables of the economy and stays at natural level. Because of flexible prices and wages changes in prices causes proportional changes in the wages, making real profit unchanged. Firms therefore have no incentive to increase output as a result of changes in prices. One distinction between Keynesian and neoclassical perspectives is that Keynesians assume sticky wages and imperfect markets. Whereas neoclassical 13 Perhaps the issue of aggregate supply and the trade-off between inflation and growth is widely examined in the Phillips curve relationship, both theoretically and empirically.21 The original Phillips curve, in reference to the work of Phillips (1958), depicts a negative relationship between the rate of growth of the nominal wage and the unemployment rate.22 Phillips’ empirical work was further strengthened by another empirical work by Samuelson and Solow (1960) and by Lipsey (1960) which attempted to provide a theoretical framework to this relationship. The relationship also implied that policy makers trying to control aggregate demand, using monetary and fiscal policy, faced a usable trade-off, by which lower unemployment rates could be obtained at the cost of higher inflation rates and vice versa (Tobin 1987, p. 315).23 When expansionary policies shift the aggregate demand curve, the price level rises. This reduces real wage and hence induces more employment along a downward sloping labour demand curve as long as there is unemployment. When full employment is achieved, such policies will cause only inflation. The negative relationship between inflation and unemployment can also be explained by using political economy approach in terms of workers’ bargaining power. When unemployment is low, workers have greater bargaining power to increase nominal wages (see Blanchard & Sheen 2009, p. 137) and thus it may lead to an inflationary pressure.24 Therefore, high employment and growth may lead to higher inflation.25 idea rests on the assumption that prices and wages are fully flexible and market clearing forces would take output to its natural level always. 21 Nobel Laureate economist George Akerlof (cited in Mankiw & Ball 2011, p. 333) considers the Phillips curve relationship as, perhaps, the single most important macroeconomic relationship. 22 Although Phillips curve traditionally tries to establish a relationship between inflation and unemployment rate, it can also be reinterpreted to show a relationship between inflation and output by linking output to unemployment. Modern day Phillips curve can be expressed as price inflation in terms of expected inflation, output gap (the difference between actual output and potential output), and supply shock. The inclusion of expectation is due to the works of Nobel Laureates Friedman (1968) and Phelps (1967) as explained in the text, while the supply shock is incorporated because of its importance felt during the oil price crises of the 1970s and a failure of traditional Phillips curve to explain the inflation output trade-off. 23 According to Mankiw (2001) the trade-off does not imply a stable downward sloping Phillips curve but rather the effects of monetary policy, the fact that changes in monetary policy takes the two variables in opposite directions. 24 The bargaining power is expected to increase as competition amongst employers is intensified to attract workers when unemployment is low. 25 It is in this sense a trade-off exists between inflation and growth since in order to achieve high growth, low inflationary environment may have to be sacrificed. 14 The theoretical foundation of the Phillips curve relationship was challenged by Friedman (1968) and Phelps (1967), both working independently. They argue that it is not correct to think that workers observe prices imperfectly and are not capable of correctly assessing their real wages in the event of inflation. Assuming adaptive expectations, they argue that real wages should change in line with changes in the price level. Therefore, according to Friedman (1968) there may be a short run tradeoff, but no long run trade-off between inflation and unemployment (and therefore growth). This implies that long run Phillips curve is vertical at an unemployment rate known as natural rate and demand side policies are, therefore, ineffective in the long run.26 Incorporating rational expectations arguments, replacing Friedman’s adaptive expectation, Lucas (1972, 1973) and Lucas and Sargent (1978) went on one step further in claiming no short run trade-off between the two variables. They argue that when policy makers try to employ demand side policies systematically it fails to have any real effect as agents have rational expectations and hence incorporate these inflation expectations in any contract that they complete.27 The Friedman-Phelps-Lucas approach is predicated on rational expectations formation by economic agents and natural rate hypothesis.28 In essence this framework does not believe that there is a trade-off between inflation and growth. There is, therefore, no gain in terms of higher output and employment by causing higher inflation. This approach, however, has several opposing views. First, on the grounds of expectations, Tobin (1971a, 1971b) proposed models incorporating inflation expectations rather than formation of inflation expectations. His model showed that the long run Phillips curve is negatively sloped and therefore there is a permanent trade-off between inflation and unemployment (Palley 2012, p. 10). Second, the natural rate hypothesis is questioned by some economists, for instance Ball (1997) and Blanchard and Summers (1988). These economists suggest that 26 This is analogous to long run aggregate supply being at the natural level of output. Real effect, according to the Lucas Critic, may happen in the short run only if there is an unexpected demand side policy. 28 According to Mankiw and Ball (2011, p. 353) natural rate hypothesis can be summarised as fluctuations in aggregate demand that affect output and employment only in the short run but in the long run the economy returns to the level of output, employment, and unemployment described by the classical model. 27 15 because of the hysteresis effect recessions can raise the natural rate of unemployment thus having a lasting damaging impact on the economy.29 Policy intervention using aggregate demand is also believed to affect the output and employment even in the long run in this case. In another paper Tobin (1972) argued that positive inflation helps reduce the natural rate of unemployment because it provides a necessary mechanism for lowering real wages without nominal wage cuts. Later in another influential study by Akerlof et al. (1996) also reiterated the importance of it.30 Using the post-Keynesian approach, Arestis and Sawyer (2004, 2005, 2009), Rowthorn (1995), Sawyer (2001, 2002, 2005) and Storm and Naastepad (2007) also argue that natural rate of unemployment can change by changes in labour productivity and in the capital stock. A rise in the employment according to their argument might not be associated with an increase in the pace of inflation (see Argitis & Dafermos 2011, p. 1016). In the context of developing economies, the debate on the role of demand side policies or inflation and employment (or growth) relationship is not new. Back in the early 1950s, for example, economists debated this issue within the Indian Planning Commission. Although they did not use the argument of expectation formation, they came to a very similar conclusion in favour of a vertical relationship based on two different arguments. Rao (1952) argued that the Keynesian fiscal multiplier works only in nominal terms in developing countries. That is, expansionary policies only generate inflation and no increase in output or employment. He based his argument on price inelasticities of supply, so that any increase in demand will only generate inflation – as if the economy is operating at full employment. Dasgupta (1954), on the other hand, believed that the problem was not due to supply rigidity caused by various structural problems in the commodity market, but due to subsistence real wage. That is the real wage in developing countries is already too low, and it cannot be lowered any more for fiscal policy to work for creating more employment. In other words, every time prices go up due to expansionary policies, nominal wages need to be raised so that the real wage does not fall below the subsistence level. As a 29 Hysteresis, as stated in Blanchard and Sheen (2009, p. 670), is the proposition that the equilibrium value of a variable depends on its history. In relation to unemployment what it means is that a prolonged period of actual unemployment would raise the equilibrium (natural) rate of unemployment. 30 Poole (1999) provides a rebuttal on this. 16 result, the employment level – determined by the real subsistence wage – remains unchanged, but at a higher price level. The empirical literature also provides a mixed result on the issue of causality, that is, whether there is unidirectional causality running from inflation to growth or from growth to inflation, or a bi-directional causality. Milton Friedman states, as noted in Friedman (1973) and Friedman and Schwartz (1963), that historically all possible combinations have taken place – economic development in the presence as well as in the absence of inflation. Relatively earlier empirical works such as Tun Wai (1959) and Bhatia (1960-61) could not show a meaningful relationship between inflation and growth. Sidrauski (1967) is amongst the first to show, using super-neutrality of money, that inflation has no effect on growth. Odedokun (1991), on the other hand, shows that the Lucas hypothesis of no inflation output trade-off can be rejected in the case of developing country. Paul et al. (1997) captures this mixed findings in terms of causality. It finds non causal relationship in 40 per cent of the countries studied, bi-directional causality in about 20 per cent of the countries, and unidirectional relationship in about one third of the countries. The issue of which way the causality runs has been part of an influential debate, generally known as the Structuralist – Monetarist Controversy since the 1950s, originating against the backdrop of Latin American situation. The structuralist argument states that causality may run from inflation to growth as well as from growth to inflation. Based on Keynesian type theoretical propositions outlined above, they argue that inflation promotes real economic growth (Baer 1967; Felix 1961; Georgescu-Roegen 1970; Seers 1962; Taylor 1979, 1983). Dorrance (1964) and Lewis (1964), on the other hand, explain that inflation may be an inevitable companion to economic growth which is also supported by Harberger (1963) and Vogel (1974). In a growing economy, because of the structural bottlenecks, supply may not be able to keep pace with the rising demand. Inflation as a result is unavoidable in a developing economy. Bruno (1995) suggests that inflation that results directly from economic expansion does not create any significant barriers to expansion. In support of this, moderate inflation is found to be helpful for growth by Mallik and Chowdhury (2001). They however also warn that growth feeds back into inflation; thus, it may result into an overheating situation. 17 The causality runs from inflation to growth, negatively, is argued by the monetarist view, primarily based on Friedman’s hypothesis that inflation creates uncertainty. A number of studies such as Grier and Tullock (1989), Kormendi and Meguire (1985), and Jung and Marshall (1986) assume that causality runs from inflation to growth. Grier and Perry (2000) and Tommassi (1994) show that inflation uncertainty has a negative impact on growth. However, Dotsey and Sarte (2000) support a positive association in this regard. Stockman (1981), whose work finds empirical support from Zhang (2000), shows that in an economy with cash in advance constraint on consumption and investment, expected inflation reduces the demand for real balances thereby negatively affects growth. The causality between inflation and growth perhaps is not uniform because of the non-linear nature of the inflation growth relationship. Fischer (1993) is amongst the first to examine the possibility of non-linearity in the relationship. In the case of a non-linear relationship, the interest is to see if there is a threshold level up to which inflation and growth are positively related, but beyond which the relation switches sign and becomes negative. In recent years, research interest has paid attention to finding such a threshold level. Empirical findings of the studies in this respect also provide different threshold levels for cross country as well as country specific cases. In the case of cross-country studies, Sarel (1996) and Burdekin et al. (2004) find a single digit threshold level while other studies such as Khan and Senhadji (2001), Kremer et al. (2009), Pollin and Zhu (2006), and Sepehri and Moshiri (2004) find the threshold level between 11 per cent and 18 per cent. In an influential study, Bruno and Easterly (1998) argue that the case for growth effects of low to moderate rates of inflation remain very ambiguous. They find that only in the case of discrete high inflation, beyond 40 per cent, inflation becomes harmful for growth. In the country-specific cases, studies such as Chowdhury and Ham (2009), Kheir-ElDin and Abou-Ali (2008), Mubarik (2005), Phiri (2010), and Salami and Kelikume (2010) suggest a threshold level between 8 per cent to 15 per cent. While other studies such as Ahmed and Mortaza (2005) are more on the conservative side and suggest a threshold level around 6 per cent. 18 The differences in threshold point for developing countries can be due to a number of reasons including country-specific circumstances and the stage of development an economy is in, for instance, initial level of output and growth rate. In this connection, Drukker et al. (2005) in a panel study shows that if the initial inflation rate is below 19.16 per cent, increases in inflation do not have statistically significant effect on growth. However when initial inflation is above 19.16 per cent, further increases in inflation lead to a decrease in long run growth. 1.5 The Research Objectives of the Thesis In the light of the above discussion the present study is motivated to examine macroeconomic policies of low inflation in developing countries. The three most fundamental questions that we come across are as follows Research question 1: Is there a threshold level of inflation at 5 per cent for the Asian developing countries beyond which inflation is harmful for economic growth? Research question 2: Does the threshold level vary depending on the level of development within the Asian developing countries with poorer countries tend to have higher threshold? Research question 3: Is there a consistency between the need for a change in the low inflation policy suggested by leading economists and country-policy reports by the IMF? The investigation is expected to shed some light on a number of important issues such as in a growing and dynamic economy whether moderate inflation is perhaps required for efficient reallocation of resources and for faster structural change to take place. Restrictive policies, overly concerned with keeping inflation at low-single digits, may stall economic growth and plunge the economies into stagnation and delay the progress in terms of poverty. 19 The critical review and the empirical investigation of the study take place in five different chapters. Chapter 2 provides a detailed content analysis of the Article IV country-specific consultation reports prepared by the IMF which acts as a benchmark for their policy advice for these countries. This is carried out in the case of selected developing countries of Asia. We have selected 12 countries from South and Southeast Asia including China. The countries within this region are of special interest to us because of their recent economic performance. Besides, there are enough variations within the selected countries in terms of socioeconomic development, allowing us to determine whether policy advice on inflation and growth by the IMF varies according to country-specific circumstances. We are also interested to find out the nature of policy advice in the aftermath of global crisis 2009. So the selection of counties is also dictated by that availability and release of country reports by the IMF around the time of 2009-2010. The list of countries for which the reports are reviewed are Bangladesh, Bhutan, Cambodia, China, India, Indonesia, Lao PDR, Malaysia, Maldives, Nepal, Pakistan, and The Philippines. Chapter 3 takes a look at the cross-country evidence based on the existing literature. It provides a critical review of the major recent works relating to the growth-inflation nexus. Chapter 4 provides a critical review of the country-specific literature for Asian developing countries under investigation. Here, again we have paid attention to 15 countries from South and Southeast Asia. However, the selection depends on the availability of literature on these countries. The countries are Bangladesh, Bhutan, Cambodia, China, India, Indonesia, Lao PDR, Malaysia, Maldives, Nepal, Pakistan, The Philippines, Sri Lanka, Thailand, and Vietnam. Chapter 5 carries out a cross-country panel study to reveal findings relating to the research questions. A list of 14 countries has been selected based on the availability of data to perform the empirical analyses. These countries are Bangladesh, Cambodia, India, Indonesia, Lao PDR, Malaysia, Pakistan, The Philippines, Thailand, Vietnam, Kazakhstan, Kyrgyz Republic, Papua New Guinea, and Tajikistan. Our selection of countries shows enough variations and a comprehensive picture about the developing countries of Asia. We have excluded the Arab countries intentionally because we believe their history and socioeconomic development are quite different from the rest of the Asian countries. Chapter 6 makes an attempt to empirically investigate the issues in the context of a particular country, Bangladesh. The purpose of the time series analysis in a country-specific case is to strengthen the case we examine. 20 Chapter 2 The IMF and the Policy of Low Inflation: A Review of Article IV Consultations for Selected Developing Countries of Asia 2.1 Introduction The discussion of Chapter 1 notes the tendency of macroeconomic policies in developing countries emphasising strongly on keeping inflation at a low single digit level. This is primarily spearheaded by the fact that as recipients of the IMF loan, developing countries are required to focus on attaining macroeconomic stability by keeping inflation low. As explained in the discussion of Chapter 1, the focus of policy on reining in inflation is based on the assumption that stabilisation will ultimately result in long term sustainable economic growth, employment generation, and poverty alleviation. What is also important, as noted earlier, is that the IMF seems to assess this macroeconomic stability based on a country’s ability to achieve and maintain certain nominal targets. In the context of inflation the IMF policy guideline, explicitly or implicitly, tends to suggest a target of 5 per cent or below irrespective of country-specific circumstances.31 For instance, an inflation target of 5 per cent or less was suggested to 22 out of 32 programme countries between 1995 and early 2007 (Goldsbrough, Adovor & Elberger 2007, Table 1, p. 5). According to IMF’s Independent Evaluation Office (IEO 2007) an inflation target of less than 5 per cent was suggested to 29 Sub-Saharan African countries during the 2000s. Evidence from the cross-country literature, discussed in Chapter 1, shows that the threshold levels of inflation beyond which inflation negatively affects growth vary according to the level of development. Generally, poorer countries tend to have a higher threshold level and the harmful effect of inflation on growth for developing countries is found to be well above 5 per cent. 31 In addition, as part of the stabilisation programme, nominal targets for fiscal deficits at 5 per cent (but preferably primary surplus), debt GDP ratio at 40 per cent, and a sustainable current account balance are suggested. These targets are also expected to be helpful in containing inflation at a low level. 21 As observed earlier that there has been a rethinking about this stereo-type policy suggestion in the wake of the global financial crisis of 2007–2008 and the subsequent Great Recession of 2008–2009 it led to.32 As mentioned in Chapter 1, a former Managing Director of the IMF argued the need for a “wholesale re-examination of macroeconomic policy principles” and questioned the pre-crisis advice of “keeping inflation low and stable was the best way to secure optimal economic performance”.33 In addition, the current Director of Research at the IMF points out that the crisis “has exposed flaws in the pre-crisis policy framework”–particularly about one target, that is low and stable inflation, and one instrument, the policy rate, to achieve it (Blanchard, Dell'Ariccia & Mauro 2010). In short the orthodox approach sponsored by the IMF, that has narrowed the goals and tools of macroeconomic policy, has been questioned by its own key figures within the organisation.34 Against this backdrop, the key objective of this chapter is to find out whether the recent shift in the thinking process by the heads and the research staff of the IMF has any reflection on the policy advice in practice. It provides an in depth examination of the nature of IMF’s policy advice in selected developing countries of Asia. The chapter also attempts to seek the rationale behind keeping inflation at a low single digit level by looking at the historical evidence on country-specific cases. The methodology we use in this chapter is as follows. In order to investigate the nature of IMF’s policy advice we provide content analysis of the IMF Article IV staff consultation reports in the aftermath of the global crisis of 2007-2008. These reports contain IMF’s recommendations to the individual countries as part of its surveillance 32 There has been in fact a triple crises, energy and food price shocks included, as noted by Chowdhury and Islam (2010), due to oil price shock of 2007 and international food price hike around that time. As developing countries are particularly prone to such supply shocks it is important to take this into consideration as well. 33 See the opening remarks by Dominique Strauss-Kahn, at the IMF conference on ‘Macro and Growth Policies in the Wake of the Crisis’, Washington D.C., March 7, 2011, available via the internet at http://www.imf.org/external 34 In fact, raising the concern about the IMF’s international monetarism approach that is not explicitly concerned about economic growth (see Fischer (1987, p. 165) and Fasano-Filho (1996, p. 135)), is not new. In the mid-1990s the then Managing Director urged that the IMF should help its member countries to achieve conditions for high quality growth that brings lasting full employment and poverty reduction (Fasano-Filho 1996, p. 136). In 1999, the IMF launched the Poverty Reduction and Growth Facility (PRGF) the objective of which was to make poverty reduction and growth more central to lending operations in its poorest member countries (Selassie et al. 2006, p. vi). Nevertheless, the basic assumption that stable and low single digit inflation is both necessary and sufficient conditions persisted in the IMF’s policy advice. 22 responsibilities. These country reports are released periodically often in a 12 or 24month-cycle. We select the time-frame 2009-2010 to evaluate the policy prescriptions in the aftermath of the global crisis. Reports of 12 Asian developing countries released, and therefore available in public around that time, are selected. These countries are Bangladesh, Bhutan, Cambodia, China, India, Indonesia, Lao PDR, Malaysia, Maldives, Nepal, Pakistan, and the Philippines. We highlight the staff appraisal on growth and inflation outlook and analyse the recommendations on monetary and fiscal policy in view of the country-specific circumstances. The analysis is aided by the time series data on growth and inflation to reflect the medium and long term picture as well as forecasts on future trends. The detailed content analysis with the help of historical data seeks to explain whether the medium and long-run trends provide a rationale for targeting low single digit inflation irrespective of country-specific circumstances. The rest of the chapter is divided into three sections. Section 2.2 attempts to provide a review of recent related literature. Some of these studies are mostly concerned with the policy advice of an inflation target of 5 per cent or less in the case of developing countries. However, majority of these studies cover pre-crisis period as the research on this topic in the post crisis period is still very limited. Section 2.3 provides an indepth content analysis of the country-specific reports prepared by the IMF staff for the Article IV consultation in the context of 12 countries selected by this chapter. The aim is to examine the extent to which the revised position of the IMF research staff has been translated into policy recommendations in the post crisis era. This is timely and appropriate since not much research has been done on this topic so far. Finally, section 2.4 draws conclusions on the IMF’s policy stance on inflation based on the literature reviewed and historical country-specific experiences presented in the previous sections. It tries to provide an answer whether the policy advice suggested in the Article IV consultation reports is consistent with the historical experience and the findings of the research done by the IMF’s own staff and others. 2.2 A Review of Related Literature Existing literature providing a post crisis analysis on the IMF policy stance particularly in respect of inflation in developing countries is inadequate. Table 3.1 23 provides a summary of the studies selected to be reviewed in this section. Perhaps the only two comprehensive studies in the case of developing countries involving a content analysis using the Article IV consultation staff reports are Roy and Ramos (2012) and Anwar and Islam (2011). The study by Roy and Ramos (2012) reviews the 2010 reports for 26 developing countries to assess whether IMF’s policy recommendations are based on an understanding of country-specific circumstances and whether they reflect the shift in the thinking as proposed by the senior members and the research works carried out by the IMF. The study selects four main areas for assessment: exchange rate, inflation, fiscal consolidation, and employment and other social policies. In relation to the recommendations on inflation the study finds little evidence of policy shift in line with the need for a change suggested above. There has been no cost benefit analysis to support recommendations for reducing inflation and recommendation still focuses on the use of monetary policy only, regardless of the source of inflation (Roy & Ramos 2012, p. 20). One drawback of the study is that it limits itself to content analysis only and does not provide any empirical (historical) evidence in support of its conclusion. This chapter aims at fulfilling this gap by examining whether there is any relationship between the IMF’s advice and historical median values of inflation and growth in section 2.3. Table 2.1: Inflation Targets and IMF Policy, Selected Recent Studies Studies Countries Period Findings Relevant to This Study Roy and Ramos 26 2010 Policy recommendations do not reflect IMF’s (2012) developing revised position and in several cases are not countries appropriate due to country specific circumstances. Anwar and Islam 12 IT and 12 1961 – Inflation targets – on average less than 5 per (2011) non-IT 2009 cent – adopted by the IT regimes are too low, developing perhaps influenced by the declining trends of countries the 2000s, and not in accordance with historical trends. The content analysis does not provide a clear evidence of change in policy advice. Goldsbrough, Adovor, and 32 countries 1995 – 22 cases out of 32 arrangements targeted 2007 inflation rate at or under 5 per cent. Conclusion 24 is that a number of programs went beyond the Elberger (2007) limit of theoretical and empirical claim that pushing inflation below 5 per cent is harmful. IEO (2007) Chowdhury (2006) 29 Sub- 1999 – Despite the inconclusive nature of empirical Saharan 2005 relationship between inflation and growth as African acknowledged by the IMF staff, an inflation developing target of less than 5 per cent has been suggested countries to these countries. 8 Asian 1970 – Both very low and high inflation are harmful developing 2002 for growth and poverty reduction. An inflation rate between 3 and 5 per cent may run the countries danger of deflation leading to a stabilisation trap. Moderate inflation in the range of 10 to 15 per cent is found to stabilise employment. The IMF study by 35 Selassie (2006) 15 2000 – On balance it supports the use of single-digit developing 2003 inflation targets, however suggesting caution countries on setting inflation below 5 per cent. Note: IT stands for inflation targeting. An influential study providing a critique to the IMF’s post-crisis policy stance favouring low inflation strategy is Anwar and Islam (2011). Against the backdrop of a need for “wholesale re-examination” of the IMF policy, the study investigates how appropriate the target of low single-digit inflation is, to promote growth and employment in developing countries. The study finds that amongst the 12 developing countries, currently adopted inflation targeting (IT), the median inflation target is 4.25 per cent, below 5 per cent (see Table 3.2). The individual country targets are also at or below 5 per cent except for Ghana and Turkey. The targets set for the countries, as the study shows, are not derived or consistent with the long run historical trend of the countries. It argues that these inflation targets appear to be influenced by the declining trend in inflation of the 2000s. Inflation in general came down in that decade throughout the world and as such a target based on a temporary phenomenon is questioned by the study. Moreover, this declining trend could be a 35 This study is often referred to as IMF (2006) by others, although the usual disclaimer applies stating that the views expressed in the paper are that of the authors and do not necessarily reflect the IMF (see p. vi). 25 result of the dominance of inflation targeting and inflation first policies pursued since the 1990s. Table 2.2: Inflation Targeting Countries (Emerging and Developing Countries) 2011 Inflation Targeting Country Median Target (percentage) Armenia 4 Brazil 4.5 Chile 3 Columbia 3 Czech Republic 2 Ghana 9.2 Guatemala 5 Hungary 3 Indonesia 5 Mexico 3 Peru 2 Philippines 4 Poland 2.5 Romania 3 Serbia 4.5 South Africa 4.5 Thailand 1.75 Turkey 5.5 Median inflation target (excluding economies in transition: 4.25 Armenia, Czech Republic, Hungary, Poland, Romania, and Serbia, 12) 26 Median inflation target (all developing countries and economies 3.5 in transition, 18) Source: Adapted from Anwar and Islam (2011, Table 1, p. 3). Based on the content analysis of recent Article IV consultation reports for 19 developing countries, the Anwar and Islam (2011) find evidence – similar to Roy and Ramos (2012) – of IMF’s preference towards low single digit inflation mostly around 5 per or below, with no clear rationale for it. Anwar and Islam (2011) also argue that implementation of IT in the presence of supply shocks is challenging in the case of developing countries. In addition to these findings, Anwar and Islam (2011) show no significant achievement made by the IT adopted developing countries in terms of labour productivity, vulnerable employment, working poverty, and growth compared to their non-IT counterparts. In short, the study provides an important conclusion that the argument of low inflation being beneficial for growth is weak. The studies, immediately preceding the crisis, observe a strong preference of inflation targets towards a rate of 5 per cent or less. Goldsbrough et al. (2007) and IEO (2007) reveal that countries with IMF supported programmes had to target an inflation rate less than 5 per cent.36 The study by Selassie et al. (2006), carried out at the IMF, also support a single digit target, although the authors note that a target less than 5 per cent should be pursued with caution as it may entail a loss of output (p. 24). Chowdhury (2006) reviews post 1997-1998 Asian crisis monetary policy in selected Asian countries. Based on post-Asian crisis economic performance, Chowdhury (2006) warns that low inflation, in as much as high inflation, can be harmful for growth. Too low inflation may lead to a stabilisation trap, a situation of low inflation and insufficient growth for poverty alleviation. Goldsbrough et al. (2007) think that some of the IMF programmes suggested too low an inflation target in the respective countries. The binding constraint of conditionality is also revealed by the empirical analysis of IEO (2007). It states that country macroeconomic conditions, as proxied by the inflation rate, are the main determinants of whether and to what extent the IMF programmes permit the spending of incremental aid to Sub36 IEO (2007, p. 18) states that maintaining inflation in the low single digits, 4 – 6 in general, is an important feature of PRGF-supported programmes of the IMF. 27 Saharan African countries (p. 9). The power to spend the anticipated aid increases has been strongly biased towards countries with inflation rate below 5 per cent (in which case they were allowed to spend 79 per cent as opposed to only 15 per cent where inflation rate was higher). In summary, the pre-crisis period shows a clear inclination towards targeting a low inflation rate at or below 5 per cent by the IMF policy in the context of the developing countries. It is important to note that despite this, real GDP growth remained short of 7 per cent which is considered necessary to meet the Millennium Development Goal (MDG) target of halving poverty by 2015 (see evaluation by Selassie 2006, p. 8). In sum, the above review shows that post-crisis policy advice of the IMF continues to be the same as the pre-crisis advice and does not show any change despite IMF’s senior officials’ calls and overwhelming research findings pointing to the need for a change. 2.3 A Review of the IMF Article IV Staff Consultation Reports in Selected Developing Countries of Asia The discussion of the above section reveals the prevalence of low inflation policy strategy in general for developing countries, notwithstanding the suggestions made by the IMF’s own and other leading economists. The discussion now turns to an investigation in the case of 12 selected developing countries of Asia. The content analysis is more extensive unlike the existing studies reviewed in the sense that it tries to analyse the specific recommendations on monetary and fiscal policy in relation to inflation and economic growth supported by the trends revealed by the time series data on each country. Besides, some of these country reports have not been reviewed in the existing literature. The analysis of country-specific reports, in this chapter, is carried out in the following manner. We have selected 12 countries which are Bangladesh, Bhutan, Cambodia, China, India, Indonesia, Lao PDR, Malaysia, Maldives, Nepal, Pakistan, and the Philippines. The selection of these developing countries in Asia shows enough variations in the country-specific circumstances and the levels of economic and social development. Such variations would help us understand if the country28 specific circumstances matter in explaining the historical trends in growth and inflation and whether the IMF policy advice varies accordingly. For instance, in the list we have China, the largest in Asia in terms of size and population with predominantly a command economy where markets are allowed to function within a tight regulatory environment; Bangladesh, a geographically small, but densely populated country with a population of 160 million and growing middle class. It initially followed a socialist path, but rapidly moved into a more market oriented system. The selection also includes Nepal, a land-locked country emerging out of social conflicts. The country consultation reports of these countries in the years 2009 and 2010 are selected (depending on their availability). This is to capture the policy advice around the time of the crisis. To provide some empirical support to the content analysis on each country the data on inflation and growth are presented in a table. In each case, the mean values of 2009 and 2012/2013 are shown to reflect the current and forecasts on inflation and growth. In addition, long-run median and decadal median values between 1960 and 2009 are included to reveal the pattern of historical experience on growth and inflation in each country. The median values are used to avoid outliers resulting from extreme episodes of inflation. The data are gathered from the standard sources of the World Development Indicators of the World Bank and International Financial Statistics and World Economic Outlook of the IMF. The country summaries are provided in the following subsections. 2.3.1 Bangladesh Table 2.3.1: Average Growth-Inflation Figures in Bangladesh Country Bangladesh Indicators Long run Decadal median median 2009 Growth 5.58 2012 5.9 * 2013 6.4 * 1971-2009 4.72 1960- 1970- 1980- 1990- 2000- 69 79 89 99 09 - 3.27 3.42 4.78 5.77 29 Inflation 5.42 10.4 7.9 8.40 - 16.60 10.25 5.75 6.22 Source: World Bank, World Development Indicators; IMF, International Financial Statistics, and World Economic Outlook, and author’s calculations. Notes: Growth and inflation refers to CPI inflation and real GDP growth respectively. The data for the year 2009 are stated because it best reflects the current condition at the time of the release of Article IV Consultation-Staff Report. * refers to projected figures taken from the IMF, World Economic Outlook. The figures for the decade 1960s are not reported as Bangladesh was a part of Pakistan during the time. Long-run growth performance in Bangladesh, based on historical evidence, remains well below the required rate of 8 per cent to graduate to a middle income country by 2021 and to have a significant dent on poverty.37 The long-run median growth is only 4.72 per cent (see Table 3.3.1) and it seems to be held back by infrastructural bottlenecks (see IMF 2010a, p. 15). Despite the fact that inflation remained below 6 per cent in 2009, the consultation report (IMF 2010a) warned that monetary policy was “too accommodative” for domestic conditions and preventing an increase in inflation is an “immediate policy concern.” This perhaps came from the fear that inflation would reach a double digit figure in the near future, 10.4 per cent in 2012. This advice was offered despite the fact that the inflationary pressure was due mainly to food and fuel price rises in the international markets, exchange rate depreciation (from around 70 taka a dollar to about 83 taka a dollar) and the removal of subsidies or administrative adjustment of prices upward. Although the report identified that external supply shocks such as international food and commodity prices were contributing factors to inflation it suggested for employing tight monetary policy. The IMF staff suggested that the central bank should take “a pre-emptive and bold action” to contain “signs of a pickup in prices” (p. 15). The higher interest rate is believed to be a “small price to pay to prevent a harmful acceleration of inflation” (p. 15). 37 A recent report by the World Bank (World Bank 2012) suggests that the country needs to grow at 8 per cent to become a middle income country by 2021. The report notes that the decline in population growth is impressive, from nearly 3 per cent per annum in the 1970s to 1.3 per cent in the 2000s, and increased labour productivity has helped to achieve the current growth rate. However, to achieve middle income country threshold of US$1450 GNI per capita by 2021 from Bangladesh’s current US$784 per capita, the required growth rate in GDP is 8 per cent assuming share of remittances is 5 per cent and a constant real exchange rate (ibid p. 4). 30 This fear of inflation going out of hand was not justified, given the inflation history of Bangladesh. Table 2.3.1 shows that the inflation in 2009 was well below the longrun historical inflation at 8.4 per cent. The figure was also lower than the decadal medians for 1990s and 2000s. The forecast values of inflation do not show signs of runaway inflation either (inflation is expected to go down in 2013; see Table 2.3.1). On the other hand, growth performance was seen by the report as “remarkably well” in the face of global recession (p. 14). Fiscal conservatism of the government was said to be responsible for this and maintaining “fiscal prudence and sustainability” was advised to unleash Bangladesh’s potential (p. 15). However, one wonders how fiscal conservatism is going to resolve the country’s huge infrastructure shortage which is also a contributory factor for higher inflation in addition to a major obstacle to growth. So far the public-private partnership (PPP), which is generally touted as a solution, is not showing much promise. The PPP experience in the energy sector does not auger well when shifting contingent liabilities to the government is a major cause of deteriorating fiscal position. 2.3.2 Bhutan Table 2.3.2: Average Growth-Inflation Figures in Bhutan Country Bhutan Indicators Long run Decadal median median 2009 2012 * 2013 * 1980-2009 1960- 1970- 1980- 1990- 2000- 69 79 89 99 09 Growth 6.51 7.0 9.9 6.75 - - 7.50 5.33 7.18 Inflation 4.36 8.4 7.3 7.03 - - 9.90 9.74 4.48 Source: World Bank, World Development Indicators; IMF, International Financial Statistics, and World Economic Outlook, and author’s calculations. Notes: Growth and inflation refers to CPI inflation and real GDP growth respectively. The data for the year 2009 are stated because it best reflects the current condition at the time of the release of Article IV Consultation-Staff Report. * refers to projected figures taken from the IMF, World Economic Outlook. 31 Bhutan, except for the 1990s, shows strong growth performance at around 7 per cent and growth is expected to reach nearly 10 per cent in 2013 (see Table 3.3.2). On the other hand, inflation in 2009, at 4.36 per cent, remained well below its long-run median, at 7.03 per cent as well as the common 5 per cent benchmark of the IMF. Inflation in Bhutan is driven by inflation in India as Bhutan’s currency is pegged to the Indian currency. This is also evident from the historical median-inflation for both the countries, at around 7 per cent. The country report on Bhutan (IMF 2009a, p. 8) praises this robust growth without inflationary pressure. The rapid credit growth is believed to help faster private sector growth and is expected to result in higher productivity in the country. Although inflation is stable and future inflation is expected to remain close to its historical median (see Table 2.3.2) the staff appraisal warned of a potential overheating in the economy. This is, as the report suggests, due to “spill-overs from the hydropower and development spending as well as rapid credit growth, financial sector vulnerabilities, and concerns about debt sustainability” (p. 9). The government, according to the report, should be careful about fiscal management as the budgeted fiscal deficit for 2009-2010 exceeds the government’s implicit policy to limit the deficit under “5 per cent of GDP” (see ibid., p. 10). In line with this, the staff felt that a near-term “tightening bias” is necessary to ensure macroeconomic stability. However, it is not clear if public investment in hydropower and development spending are growth promoting, how debt can become unsustainable or the economy can overheat. Instead of advising for a tightening bias, the advice should have been on ensuring spending on right sectors with greatest growth potential and on revenue collection by enhancing progressivity of the tax structure and efficiency of tax administration. If this can be done, the debt can be repaid from higher growth and consequent revenue dividends. Thus, the IMF’s advice seems short-termist and follows from the belief that macroeconomic stability in a narrow sense is both necessary and sufficient conditions for growth and development. 32 2.3.3 Cambodia Table 2.3.3: Average Growth-Inflation Figures in Cambodia Country Cambodia Indicators Long run Decadal median median 2009 2012 * 2013 * 1990-2009 1960- 1970- 1980- 1990- 2000- 69 79 89 99 09 Growth -1.89 6.2 6.4 7.73 - - - 5.86 8.16 Inflation -0.66 4 3.6 3.92 - - - 4.00 3.57 Source: World Bank, World Development Indicators; IMF, International Financial Statistics, and World Economic Outlook, and author’s calculations. Notes: Growth and inflation refers to CPI inflation and real GDP growth respectively. The data for the year 2009 are stated because it best reflects the current condition at the time of the release of Article IV Consultation-Staff Report. * refers to projected figures taken from the IMF, World Economic Outlook. A strong growth performance has been registered by the Cambodian economy for several years, the 2000-2009 average standing at 8 per cent (see Table 2.3.3). The report on Cambodia (IMF 2009c, p. 3) acknowledges the country’s strong growth performance and significant reduction in poverty.38 But the global recession seems to have buffeted the economy causing growth to plummet to -1.89 in 2009. At the same time Inflation was negative in 2009 and forecasts also show that it would remain below 5 per cent in the near future. The decline in both growth and inflation reveals that expansionary macroeconomic policies were required to boost aggregate demand for growth to recover. Yet, the staff-report saw an upside risk due to “overly expansionary” fiscal stance (p. 19), exerting “pressure on inflation” (p. 3), and undermining “stability” (p. 10).39 The report, however, admitted that fiscal easing in the face of weak demand in 2009 was appropriate but raised concern that 2009 fiscal deficit was expected to widen to 6.75 38 According to the country report (see p. 4) poverty level went down from 50 per cent in the mid1990s to 30 per cent in 2007. 39 In fact the report underscores the need to keep the budget deficits within 5.75 per cent of GDP which it believes will be “appropriately accommodative in providing adequate space for development objectives, but eliminating the domestic financing requirement and setting the stage for medium-term fiscal consolidation” (p. 11). 33 per cent of GDP against the target of 4.25 per cent (p. 10). The report was also against further easing of monetary policy particularly lowering reserve requirement. It is “neither warranted based on monetary conditions nor desirable from a prudential perspective”, according to the report (see p. 14). However, the report does not offer any advice as to how to support recovery. It displays unwavering faith in macroeconomic stability in terms of inflation below 5 per cent and budget deficits below 3 per cent to deliver growth and economic recovery. 2.3.4 China Table 2.3.4: Average Growth-Inflation Figures in China Country China Indicators Long run Decadal median median 2009 2012 * 2013 * 1960-2009 1960- 1970- 1980- 1990- 2000- 69 79 89 99 09 Growth 8.8 8.2 8.8 8.84 9.84 7.32 9.53 9.21 9.53 Inflation -0.70 3.3 3.0 2.15 - 1.05 6.25 4.94 1.31 Source: World Bank, World Development Indicators; IMF, International Financial Statistics, and World Economic Outlook, and author’s calculations. Notes: Growth and inflation refers to CPI inflation and real GDP growth respectively. The data for the year 2009 are stated because it best reflects the current condition at the time of the release of Article IV Consultation-Staff Report. * refers to projected figures taken from the IMF, World Economic Outlook. Fiscal stimulus and loosening of monetary policy in the face of global recession have played an instrumental role in mitigating the effects of the shocks in the Chinese economy, the IMF report claims (see IMF 2010b, p. 4). According to Table 2.3.4, there does not seem to be a tendency of inflationary pressure amidst this strong growth performance. Inflation was in negative territory in 2009 and would remain around 3 per cent in the upcoming years. Despite this, the country-report raised concern about inflationary pressure in the period ahead because of rapid expansion of monetary aggregates and credit growth (ibid., p. 8). The Chinese authority however 34 opposed the view of raising nominal interest rate at this point as they believe that the inflation outlook was still benign (p. 8). 2.3.5 India Table 2.3.5: Average Growth-Inflation Figures in India Country India Indicators Long run Decadal median median 2009 2012 * 2013 * 1960-2009 1960- 1970- 1980- 1990- 2000- 69 79 89 99 09 Growth 8.71 6.9 7.3 5.47 6.08 2.45 5.44 5.69 8.03 Inflation 10.87 8.2 7.3 7.16 3.58 5.99 8.76 9.59 4.32 Source: World Bank, World Development Indicators; IMF, International Financial Statistics, and World Economic Outlook, and author’s calculations. Notes: Growth and inflation refers to CPI inflation and real GDP growth respectively. The data for the year 2009 are stated because it best reflects the current condition at the time of the release of Article IV Consultation-Staff Report. * refers to projected figures taken from the IMF, World Economic Outlook. A strong growth performance, at about 8 per cent, in recent decade has made India a major emerging country of the world. According to the staff-appraisal, the economy has rebounded ahead of most countries in the world but inflation is intensifying (IMF 2010c, p. 3). Table 3.3.5 shows that inflation in 2009 was at 10.87 per cent but was expected to come down and remain close to its historical long-run median around 7 per cent. Still the report identifies “elevated inflation” as a major risk which could “stall the recovery” (ibid., p. 4). An exit strategy from the accommodative policy by reducing fiscal deficits and gradually tightening monetary policy to anchor inflation was recommended. In addition, the government was advised to work under a framework anchoring debt target to reduce it at 60-65 per cent of GDP by 2015 (see p. 3). To achieve this target, the report suggested that more privatisation and reduction in public sector’s claim on resources were required (p. 3). It is interesting to note that the report also observed that “there is a risk that private demand may not be broaden enough to carry the growth momentum” (p. 5). 35 2.3.6 Indonesia Table 2.3.6: Average Growth-Inflation Figures in Indonesia Country Indonesia Indicators Long run Decadal median median 2009 2012 * 2013 * 1960-2009 1960- 1970- 1980- 1990- 2000- 69 79 89 99 09 Growth 4.47 6.1 6.6 5.96 2.81 7.71 6.54 7.14 4.91 Inflation 6.38 6.2 6.0 10.46 128.84 14.30 9.38 8.97 8.34 Source: World Bank, World Development Indicators; IMF, International Financial Statistics, and World Economic Outlook, and author’s calculations. Notes: Growth and inflation refers to CPI inflation and real GDP growth respectively. The data for the year 2009 are stated because it best reflects the current condition at the time of the release of Article IV Consultation-Staff Report. * refers to projected figures taken from the IMF, World Economic Outlook. Growth in Indonesia was moderate at 4.47 per cent in 2009. In fact, the median growth between 2000 and 2009 was 4.91 per cent (see Table 2.3.6) and was about 2 percentage points lower than the median figures of the previous three decades. The economy has managed to grow at around 7 per cent, on average, in the 1970s, 1980s, and 1990s. The relatively poor performance in the 2000s is due to the fact that Indonesia was hard hit by the Asian Currency Crisis of 1997. The impact of the crisis is reflected in its growth performance in the following decade. The country appears to find it difficult to regain its past robust growth path. On the other hand, inflation was around 6 per cent and is expected to remain at that level in 2012 and 2013 (see Table 2.3.6). According to Table 2.3.6, this inflation figure does not seem to impose any threat. The figure was still at least 2 percentage points less than the decadal median values since the 1980s. The Indonesian economy has the experience of growing faster with higher inflation rates (see the decadal median values of the 1980s, 1990s, and 2000s). The country report on Indonesia (IMF 2009b, p. 11) found the fiscal stimulus and monetary easing in the aftermath of the global financial crisis as appropriate as it has supported the “weaker domestic demand.” It is important to note that Indonesia, 36 before the fiscal stimulus package for 2009, followed several years of prudent fiscal management and achieved a primary fiscal surplus of about 2 per cent of GDP per year (see ibid., p. 6). This is the reason why the report was also positive about the fiscal stimulus since it would not hamper the medium term public debt consolidation. Although inflation figures in Table 2.3.6 do not appear to be alarming, the staffappraisal suggested that the central bank should be more cautious. It argued that there was “ample stimulus in the pipeline” (p. 19), in the form of interest rate cuts and “abundant liquidity in the banking sector” (p. 10), “leading to higher credit growth and inflationary expectations” (p. 10). In addition, referring to an estimated Taylor rule40 the report indicated that the central bank’s “policy stance has generally tended to have an expansionary bias over time” (p. 10). Therefore, it underscored a need for a concerted effort “based on a more credible and objective inflation targeting framework to consistently lower inflation and guide inflationary expectations towards a medium-term target range of 3-4 per cent” (p. 11). This is believed to “enhance policy credibility and lower economic costs” (p. 11). 2.3.7 Lao PDR Table 2.3.7: Average Growth-Inflation Figures in Lao PDR Country Lao PDR Indicators Long run Decadal median median 2009 2012* 2013* 1980-2009 1960- 1970- 1980- 1990- 2000- 69 79 89 99 09 Growth 7.21 8.4 7.1 6.07 - - 4.77 6.19 6.86 Inflation 0.03 6.7 5.3 10.63 - - 61.33 16.51 7.72 Source: World Bank, World Development Indicators; IMF, International Financial Statistics, and World Economic Outlook, and author’s calculations. Notes: Growth and inflation refers to CPI inflation and real GDP growth respectively. The data for the year 2009 are stated because it best reflects the current condition at the time of the release of Article 40 Taylor Rule was developed by John B. Taylor of Stanford University in the early 1990s (see Taylor 1993). It provides a guideline for the central bank on how to set the short term real interest rate in order to stabilise the economy and achieve inflation target. The rule can be written as: r = 0.01 + 0.5y + 0.5π, where r is the real interest rate set by the central bank, y is the percentage deviation of real GDP from a target, and π is the inflation rate. 37 IV Consultation-Staff Report. * refers to projected figures taken from the IMF, World Economic Outlook. As a transitional economy Lao PDR has performed well in recent decades with growth remaining above 6 per cent on average (see Table 2.3.7). Acknowledging this, the consultation report on the country identified that the expansionary policies were responsible for this robust growth particularly in the face of the global recession (see IMF 2009d, p. 3). According to the report, inflation was expected to remain low and stable assuming no significant rise in commodity prices (p. 3). Table 3.3.7 reveals that indeed inflation was almost zero in 2009 and forecast figures were well below the country’s historical median inflation at 10.63 per cent. Yet, when it comes to policy advice the staff-appraisal took the typical stance. It concluded that “overly expansionary fiscal and credit policies pose a risk to macroeconomic stability” (ibid., p. 3). The authorities have been advised of a “prompt and determined tightening of monetary and fiscal policy” (p. 15). The report feared that continued credit growth “could fuel inflation down the road.” It was therefore imperative, as the report suggested, to send strong signals to the banks to check credit growth by significantly raising required reserve ratio (p. 3). 2.3.8 Malaysia Table 2.3.8: Average Growth-Inflation Figures in Malaysia Country Malaysia Indicators Long run Decadal median median 2009 2012* 2013* 1960-2009 1960- 1970- 1980- 1990- 2000- 69 79 89 99 09 Growth -1.72 4.4 4.7 6.81 7.08 7.72 6.38 8.72 5.62 Inflation 0.58 2.7 2.5 2.81 -0.10 4.07 3.25 3.51 1.67 Source: World Bank, World Development Indicators; IMF, International Financial Statistics, and World Economic Outlook, and author’s calculations. Notes: Growth and inflation refers to CPI inflation and real GDP growth respectively. The data for the year 2009 are stated because it best reflects the current condition at the time of the release of Article IV Consultation-Staff Report. * refers to projected figures taken from the IMF, World Economic Outlook. 38 Malaysia posted negative growth in 2009 (see Table 3.3.8) and the country report stated that “forceful counter-cyclical policies” have helped the country to emerge from “the global recession with strong forward momentum” (see IMF 2010d, p. 4 and 19). The country’s growth forecasts, around 4 per cent, were still well below it historical standard, at 6.81 per cent. There was also no sign of overheating as inflation was expected to remain at about 2 per cent. Therefore, the full recovery did not seem to have taken place. Yet, the IMF report stated that the policy of fiscal consolidation and normalisation of monetary policy should be underway. The staffappraisal suggested that raising policy rate after aggressive loosening in 2009 was consistent with the Taylor rule calculations (ibid., p. 10). The staff is concerned about “side-effects of abnormally low interest rates” (p. 10). Higher interest rate was believed to send a signal against “investor’s complacency in the pricing of risks after a year and a half of loose monetary conditions” (p. 10). 2.3.9 Maldives Table 2.3.9: Average Growth-Inflation Figures in Maldives Country Maldives Indicators Mean Long run Decadal median median 2009 2012* 2013* 1990-2009 1960- 1970- 1980- 1990- 2000- 69 79 89 99 09 Growth -2.32 4.4 3.5 6.97 - - - 9.01 5.98 Inflation 3.98 11.5 8.3 4.73 - - - 5.84 2.51 Source: World Bank, World Development Indicators; IMF, International Financial Statistics, and World Economic Outlook, and author’s calculations. Notes: Growth and inflation refers to CPI inflation and real GDP growth respectively. The data for the year 2009 are stated because it best reflects the current condition at the time of the release of Article IV Consultation-Staff Report. * refers to projected figures taken from the IMF, World Economic Outlook. The external shocks due to the global recession have prompted a severe decline in economic activity in Maldives and growth dipped by about 4 per cent reaching at 2.32 per cent in 2009 (see IMF 2010e, p. 3 and Table 2.3.9). Future economic 39 outlook of the economy also indicated that growth would be modest at around 4 per cent in the years 2012 and 2013, well below its average median value, 6.97 per cent, since the 1990s (see Table 2.3.9). Consistent with a slowing economy, inflation remained below 4 per cent in 2009; but showed signs of increase in 2012 and 2013 (Table 2.3.9). This could be due to supply shocks resulting from a rise in international fuel prices and domestic electricity tariffs (IMF 2010e, p. 3). In short the key problem is that the country is in recession with problems arising mainly from supply-side shocks. Despite this grim prospect of growth and slow recovery, the country report on Maldives noted that the crisis has led to “an unsustainable fiscal expansion” turning it into “a serious near-term risk to macroeconomic stability” (ibid., p. 3). Monetary policy, on the other hand, according to the staff, has been constrained by fiscal dominance because of the government borrowing without limit from Maldives Monetary Authority (p. 6). It seems keeping inflation at around 3 per cent was viewed as the benchmark for attaining macroeconomic stability by the report. Thus, the report suggested to pursue tighter monetary and fiscal policy to “help inflation decline gradually to about 3 per cent in 2012-2014” (p. 10). 2.3.10 Nepal Table 2.3.10: Average Growth-Inflation Figures in Nepal Country Nepal Indicators Long run Decadal median median 2009 2012 * 2013 * 1960-2009 1960- 1970- 1980- 1990- 2000- 69 79 89 99 09 Growth 4.31 4.2 3.8 3.95 1.89 2.75 4.35 4.42 4.31 Inflation 11.6 7.8 7.4 8.24 4.05 7.98 10.94 8.29 5.90 Source: World Bank, World Development Indicators; IMF, International Financial Statistics, and World Economic Outlook, and author’s calculations. Notes: Growth and inflation refers to CPI inflation and real GDP growth respectively. The data for the year 2009 are stated because it best reflects the current condition at the time of the release of Article IV Consultation-Staff Report. * refers to projected figures taken from the IMF, World Economic Outlook. 40 Nepal is one of the poorest countries in Asia with 25.2 per cent people living below the poverty line.41 Its growth performance is also poor with historical median value, since the 1960s, standing below 4 per cent (see Table 2.3.10). The IMF itself admits that the structural problems of the economy has stopped Nepal to achieve higher growth path (IMF 2010f, p. 8). In spite of the lacklustre growth performance, the country report on Nepal praised the fiscal discipline maintained by the country and it was seen as a “remarkable achievement” (ibid., p. 4). Fiscal policy was said to have continued to be “broadly prudent and public debt has declined to 40 per cent of GDP” (p. 4). The report’s attention to inflation over growth was also noticeable. It noted that loose monetary conditions of 2008-2009, which led to rapid credit growth, were responsible for double digit inflation in 2009 (see p. 3 and 7). In 2009, inflation reached at around 11 per cent (see Table 2.3.10). The staff-appraisal, therefore, thought that it was a challenging task to manage the economy since tighter monetary policy would contain inflation but hurt economic growth (see p. 3). The case of Nepal is a classic example where a country sacrifices growth opportunities in order to keep inflation at a single digit level in line with the IMF prescription. 2.3.11 Pakistan Table 2.3.11: Average Growth-Inflation Figures in Pakistan Country Pakistan Indicators Long run Decadal median median 2009 2012 * 2013 * 1960-2009 1960- 1970- 1980- 1990- 2000- 69 79 89 99 09 Growth 3.56 3.4 3.5 4.98 5.81 3.99 6.44 4.01 4.26 Inflation 13.64 12.0 12.5 7.15 3.18 7.71 6.22 10.17 7.52 Source: World Bank, World Development Indicators; IMF, International Financial Statistics, and World Economic Outlook, and author’s calculations. Notes: Growth and inflation refers to CPI inflation and real GDP growth respectively. The data for the year 2009 are stated because it best reflects the current condition at the time of the release of Article IV Consultation-Staff Report. * refers to projected figures taken from the IMF, World Economic Outlook. 41 Poverty headcount ratio at national poverty line (% of population) taken from the World Bank, World Development Indicators. 41 Pakistan’s recent and historical median growth performance does not show a promising sign. According to Table 2.3.11, growth was 3.56 per cent in 2009 while the long-run median is 4.98 per cent. Both figures are well below the 7 per cent threshold required to take a country into the trajectory of higher development. Inflation reached a double digit figure in 2009 and is expected to remain so in the near future. Supply shocks resulting from increased oil and food prices and energy subsidy financed by the central bank have fuelled the inflation (IMF 2009e, p. 5). Adverse security developments are also responsible for deteriorating condition in Pakistan. Despite the dismal growth performance, the authorities have embarked on a stabilisation programme incorporating “a significant tightening of fiscal and monetary policies to bring down inflation and strengthen the external position” (ibid., p. 5). The authorities believe that the scope for countercyclical fiscal policy is limited given the inflation situation of the country and they remain committed to achieving fiscal deficit of 4.3 per cent of the GDP (see IMF 2009e, p. 15). In the case of monetary policy, the authorities think that it is “premature to reduce the discount rate at this juncture, as core inflation remains elevated” (ibid., p. 16). These steps taken by the authorities have received approval in the staff-appraisal which states that “Pakistan’s stabilisation program is on track” (ibid., p. 21). 2.3.12 The Philippines Table 2.3.12: Average Growth-Inflation Figures in The Philippines Country Pakistan Indicators Long run Decadal median median 2009 2012 * 2013 * 1960-2009 1960- 1970- 1980- 1990- 2000- 69 79 89 99 09 Growth 1.14 4.2 4.7 4.61 4.82 5.35 3.46 3.01 4.66 Inflation 3.22 3.4 4.1 7.51 5.40 12.14 10.40 7.93 4.96 Source: World Bank, World Development Indicators; IMF, International Financial Statistics, and World Economic Outlook, and author’s calculations. Notes: Growth and inflation refers to CPI inflation and real GDP growth respectively. The data for the year 2009 are stated because it best reflects the current condition at the time of the release of Article 42 IV Consultation-Staff Report. * refers to projected figures taken from the IMF, World Economic Outlook. The historical evidence suggests that the Philippines have yet to show strong growth performance (see Table 2.3.12). Its long run median growth is below 5 per cent and the figure dropped at 1.14 per cent in 2009. The IMF country report stated that the slowdown in growth had been “cushioned by supportive fiscal and monetary policies and resilient remittances” (see IMF 2010g, p. 3). The report praised the inflation condition of the country noting that the credit growth decelerated and inflation remained low and inflation expectation was well anchored at a level consistent with the central bank’s inflation target (ibid., p. 12). Although the recovery was not promising the report suggested “fiscal withdrawal” and capping the fiscal deficits at 3.5 per cent of GDP to contain the high public debt level and to avoid jeopardizing investor confidence (see p. 9).42 A summary of the country reports analysed above shows little evidence of change in the advice on policy by the IMF staff. This finding is mostly consistent with that of Anwar and Islam (2011). The decadal median values show enough variation and therefore it is difficult to suggest a unique inflation target that can be followed for all the countries. It provides a strong case for following what is stated in the preamble of IMF’s Article IV that country specific circumstances should be taken into consideration as reiterated by studies such as Roy and Ramos (2012) and Anwar and Islam (2011). To end this section we provide an additional analysis based on two regimes – pre and post structural adjustment programme started in the 1980s by the World Bank and supported by the IMF. Table 2.4 presents average inflation and growth data for the two regimes, 1960-1979 and 1980-2009 for South Asian countries (they are considered because of the availability of relatively longer time series data). These countries started to embrace the reform programmes since the 1980s. However, post 1980 era records a higher inflation rate for all the countries (ignoring Bangladesh as it went through an unusual phase in its history in the 1970s) at above 5 per cent (the 42 This suggestion comes in view of the government’s plan to target a deficit of 3 per cent of GDP in 2010 so as to attain a balance-budget by 2013 (see IMF 2010g, p. 9). 43 preferred rate as one would expect after going through the reform programmes). Growth figures for all these countries, except for Pakistan, are also higher in the post 1980 era. In short, such evidence provides a clear argument against targeting a low single digit inflation rate. Table 2.4: Median CPI Inflation and Growth in Countries in South Asia, Pre and Post 1980 Country Variable 1960-1979 1980-2010 Bangladesh Inflation 16.60 6.90 Bangladesh Growth 3.27 4.80 India Inflation 5.73 8.33 India Growth 3.79 5.82 Nepal Inflation 7.58 8.29 Nepal Growth 2.34 4.31 Pakistan Inflation 5.56 7.88 Pakistan Growth 5.36 4.84 Sri Lanka Inflation 3.19 11.10 Sri Lanka Growth 4.13 4.97 Note: Pre 1980 period for Bangladesh is 1971 – 1979 a turbulent period in its history. 2.4 Conclusion The findings of this chapter clearly show that the IMF policy advice is not consistent with the recent suggestions, particularly made by its own leading influential economists. Nor does it comply with the historical country-specific experience. Both content analysis using the Article IV consultation reports and historical evidence from time series data reaffirm this. The IMF even in the wake of the global crisis seems to continue to advocate for low inflation strategy without giving due consideration to the country specific circumstances. Our critique on the continued policy advice by the IMF has been carried out from 3 different directions so far. We find inconsistency at all 3 levels (a) with the shift in the IMF’s own thinking as revealed by its senior economists, (b) with the history of inflation and growth and country circumstances, and (c) findings from the literature. The IMF policy still 44 tends to follow a pre-set view with predictable conclusions that do not allow for alternative perspectives. The analysis of this chapter reiterates the need for a change in the narrow policy advice and put it firmly on its already stated principle of “fostering economic growth with reasonable price stability.” The next chapter examines the country-specific experience of inflation and economic growth in some selected developing countries of Asia from the existing literature. 45 Chapter 3 Cross-Country Evidence on Inflation-Impact on Growth: A Review of Related Literature 3.1 Introduction This chapter provides a critical review of the cross-country panel studies on inflation and growth in the context of developing countries. The key purpose of this chapter is to examine whether the existing literature provides a strong empirical support to keep inflation within 5 per cent level in the case of developing countries as suggested by the IMF. In the following section the evidence on issues relating to growth-inflation nexus are reviewed followed by a section with concluding remarks. 3.2 Cross-Country Evidence of Growth-Inflation Nexus On the issue of whether inflation is harmful to growth one of the most influential study by Bruno and Easterly (1998) states that “the ratio of fervent beliefs to tangible evidence seems unusually high on this topic, despite extensive previous research” (p. 3). They find robust evidence that growth falls sharply at discrete high inflation (which they propose to be 40 per cent per annum) crises, then recovers rapidly and strongly after inflation falls. Like Bruno and Easterly (1998), Barro (1996) also notes that the clear evidence for adverse effects of inflation comes from the experience of high inflation. He, however, points out that although the magnitudes of effects are not that large, over long periods these changes in growth rates have dramatic effects on the standard of living and, therefore, justify the interest in attaining price stability. The important question therefore is to find out what level of inflation can be considered not harmful for economic growth. The next few paragraphs address this issue and examine the findings from the literature in relation to developing countries. Table 3 provides a list of selected cross-country studies investigating the growthinflation relationship incorporating developing countries. 46 Table 3: Selected Cross-Country Studies on the Relationship between Growth and Inflation Study Period Number of Method Major findings Rolling- Inflation threshold is window two- between 8 and 15 per cent. stage-least- A positive effect of trade square and human capital on method. growth is significant when countries Yilmazkuday 1965-2004 84 countries (2013) inflation is below 8 and 15 per cent respectively. Lopez- 1961-2007 44 countries Panel smooth Inflation has a nonlinear Villavicencio and transition impact on growth. Mignon (2011) model Threshold inflation differs (PSTR) and strongly between dynamic developed and developing generalised countries. For developing method of countries it is 17.5 per cent moments below which the (GMM) relationship is nonsignificant. Bick (2010) 1960-2004 40 developing Generalised Inflation thresholds are countries panel found to be at 12 and 19 threshold per cent with and without model. regime intercept respectively. Espinoza et al. 1960-2007 165 countries (2011) Logistic Inflation threshold lies smooth between 7 and 13 per cent transition for developing countries. regression model. Pollin and Zhu (2006) 1961-2000 80 middle-income Pooled OLS, Inflation threshold roughly and low-income Fixed effects, between 15 and 18 per countries Random cent. effects, and Between effects panel estimation 47 models. Drukker et al. 1950-2000 138 countries (2005) Non-dynamic, Threshold level of inflation fixed effects is around 19 per cent in the panel data full sample. models. Sepehri and 1960-1996 Moshiri (2004) 92 countries with Spline Inflation threshold is at 15 26 lower-middle- regression and 11 per cent for lower- income and 28 technique. middle-income and low- low-income income countries countries. respectively. Burdekin et al. 1965-1992 21 industrial and Generalised Higher inflation threshold (2004) and 1967- 51 developing least squares for more advanced 1992 for countries. (GLS) countries; 8 and 3 per cent developed estimator with for industrial and and fixed effects. developing countries developing respectively. countries respectively Khan and 1960-1998 140 countries Senhadji (2001) Conditional Inflation threshold is at 1-3 least squares and 11-12 per cent for industrial and developing countries respectively. The positive effect of inflation on growth is present up to 18 per cent for developing countries when threshold level varies from 1 to 50 per cent. Ghosh and 1960-1996 145 countries Phillips (1998) Binary Inflation threshold at 2-3 recursive per cent. trees method. Bruno and 1961-1992 26 countries Easterly (1998) Descriptive There is no cross-sectional analysis correlation between longrun averages of growth and inflation below inflation rate of 40 per cent. Paul et al. (1997) 1960-1989 70 countries Granger Low world inflation regime methodology is expected to redistribute real growth opportunities benefit from the developing 48 countries in favour of industrialised countries. No causal relationship in 40 per cent, bidirectional causality in 20 per cent, and unidirectional causality in one third countries are found. Causality, uni or bidirectional, is more prominent in industrial countries. Barro (1996) Sarel (1996) 1960-1990 1970-1990 100 countries 87 countries Regression Clear evidence for adverse with effects of inflation comes instrumental from the experiences of variables high inflation. Spline Threshold level at 8 per regression cent. There is strong empirical evidence, based on most of the recent cross-country panel studies, that the relationship between growth and inflation is nonlinear. In other words, inflation may have both positive and detrimental effects on growth. These studies are both recent and relatively old. They are also carried out inside and outside the IMF and the World Bank. For instance, Sarel (1996), a study carried out at the IMF, points out the importance of nonlinearity ignoring which may lead to estimated effect of inflation on growth being biased by a factor of three. The study using annual data for 87 countries over 1970-1990 finds evidence of a structural break of inflation rate at 8 per cent. Sarel (1996), however, does not differentiate between developed and developing countries and, therefore, suffers from the potential bias in the estimation due to combining various countries at different levels of development (see Sepehri and Moshiri 2004). Other IMF studies, including Ghosh and Phillips (1998) and Espinoza et al. (2011), also support a nonlinear relationship. Ghosh and Phillips (1998) also argue that this relationship is convex in nature. So, for instance, the decline in growth associated with an increase in inflation from 10 per cent to 20 per cent is much higher than that associated with an increase from 40 per cent to 50 per cent. Espinoza et al. (2011) revisit the impact of inflation on growth. They justifiably use a nonlinear framework – logistic smooth transition regression (LSTR) 49 model – to estimate the inflation threshold incorporating 165 countries over the period 1960-2007. They find threshold levels in the range of 7 to 13 per cent for developing countries. Examples of studies outside the IMF include Pollin and Zhu (2006) and LopezVillavicencio and Mignon (2011). The study by Pollin and Zhu (2006) incorporates a squared term of inflation in their estimation of the inflation threshold for 80 countries for the period 1961-2000. On the other hand, Lopez-Villavicencio and Mignon (2011) employs panel smooth transition (PSTR) model as well as generalised method of moments (GMM) with a quadratic interaction term of inflation, similar to Pollin and Zhu (2006), to check the robustness of their results. In short, these findings show the importance of nonlinearity in the relationship between growth and inflation, ignoring which might “significantly underestimate the impact of inflation on growth” (Espinoza, Leon & Prasad 2010, p. 6). Studies also find that the degree of correlation between growth and inflation may vary according to circumstances. For instance, the influential study by Bruno and Easterly (1998) finds correlation with high frequency data and with extreme inflation observations. Pollin and Zhu (2006) draw the conclusion that the relationship between growth and inflation is more strongly correlated in the case of policy aiming at demand management to stimulate growth. Both these results have relevance in explaining the correlation between growth and inflation in developing countries. Developing countries are more prone to supply shocks and crises resulting in episodes of high inflation. At the same time since private sectors are not well developed, these economies often resort to government initiatives to stimulate demand and thereby growth. Although there exists a general consensus on the relationship between growth and inflation being nonlinear, there is less agreement over the issue of causality. In spite of their finding that inflation hurts growth Ghosh and Phillips (1998, pp. 708-709) state that they cannot “claim to have shown that inflation causes growth…[as] it is difficult to conceive of any methodology that would decisively prove causality from inflation to growth.” In an earlier and most widely cited study, Barro (1996) suggests that effects of inflation on growth and investment are significantly negative and the 50 causation is believed to be from higher long-term inflation to reduced growth and investment. His empirical analysis covers 100 countries from 1960 to 1990. Using a smaller sample and timeframe, but with a more formal econometric technique – Granger causality method – Paul et al. (1997) examine the direction and pattern of causality between inflation and economic growth in 70 countries over the period 1960-89. Their results find unidirectional causality in 26 countries, bidirectional causality in 16 countries, and no causal relationship in 28 countries. The other important finding of the study is that majority of the countries which show either uni or bidirectional causality belong to the industrial country group. In other words, for developing countries the causality between inflation and growth is not much evident. Paul et al. (1997) also note that majority of the developing countries showing no trade-off between inflation and growth are medium to high inflation economies (ibid., p. 1393). This finding by Paul et al. (1997) seems to contradict the conclusion drawn by Bruno and Easterly (1998) who state that harmful effect of inflation is evident only in the presence of extreme inflation.43 The existence of nonlinearity in the relationship suggests that there is a threshold level of inflation beyond which inflation has an adverse effect on growth. In a study by Amato and Gerlach (2002, p. 788), the authors state that “there is little hard evidence to suggest that the level of inflation targets [in developing countries]…should be much different than in more advanced economies.” Evidence based on the cross-country studies, however, tends to support the idea that the threshold level of inflation varies between developed and developing countries with a relatively higher level for the latter. Sepehri and Moshiri (2004) point out that the inverted U shaped relationship between inflation and growth vary across countries of various stages of development. The study by Ghosh and Phillips (1998, p. 709), carried out at the IMF, also believes that the threshold level of inflation is “expected to differ, at least somewhat, across countries.” Because of the existence of such degree of heterogeneity across countries in terms of development, Sepehri and Moshiri (2004) suggest that it is inappropriate to set “a single numerical policy target” uniformly for all developing countries (ibid., p. 202). Their basic model is based on Solow’s augmented production function which is estimated using Sarel’s 43 We, however, know that correlation and causation are not the same. Bruno and Easterly (1998) looked at the correlation while Paul et al. (1997) examined the causation. 51 (1996) spline regression technique. The dataset contains annual data covering 92 countries (including 26 and 28 lower middle income and low income countries respectively) over the period 1960-1996. In general, for developing countries the study finds a threshold at a double digit level and higher than for advanced economies. They, however, find a higher threshold for lower middle income countries compared to low income countries (see Table 3) which is inconsistent with other studies. Generally, as mentioned above, poorer countries are likely to have higher threshold of inflation affecting growth. This higher inflation tolerance of developing countries might be due to a number of reasons such as the BalassaSamuelson effect, indexation systems, exchange rate policies, and experience of high bouts of inflation in these countries (see Lopez-Villavicencio & Mignon 2011, p. 462) In a similar study around the same time, Burdekin et al. (2004) use a variant form of Sarel’s (1996) econometric technique and treat developed and developing countries separately to estimate the effect of inflation on growth. The study uses annual data for 21 industrial countries from 1965-1992 and 51 developing countries from 19671992. The inflation threshold for developing country, according to this study, is 3 per cent which is significantly lower than 8 per cent found for the advanced economies. This finding is at odds with the results found by Sepehri and Moshiri (2004) as well as other major recent studies. Perhaps a rationale behind this result by Burdekin et al. (2004) can be found in the explanation by Dorrance (1966). Prices are more inflexible downwards as a result of more organised trade unions in rich countries compared to poor countries. Developing economies are also mostly dependent on a few primary products for exports and agricultural products. Therefore, Dorrance (1966) argues that the appropriate rate of inflation needed to achieve relative price flexibility in advanced economies is higher than that of developing countries. One serious limitation of the study by Burdekin et al. (2004) is that it uses previous period real GDP per capita as a regressor. The level of real GDP per capita is likely to be nonstationary and thus using this variable may lead to a spurious relationship. Not only does the threshold vary between developed and developing countries, the difference between the figures is quite substantial. One of the most influential studies in this area, Khan and Senhadji (2001), finds inflation threshold for developing 52 countries at 11-12 per cent. This double digit figure is much higher than the 1-3 per cent threshold found for the advanced economies. The dataset of this study, done at the IMF, covers 140 countries over the period 1960-1998. More importantly, the study conducts sensitivity analysis to see the effect of inflation on growth when the threshold level varies from 1 to 50 per cent. The sensitivity analysis shows a positive effect of inflation on growth up to 18 per cent for developing countries (p. 16). In a more recent study (outside the IMF), Lopez-Villavicencio and Migon (2011) also find a wide difference between the threshold figures for developed (2.7 per cent) and developing countries (17.5 per cent). The study shows that for developing countries the relationship between inflation and growth is non-significant when inflation is below 17.5 per cent. The results are consistent using both PSTR model and GMM. Based on their results, the authors question the validity of long-run neutrality of money since monetary policy may have different effects on output depending on the inflation level. The findings of both these studies, in short, tell us that governments in developing countries need not worry too much about runaway inflation and its marginal impact on growth when inflation is below 18 per cent. This is in sharp contrast to what IMF advocates for – keeping inflation at 5 per cent level – and therefore may have profound implications on the policy making in developing countries. A low inflation threshold such as 5 per cent may be applicable for advanced economies but not in the case of developing and emerging economies. Apart from the two studies mentioned above, several other studies find the threshold for developing countries generally between 7 and 19 per cent. Pollin and Zhu (2006) estimate the inflation threshold for 80 countries over the period 1961-2000. The authors, in addition to entire sample period, consider four separate decades and consistently find that higher inflation is associated with moderate gains in growth up to around 15-18 per cent inflation. Based on their results, they have questioned the justification of inflation targeting policies to keep inflation at 3-5 per cent levels. There are, however, several limitations of this study. First panel estimation models – pooled OLS, fixed effects, random effects, and between effects – used are not appropriate in a dynamic setting and fail to take care of the issue of endogeneity problem. Such problems can be taken care of by using GMM estimations which some of the other studies relied on. The study also does not include variables relating 53 to money as a control variable in the model.44 Money is an important covariate in explaining the growth-inflation relationship. Moreover, in the case of regression by decades, estimates of linear and squared terms of inflation are not significant at 5 per cent in a number of cases. Drukker et al. (2005), using an unbalanced panel dataset covering 138 countries spanning over 1950-2000, find a well-defined threshold level at 19.16 per cent for the full and non-industrialised samples.45 One limitation of this study is that the inference made is based on a non-dynamic fixed effects model instead of more appropriate dynamic models such as GMM. Bick (2010) also finds a similar result, but argues that the inclusion of regime intercept, used in his study, lowers the threshold from 19 per cent to 12 per cent. Inflation rates less than 12 per cent are associated with a significant positive effect on growth. Bick’s (2010) study consists of a balanced panel dataset46 of 40 developing countries covering the period 19602004. Instead of suggesting a precise threshold point the study by Espinoza et al. (2011), done inside the IMF, shows that inflation threshold for developing countries lies between 7 and 13 per cent. For greater precision of the estimates they have used bootstrapping and Monte Carlo simulations. Finally, Yilmazkuday (2013) reveals that the positive impact of some of the determinants of growth on growth depends on level of inflation. The positive effect of human capital on growth is present and significant when inflation is below 15 per cent. Financial development is effective only when inflation is below 10 per cent and trade has a positive impact on growth when inflation remains below 8 per cent. An interesting finding of the study is that government size negatively affects growth when inflation is below 10 per cent. This shows that in a low inflationary environment government expenditure may not be able to positively affect growth. Thus keeping inflation at a low single digit level may conflict with the need for government expenditure which most of the developing countries resort to for development, in the absence of a well-developed private sector. Although the 44 The paper by Khan and Senhadji (2001) also suffers from the same problem. They also find two but lower threshold levels, 2.57 and 12.61 per cent, for industrial countries. 46 One may question this aspect of the dataset as it is difficult to create a balanced dataset for developing countries due to poor availability of data. 45 54 threshold figures for inflation vary, the author takes a conservative approach and suggests that inflation hurts growth when it goes beyond 8 per cent. One of the limitations in the methodology of the above mentioned studies is that they usually take five-year averages to smooth out business cycle fluctuations. It is not appropriate to assume uniform business cycle for all the countries within crosscountry panel studies. Their business cycle may not follow the same pattern. Besides, averaging out data by taking five-year averages causes reduction in sample which might make the variables statistically insignificant. These issues are discussed in detail in the empirical part of the thesis, Chapter 5. We also notice that certain control variables chosen in the models of the studies mentioned above may not be appropriate in examining the growth-inflation relationship. The issue here is how inflation affects growth conditioned upon other covariates. The interest, therefore, is not in finding the determinant of growth or how they impact growth. The covariates should be chosen not because they are the potential candidates of growth determinants but because they may influence the condition and the way in which inflation affects growth. In this regard, one can argue that for developing countries consumption is a better regressor than investment (as is used in most of the studies cited above) since the former has a greater explanatory power in the changes in aggregate demand and thus both growth and inflation. At the same time, population growth used as a proxy for labour supply growth may not be effective as a left hand side variable. In most developing countries population control in recent decades is a dominant policy strategy; as a result, it has become nearly constant. Therefore, how powerful it is in explaining the growth-inflation relationship becomes questionable. 3.3 Conclusion The review of the existing literature on cross-country panel studies provides strong evidence that inflation threshold in developing countries is well above 5 per cent level. Broadly, the threshold varies between 7 and 19 per cent with majority of the studies supporting a double digit level for developing countries. The findings clearly reveal that the IMF’s suggestion to keep inflation at 5 per cent or below may be 55 applicable for developed countries, but certainly not for developing countries. Most studies including the influential ones carried out at the IMF suggest that the threshold for developing countries is significantly higher than that of advanced economies. Since the growth-inflation relationship is non-linear, developing countries can gain from moderate levels of inflation and should not be alarmed when inflation crosses the 5 per cent benchmark set by the IMF. In fact, Anwar and Islam (2011), as discussed in Chapter 2, conclude that considering the threshold effects setting too low an inflation target, particularly at single digits, can impose opportunity costs in terms of foregone growth and employment creation. Anwar and Islam (2011) also find little evidence of low inflation being translated into benefits of reduced cost of borrowing. This is because, as the authors argue, such costs are likely to be determined by structural factors of the economy. Epstein and Yeldan (2008) also draw a similar conclusion and state that the IMF prescribed stabilisation program to attain price stability to ensure growth has not been materialised. They suggest to design a more socially desirable macroeconomic policy environment to promote employment, sustained growth, and improved income distribution. 56 Chapter 4 An Enquiry into the Relationship between Inflation and Growth in Selected Developing Countries of Asia 4.1 Introduction There is much evidence, from the discussion of the previous 3 chapters, that moderate inflation in the range between 10 and 15 per cent in the developing countries may not signal a worrying sign. In other words, inflation within this range may not cause acceleration in inflation expectations which is feared to affect economic growth adversely. The critical review and historical analysis also provide arguments that maintaining inflation at a low single digit level may entail a high sacrifice for the developing countries in terms of growth, employment generation, and poverty reduction. Monetary policy aimed at a low inflation target may have such unwarranted effects especially when inflation is caused by supply side factors. In short, the fear of runaway inflation – believed to be triggered when inflation reaches beyond 5 per cent – and the policies targeting a low inflation may create constraints on the performance and progress of developing countries, which might otherwise grow at a faster pace. Against this background, this chapter critically reviews country-specific studies on the relationship between growth and inflation. A list of 15 developing countries of Asia is selected to review the country-specific cases. These countries are Bangladesh, Bhutan, Cambodia, China, India, Indonesia, Lao PDR, Malaysia, Maldives, Nepal, Pakistan, The Philippines, Sri Lanka, Thailand, and Vietnam. The time series data used for supporting the critical review are collected from the World Bank’s World Development Indicators (WDI) and the IMF’s World Economic Outlook (WEO). The periods considered for inflation and growth are 1960-2009 and 1960-2010 respectively. The difference in the periods is due to the availability of data from the WDI 2011 at the time of the analysis. In addition, the periods for some of the countries are much smaller because of poor availability of data. 57 The rest of the chapter proceeds as follows. Section 4.2 looks at trends in inflation and growth in the countries in question. Section 4.3 provides evidence on the nature of relationship between inflation and growth from the literature. A discussion on findings about threshold levels of inflation takes place in Section 4.4. Section 4.5 reviews experiences of policies targeting low inflation in the selected countries. The debate on sources of inflation in the selected countries is reviewed in Section 4.6. Finally, Section 4.7 draws conclusion. 4.2 Trends in Inflation and Growth in Some Selected Developing Countries of Asia A summary of descriptive statistics of inflation in the 15 selected countries is presented in Table 4.1. Long term data over the period 1961–2009 show that median inflation rate varies from a relatively moderate 7–10 per cent to a relatively low 2–4 per cent. For instance for countries such as India, Indonesia, Pakistan, The Philippines, and Sri Lanka the rate ranges between 7 and 10 per cent while for Malaysia and Thailand it varies between 2 and 4 per cent. Inflation volatility, measured by the coefficient of variation (CV), also reveals differences amongst these countries. For example, while India has median inflation rate of around 7 per cent, higher than China’s at around 2 per cent, its inflation is less volatile compared to China’s. This shows that low long-run inflation may not necessarily ensure less volatility in inflation and therefore less uncertainty arising from it. Such differences confirm that historical inflation experiences vary across countries. Considering an identical inflation threshold point such as 5 per cent for all, therefore, is not appropriate (see for instance Sepehri & Moshiri 2004). Table 4.1: Summary Statistics of CPI Inflation in 15 Developing Countries of Asia Country Period Mean Median Max Min SD CV Bangladesh 1971 – 2009 10.82 8.40 54.60 -2.40 11.88 1.10 Bhutan 1981 – 2009 7.77 7.03 18.02 1.88 3.87 0.50 Cambodia 1995 – 2009 5.64 3.92 25.00 -0.79 6.87 1.22 China 1970 – 2009 4.56 2.15 24.24 -1.41 6.03 1.32 India 1961 – 2009 7.66 7.16 28.60 -7.63 5.28 0.69 Indonesia 1961 – 2009 52.87 10.46 1136.25 3.72 166.91 3.16 58 LAO PDR 1989 – 2009 24.21 10.63 128.42 0.04 32.01 1.32 Malaysia 1961 – 2009 3.22 2.81 17.33 -0.41 3.11 0.97 Maldives 1988 – 2009 5.72 4.73 20.18 -2.86 5.89 1.03 Nepal 1965 – 2009 8.18 8.24 19.81 -3.11 5.28 0.65 Pakistan 1961 – 2009 8.10 7.16 26.66 -0.52 5.56 0.69 Philippines 1961 – 2009 9.72 7.51 50.34 0.75 8.66 0.89 Sri Lanka 1961 – 2009 8.99 8.45 26.15 -0.16 6.01 0.67 Thailand 1961 – 2009 4.80 4.14 24.31 -0.85 4.85 1.01 Vietnam 1996 – 2009 6.22 6.36 23.12 -1.71 5.79 0.93 Source: Author’s calculations based on the data from the IMF, World Economic Outlook and World Bank, World Development Indicators. Table 4.2 records the descriptive statistics relating to growth in the selected countries. In terms of long-run growth experiences, between 1961 and 2010, China and Malaysia post relatively good growth performance with, median rates at 8.85 and 6.81 per cent, respectively. Nepal, one of the poorest countries in the region, posts the worst growth record with a median rate at 3.95 per cent.47 Output growth volatility also reveals some interesting stylised facts. For example, high growth rate can be associated with high volatility. This is evident from the Chinese experience. While its median growth rate, during the period 1961–2010, is the highest amongst the countries under study, volatility of its growth is also relatively high. On the other hand, growth variability in Pakistan and Sri Lanka, for example, is relatively low, but so are their growth rates, standing below 5 per cent. We also note an important fact from the growth figures shown in Table 4.2. The majority of countries, except for China and Cambodia (for the latter, relatively shorter time series data are available), fall short of the required 7 per cent growth to meet their Millennium Development Goals (MDGs) targets as suggested by the IMF study Selassie et al. (2006). Table 4.2: Summary Statistics of Real GDP Growth in 15 Developing Countries of Asia Country Period Mean Bangladesh 1971 - 2010 3.70 Bhutan 1982 - 2010 Cambodia China Median Maximum Minimum SD CV 4.72 9.16 -15.05 3.99 1.08 7.32 6.75 18.42 2.06 3.59 0.49 1994 - 2010 7.41 7.73 12.44 -1.89 3.24 0.44 1961 - 2010 7.66 8.85 17.73 -31.61 7.55 0.98 47 Nepal’s rank is at 157, at the bottom of the ranking amongst these countries according to Human Development Index (HDI) 2011. 59 India 1961 - 2010 5.47 5.47 19.57 -5.38 3.54 0.65 Indonesia 1961 - 2010 5.40 5.96 11.36 -14.07 3.84 0.71 Lao PDR 1985 - 2010 5.93 6.07 13.27 -2.03 2.84 0.48 Malaysia 1961 - 2010 6.20 6.81 11.08 -7.64 3.41 0.55 Maldives 1996 - 2010 6.24 6.98 16.53 -4.76 5.04 0.81 Nepal 1961 - 2010 3.56 3.95 9.24 -3.02 2.71 0.76 Pakistan 1961 - 2010 5.21 4.98 10.75 0.47 2.30 0.44 Philippines 1961 - 2010 3.94 4.62 8.54 -7.61 3.04 0.77 Sri Lanka 1961 - 2010 4.60 4.89 7.71 -1.56 1.88 0.41 Thailand 1961 - 2010 6.13 5.94 12.48 -11.10 3.81 0.62 Vietnam 1985 - 2010 6.55 6.75 9.11 2.75 1.72 0.26 Source: Author’s calculations based on data from the IMF, World Economic Outlook and World Bank, World Development Indicators. Time series graphs of inflation and growth of the countries under study also help decipher some of the stylised facts about the relationship between the two variables. These graphs are categorised in terms of countries within the region and are presented in Figures 4.1a – 4.1.d. In general, trends in inflation and growth in these countries reveal that over a long period the two variables seem to move together. There is, however, no particular pattern in the relationship that can be observed from the graphs across countries. For example, in the case of four countries of South Asia – Bangladesh, India, Pakistan, and Sri Lanka – there are traces of inflation becoming both lead and lag variable, implying that the causality between inflation and growth may run in both ways – inflation causing growth as well as growth causing inflation (Figure 4.1.a). In the early 1980s, India’s rising growth rates seemed to be preceded by rising inflation. On the other hand, in the early 1990s rising inflation seemed be followed by declining growth. We cannot, however, draw conclusion about the causality between the two based on such casual observations. This requires econometric techniques such as Granger causality test to determine the direction and nature of causality. We also notice some interesting facts about Malaysia between the mid-1980s and mid-1990s (Figure 4.1.c). High and sustained growth during this period seems to be associated with rising inflation. The economy of Thailand shows a similar trend in growth and inflation in the latter half of the 1980s. The Chinese economy, between the period 1980 and 2010 as illustrated in Figure 4.1.d, also appears to show that inflation is mostly a lag variable. 60 The evidence from the above discussion lends support to the discussion in Chapter 1. That is, there are past instances when developing countries had economic growth with rising inflation. It is also consistent with Bruno and Easterly’s (1996) observation discussed in Chapter 1. The co-movement of inflation and growth reveals an interesting feature. When growth rises, so does inflation. But once inflation reaches a certain level growth starts to decline causing inflation to fall in turn. This feature to a great extent provides support for the structuralist view – inflation is an inevitable consequence of economic growth. What is perhaps also implied that causality mostly runs from growth to inflation. Figure 4.1.a: Inflation and Growth in South Asia – Bangladesh, India, Pakistan, and Sri Lanka Bangladesh India 60 30 40 20 20 10 0 0 -20 -10 60 65 70 75 80 85 90 95 00 05 10 60 65 70 75 Pakistan 80 85 90 95 00 05 10 95 00 05 10 Sri Lanka 30 30 25 25 20 20 15 15 10 10 5 5 0 0 -5 -5 60 65 70 75 80 85 90 95 00 05 10 CPI Inflation 60 65 70 75 80 85 90 Real GDP Growth Source: IMF, World Economic Outlook and World Bank, World Development Indicators. 61 Figure 4.1.b: Inflation and Growth in South Asia – Bhutan, Maldives, and Nepal Bhutan 20 16 12 8 4 0 60 65 70 75 80 85 90 95 00 05 10 90 95 00 05 10 90 95 00 05 10 Maldives 25 20 15 10 5 0 -5 60 65 70 75 80 85 Nepal 20 16 12 8 4 0 -4 60 65 70 75 80 CPI In f la t io n 85 Real GDP Growth Source: IMF, World Economic Outlook and World Bank, World Development Indicators. 62 Figure 4.1.c: Inflation and Growth in East Asia – Indonesia, Malaysia, Philippines, and Thailand Indonesia Malaysia 60 20 15 40 10 20 5 0 0 -5 -20 1970 1975 1980 1985 1990 1995 2000 2005 2010 -10 1970 1975 1980 1985 Philippines 1990 1995 2000 2005 2010 2000 2005 2010 Thailand 60 30 50 20 40 30 10 20 0 10 -10 0 -10 1970 1975 1980 1985 1990 1995 2000 2005 2010 CPI Inflation -20 1970 1975 1980 1985 1990 1995 Real GDP Growth Source: IMF, World Economic Outlook and World Bank, World Development Indicators. 63 Figure 4.1.d: Inflation and Growth in China and East Asia – Cambodia, Lao PDR, and Vietnam Cambodia China 30 25 25 20 20 15 15 10 10 5 5 0 0 -5 1980 1985 1990 1995 2000 2005 2010 -5 1980 1985 1990 Lao PDR 1995 2000 2005 2010 2000 2005 2010 Vietnam 160 25 120 20 80 15 40 10 0 15 10 5 5 0 0 -5 1980 1985 1990 1995 2000 2005 2010 CPI Inflation -5 1980 1985 1990 1995 Real GDP Growth Source: IMF, World Economic Outlook and World Bank, World Development Indicators. 4.3 The Nature of Relationship between Inflation and Growth High episodes of inflation distort macroeconomic stability and adversely affect economic growth. As a result a strong inflation aversion develops and an antiinflationary psyche starts to dominate the mindset of the policy makers in such inflation hit countries (see, for instance, Chowdhury & Ham 2009; Chowdhury & Siregar 2004).48 As Bruno (1992) puts it clearly: “…stabilization may, for a time, reduce inflation dramatically, but it will not …eradicate inbred inflation mentality…on the part of all relevant agents, both public and private.” Such high bouts of inflation, however, are not frequent in economic history and therefore do not 48 A classic example is Indonesia within this region. In view of hyperinflation in the mid-1960s, which coincided with economic stagnation, inflation was regarded as ‘enemy number one’ (Chowdhury 2002, p. 21). 64 provide a clear picture of the relationship between inflation and growth. To reiterate the findings of Bruno and Easterly (1998), inflation has a clear detrimental effect on growth at a very high rate such as 40 per cent. The nature of relationship is not conclusive below such high levels of inflation in the literature. This section reviews the issue in the context of the sample countries of Asia. Table 4.3 lists some of the recent studies investigating the nature of relationship between inflation and growth in the countries selected. Table 4.3: Studies on Inflation-Growth Causality Study Empirical Dependent Regressor Approach Variable s 1971 – Vector Growth Lagged In the short run Mukhopadhya 2007 autoregressiv and values of inflation y (2011) (annual e (VAR) inflation growth Granger causes ) model and growth but in inflation the long run Datta and Country Malaysia Period Findings growth Granger causes inflation. Hussain and Pakistan Malik (2011) 1960 – Cointegratio Growth Growth Inflation is 2006 n and error and and positively (annual correction inflation inflation related to ) model growth. Hossain Banglades 1973 – Cointegratio Real Real A cointegral (2010) h 2008 n and error balances of GDP, relationship (annual correction broad domestic exists amongst ) framework money and nominal money supply, CPI interest growth, and inflation rate, inflation. foreign nominal interest rate, nominal effective exchange rate Mahadevan India, 1966 – Multivariate Labour CPI and No causality 65 and Asafu- Indonesia, 1997 vector error productivit money between Adjaye (2006) Malaysia, (annual correction y supply inflation and Philippines ) model productivity , Sri growth in Lanka, Thailand, Thailand Indonesia, (and Philippines, and others) Sri Lanka; one way causality from inflation to productivity growth in Malaysia; and bidirectional causality in India. Ahmed and Banglades 1981 – Cointegratio Growth Growth A statistically Mortaza h 2005 n and error and CPI and CPI significant (annual correction inflation inflation negative ) models (2005) relationship between inflation and growth exists in the long run. Hossain (2005a) Indonesia 1954 – Granger CPI Money No short run 2002 causality inflation supply causality from (annual model and money growth, inflation to supply CPI growth exists. growth inflation, But a short run permanen bi-directional t income causality ) between inflation and money supply growth exists suggesting that the former having a feedback effect on the latter 66 during a hyperinflationar y period. Hossain Indonesia (2005b) 1952 – Cointegratio 2002 CPI Narrow Price level, n and error money money supply, (annual correction (M1), and real ) (ECM) broad permanent model money income form a (M2), real weakly GDP, cointegral nominal relationship. exchange rate Mubarik Pakistan (2005) 1973 – Granger 2000 causality test Growth CPI A unidirectional inflation, Granger (annual populatio causality exists ) n growth from inflation rate, and to growth. investmen t growth rate 1950 – Bivariate and Siregar 1997 (2004) Chowdhury Indonesia Growth Inflation A bi-directional vector and causality (annual autoregressiv squared between ) e model term of inflation and (VAR) inflation growth exists but in the long run inflation is unlikely to negatively affect growth because of the evidence of long run neutrality of money. Chowdhury (2002) Indonesia 1950 – Descriptive 1997 Analyses n.a. n.a. No statistical significant (annual relationship ) between 67 inflation and growth and therefore provide an argument against tightening of policies aiming at low inflation. Mallik and Banglades 1974 - Cointegratio Growth Growth Positive Chowdhury h, India, 97 n and error and and relationship (2001) Pakistan, (BGD) correction inflation inflation between and Sri 1961- model Lanka 97 inflation and growth. (IND) 195797 (PAK) 196697 (LKA) (annual data) Notes: The studies are arranged first chronologically and then in alphabetical order. As noted above, avoiding the influence of few outliers of inflation in the analysis of inflation-growth relationship is important and appropriate. Chowdhury and Siregar (2004) argue that once the extreme cases of inflation – a rate greater than 100 per cent – are removed one observes a positive relationship between the two variables in the case of Indonesia.49 However, there seems to be no relationship between the two when inflation is in the range between 8 and 15 per cent. A similar conclusion is drawn in an earlier study by Chowdhury (2002). In both these studies, initial observations using scatter plot diagrams show that both high and low growth rates occur when inflation is in the range of 5–18 per cent. 49 Removing such outliers to look at the relationship between inflation and growth is important in the case of Indonesia since the country has experienced very high episodes of inflation, particularly in the 1960s when inflation averaged above 196 per cent (Chowdhury & Siregar 2004, p. 140), within the sample period (1950 – 1997) of the study. 68 The key motivation of Chowdhury and Siregar’s (2004) study is to examine if there is any scope for a moderate inflation rate above what is targeted in Indonesia around that period. So the study sets itself to investigate the two important issues – the nature of relationship between inflation and growth and a range of inflation at which growth will not be negatively affected. It is set against the backdrop when Indonesia embraced inflation targeting in its monetary policy after the Asian Currency Crisis of 1997–1998. The inflation target level during the 2000s was set between 3 and 10 per cent in the country (see Siregar and Goo, 2010, Table 2, p. 115).50 Chowdhury and Siregar (2004) argue that “over-fixation with a single digit inflation target cannot be justified based on the fear of inflation going out of control once it is allowed to go beyond, say 10 per cent” (p. 139). According to them such tight inflation targets may have hampered economic growth and stalled the nascent recovery process.51 Chowdhury and Siregar (2004) emphasise that the understanding of the nature of relationship between inflation and growth is important in the presence of excess capacity and persistent high unemployment or underemployment. The authors argue, based on their findings, that while policy makers should be cautious against high inflation they should also be aware that moderate inflation may be required in a country such as Indonesia to sustain economic growth. In support of the case, Chowdhury and Siregar (2004) also provide results from econometric tests. The Granger causality test shows presence of bi-directional effects between economic growth and inflation. The findings from the impulse response functions show that a positive supply shock, which the authors have interpreted as a rise in the economic growth, leads to a decrease in inflation. In addition, a shock to inflation does not have any effect on economic growth at the impact period. Thus, Chowdhury and Siregar (2004) conclude that inflation can act as a “lubricant to the engine” of the economy. They also suggest to inflate the economy to induce growth during the recovery period. Chowdhury and Siregar (2004) make an important contribution in establishing the cases that excluding outliers of inflation there exists a positive relationship between 50 See also Chowdhury and Siregar (2004, Endnotes 3, p. 150). Indonesia was hard hit by the Asian Currency Crisis 1997 – 98, suffering from severe inflationary recession. To capture the severity of the crisis Chowdhury and Siregar (2004, p. 138) also provide some poverty statistics – the rate of poverty went up from a historical low of 11 per cent in 1997 to around 40 per cent in 1998. 51 69 the two variables and at moderate rates inflation may not be harmful for economic growth.52 Perhaps a limitation of the empirical method of the study is that it considers a bivariate framework, incorporating only inflation and growth, for their analysis. The bivariate vector autoregressive model (VAR), used in the study, would suffer from omitted variable bias leading to weak explanatory power in explaining the causality. A similar positive relationship between inflation and growth is found in four South Asian countries namely Bangladesh, India, Pakistan, and Sri Lanka in a study by Mallik and Chowdhury (2001).53 The empirical analysis of the study, using cointegration and error correction framework, also shows that sensitivity of inflation to changes in growth is larger than that of growth to changes in inflation. This implies that moderate inflation may be helpful for growth. Mallik and Chowdhury (2001), however, provide a cautionary note on possible overheating situation. This is because economic growth feeds back into inflation causing economy to overheat. A different finding is suggested in a study by Datta and Mukhopadhyay (2011) in the context of Malaysia for the period 1971 – 2007. Both short run and long run causation is found between inflation and growth, similar to Chowdhury and Siregar (2004). But in this case, Datta and Mukhopadhyay (2011) note a short run negative causality running from inflation to growth and a long run positive causality running from growth to inflation. There is also a long run equilibrium relationship between the two variables, as they are found to be cointegrated. These findings suggest that inflation is harmful for economic growth in the short run, but growth in the long run is inflationary. This conclusion is in contrast with that of Chowdhury and Siregar (2004). Perhaps the historical experience of inflation and growth in the two neighbouring countries, Indonesia and Malaysia, helps explain the differences. Long run growth performance of Malaysia is relatively better than that of Indonesia (median rates 6.81 and 5.96 per cent for Malaysia and Indonesia respectively). Long run inflation is significantly higher in Indonesia compared to Malaysia (median rates 52 The other significant finding of the study is the threshold level at around 20 per cent for Indonesia which is discussed in the following section. 53 The findings of this paper are discussed in detail in Chapter 2. 70 10.46 and 2.81 per cent respectively).54 The comparison of long-run median inflation and median growth in the two neighbouring countries may imply a negative long-run relationship between inflation and growth. However, it is also important to note that Indonesia grew in an inflationary environment.55 One could, therefore, ask whether Indonesia could grow at a faster rate had its inflation rate been lower, or similar to that of Malaysia. Unfortunately, that counter factual is not available. But perhaps, Indonesia needed higher inflation (more lubricants) given the lower level of development and larger infrastructure deficits than Malaysia. Additionally, Datta and Mukhopadhyay’s (2011) study seems to have serious methodological limitations. They conclude that inflation and growth are stationary variables and therefore they are cointegrated. This conclusion appears to be incorrect based on the standard practice in econometric theory. First, it is very likely that firstdifferenced variables such as inflation and growth rates would be stationary in a finite horizon, as the study have found. This however does not ensure that the two variables are cointegrated, as the paper concludes. In order to test for cointegration, standard practice requires unit root test for variables at their levels – real output and price level in this instance. If both series (variables at their levels) are non-stationary (possessing unit root) and error from the regression between the two variables is stationary (does not have unit root) only then cointegration exists between the two variables – output and price level. So, the conclusion of Datta and Mukhopadhyay (2011) that inflation and growth are cointegrated is methodologically flawed. Unlike the studies above, Hossain (2005a) finds no short run causality from inflation to growth in complete and subsample periods in the case of Indonesia between 1954 and 2002. Perhaps the difference in the causality results between the two studies – Hossain (2005a) and Chowdhury and Siregar (2004), both in the context of Indonesia – is due to the issues relating to modelling and time period considered. Hossain (2005a) performed the Granger causality test in a trivariate framework incorporating 54 For figures and the discussion on inflation-growth historical experience, both long run and medium run, see Chapter 3. The long run average is considered taking a five decade period between 1960 and 2009. 55 Even during the high episodes of inflation of the 1950s and 1960s the Indonesian economy grew, making Mackie (1967) to comment that the economy is somehow immune from adverse consequences of inflation. Hossain (2005a, p. 64) suggests a possible reason could be the subsistence nature of the economy during that period. Another important difference between the two countries is that unlike Indonesia Malaysia has not experienced any hyperinflationary period. 71 money supply growth. The expanded model, as Hossain (2005a) suggests, is appropriate to examine the effects of inflation on economic growth conditional on growth in money supply. The overall implication of the result of Hossain’s (2005a) study is, however, similar to that of Chowdhury and Siregar (2004) – inflation may not be harmful for growth – and Hossain (2005a) notes that the Indonesian economy did not suffer much from moderately high or volatile inflation since the early 1970s. This fact, he comments, is reassuring but requires further investigation. In a different study Hossain (2005b) provides similar comments on Indonesia’s remarkable economic growth and structural transformation amidst moderately high and fluctuating inflation. The empirical results in this case suggest a weak cointegrating but stable relationship among real income, price level, and money stock; therefore one can expect a causal relationship amongst them. The cointegrating relationship amongst the three variables, covering the period 1973– 2008 for Bangladesh, helps to infer that there are forces within the economic system that brings these variables together if they drift apart temporarily (Hossain, AA 2010, p. 574). Mahadevan and Asafu-Adjaye (2006) investigated the causal relationship between inflation and productivity growth in Asian economies.56 The empirical analysis of this study employs a multivariate vector error correction model covering the period 1966–1997. Based on the findings, they have grouped the countries in three different categories. Inflation and productivity growth do not have any causal relationship in Thailand, Indonesia, the Philippines, and Sri Lanka. A unidirectional causality runs from inflation to productivity growth in Malaysia and a bi-directional causality runs in India. Thus Mahadevan and Asafu-Adjaye (2006) conclude that a two-way causal relationship between inflation and productivity growth does not necessarily imply that monetary policy must be pursued using specific inflation targets, for instance, in the case of India. On the contrary, aggressive demand boosting fiscal and monetary policies may not trigger an inflationary spiral if, for instance, the observation of Mishra and Mishra (2012a) that the supply curve in India is rather flat showing evidence of an excess capacity in the economy is true. 56 An inference can be drawn on the impact of economic growth. 72 In the case of Malaysia, Mahadevan and Asafu-Adjaye’s (2006) results suggest that there is some unexploited scope for monetary policy to target inflation for productivity growth gains. However, they believe that if the country manages to keep prices stable in the future then the expected positive effect on productivity growth may not be significant if inflation targeting is adopted. Unlike its neighbours Indonesia, The Philippines, and Thailand who sought help from the IMF after the Asian Currency Crisis 1997-98, Malaysia refused to comply with the IMF’s suggestions to implement inflation targeting. In sum, the above discussion shows that inflation and growth may move together in developing countries. This is because of structural reasons such as inevitable budget deficits due to the need to finance the effort to close large physical and social structural deficits with a low tax base, subsistence agriculture, and various other supply rigidities or bottlenecks. To end this section, we perform a simple exercise using scatter plots and Polynomial Kernel Fit curves to provide empirical evidence on the relationship between growth and inflation from two groups of countries. The plots are presented in Figures 4.2.a– 4.2.d. The two groups are South Asia which includes Bangladesh, India, Pakistan, and Sri Lanka and Southeast Asia which includes Indonesia, Malaysia, The Philippines, and Thailand. These countries are chosen because of their better availability of data on inflation and growth. For South Asia a linear analysis shows inflation and growth have a positive relationship up to an inflation rate of 28 per cent (Figure 4.2.a). When inflation reaches 29 per cent the relationship starts to become negative. The nonlinear analysis using Polynomial Kernel Fit curve shows inflation can have both positive and negative effects on growth (Figure 4.2.b). This curve is drawn for the whole sample without any restriction imposed on inflation or real GDP growth. The inverted U shaped nature of the curve is consistent with the findings from the cross-country literature discussed in Chapter 2. From the Kernel Fit graph it appears that a turning point exists in the range between 10% and 15% of inflation. In the case of Southeast Asia, the linear analysis tells us that the relationship is positive up to 19 per cent of inflation (Figure 4.2.c). The relationship becomes 73 negative at a higher inflation rate. Comparison between the two groups of countries provides an important insight. Negative impact of inflation on growth starts at a higher inflation rate in less developed South Asia compared to more developed Southeast Asia. This is consistent with the empirical literature, reviewed in Chapter 2, that inflation starts to have negative impact at a higher rate for less developed economies. The nonlinear analysis using Polynomial Kernel Fit curve shows no clear relationship between inflation and growth between zero and 10 per cent of inflation (Figure 4.2.d). But there is a steep decline in growth taking place when inflation goes beyond 40 per cent. This result is consistent with Bruno and Easterly’s (1998) observation. The graph is drawn by imposing a restriction on inflation, that is, only inflation rates less than 100 per cent are considered. This simple exercise reveals some important results which are consistent with the findings from existing literature. Figure 4.2.a: Scatter Plot and Linear Fitted Line in Four Countries of South Asia 20 Real GDP Growth 15 10 5 0 -5 -10 -8 -4 0 4 8 12 16 20 24 28 CPI Inflation Linear Fitted Line 0.95 Ellipse 74 Figure 4.2.b: Scatter Plot and Polynomial Kernel Fit Curve in Four Countries of South Asia 20 16 Real GDP Growth 12 8 4 0 -4 -8 -12 -16 -10 0 10 20 30 40 50 60 CPI Inflation Figure 4.2.c: Scatter Plot and Linear Fitted Line in Four Countries of East Asia 15 Real GDP Growth 10 5 0 -5 -10 -15 -8 -4 0 4 8 12 16 20 CPI Inflation Linear Fitted Line 0.95 Ellipse 75 Figure 4.2.d: Scatter Plot and Polynomial Kernel Fit Curve in Four Countries of East Asia 15 Real GDP Growth 10 5 0 -5 -10 -15 -10 0 10 20 30 40 50 60 CPI Inflation 4.4 Identifying a Threshold Level of Inflation The recognition, at the operational level, that the relationship between inflation and growth depends on the level of inflation is translated into the use of threshold identification in the academic literature (see Espinoza, Leon & Prasad 2010, p. 3). However, the rate of inflation at which its negative effects on growth dominate its positive effects, making inflation detrimental to growth, remains a controversial issue (Sepehri & Moshiri 2004, p. 192). Cross country studies carried out at the IMF such as Khan and Senhadji (2001) and Espinoza et al. (2010) show the existence of a threshold level of inflation in the context of developing countries. While Khan and Senhadji (2001) suggest a level between 11 and 12 per cent, a slightly wider range of 7–13 per cent is suggested by Espinoza et al. (2010).57 Outside the IMF, studies such as Sepehri and Moshiri (2004) notes that the range could vary between 11 per cent and 15 per cent depending on the stage of development a country is in. These 57 Khan and Senhadji (2001, p. 16) also notes a positive effect of inflation on growth only present for inflation rates lower than 18 per cent for developing countries. This is found by performing a test of the sensitivity to the location of the threshold. 76 findings are discussed in Chapter 2. Generally, the cross-country literature suggests a higher threshold for poorer countries. In general, as summed up in Hwang and Wu (2011, p. 69) the techniques used to identify the turning point of inflation in the nonlinear relationship between inflation and growth include (a) certain benchmarks of inflation chosen arbitrarily, for instance by Fischer (1993), (b) spline regression method, such as Sarel (1996), (c) method of bootstrapping, for example by Khan and Senhadji (2001), and (d) incorporating a quadratic term of inflation, as used by Pollin and Zhu (2006). Findings of these cross-country studies provide a general view of threshold level in developing countries. Policy makers in a particular country would be more interested to know the evidence on a country specific context. The knowledge of such a turning point of inflation helps maximise the high growth-low inflation objective. Table 4.4 lists some of the recent country-specific studies attempting to gauge the threshold level of inflation in developing Asia. The selection of the studies is based on the sample countries under investigation. A critical review of the analyses and findings of the studies are provided below. Table 4.4: Studies on Threshold Level of Inflation Study Empirical Dependent Approach Variable 1960 – Cointegration Malik 2006 (2011) (annual) Hussain and Country Pakistan Period Regressors Findings Growth Growth and Threshold and error and inflation level at 9 per correction inflation cent. model Hwang and Wu (2011) China 1986 – Spline 2006 (annual) Growth Capital Threshold function growth rate, level at 2.5 regression labour per cent. growth rate, human capital growth rate, and CPI inflation rate 77 1972 – Threshold Growth Lagged Threshold and Ham 2007 vector and values of level at 10 (2009) (annual) autoregressive inflation growth and per cent is inflation suggested. Chowdhury Indonesia model (TVAR) Hayat and Bangladesh Kalirajan 1976- Vector Per capita Literacy No particular 2005 autoregressive growth and rate, life threshold (VAR) model inflation expectancy, found. Rise in agriculture inflation from value added, any level is gross capital harmful for formation, growth (2009) export, and government consumption as a percentage of GDP, growth in broad money supply, and dummy variables. Munir et al. Malaysia (2009) 1970- Threshold 2005 Growth Inflation, Threshold at autoregressive financial 3.89 per cent (TAR) model depth, gross capital formation 1981 – Cointegration Growth Growth and Threshold Mortaza 2005 and error and CPI CPI inflation level at 6 per (2005) (annual) correction inflation Ahmed and Bangladesh cent. models Hussain (2005) Pakistan 1973 – Threshold 2005 model (annual) Growth CPI No clear inflation, evidence of a population threshold growth rate, level is investment found. A income range of 4–6 ratio, and per cent is 78 M2 GDP suggested. ratio Mubarik Pakistan (2005) 1973 – Threshold 2000 models Growth (annual) CPI Threshold inflation, level at 9 per population cent. growth rate, investment growth rate 1950 – OLS using a and Siregar 1997 (2004) (annual) Chowdhury Indonesia Growth A linear and Threshold quadratic a squared level at 20 specification term of per cent is inflation suggested. 1971 – Cross Per capita Wholesale No threshold Kalirajan 1998 sectional growth price index level of (2003) (annual) regression (WPI) inflation is inflation, found. Any rate of increase in rainfall, inflation from literacy rate, the previous population period growth rate, negatively private affects sector growth. domestic Bringing investment down (% of GDP) inflation to rate, public single digit sector level, at par domestic with its trade investment partners’ (% of GDP) inflation, rate, would help government increase the consumption per capita expenditure growth by (% of GDP) about 2 rate, terms percentage of trade points. Singh and India Notes: The studies are arranged first chronologically and then in alphabetical order. 79 Among the studies on inflation threshold, presented in Table 4.4, the highest level of about 20 per cent is suggested by Chowdhury and Siregar (2004) in the case of Indonesia.58 The study estimates the turning point in a nonlinear framework, expressing growth as a quadratic function of inflation.59 However, the singleequation-OLS framework used in the study is subject to reverse causality problem particularly when the study itself identifies a bi-directional causality between the two variables. Thus, the study may overestimate the threshold level and the estimates using a better methodology may produce a lower threshold. For example, a much lower turning point between 8.5 and 11 per cent is found by Chowdhury and Ham (2009) for Indonesia by using a threshold Vector Autoregressive (TVAR) model. They have used annual data between 1972 and 2007, acknowledging that the results are limited by small sample size and, therefore, their investigation was exploratory in nature. Nevertheless, in light of the findings, the authors argue that targeting a rate below 5 per cent, as set by the central bank of the country, is “too much on the cautionary side…(appears) an overkill, and hence…harmful for the economy” (p. 653). Additionally, they point out that investment in Indonesia has declined as a result of high lending rates – a common instrument to fight inflation. However, they did not investigate causes of the decline in investment. Investment depends on both real cost of borrowing and expectations. It is not clear whether investment has declined due to high real cost of borrowing, as claimed by Chowdhury and Ham (2009), or due to poor expectations resulting from uncertainty in economic recovery. On the empirical side, the achievement of Chowdhury and Ham (2009) is to employ a superior econometric method (compared to OLS) – the threshold VAR trying to capture the structural break – the level at which inflation turns detrimental. The authors carefully take the outliers in the inflation series into consideration by incorporating a pulse dummy. However, there seems to be a number of limitations in the empirical methods used by them. For instance they employ inflation and growth 58 The study, as mentioned in the previous section, finds a positive relationship between inflation and growth, excluding inflation outliers, and no relationship when inflation is in the range of 8 – 15 per cent. 59 Pollin and Zhu (2006) also used this standard technique of finding the turning point of a quadratic function. 80 rates in an unrestricted VAR framework. It is argued as appropriate on the ground that the rates are stationary variables and, therefore, spurious regression is avoided. However, it does not provide any information on the nature of stationarity of the level variables of inflation and growth – price level and output level respectively. If the price level and output level are non-stationary but cointegrated then the appropriate framework is a restricted model – Vector Error Correction Model (VECM). An unrestricted model using the first differences (inflation and growth), even though they are stationary, is not considered a correctly specified model in the presence of cointegration. Secondly, Chowdhury and Ham (2009) argue for selecting a short lag-order so that the estimation does not lose degrees of freedom. It does not follow the standard lag order selection criteria such as AIC or BIC to determine the appropriate lag length to remove autocorrelation. The model’s adequacy, therefore, becomes rather weak. The difference between the suggested threshold levels in a particular country context is indeed noticeable. Considering that Indonesia’s historical median inflation rate is 10.46 per cent the findings of Chowdhury and Ham (2009) appear to be more consistent than what is suggested by Chowdhury and Siregar (2004). However, Indonesia is an inflation prone country; yet, showed the capacity to expand in a relatively high inflationary situation in the past. Therefore, the stylised fact that countries in a higher development phase should have a lower threshold level of inflation may not hold in the case of Indonesia. A comparison between Indonesia’s long run median inflation rate and that of three South Asian countries reveals some support for this (Figure 4.3). Indonesia is ahead of these countries in terms of economic development (a higher average per capita income and long run median growth rate as well) but has a relatively higher inflation experience. 81 Figure 4.3: Inflation and Growth in Selected Countries (Long Run Median in per cent) 4.98 Pakistan 7.52 7.15 5.96 Indonesia Long Run Median Growth 8.34 10.46 Median Inflation in the 2000s 5.47 4.32 India Long Run Median Inflation 7.16 4.72 6.22 Bangladesh 8.4 0 5 10 15 Source: Author’s calculation based on data gathered from IMF, International Financial Statistics and World Bank, World Development Indicators. A threshold level of 6 per cent is suggested by Ahmed and Mortaza (2005) for Bangladesh, using annual data between 1981 and 2005. The measure of threshold is an important contribution, perhaps first of its kind in the case of Bangladesh. This initiative provides some clear empirical evidence to the policy makers, helping them to conduct monetary policy. To achieve better results, its empirical estimation carefully ignores the outliers in the inflation and growth series of the 1970s. The outliers arise from unusual circumstances in the history of the country around that time.60 The empirically derived threshold level, however, is not consistent with the historical experience of the country. Long run median inflation for Bangladesh is 8.4 per cent, (see Figure 4.3). Besides, the experience of the last two decades shows that growth has gone up steadily from 4.78 to 5.77 per cent, to a historical high, when at the same time inflation has increased from 5.75 to 6.22 per cent. In other words, there is no strong evidence from the historical data that growth has declined when inflation has crossed the 6 per cent margin. The suggested threshold level of 6 per cent or below appears to be very conservative also, considering the findings for other countries for instance India, Indonesia, and Pakistan (see Table 4.4). Bangladesh falls 60 In 1971 after a war Bangladesh became independent. In 1974 the country suffered from famine and severe flooding. 82 at a much lower path of development among these countries and therefore likely to have a higher threshold level (see Figure 4.4 which shows that countries with lower per capita GDP tend to have higher long run inflation).61 For the same reasons the comments by Hossain (2010) is also not well founded. Hossain (2010) comments that for Bangladesh the level of threshold would be much lower than what is proposed by Khan and Senhadji (2001) for developing countries (11–12 per cent). Figure 4.4: Scatter Plot between Average Real GDP Per Capita and Long Run Inflation 12 Indonesia Long Run Median Inflation 10 Bangladesh Sri Lanka 8 The Philippines Pakistan 6 India 4 Nepal Malaysia Thailand China 2 0 0 500 1000 1500 2000 2500 3000 Average Real GDP Per Capita Source: Author’s calculation using data from World Bank, World Development Indicators and International Monetary Fund, International Financial Statistics. Note: Only the countries for which long data series are available are included. Both median inflation and average real GDP per capita are calculated over the period of 1960 – 2009. For Bangladesh the period is 1971 – 2009. On the empirical side, Ahmed and Mortaza (2005) have used the techniques developed by Khan and Senhadji (2001) – a bivariate model using dummy variable to capture the structural break. Using a bivariate model would be subject to the problems of omitted variable bias as mentioned above and perhaps a superior method 61 6 per cent is also lower than what is suggested by Khan and Senhadji (2001) and Espinoza et al. (2010) for developing countries. 83 would have been to estimate a multivariate model using instruments to avoid simultaneity bias. Evidence from the other countries of the South Asian subcontinent also provides some interesting facts. For Pakistan two studies carried out at the same time find contrasting results. For example, Mubarik (2005) suggests a turning point of inflation at 9 per cent. Hussain (2005), on the other hand, criticises Mubarik’s (2005) results and finds no clear threshold level. He suggests a “tolerable” level would be in the range of 4–6 per cent in the presence of both supply and demand shocks. Like many contemporary studies Mubarik (2005) too is inspired by the work of Khan and Senhadji (2001) and follows its framework. The measure of threshold point, 9 per cent, appears to be close to the findings (11-12 per cent) of Khan and Senhadji (2001) whose sample includes Pakistan. Mubarik’s (2005) empirical model consists of four variables – economic growth, inflation, population, and total investment growth rates. He, however, does not include any policy variables, monetary or fiscal, in the model. As a result the dynamics of how inflation and growth gets affected by policy instruments, such as money supply, interest rates, exchange rate, and fiscal deficits, remains unclear. In the case of data transformation, Mubarik (2005) uses Hodrik Prescott (HP) filter to further smoothen the data.62 He argues that there is still enough volatility in the variables even after log transformation.63 In the other study on Pakistan, Hussain (2005) criticises the use of HP filter to smoothen the data series and measure the output gap. Although the use of HP filter to remove volatility in growth is a common technique in econometric literature, Hussain (2005) points out that it lacks substantive economic content and often embodies problems relating to estimation in the presence of autocorrelation. Unlike Mubarik (2005), Hussain (2005) includes policy variables such as money supply GDP ratio within its multivariate framework and, therefore, provides somewhat a stronger basis in explaining the effects of monetary instruments. However, relying only on a monetary framework, which the model does, may suffer from problems of 62 Using HP filter to remove volatility in growth is a common technique in economic literature. Log transformation is particularly useful when data series are erratic in nature. Smoothing using log helps reduce the influence of outliers. 63 84 sufficient explanatory power. For instance, Khan and Sakib (2012) suggest that monetary determinants are rather marginal and discard monetary growth as a potential cause of high inflation in Pakistan. A variable such as political instability has a much greater power in explaining inflation in the country, as they suggest. The other key shortcoming of the study by Hussain (2005) is perhaps the suggestion of the lower bound of threshold level at 4 per cent. Given the historical median inflation rate at around 7 per cent and 11–12 per cent threshold suggested by Khan and Senhadji (2001) (whose sample includes Pakistan) this seems to be quite a conservative suggestion. The upper bound of the threshold range at 6 per cent may find some rationale from the long run experience of Pakistan. The highest decadal growth, in Pakistan’s history, was registered in the 1980s at 6.4 per cent when inflation remained at 6.2 per cent. Other high inflationary episodes recorded a lower associated growth. Another limitation of this study is that it fails to differentiate between the sources of inflation while suggesting the threshold. The discussion of the possible unwarranted effects of keeping inflation at such a low level in the presence of supply shocks remains absent in the study.64 Similar to the findings of Hussain (2005), Singh and Kalirajan (2003), using annual data from India for the period 1971–1998, strongly suggest that there is no threshold level. Inflation is said to be harmful and any increase in inflation from any level has negative effect on growth. This is indeed a very strong statement, not found in the wider literature, and contrary to the historical experience of the country. Singh and Kalirajan (2003) suggest that monetary policy in India should be focused towards maintaining price stability. This is in sharp contrast to the suggestion made by Jha (2008). Jha (2008) argues that in the presence of widespread poverty monetary policy targeting low inflation cannot be appropriate for a country such as India. The empirical model of Singh and Kalirajan (2003) uses cross sectional regression, within a multivariate framework, and suggest that policy should always create a downward pressure on inflation, “without having to worry about what the threshold level is” (p. 393). The authors note that average annual inflation in India during the 64 Even studies carried out at the IMF such as Selassie et al. (2006) underscore the importance of this aspect of countercyclical monetary policy (see Chapter 3). 85 sample period was 8.2 per cent, higher than that of most of its major trading partners (all of which are developed countries). Their results indicate that bringing down inflation at a very low level, close to that of its trading partners in the developed world, would help increase the per capita growth by about 2 percentage points. However, they fail to address the idea that inflation-growth profiles vary significantly between developing and developed countries, for instance as suggested by Sepehri and Moshiri (2004). Therefore, targeting the same inflation rates of the trading partners such as the USA and Germany may have serious negative consequences for India. The result by Singh and Kalirajan (2003) that any increase in inflation is harmful also contradicts with nonlinear nature of relationship between inflation and growth. Historical evidence suggests that India managed to sustain growth rates higher than 5 per cent at a time when inflation was around 9 per cent.65 Both Hussain (2005) and this study appear to have found similar results – no clear threshold level and inflation having a universal negative impact on growth – from their empirical results. The methods used in both studies do not explain clearly how they take care of the extreme values of inflation due to, for instance, a crisis or a shock. It is evident from Figure 4.1a that both India and Pakistan experienced high inflationary episodes during the early 1970s, at the time of first oil price shock. The results investigating the growth-inflation relationship could have been driven by the outliers in inflation data, as noted and explained by Bruno and Easterly (1998). The discussion on threshold level provides some interesting insights. The studies suggesting low inflation threshold level, for instance Ahmed and Mortaza (2005), Hussain (2005), and Singh and Kalirajan (2003), seem to be influenced by the decadal inflation experience of the 2000s. Figure 4.2 shows that in all cases inflation in the 2000s was less than its historical figures (except for Pakistan where both figures were close). The countries, in general, experienced low inflation in the 2000s, but this should not be the benchmark for setting the threshold standard (Anwar & Islam 2011). In addition, for most of the countries (except for India) the inflation 65 The decadal average growth rates of the 1980s and 1990s in India were 5.44 and 5.69 per cent (increasing) respectively, when inflation rates were 8.76 and 9.59 per cent (also increasing) respectively. 86 rates in the 2000s were higher than 5 per cent. The findings of most of the above studies reviewed also suggest a threshold above 5 per cent. In short, there is no strong evidence in support of the IMF’s claim, that countries require their inflation to be below 5 per cent to generate long term growth. The findings also provide strong evidence against one size fits all policy. The advice to keep inflation at a very low level for all countries is not supported by different threshold findings for different countries. It is also interesting to note different threshold findings for the same country. This could be due to modelling issues such as univariate or multivariate framework and selection of explanatory variables; and methods of empirical estimation taking into account reverse causality and effects of outliers or the sample size and period covered. 4.5 Experiences with Policies Targeting Low Inflation Emerging and developing economies, following the experience of developed countries, have started to adopt an inflation targeting approach to monetary policy to maintain price stability in recent years (Mishra & Mishra 2012a). There is a pressure on developing countries from the IMF to target low inflation and gear policy instruments accordingly. This section reviews the issue of inflation targeting at a low level in the context of the sample countries. Table 4.5 lists some of the recent studies incorporating the experiences of low inflation targeting in developing Asia. Table 4.5: Studies on Low Inflation Targeting Study Country Period Empirical Dependent Approach Variable Inflation Regressors Findings Lagged Inflation Gerlach Indonesia, 1985: 1 – Hansen’s mean and Philippines, 2010: 1 unbiased values of persistence has Tillmann Thailand, (quarterly) estimator of inflation declined (2012) China, and the sum of because of Malaysia autoregressive pursuing (among coefficients inflation others) targeting and therefore IT 87 can be successfully adopted by emerging market economies. 1996:1 – Instrumental WPI Output gap, Flexible and 2007:3 variable- inflation WPI domestic Mishra (monthly) generalised and output inflation, inflation method of gap expected targeting is moments (IV- WPI seen as a better GMM) inflation, alternative real compared to effective strict one in exchange the presence of rate, real insufficiently interest rate integrated Mishra India (2012a) financial markets and rigidities in the economy. 1996: 1 – Vector WPI World oil Inflation and 2005: 3 autoregressive inflation prices, targeting Mishra (monthly) model (VAR) federal framework is funds rate, suggested to output gap, achieve a WPI better inflation, monetary exchange management. rate, The pass monetary through of the policy international instruments supply shocks (growth to domestic rate of inflation is reserve fast. Mishra (2012b) India money, M0, yield of 91 day treasury 88 bill rate, and call money rate), gross bank credit, and broad monetary aggregates (M3) Siregar Indonesia 1990s: 1 – Autoregressive and Goo and 2008: 12 (2010) Thailand (monthly) Inflation Nominal Pursuing the distributed lag effective inflation (ARDL) model exchange targeting and Mark rate, world policy is seen switching inflation as successful. model rate, and Inflation has openness gone down in the post-IT period. The pass through effects has declined considerably except for tradable goods prices of Indonesia. Jha (2008) India 1992: 4 – Vector WPI Industrial The rationale 1998: 3 autoregressive inflation production for introducing (monthly) model (VAR) growth, inflation and Granger real targeting as an causality test effective exclusive exchange concern for rate, monetary narrow policy is money, call inappropriate money rate, particularly in CPI for the presence of industrial widespread workers. poverty. IT would not be 89 effective, even if adopted in the monetary policy, since short term interest rate does not have a significant effect on inflation. Notes: The studies are arranged first chronologically and then in alphabetical order. In the presence of widespread poverty the objective of monetary policy cannot be exclusively inflation control. This argument is put forward by Jha (2008) which evaluates the case for IT in India. Jha argues that poverty and its alleviation should be the “cornerstone of success of any policy” in a country which has a long standing problem of poverty. He believes “higher economic growth along with some supporting redistribution measures” can play a crucial role in reducing poverty. Crucially, Jha (2008) notes that the rate of poverty reduction was higher in the 1980s compared to the reform period of post 1991.66 Although the trend has been reversed later, he believes that there exists enough scope for poverty reduction through higher growth. A rapid reduction in poverty in India is attainable through a sustained economic growth of 8 per cent (Jha 2008, p. 260). The two important issues the monetary policy should address in this regard are interest rate and exchange rates. Jha (2008) suggests that low interest rate, to generate investment, and a slightly undervalued exchange rate with low volatility, to boost exports, are critical to sustaining high growth rates (p. 260). Jha (2008) also argues that the rationale for IT is incomplete in India. There is evidence of sacrifice in terms of output loss as a result of pursuing IT. Output movements in transition countries adopting IT have been higher than in developed market economies, Jha notes. Therefore, following the footsteps of the developed countries in pursuing IT cannot be justified. In addition, he argues that although IT may have been responsible for maintaining a low inflation regime, it has not brought 66 India suffered a balance of payment crisis in 1991. 90 down the inflation rate itself substantially or changed the volatility of the exchange rate. The potency of monetary policy in targeting inflation is also questioned by Jha (2008). He identifies price level stabilisation as an overriding short term concern of monetary policy in India. However, the causality tests show that short term interest rate – the key instrument with which monetary policy targets inflation – does not have a significant effect on inflation.67 Therefore, Jha (2008) concludes that the central bank of India could not implement IT effectively even if it wants to. The reason behind this result, Jha (2008) argues, is that monetary policy in India is constrained by a number of issues, such as incomplete liberalisation of financial markets. The interest rate transmission channel is incomplete due to these limitations. Because financial market liberalisation is not complete a strong monopoly elements still exists in the banking system. Mishra and Mishra (2012a, p. 1053) also point out that financial markets in India are still not sufficiently integrated to ensure quick transmission of interest rate impulses (supported by low sensitivity of demand) and there are still rigidities in the economy, as indicated by a flat Phillips curve. In the presence of such anomalies Jha (2008) supports the multiple indicator approach of monetary policy adopted by the central bank of India in April 1998.68 Such policy direction, as he suggests, would help maintain stable interest and inflation rates and a slightly undervalued currency in order to promote higher export led growth. The above discussion reveals why rule based interest rate policy, particularly of Taylor type, may not be efficient to stabilise a developing economy. This is also supported by Mishra and Mishra (2012a). The study attempts to answer the question of the probable consequences of shifting to an inflation targeting framework of monetary policy and how different shocks will affect the economy under such framework. It uses a general linear model of the economy with quadratic loss function to be minimised by the central bank for India. The results of the study 67 Granger causality tests indicate a weak relation between call money rate and measures of inflation. Empirical investigation of the paper raises a question though. Call money rate is found to be I(1), a non-stationary series possessing unit root. In a finite sample (April, 1992 – March, 1998) it is unusual that the rate has unit root. 68 The key objectives of this policy include (1) a stable inflation environment, (2) appropriate liquidity condition to support higher economic growth, (3) orderly conditions in the exchange rate market to avoid excessive volatility in exchange rate, and (4) stable interest rates (RBI 2002 cited in Jha (2008, p. 264)). 91 indicate a short run trade-off between inflation and output because of the presence of rigidities in the price setting behaviour. In light of the above results, they suggest that flexible domestic inflation targeting and discretionary optimisation in monetary policy would work better for economic stabilisation in India. However, in another study, Mishra and Mishra (2012b) suggest that the IT framework helps achieve better monetary management. They build a short run comprehensive VAR model of monetary policy for the Indian economy to examine a hypothetical inflation targeting monetary policy regime.69 The authors note that the multiple indicator approach of monetary policy is not efficient because “the multiplicity of objectives leads to inherent conflict among such objectives … in particular between exchange rate stabilization and inflation stabilization” (p. 87). It creates confusion in the market which remains unsure of which variable the central bank will choose to defend. The empirical estimation of Mishra and Mishra (2012b) is based on two models. First a benchmark case is constructed in line with the multiple indicator approach currently practiced by the central bank of India. In this case, the monetary policy instrument is set after looking at current values of inflation, output, and exchange rate. The second model is a pure inflation targeting case. Here, monetary policy instruments are allowed to react to inflation only contemporaneously. The results suggest that demand effects of changes in interest rate are stronger than exchange rate effects. The increase in interest rate (or negative monetary shocks) does not appreciate the exchange rate as much in the inflation targeting scenario as in the benchmark case. This helps mitigate the potential conflict between exchange rate and interest rate, one of the main monetary policy dilemmas in inflation targeting, as the study suggests. However, the interpretation is not consistent with the results presented in their study. The impulse response functions under both baseline and inflation targeting models (see Figure 1 and 2, p. 93 and p. 95, respectively) show an important fact. Of the five variables – inflation, output gap, nominal effective exchange rate, growth in bank credit, and money supply growth – on which response of call money rate shock is tested, only the response in growth in bank credit appears 69 The study is an extension of their work, Mishra and Mishra (2009), in which they suggest that the Indian economy satisfies the preconditions for inflation targeting. 92 to be significant. None of the responses in other variables appear significant. In short, interest rate does not appear to be effective in controlling inflation. This is consistent with the conclusion drawn by Jha (2008) which also finds interest rate being ineffective in creating a response to inflation in India. Monetary policy in developing countries grapples with the dilemma relating to inflation, interest rate, and exchange rate, as mentioned above. This is evident in the case of Indonesia and Thailand as noted in Siregar and Goo (2010). In the case of Thailand, the monetary authority shifted to a full IT rule from the IT framework – defined as a flexible approach to the IT policy, during the volatile period. As for Indonesia, managing exchange rate volatility continues to be one of the objectives of interest rate policy in addition to anchoring inflationary expectation. Siregar and Goo (2010) evaluate the overall performance of the IT regime adopted by Indonesia and Thailand after the Asian Currency Crisis 1997-98. The results demonstrate that the IT regime in these two economies has had some success, but not during the immediate aftermath of the Lehman Brothers’ collapse in 2008, in the run up to the global financial crisis and the Great Recession. Their test results seem to suggest that during the volatile period, covering the era of Asian Currency Crisis 1997-98 and its aftermath and the Global Financial Crisis 2007-08, the output gap did not significantly influence the interest rate policy of both Bank of Indonesia and Bank of Thailand. Siregar and Goo (2010) also note that there seems to be no trade-off between inflation and economic growth, covering monthly data for the two periods – two years before and after the implementation of IT. The average GDP growth rates as they note of Indonesia and Thailand after two years of implementing IT were significantly higher and less volatile (smaller standard deviations) than the rates in two years prior to IT. These results, however, do not seem to be consistent with the descriptive statistics using annual data presented in Table 4.6. To test the validity of the claims made by Siregar and Goo (2010), Table 4.6 presents some descriptive statistics (5 year averages) from two periods. The two periods are 1992–96 and 2002–06, pre and post IT-start in the two countries.70 A five year period average is 70 Both countries started implementing IT in the year 2000. 93 considered, assuming that it will reasonably capture the effects. The turbulent periods of Asian Currency Crisis (1997-98), Global Financial Crisis (2007-08) and the Great Recession (2008-09) are omitted to avoid any outlier effects. Table 4.6: Inflation and Growth Statistics in Indonesia and Thailand Pre and Post IT Implementation 1992 – 1996 2002 – 2006 (5 year average) (5 year average) Country Indonesia Thailand Statistics Inflation Growth Inflation Growth Mean 8.63 7.33 9.65 4.97 SD 0.92 0.44 3.11 0.47 CV 0.11 0.06 0.32 0.09 Mean 4.82 7.78 2.89 5.54 SD 1.09 1.23 1.72 0.97 CV 0.23 0.16 0.60 0.18 Source: Author’s calculations based on data from World Bank, World Development Indicators and IMF, International Financial Statistics. The superior growth performance after the implementation of IT in Indonesia and Thailand as claimed by Siregar and Goo (2010) does not seem to be valid when we analyse the growth averages using annual data.71 Growth rates in both countries are less in the post IT period. In Indonesia, it has gone down from 7.33 per cent to 4.97 per cent while in Thailand, the figure drops from 7.78 per cent to 5.54 per cent (Table 4.6). Coefficient of variation, which is a better measure of volatility in variables compared to standard deviation, shows similar results. Output growth volatility has gone up in both countries in the post IT era (see Table 4.6). Although the mean inflation has dropped in Thailand in the post IT era, the same did not happen in Indonesia. Perhaps a more strict IT rule followed by Thailand is responsible for the decline in average inflation.72 Interestingly enough, IT does not seem to cause inflation volatility to go down. In fact, the volatility, as measured by 71 Siregar and Goo (2010) use monthly data set. As Siregar and Goo (2010) note that Indonesia pursued a less strict IT rule, paying attention to managing its exchange rate volatility also. Developing countries do so to maintain its competitiveness in export market. The figures of inflation target in the two countries reveal that Thailand followed a strict IT compared to Indonesia. According to Table 2 (p. 115) of the paper inflation target level during the 2000s was set between 3 per cent and 10 per cent in Indonesia and between 0 and 3.5 per cent in Thailand. 72 94 the coefficient of variation, has risen in both countries after the implementation of IT. In short, the statistics presented in Table 4.6 show a considerable loss in output (evident from declining growth rate and higher output growth volatility) after pursuing a narrow IT based policy. Similar to Siregar and Goo (2010) a cross country study by Gerlach and Tillmann (2012) suggests that inflation persistence has declined in the economies using IT but not elsewhere. They explore how successful monetary policy incorporating inflation targeting is, in terms of bringing down persistent inflation shocks, in a sample of economies in Asia-Pacific. For the purpose of comparison, the sample includes countries practising IT, such as Indonesia, Philippines, and Thailand (among others), and countries without IT, including China and Malaysia (among others). The study is motivated by the observation that the introduction of monetary policy strategies focused on achieving low and stable inflation is associated in many countries with a sharp decline in the persistence of inflation shocks. The important conclusion is that inflation targeting outperforms alternative strategies, such as exchange rate pegs and ‘eclectic’ strategies, when performance of monetary policy is measured in terms of inflation persistence rather than the level of inflation. Gerlach and Tillmann (2012), therefore, support that IT can successfully be adopted by emerging market economies. A significant innovation of the study is that it measures inflation persistence by the sum of the coefficients in an autoregressive representation of inflation using the median unbiased estimator developed by Hansen (1999). Thus far, as they argue, the literature on inflation persistence in emerging market countries mostly relies on OLS estimates of the AR(1) coefficient. The least squares estimate suffers from a bias as the sum of the autoregressive coefficients approaches unity. Moreover, confidence bands based on a normally distributed estimator do not have correct coverage. Gerlach and Tillmann (2012) underscore the importance of reliable confidence bands to check whether persistence falls over time. The use of Hansen’s (1999) grid-bootstrap estimator solves these issues. In the sample Asian economies, after the Asian Financial Crisis of 1997 – 1998, the drop in the persistence is particularly large in those economies that formally adopted IT as a monetary policy strategy. It is important to note from the findings of Gerlach and Tillmann (2012) that the fall in the inflation persistence is not synchronised 95 across IT countries. Inflation persistence in Thailand falls immediately after the new monetary regime became effective. For the Philippines, however, a substantial reduction occurs only at the end of the sample period. Likewise persistence in inflation in Indonesia falls much later than in other countries. These facts raise a number of questions about the conclusion drawn by Gerlach and Tillmann (2012). First, even if a country follows IT the fall in persistence may take time. In that sense policies, such as higher interest rate and tighter credit control aimed at reducing the persistence, may involve a high sacrifice ratio in terms of loss in output. Secondly, one may ask how important this result is in a general macroeconomic perspective. Even though inflation has come down more rapidly in the IT countries (for instance Indonesia, Thailand, and The Philippines) their economic performance is less impressive compared to non-IT counterparts (for example China and Malaysia).73 So reducing inflation persistence cannot be an end in itself. It should be a means to reach the objective of growth and economic development. In sum, the above discussion provides insights on a number of issues. There is much evidence that countries trying to keep inflation at a very low level have to sacrifice in terms of growth. This is evident from higher growth volatility. Monetary policy also faces a dilemma in controlling inflation and maintaining export led growth. Tight monetary policy using higher interest rate trying to curb inflation may cause appreciation of domestic currency. This negatively affects competitiveness in the international market and hampers export led growth. 4.6 Determinants of Inflation: Demand Driven versus Supply Side The cause of inflation whether it arises from demand or supply side factors is important in determining monetary policy stance. In the event of a supply shock inflation which also causes drop in output, a tight monetary policy to fight inflation may have unwarranted consequences. It may cause output to fall further, leading to rise in unemployment and poverty. With reference to India’s attempt to address recent food price inflation, a former World Bank economist Surjit Bhalla says “Let’s 73 Malaysia is ranked highest amongst these Asian countries according to the Human Development Index (HDI) 2011. 96 kill GDP, inflation will fall.”74 Table 4.7 records some of the recent studies relating to this issue in the context of the sample countries of Asia. Table 4.7: Determinants of Inflation Study Empirical Dependent Approach Variable Computable Macro wa and general variables (supply side Bandara equilibrium such as factor) would (2012) model (CGE) employmen have an t, real overall export, real negative imports, impact on the real country’s household growth and consumptio poverty. n, real Complementa GDP, terms ry policies of trade, such as fuel and subsidy household scheme, level targeting low absolute income poverty groups are Naranpana Country Sri Lanka Period n.a. Regressors Findings Oil price High oil price suggested as necessary. 1951 – Generalised CPI Lagged The effects of Saqib 2007 Method of inflation inflation monetary (2011) (annual) Moments rate, broad determinants (GMM) money on inflation supply (M2) are ‘marginal’ growth rate, and political growth rate instability is of credit to positively private associated sector, with inflation. Khan and Pakistan 74 Source: The Financial Express, 15 October 2011; available via the internet at http://www.financialexpress.com/news/column-let-s-kill-gdp-inflation-will-fall/860148 97 growth rate of budget deficits, agricultural output, trade share, growth rate of real GDP per capita, growth rate of oil price, government crisis, and cabinet changes 2000 – Generalised Nominal Output gap, Loose and Rosario 2010 method of interest rate long run monetary (2010) (quarterl moments inflation policy is seen y) (GMM) target, other as a cause variables behind high such as inflation exchange during the rate, money GFC of 2007 growth, and – 2008. Ramayandi Indonesia international interest rate Jongwanich Developin 1996: 1 – Vector Oil and Shocks Against the and Park g Asia 2009: 1 autoregressiv food price correspondi backdrop of (quarterl e (VAR) inflation, ng to GFC 2007- y) model output gap, demand, 2008, Asia’s change in supply, inflation is exchange exchange largely home- rate rate, import, grown and has producer arisen due to and excess consumer aggregate price demand and inflation inflationary (2009) expectations rather than 98 external price shocks. Monetary policy is seen as a powerful tool to combat inflation. Islam Banglades 1987 – Descriptive (2008) h 2007 analyses n.a. n.a. High commodity (annual) price, supply side constraints in the commodity markets, and undervaluatio n of currency are the main causes of inflation. Noneconomic factors such as corruption and political uncertainties are also responsible for inflation. Lim (2008) Philippine 1950 – Descriptive s 2007 analysis (annual) n.a. n.a. Inflation is mostly due to supply side causes and an alternative monetary policy to inflation targeting is required, encompassing a more 99 development oriented programme. Siregar and Thailand 1985 – Autoregressi CPI Nominal Loose Rajaguru and 2001 ve distributed inflation effective monetary (2005) Indonesia (annual) lag (ARDL) exchange policy, model rate, GDP, particularly domestic supply of interest rate, narrow foreign money, interest rate, explains the money severe supply (M0) inflationary (and others) pressure in Indonesia during the Asian Financial Crisis of 1997-98 compared to other countries studied in the paper. Management of base money and exchange rate volatility is important in achieving overall price stability. 1984: 2 – Structural Change in Lagged Adoption of Ward 1999: 1 vector nominal values of the IT in (2002) (quarterl autoregressiv exchange dependent monetary y) e (SVAR) rate, money variables policy may be model growth, effective in change in future. Since short term demand Siregar and Indonesia 100 interest shocks are rate, real more output important growth, than supply inflation, shocks, change in government world short can play a term greater role nominal using fiscal interest policy. rate, and world inflation rate Notes: The studies are arranged first chronologically and then in alphabetical order. Lim (2008) supports the view that the inflation experience in the Philippines has been mostly a supply led and cost push phenomenon. He argues this just by using two types of supply side shocks – currency depreciation and oil price shock – one can explain practically all the above-10 per cent inflation in modern Philippine history. High inflation periods are not triggered by high domestic demand but by supply shocks. Lim (2008) also argues that if one uses monthly or quarterly data on inflation one will also see that agricultural price shocks make their impact on price inflation due to weather disturbances leading to food shortages. He documents that even with a lax monetary policy, lending to the private sector has not increased adequately amid fiscal difficulties; this contributes to lagging investment and employment creation. Monetary policy is not independent from the other macro sectors as well as the real and external sectors of the economy. Lim (2008) therefore proposes alternative monetary policies to IT that take into consideration the bigger and more complex role of monetary policy in an economy that requires a more development promoting programme. He argues that high liquidity and large monetary expansion in the mid-1990s did not have any impact on inflation (which remained below double digits and continued to decline until the Asian Currency Crisis 1997-98) because of the financial liberalisation. As a result of financial liberalisation the link between quantitative monetary targets and inflation had 101 weakened. This is because of the structural breaks in the income velocity of money and volatilities and instabilities in the money multiplier. A contrasting view of the cause of inflation is expressed by Jongwanich and Park (2009). They test whether inflation surge in developing Asia during 2007 – 2008 was primarily the result of external price factors such as oil and food shocks. Their empirical result suggests that inflation is largely home-grown and has arisen because of excess aggregate demand and inflationary expectations, rather than external price shocks. They, therefore, suggest that the monetary policy is a powerful tool in fighting inflation in Asia. This argument is based on the premise that monetary policy is potent in controlling inflation expectation regardless of the cause of it, demand pull or cost push factors. Jongwanich and Park (2009) divide the inflation drivers into cost push factors, mainly international oil and food prices, and demand pull factors, mainly excess aggregate demand and inflation expectations proxied by output gap and lagged domestic inflation respectively. Using variance decomposition analysis they show that excess aggregate demand and inflation expectations account for much of the consumer price inflation in the nine Asian countries studied.75 For instance over a year more than 20 per cent consumer price inflation variation in Indonesia and Malaysia resulted from demand pressure. Inflationary expectations explain around 60 per cent of consumer price variations in Malaysia. On average the two non-external factors jointly explain about 60 per cent of consumer price inflation in the region as a whole. The external cost push factors, oil and food prices, on the other hand explain only around 20 per cent of the variation in consumer price inflation. Jongwanich and Park (2009) note that the emergence of excess demand during this period was in part caused by lax monetary policy and sustained balance of payments surplus in many Asian countries. The resulting expansion of domestic liquidity helped fuel both the growth of aggregate demand and an increase in the output gap ratio. Just as importantly easy monetary policy, as evident from declining nominal 75 The phenomenon has a rationale which the authors have not mentioned. The rising aggregate demand makes sense since these economies are growing as they are trying to come out of the adversities of the crisis. 102 lending rates, has eroded the anti-inflationary credibility of monetary authorities and thus contributed to the formation of higher inflationary expectations. Empirical evidence on the relative importance of aggregate demand shocks over aggregate supply shocks in affecting macroeconomic fluctuations is also found in Siregar and Ward (2002). Based on an open economy model, they investigate the issue in the case of Indonesia. In an earlier study, Siregar and Ward (2001), based on a closed economy model, noted the opposite result. In Siregar and Ward (2001) the fluctuations are predominantly explained by shocks to aggregate supply with aggregate demand shocks playing a lesser role. The drawback of this study, as their later study notes, is the failure to capture the explicit transmission mechanisms for foreign shocks to affect the economy. The later study also suggests that monetary policy would not be totally effective as a measure for targeting output levels or economic growth. The results show that output effects of monetary policy shocks are small and statistically significant only in the short run. It may however, be useful in stabilising demand side variables, particularly inflation rate. The limited role of aggregate supply shocks in determining fluctuations in the variables may be a reflection of serious supply side bottlenecks in the economy. Besides, due to data limitations the study does not cover the two major oil price shocks (1973 and 1979) which had been favourable to the Indonesian economy to some extent. Although the findings of the above studies do not put much emphasis on the impact of food and oil prices, they may have considerable effects on inflation in food and oil importing developing countries. Naranpanawa and Bandara (2012) note that oil and net food importing developing countries were severely affected by food and oil price shocks in 2007 and 2008 creating negative impact on poverty, growth, and inflation. Using simulated results they find that high oil price in Sri Lanka has led to a decline in demand for imported and domestic commodities and a fall in investment, leading to a reduction of real GDP by 3.14 per cent. Higher oil prices in combination with associated higher import prices tend to increase cost of production leading to reduction of business profitability. In addition this will unavoidably lead to higher inflation which will in turn lower the purchasing power of consumers. Change of prices of basic needs commodities are given by the percentage change in the poverty line which shows an increase of 1.05 per cent, thus affecting low income households 103 adversely. Hence, reduction of business profitability, together with a sharp decline in aggregate demand results in a massive drop in employment by 5.8 per cent. Higher oil prices have fed into significant increase in domestic prices in oil importing developing countries. Therefore, many developing countries are vulnerable to the oil price shock. The results suggest that in the short run high oil prices would have an overall negative impact on Sri Lanka’s economic growth and also exacerbate poverty. In support of supply side causes of inflation, Islam (2008) notes that supply side constraints in the commodity markets in Bangladesh have instigated price hikes. He also notes that as a result of undervalued currency Bangladesh is suffering from currency induced inflation particularly in the presence of high international oil and food prices. Islam (2008) identifies the difficulty in allowing appreciation of the currency as the export sector might lose its competitiveness as a result of this. According to this study, there are both monetary and structural reasons for inflation in Bangladesh. Siregar and Rajaguru (2005) explain that in the case of Indonesia their results robustly confirm the significant roles of expected depreciation of currency, money supply, and domestic interest rate in explaining inflation rate in the country. Among these significant factors, the test results also suggest that the base money is the most significant and persistent contributor to the substantial increase in the overall price level in Indonesia during the post crisis period. The test results also indicate that the role of money supply is the most significant one for the Indonesian case, contributing as much as 37 per cent of the variation in the inflation rate. In Thailand the role of exchange rate variable clearly dominates contributing around 9 – 23 per cent of the variation in the inflation rate. Preceding findings have suggested that while the loose management of the base money in Indonesia has been the primary cause of high and persistent inflation, the volatility of the local currency contributed the most to the price changes in Thailand during the post 1997 crisis. For example from December 1997 to December 1998, Indonesia had expanded (year on year) its monthly base money by an average of about 75 per cent. For the same 12 month period, Thailand on the other hand had successfully tightened their monetary policy and reported a monthly average year on year contraction of base money by around 0.35 per cent. As 104 for Thailand weak and volatile local currency contributed significantly more to the price fluctuations than the base money. The case of Indonesia requires some more explanation from the events that took place around that time. Excess liquidity was created in order to provide substantial support to the troubled banks. The day after the first post crisis agreement was signed with the IMF, the government of Indonesia announced the liquidation of 16 banks. It created a shock wave that resulted in a total loss of confidence in the banking system. Within a month after the announcement of the closures of 16 banks, the level of base money has grown by more than 36 per cent largely due to Bank Indonesia Liquidity Supports to troubled banks and to lessen the impact of depositor runs on banks. So the rise in the base money was in effect an attempt to stop bank run which could have led to another crisis. One questionable issue relating to empirical investigation of the study is that it finds interest rate, both domestic and foreign, to be non-stationary. The rate variables having unit roots in a finite sample is a very unlikely case. In a finite sample such series should show mean reversion process. There are other studies which support the view that expansionary monetary policy through rise in aggregate demand causes inflationary situation in developing countries. Ramayandi and Rosario (2010) argue that the lack of discipline in monetary policy in Indonesia is responsible for the surge in inflation rate. The expansionary monetary policy, the authors argue, led to rise in aggregate demand with a growing and unchecked private credit, all proving to be unproductive in stabilising the economy. They suggest a need for monetary policy discipline to safeguard the country’s economic stability. The authors present a comparison between the actual policy rate and its benchmark path derived from an estimated policy reaction function. They notice two episodes – between 2005Q2 and 2006Q3 and between 2007Q4 and end of 2008 – when huge deviations are observed and the actual policy rate remains below the benchmark rate. Notably inflation peaked in both cases. Output gap behaved differently in the two episodes. Within episodes 1, inflation surged about 17 per cent while there was no significant pressure observed in the output gap. In contrast, episode 2 shows that a peak in inflation was accompanied by a systematic pressure in the output gap. The authors suggest that the actual policy rate should have been increased in response to a rise in inflation expectation and or 105 output gap. In the case of episode 1, this argument does not seem to be valid. The inflation rate came down in the subsequent quarters even though there was a difference between the actual and benchmark rates. So the difference cannot be held responsible for creating a problem of inflation. It is also important to note that during this period the output gap was below zero, perhaps the reason why the authorities were reluctant to raise interest rate. In the case of the second episode, inflation seems to be a lag variable – output gap has an increasing trend followed by a similar trend in inflation. Considering that inflation is rising and output gap is also positive there is perhaps some justification for raising the interest rate (as reflected by a greater increase in the benchmark rate). However, one notices that both variables, inflation and output gap, dipped in the last quarter of 2008. Whether a rise in the actual policy rate (in line with the benchmark rate) – a single instrument – would have managed to curb the rise in inflation rate in the midst of the global financial crisis remains a question. The authors’ suggestion seems to follow the “one target one instrument” and strict rule based monetary policy which has received criticism in the face of the crisis. An interesting departure from these conventional studies to find the causes of inflation is the study by Khan and Saqib (2012). It investigates the effects of political instability on inflation in Pakistan. Applying the GMM technique and using data from 1951 to 2007, they have examined this link through two different models. The results of the monetary model suggest that the effects of monetary determinants are rather marginal and inflation depends on the political environment of Pakistan. They have discarded monetary growth as a potential cause of high inflation in Pakistan (high inflation is defined as inflation rate more than 7 per cent – average annual inflation rate in the period 1951 – 2007) on the grounds that effect of money growth on inflation is dependent on political environment. The nonmonetary model’s findings explicitly establish a positive association between political instability and inflation. This is further confirmed through an analysis based on interactive dummies that reveal political instability significantly leading to high (above average) inflation. Khan and Saqib (2012) argue that political instability undermines the competence of a government and diminishes its ability to deal with shocks that eventually result in macroeconomic disequilibrium such as inflation. The authors note that the 106 conventional view that political instability leads to high inflation due to government’s excessive reliance on seignorage may not hold in a low to moderately high inflation country such as Pakistan. Government crisis is found to be more important than oil price in explaining inflation. Government crisis has a stronger effect on inflation in situation of high inflation. When inflation is above average an additional government crisis increases inflation by 0.715 percentage points from its average. The study suggests that unless political reforms aimed at mitigating government crises and cabinet changes are not undertaken, inflation stabilization efforts by the technocrats would fail to yield long term price stability. 4.7 Conclusion This chapter provides a critical review of country-specific studies on growthinflation relationship in some selected developing countries of Asia. Stylised facts from growth-inflation data and the review of literature reveal some interesting insights about the relationship. The discussion shows that inflation and growth may move together in the developing countries of Asia. There are a number of reasons behind this. They include structural causes such as inevitable fiscal deficits due to the need to finance the effort to close large physical and social infrastructure deficits with a low tax base, subsistence agricultural practices, and various other supply rigidities or bottlenecks. There are instances when some of these countries grew amidst inflation. The findings from the review of literature and the historical experience of inflation in the Asia-Pacific region (as evident from mean inflation in Table 4.1) also show that for majority of the countries mean inflation threshold lies above 5 per cent. In addition we find strong evidence against a one size fits all policy. Thus the advice by the IMF to keep inflation at a very low level for all countries is not supported by different threshold levels found for different countries. It is also interesting to note findings of different threshold levels for the same country. This could be due to modelling issues such as univariate or multivariate framework and selection of explanatory variables; and methods of empirical estimation taking into account reverse causality and effects of outliers or sample size and periods covered. 107 The discussion from the literature also provides evidence that the countries trying to keep inflation at a very low level sacrifice in terms of growth. This is evident from higher growth volatility. Monetary policy in the developing countries also faces a challenge in controlling inflation and maintaining export-led growth. Contractionary monetary policy using higher interest rate to fight inflation is responsible for this as higher interest rate causes currency appreciation and thus negatively affects competitiveness in the international market. Tight monetary policy could also adversely affect growth when inflation is caused by supply side factors. In the event of negative supply shock fighting inflation using higher interest causes a further decline in output and growth. 108 Chapter 5 A Panel Study on Growth-Inflation Relationship in Selected Developing Countries of Asia 5.1 Introduction This chapter attempts to employ cross-country panel estimation techniques to investigate the growth-inflation relationship in selected developing countries of Asia. The objective of the chapter is to provide empirical support to the critical review of literature presented in Chapters 2, 3, and 4. A key issue we examine here is whether there is significant evidence of an inflation-threshold close to 5 per cent, beyond which inflation becomes harmful for economic growth. We also investigate if the threshold varies under different structural and economic circumstances. Our intention is to establish the argument empirically that keeping inflation at 5 per cent in the developing countries is not justified. The previous chapters have established the case strongly by analysing the extant literature and descriptive statistics from developing countries of Asia. Although important, empirical evidence on this issue with particular emphasis on developing countries of Asia is relatively scant. This chapter aims at fulfilling this gap. It provides empirical evidence using techniques, such as dynamic panel estimation (System Generalised Method of Moments), not employed before in analysing the case of developing countries of Asia. The use of modern panel data techniques allows us to take full advantage of time series variation as well as variations across countries. The findings from the analysis in the context of 14 developing countries of Asia for the period 1961 and 2010 help us conclude that policy prescription for keeping inflation at a low level such as 5 per cent has a very weak empirical foundation. The results using different panel estimation techniques suggest that inflation within the range of 8-15 per cent may not be harmful for economic growth. We have also noted that the threshold levels vary according to the levels of development. There is strong evidence that for poorer groups of countries threshold tends to be higher. These findings are in sharp contrast to what the IMF and the conventional wisdom of 109 mainstream economics preach – inflation below 5 per cent is beneficial for growth in the developing countries. This section of the chapter proceeds with a brief discussion on overall picture of the growth-inflation relationship evident from a selected developing countries of Asia. The rest of the chapter has three main parts. Section 5.2 explains the research methods relating to selection of variables and empirical models. The empirical results are presented in Section 5.3 and a conclusion is provided in Section 5.4. To begin with, Figure 5.1 presents pooled annual observations from a selection of 25 developing countries of Asia for the period 1961-2010.76 There does not seem to be a stable long run relationship between inflation and growth according to the figure. The real GDP growth rate shows a gradual decline as the range of inflation increases from 0-5 to 5-10 and 10-20 per cent. This happens for a relatively large number of observations; 248, 297, and 173 respectively (see Figure 5.1). However, the highest growth rate is registered in the inflation range between 20 and 40 (for a relatively small number of observations of 42). Beyond the inflation rate of 40 per cent, growth becomes negative which is consistent with what is noted by Bruno and Easterly (1998). Perhaps the information in Figure 5.1 reveals two important facts. First, evidence of high growth within the inflation range of 20-40 per cent tells us that there are instances when developing countries within this region grew amidst relatively high inflation. A number of countries such as Lao PDR, Sri Lanka, Mongolia, Kyrgyz Republic, Tajikistan, and Uzbekistan could be cited as examples. In these countries, at different points in time, growth was above 5 per cent amidst high inflation. On the other hand, growth experience in the initial ranges of inflation (0-5, 5-10, and 10-20 per cent) provides a rationale for the suggestion that inflation should be kept within 0-5 per cent. These preliminary observations, however, do not provide a case for a robust relationship between inflation and growth and that indeed an inflation-threshold exists at a very low level. Such examinations are carried out in the following sections. 76 The countries, in Figure 5.1, are selected based on their similar historical background, so that the sample is representative of the population, and at the same time to help capture variations in the developing countries of Asia. It is done keeping in mind that panel estimation techniques can take care of such country heterogeneity properties to some extent, if not fully. However, the number of countries reduces in the panel estimation as revealed in subsequent sections. This is primarily because of lack of data on regressors for the countries. 110 Figure 5.1: Inflation and Growth in Selected Asian Countries, 1961-2010 7 Real GDP growth rate 6 6.2 5.5 5.3 4.7 5 4 3 2 1 0 -1 -2 0-5 (248) >5-10 (297) >10-20 (173) >20-40 (42) >40 (40) -1.2 CPI inflation range (%) with number of observations within brackets Sources: Author’s calculations using data from the World Bank, World Development Indicators (WDI); International Monetary Fund, World Economic Outlook (WEO). Notes: Pooled annual observations based on 25 countries of Asia. The countries are Bangladesh, Bhutan, Cambodia, China, India, Indonesia, Lao PDR, Malaysia, Maldives, Nepal, Pakistan, The Philippines, Sri Lanka, Thailand, Vietnam, Fiji, Kazakhstan, Kyrgyz Republic, Mongolia, Papua New Guinea, Samoa, Solomon Island, Tajikistan, Turkmenistan, and Uzbekistan. 5.2 Estimation Methods The panel dataset initially included information from 25 developing countries in Asia for the period 1961-2010. These countries were selected from regions such as South Asia, East Asia, Asian countries within the Commonwealth of Independent States (countries in former Soviet Union), and small island nations. The list was comprised of Bangladesh, Bhutan, Cambodia, China, India, Indonesia, Lao PDR, Malaysia, Maldives, Nepal, Pakistan, The Philippines, Sri Lanka, Thailand, Vietnam, Fiji, Kazakhstan, Kyrgyz Republic, Mongolia, Papua New Guinea, Samoa, Solomon Islands, Tajikistan, Turkmenistan, and Uzbekistan. However, because of poor or no availability of data, particularly on control variables such as government consumption (% of GDP) and real household consumption per capita, we restricted regression analyses mostly to 14 countries of the list. So, the countries for panel estimations are Bangladesh, Cambodia, India, Indonesia, Lao PDR, Malaysia, Pakistan, The Philippines, Thailand, Vietnam, Kazakhstan, Kyrgyz Republic, Papua 111 New Guinea, and Tajikistan. We have excluded China despite data availability as the country is an outlier because of its size and economic structure. In short, the dataset is an unbalanced panel covering 14 countries spanning over up to 50 years (19602010).77 Under such circumstances, the potential dimension of the panel is 14×50=700 observations. But in practice because of missing observations, the dimension is smaller. The empirical approach, in essence, involves regressing two measures of growth – real GDP growth and real GDP per capita growth – on inflation, conditioned upon other variables suggested by the related literature.78 The dataset is gathered from the standard and most widely used sources – the World Bank’s World Development Indicators (WDI); the IMF’s International Financial Statistics (IFS) and World Economic Outlook (WEO). The panel estimations use annual data in order to maximise sample size and to measure the parameters of interest more precisely. This empirical approach differs from most frequent practice in the literature – smoothing data using five or ten year averages. There are two main arguments in favour of using annual observations. First, the relationship between inflation and growth is more evident in the case of higher frequency of the data as noted by Bruno and Easterly(1998). They state, “[the] results get stronger as one goes from the crosssection to ten year averages to five year averages to annual data” (p. 4). Khan and Senhadji (2001, p. 16) also note that “high-inflation effect [on growth]…is more powerful for yearly data”. The second reason is based on the argument set forth in Baltagi et al. (2009). They state that averaging out annual data over, for instance, five year periods results in reduction in sample thereby raising the possibility of making most of the variables statistically insignificant. The authors explain that “smoothing out of time series data removes useful variation from the data, which could help to identify the parameters of interest with more precision” (p. 286). Apart from these two reasons we also believe that assuming a uniform business cycle pattern for all countries in the panel, which the data-smoothing technique does, is not appropriate (see discussion in Chapter 3). 77 The growth rates are available from 1961 once we take the first difference of the natural logarithm of level variables. The dataset is still unbalanced because information on variables for the 14 countries is not available for all years. 78 The selection of variables is explained in the section that follows. 112 5.2.1 Selection of variables and summary statistics A large number of possible specifications relating to the selection of variables (regressors) exist in the cross-country growth literature (Lopez-Villavicencio & Mignon 2011, p. 458). Our choice of control variables is similar to the choice in the previous studies although it is not identical. We are interested to find out how inflation affects growth conditioned upon other covariates. We have chosen the covariates because they may influence the condition and the way in which inflation affects growth. In order to assess the nature of growth-inflation relationship, economic growth is selected as the dependent variable. Both real GDP growth and real GDP per capita growth are used as a measure of economic growth. Figure 5.2 shows the distribution of the two variables in the sample of 14 countries. The distributions of the two measures of economic growth look similar. Economists are particularly interested in looking at output per capita measure of growth as it provides a better indicator of a country’s development. Total real output per capita also allows us to take the relative economic size of the country into account (Busse & Hefeker 2007, p. 403). However in this study because we are interested to see whether the country simply produces more output or not, independent of its population growth, we choose the real GDP growth as well. Besides, growth in per capita GDP in developing countries is affected by how successful the country is in controlling its population growth, an issue which is not a prime focus of this study. So, in short, while performing the estimation we give priority to the real GDP growth as a dependent variable. Since the distribution of the two measures of growth looks similar, we do not expect much difference in the results. 113 Figure 5.2: Distribution of Real GDP Growth and Real GDP Per Capita Growth in 15 10 0 5 Density 0 5 Density 10 15 Selected Developing Countries of Asia, 1961-2010 -.4 -.2 0 Real GDP Growth .2 -.4 -.2 0 Real GDP Per Capita Growth .2 Sources: Graphs generated by author using data from World Bank, WDI. Note: Countries (N= 14) are Bangladesh, Cambodia, India, Indonesia, Lao PDR, Malaysia, Pakistan, The Philippines, Thailand, Vietnam, Kazakhstan, Kyrgyz Republic, Papua New Guinea, and Tajikistan. Next, inflation is included as the key explanatory variable of interest – to examine the impact of inflation on growth. Inflation is defined as the growth rate of consumer price index (CPI). The measure of inflation based on CPI is preferred over GDP deflator as the latter is a derived product of GDP and therefore would have an obvious association with output growth.79 In relation to inflation the first concern is the presence of outliers, very high observations of inflation for few countries, which might have strong influence on the results. In order to avoid problems due to outliers, 79 As Sarel (1996, p. 201) notes that “[i]t is better to use CPI data than implicit GDP deflators…because changes in GDP deflators are, by construction, negatively correlated with the growth rates.” 114 three different cases are considered.80 In the first instance, the models include cases where inflation rate is less than 40 per cent. Pollin and Zhu (2006) in all their estimated models exclude observations in which inflation exceeded 40 per cent. Similar practice is carried out in one of the models used in de Mendonca and Guimaraes e Souza (2011).81 The elimination of cases of high inflation (annual inflation rate greater than 40 per cent) is done in line with the observation made by Bruno and Easterly (1998) that inflation is indeed harmful for growth beyond the rate 40 per cent. Excluding the cases of inflation greater than 40 per cent, however, reduces the number of observations. In order to maintain the number of observations in the models two other cases are considered. The second case considers inflation as ln(1+π), where π is inflation. We follow this approach in line with Kumar and Woo (2010).82 Transformation of a variable to natural logarithm is expected to produce a distribution that is closer to normal. This log transformation technique to avoid the effect of outlier is common in the empirical literature. For instance Sarel (1996) suggests that the log transformation helps to reduce, at least partially, the strong asymmetry in the distribution of inflation. Finally, the third case considers a normalisation process by transforming inflation rate as π/(1+ π) following de Mendonca and Guimaraes e Souza (2011).83 Figure 5.3 shows the distribution of the different forms of inflation. In the first panel (top left) of Figure 5.3 the distribution of inflation across the full sample of 14 countries and time period (1961-2010) is shown. It is evident that the distribution is highly skewed. The second panel (top right) shows the distribution of inflation when inflation rates greater than 40 per cent are removed from the sample. The bottom two figures correspond to the distributions of the converted inflation. In all three cases, the distributions are relatively less skewed than the original one. What we also notice is that the two converted inflation distributions look more centralised than the distribution of inflation rate less than 40 80 Determination of the turning point at which a rise in inflation would cause growth to fall would vary according to the cases. We mention how the turning point is measured in the section where the results are presented. 81 See their fourth model and the explanation that follows (p. 4). 82 We take ln(1+π) instead of ln(π) because of the negative values of inflation rate. 83 The authors carry out this transformation in their fifth model following the suggestion of Cukierman et al. (1992). 115 per cent. We are interested to see whether such differences in the distributions of inflation have any implications in estimated results.84 Figure 5.3: Distribution of Inflation in Selected Developing Countries of Asia, 1961- 0 0 .2 2 4 6 Density .4 .6 Density .8 8 1 10 2010 5 10 CPI Inflation 15 20 -.1 0 .1 .2 .3 .4 CPI Inflation when the rate is less than 40 per cent 4 Density 0 0 2 1 2 Density 6 3 8 4 0 0 1 2 ln(1+CPI Inflation) 3 0 .2 .4 .6 .8 (CPI Inflation) / (1+CPI Inflation) 1 Sources: Graphs generated by author using data from World Bank, WDI and IMF, IFS and WEO. Note: Countries (N= 14) are Bangladesh, Cambodia, India, Indonesia, Lao PDR, Malaysia, Pakistan, The Philippines, Thailand, Vietnam, Kazakhstan, Kyrgyz Republic, Papua New Guinea, and Tajikistan. Figure 5.4 attempts to capture the relationship between real GDP growth and the logarithm of “1 plus inflation”.85 The graph is drawn following Khan and Senhadji (2001, Figure 1, p. 4) who smoothed out data by reducing the full sample to five observations. The arithmetic mean of real GDP growth is taken for five equal subsamples corresponding to increasing levels of inflation (ibid., p. 3).86 It is 84 For instance taking a different digit arbitrarily in the place of “1” in ln (1+π) would change the distribution. If this has implications in the estimated results it might raise questions about using such conversions. 85 Graph is similar between real GDP growth and π/(1+ π). 86 There are 449 observations in our sample of 14 countries over the period 1961-2010. The first four subsamples include 90 observations of inflation and growth each and the last one includes 89. 116 interesting to note that the observations from Figure 5.4 differ from what was noticed by Khan and Senhadji (2001). In both cases the relationship between inflation and growth is positive for low levels of inflation. But unlike Khan and Senhadji (2001) we do not see growth to decline drastically when inflation moves to a moderately high level. This also does not corroborate the findings of Ghosh and Phillips (1998) who warn of a danger of a steep drop in growth beyond the threshold level of inflation. The findings of Khan and Senhadji (2001) also show that the negative effect of inflation on growth weakens at very high inflation rates, supporting Fischer’s (1993) findings. Our findings, however, are different and more in line with what Bruno and Easterly (1998) suggested. As we see from Figure 5.4, that growth declines gradually when inflation starts to move towards a higher level. It is at a very high level of inflation that growth falls sharply. The differences between the findings are perhaps obvious. Khan and Senhadji (2001) in their analysis, displayed in Figure 1 (p. 4), did not differentiate between industrial and developing countries. Our analysis consists of a much smaller group of developing countries. This perhaps reiterates the importance of making the distinction between advanced and developing countries while performing the analysis as pointed out by Sepehri and Moshiri (2004). Figure 5.4: Relationship between Real GDP Growth and Inflation Mean of real GDP growth 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0 1 2 3 4 5 Five subsamples corresponding to increasing levels of ln(1+Inflation) Sources: Author’s calculations using data from World Bank, WDI and IMF, IFS and WEO. 117 Note: Countries (N= 14) are Bangladesh, Cambodia, India, Indonesia, Lao PDR, Malaysia, Pakistan, The Philippines, Thailand, Vietnam, Kazakhstan, Kyrgyz Republic, Papua New Guinea, and Tajikistan. Apart from inflation, a number of control variables which might influence the growth-inflation relationship are considered as regressors. The mnemonic used for some of these variables are shown in parentheses. The variables are as follows: (1) Lagged growth: Growth lagged in one period is taken on the right hand side to reflect the dynamic nature of the relationship and to capture the possibility of partial adjustment towards the steady-state. The coefficient of this variable is expected to be positive and lie between zero and one. As Greene (2003, p. 307) explains that incorporating dynamics in this way allows us to consider the “entire history of the right hand side variables, so that any measured influence is conditioned on this history.” This in turn has implications for the choice of estimator. The preferred estimator in these circumstances is dynamic Generalised Method of Moments (GMM). In standard static panel models such as Fixed Effects (FE) and Random Effects (RE), the lagged dependent variable as a regressor is not used to avoid possible bias in the estimator. We expect that previous year’s growth would have a positive impact on current year growth. Inclusion of the lagged term also helps reduce the problem of autocorrelation. (2) Real household consumption per capita (HCONPC): To measure the effect of aggregate demand on growth-inflation relationship, real household consumption per capita measured in constant US dollar is included as a control variable. The inclusion of this variable is justified since consumption is assumed to be the largest component of aggregate demand. Besides, higher household consumption per capita is assumed to be growth enhancing and an indication of poverty reduction. We prefer to use this variable to investment since investment data for poorer countries are problematic. The expected sign of its coefficient is positive. 118 (3) Financial Deepening (FD): Broad money-GDP ratio is taken as a control variable to measure the effect of financial deepening on the growth-inflation relationship. Various measures are used as a proxy for FD in the literature. We use M2/GDP which is the liquidity liabilities of the financial system. It is also known as monetisation variable which, in a growing economy, helps us capture the real size of the financial sector. M2 equals currency plus demand and interest bearing liabilities of banks and non-bank financial intermediaries. Therefore, as Guiliano and Ruiz-Arranz (2009) note, this measure is the broadest measure of financial intermediation that includes three types of financial institutions such as the central bank, deposit money bank, and other financial institutions. We are, however, aware of the limitation of the use of this proxy. For instance, Khan and Senhadji (2003) argue that M2/GDP is a poor proxy for financial development because they merely show the capacity of the financial system to provide transaction services rather than the capacity to transfer funds from savers to borrowers. Next, our task is to determine the expected sign of FD. Earlier studies such as McKinnon (1973), Shaw (1973), and Kapur (1976) argue that financial development has a positive impact on growth. So if the ratio of broad money to GDP is growth enhancing, money supply is expected to have a greater impact on growth and less impact on inflation. This is because a developed financial system absorbs the money supply and diverts it to the real sector, thus generating growth. Thus we expect a positive sign of this coefficient. The expected positive sign of the coefficient would also mean that expansionary monetary policy may not be harmful for the real sector of the economy. On the other hand, economists such as Robinson (1952) and Lucas (1988) doubt the role played by financial development in promoting economic growth. For example, Bangake and Eggoh (2011, p. 178) point out that many of the Asian economies grew fast in the 1970s and 1980s without a developed financial system.87 If the sign of the coefficient of this variable is negative it may provide support for the monetarist view – an increase in the ratio of money supply to GDP implies pressure on inflation and 87 They also note that many OECD countries engaged in financial reforms in the 1980s, yet savings, investment, and growth in them have not accelerated. 119 therefore negative impact on growth. It holds to be true provided that too much money is chasing too few goods leading to an inflationary situation. The monetary expansion may also lead to bubbles (which might result into an inflationary situation) in the financial system if monetary transmission mechanism is not effective enough to channel the funds from financial system to the real sector. In fact, developing countries may have underdeveloped financial system which may have a high ratio of money to GDP, as money is used as a store of value in the absence of other attractive alternatives (see Khan & Senhadji 2003). Baltagi et al. (2009) in this connection stress the importance of economic institutions such as property rights. If property rights are weak, financial development may not be sufficient to promote growth. Weak property rights may discourage investment even when bank loans are available. In such circumstances, we might expect a negative sign of the FD coefficient. (4) Government consumption expenditure (GOVCON): Government consumption as a per cent of GDP is taken to determine the effect of fiscal policy on growth-inflation relationship. In terms of the availability of data from standard sources, this appears to be the best proxy to measure the impact of the size of government. The sign of coefficient could be either positive or negative. A negative sign might indicate that higher government spending is inflationary (Fiscal Theory of Price Level) that is bad for growth.88 In addition, an argument in favour of the negative sign is that as the size of the government becomes larger it crowds out the private sector, thus adversely affecting growth. Arestis and Sawyer (2003), however, argue that the argument of crowding out is not valid when fiscal deficits correct a deficiency in private aggregate demand and fiscal policy is viewed in terms of functional finance. Hemming et al. (2002) identify a number of reasons why the effect of fiscal policy tends to be positive and quite large. These conditions include a demand constrained economy with excess capacity; a closed economy or an open economy with a fixed exchange rate; and households 88 The Fiscal Theory of Price posits that increased government consumption and debt adds to household wealth and hence, to demand for goods and services, leading to price pressures (Kwon, Mcfarlane & Robinson 2009). 120 with limited horizons or liquidity constraints. So, we can also expect a positive sign when government spending is beneficial in raising the productive capacity of the economy and thus growth inducing. (5) Trade openness (OPEN): Summation of export and import as a per cent of GDP is taken as a control variable to measure the trade openness. This measure of openness is considered most suitable among others. For instance, other forms of measure of openness could include capital account tracing the flow of financial products. This may not be appropriate to consider in our model as financial markets in these economies are integrated with each other to varying degrees. A more open economy is subject to shocks which could be either positive or negative. Therefore, the sign of this coefficient in the growth regression remains uncertain. The relationship between trade openness and growth is a highly debated topic but still an unresolved issue in the growth and development literature (Yanikkaya 2003). A wide range of empirical studies support that open economies grow faster than closed economies. Openness to trade favours growth because it provides access to imported inputs, which embody new technology; increases the effective size of the market facing producers, which raises the returns to innovation; and affects a country’s specialisation in research-intensive production (Harrison 1996). So, we can expect a positive sign of OPEN in our estimated model. However, Yanikkaya (2003) citing other literature89 states that if trading partners are asymmetric countries in the sense that they have considerably different technologies and endowments, even if economic integration raises the worldwide growth rate, it may adversely affect individual countries. Earlier, Rodrik (1999a) also expressed his reservations about the overwhelming support in favour of trade liberalisation and its positive impact on growth. He states that “[j]ust as the advantages of import-substitution policies were overstated in an earlier era, today the benefits of openness are oversold routinely in the policy-relevant literature and in the publications of the World Bank and the IMF” (Rodrik 1999a, p. 25). We can, therefore, expect a negative sign of OPEN as well. 89 For instance, see studies such as Grossman and Helpman (1991a, 1991b), Lucas (1988), RiveraBatiz and Xie (1993), and Young (1991). 121 (6) Agriculture’s share of GDP (AGR): Since agriculture is still the mainstay of the economy in most of the developing countries, the share of agriculture output as a per cent of GDP is considered as a control variable. In an influential study Johnston and Mellor (1961) state that agriculture productivity is vital for economic development, especially for developing countries. According to the study, agriculture contributes to economic growth by providing enough food supply, increasing agriculture exports, transferring adequate labour supply to the manufacturing sector, contributing capital for infrastructure and industrial expansion and stimulating industrial expansion by raising the net cash incomes of the farm population. However, Dowling and Valenzuela (2004) note that the experience of a broad range of countries indicates that the relative importance of the agricultural sector to the economy diminishes in favour of manufacturing and service sectors with growth over time. The accepted explanation for this relationship is the change in the composition of demand, of which the decline in the share of food (Engel’s Law) is the most notable feature (Chenery 1960, p. 624). Therefore, from the discussion, we expect that AGR would help us capture the structural changes in the economy and any impact on growth-inflation nexus as a result of the change. A greater level of development would mean that economic transformation is taking place and economies are less dependent on the primary sector. A negative sign of the coefficient, therefore, is expected which would indicate that such transformation is taking place and is beneficial for economic growth. However, we do not rule out the possibility of a positive sign of the coefficient of AGR either. As Chenery (1960) observes that this over-all relationship may not necessarily apply to every individual country. He argues that within limits, the changing composition of domestic demand for food can be offset through foreign trade. A country having a continuing comparative advantage in primary production, he points out, may, therefore, reach a high level of income without an increase in the share of industry in total output. (7) Oil and commodity price shocks: Developing economies are prone to supply shocks caused by oil and commodity prices in the international markets. Oil price directly affects the supply side of the economy and therefore growth. 122 Commodity prices in the international market may also affect domestic growth directly, especially when such commodities are important inputs in production. Additionally, fluctuations of commodity prices may act as an incentive or disincentive for domestic production and therefore affect growth. To assess such impacts two dummy variables are created. The oil dummy, DUMOIL, takes the value 1 for the years 1971, 1974, 1979, 1999, 2000, 2004, 2005, and 2008 and the value 0 otherwise. In these years, growth rate of oil price index is 20 per cent above its historical mean that is 7.44 per cent. The dummy captures the three major international oil price shocks. The commodity price dummy, DUMCOMPR, is constructed in a similar fashion. It takes the value 1 for years 1973, 1974, 2004, and 2006 when the growth rate of commodity price index is above 20 per cent of its historical mean at 3.85 per cent. The dummy takes the value zero for other years. The commodity dummy helps capture the shocks of the 2000s. Both dummies take the value 1 only when growth rates show positive 20 per cent deviation from trend. This is because our interest is to see the influence of negative supply shocks as a result of rising oil and commodity prices. Figure 5.5 shows the deviations of oil and commodity price growth rates from their trends – historical mean at 7.44 and 3.85 per cent respectively. Because of the positive correlation between oil and commodity prices, discussed later, we believe constructing such dummies to gauge their impacts is more appropriate. The expected sign of the coefficients of these dummies is negative – rising oil and commodity prices in the international markets acts as a negative shock, adversely affecting growth. However, we should also keep in mind that some of the economies may benefit from oil price rises. The economy of Indonesia, for instance as discussed in the previous chapter, seemed to have benefited from rising oil prices. In addition, the effect may not be captured in a straight forward way since governments in developing countries often subsidise oil dependent sectors to mitigate the negative shocks resulting from rising oil price. On the other hand, increase in commodity prices in the international market may positively affect economies which are net exporters of those commodities. 123 Figure 5.5: Deviation from Trend of Growth Rates in Oil and Commodity Price Indexes, 1961-2010 120 100 80 60 20 0 -20 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Percent 40 -40 -60 -80 -100 Deviation of oil price growth from historical trend Deviation of commodity price growth from historical trend Source: Authors calculations using data from IMF, IFS. Notes: Oil price index and commodity price index refer to UK Brent Crude Oil Index (2005=100) and World Bank non-energy commodity price index for lower middle income countries (2005=100) respectively. Summary statistics of the growth rates of the above variables for the period 19612010 are provided in Table 5.1. In general, long run average (unweighted) economic growth in the sample countries is below 5 per cent. The average inflation is around 17 per cent and it is significantly more volatile (as evident from the standard deviation) than economic growth over this long period. Real household consumption per capita has registered less than 2 per cent average growth over this period. Mean growth rates in financial deepening and government consumption, both as a per cent of GDP, are less than 5 and 1 per cent, respectively. The latter shows that the size of the government proportional to GDP has not grown much over the long period. Average growth rate in trade openness is around 2 per cent, reflecting the fact that the countries in the sample are becoming more open economies. It is also important to note that average growth of share of agriculture to GDP is around negative 2 per cent, revealing a declining trend in the primary sector of the economies. This is perhaps an indication that structural transition is taking place in these traditional 124 agro-based economies. In the international markets, oil price on average has increased more and remained more volatile compared to commodity prices (Table 5.1). Table 5.1: Summary Statistics of Growth Rates of Selected Variables, 1961-2010 Variable Economic growth CPI inflation Real household consumption per capita Broad money-GDP ratio Government consumptionGDP ratio Trade openness AgricultureGDP ratio Oil price index Commodity price index Observations Mean Standard deviation Minimum Maximum 538 4.63 5.35 -33.64 19.57 465 17.64 103.42 -7.63 1877.37 470 1.81 7.88 31.12 -53.96 435 4.23 10.35 -87.44 64.17 513 0.55 13.47 -65.10 152.59 529 2.05 16.90 -107.95 114.99 507 -2.14 7.39 -42.23 47.22 50 7.43 27.50 -64.86 111.20 50 3.84 13.57 -23.22 42.46 Sources: Author’s calculations based on data from World Bank, WDI; IMF, IFS and WEO. Notes: Countries (N= 14) are Bangladesh, Cambodia, India, Indonesia, Lao PDR, Malaysia, Pakistan, The Philippines, Thailand, Vietnam, Kajakhstan, Kyrgyz Republic, Papua New Guinea, and Tajikistan. The maximum number of observation for each variable is 714. The correlation matrix of the growth variables, presented in Table 5.2 also provides some interesting facts about the possible long run relationship amongst the variables. These associations, however, do not provide an understanding about the direction of causality. There exists a moderate negative association between inflation and GDP growth and it is significant at 5 per cent level. Not surprisingly, real household consumption per capita and agriculture share of GDP have significant positive and negative association with GDP growth, respectively. Higher GDP growth and higher consumption per capita (supposedly a reflection of poverty reduction) should go hand in hand while moving away from traditional sector should correlate with higher 125 growth. The other growth rates of other variables that have significant correlation with GDP growth are oil price and commodity price growth rates, both having a moderate positive relationship. It is interesting that both oil and commodity price growth rates have a positive association with GDP growth, contrary to what is perhaps expected. As explained above, it is possible since some of the countries within the sample have oil, such as Indonesia and may benefit from rising oil prices. Net commodity exporters, for instance Thailand and Vietnam which export rice, in a similar manner may benefit from commodity price rises in international markets (see discussion in Chapter 4). Table 5.2: Correlation Matrix of Growth Variables Growth Growth CPI inflation Real household consumption per capita Broad money-GDP Government consumptionGDP Trade openness AgricultureGDP Oil price Commodity price CPI inflation Real household consumption per capita Broad moneyGDP Government consumptionGDP Trade openness AgricultureGDP Oil price Commodity price 1 -0.277 1 0.493 -0.237 1 -0.02 -0.122 0.093 1 0.016 0.048 -0.086 0.266 1 -0.011 0.046 0.028 -0.032 -0.147 1 -0.139 -0.019 -0.069 0.004 -0.05 -0.102 1 0.122 0.115 0.131 -0.119 -0.079 0.3 -0.126 1 0.232 0.071 0.187 -0.127 -0.134 0.259 0.066 0.449 Sources: Author’s calculations based on data from World Bank, WDI; IMF, IFS and WEO. Notes: Bold faced figures refer to correlation significant at 5 per cent level of significance. Countries (N=14) are Bangladesh, Cambodia, India, Indonesia, Lao PDR, Malaysia, Pakistan, The Philippines, Thailand, Vietnam, Kazakhstan, Kyrgyz Republic, Papua New Guinea, and Tajikistan. Growth in Financial depth and government consumption expenditure (per cent of GDP) has a negative and positive relationship with GDP growth in the long run respectively. The rationale for such an association can be found in the economic theory. These relationships, however, are not significant. Similarly, growth in trade openness does not have a significant relationship with economic growth, even though 126 1 the sign of the correlation coefficient seem to suggest that openness is harmful for growth. Table 5.2 also shows some other significant relationships amongst variables. However, since the correlation matrix does not give us an idea about the causality, it is not clear the way in which the variables are affecting each other. Inflation has a negative association with real household consumption per capita which seems consistent if we believe higher inflation hits the poor and is harmful for poverty reduction. Its negative correlation with broad money-GDP ratio is also interesting. Both cases are possible – higher inflation leads to tight monetary policy (reducing money supply) causing the M2-GDP ratio to fall or expansionary monetary policy having a positive effect on growth and thus bringing down inflationary pressure on the economy. The moderate positive correlation between inflation and oil price growth seems to be consistent with the empirical evidence from the developing countries. Rising oil price is expected to exert pressure on inflation. Time series graphs presented in Figure 5.6 also help decipher some important facts. Mean growth rate in the sample countries shows major declines in the early 1970s and 1990s and late 1990s. There was one major hyperinflationary period in the 1960s and one inflationary spike during the Asian Financial Crisis 1997-1998. These episodes are mostly driven by the experiences of Indonesia around that time. Apart from growth and inflation experiences, one revealing fact is that real household consumption per capita shows a significant upward trend since the early 2000s. Financial deepening shows more or less an upward trend with some setbacks in the 1990s. On the other hand, government consumption-GDP ratio shows a more restrained trend. It peaks in the late 2000s perhaps due to the expansionary policy to help recover the economies from the Great Recession of 2008-09. Trade openness shows a significant upward trend in the 1990s perhaps due to globalisation policies dominant around that time. The mean of agriculture-GDP ratio shows a continuous decline over this period of time. 127 Figure 5.6: Mean of Cross-Country Data for Each Year, 1960-2010 Mean of Real GDP Growth Mean of CPI Inf lation .08 Mean of Real Household Consumption Per Capita 2.0 900 800 .06 1.5 700 .04 1.0 600 .02 500 0.5 .00 400 -.02 0.0 60 65 70 75 80 85 90 95 00 05 10 300 60 Mean of Broad Money -GDP Ratio 65 70 75 80 85 90 95 00 05 10 60 Mean of Gov ernment Consumption-GDP Ratio 18 120 60 16 100 50 14 80 40 12 60 30 10 40 20 8 65 70 75 80 85 90 95 00 05 10 05 10 70 75 80 85 90 95 65 70 75 80 85 90 95 00 05 10 60 65 70 75 80 85 90 95 Mean of Agriculture-GDP Ratio 40 35 30 25 20 15 65 70 75 80 85 90 95 00 05 10 00 05 10 20 60 45 60 00 Mean of Trade Openness 70 60 65 Source: Author’s calculations based on data from World Bank, WDI; IMF, IFS and WEO. Note: Unweighted mean values of the variables for the 14 countries which are Bangladesh, Cambodia, India, Indonesia, Lao PDR, Malaysia, Pakistan, The Philippines, Thailand, Vietnam, Kajakhstan, Kyrgyz Republic, Papua New Guinea, and Tajikistan. In summary, the selection of variable to examine the growth-inflation relationship is done with a view to incorporating economic factors relating to structural, demand and supply side shocks, and macro policies affecting the issue. From an analytical point of view the aim is to create a parsimonious model in explaining the research issues. 5.2.2 Empirical model To investigate the potential nonlinearity of the relationship between inflation and growth, the following panel model is developed. 128 yit = µi + βt + β1 πit + β2 π2 it + ∑ Zit + εit (5.1) for country, i = 1,…, N and time, t = 1,…,T. In equation (5.1), yit is the growth measured using two alternative ways – first difference of natural log of real GDP and first difference of natural log of real GDP per capita; µi is an unobservable time invariant country specific effects to capture heterogeneity in the growth-inflation relationship across countries; βt is a time specific effect incorporating dummies for different time periods; πit is the inflation rate measured as the first difference of natural log of consumer price index (CPI) or its transformed form as explained above; Zit is a vector of control variables, the regressors which help explain the growth-inflation relationship; and εit is the classical error term, assumed to be independent and identically distributed with mean zero and variance , which varies with countries and time in the regression. The country specific effects, µi, and the error term, εit, have the standard error component structure E [µi] = E [εit] = E [µi εit] = 0 (5.2) and the transient errors are not serially correlated: E [εit εis] = 0 for s ≠ t (5.3) Inspection of the time series graphs of the control variables reveals that taking the control variables in their first differenced form is appropriate to avoid spurious regression. Although, theoretically ratio variables such as M2-GDP and government consumption-GDP should show mean reversion process and therefore should be stationary, the visual inspection of the time series casts doubt on this. So following Kwon et al. (2009) the preferred form of data for the regression analyses is first differences.90 In doing so, the above general version of the model can be rewritten as follows: 90 Estimations are carried out using control variables at their level forms as well to compare differences with their first differenced counterparts. These are discussed in the next section. 129 Growthit = µi + βt + β1 ∆ ln (CPI)it + β2 (∆ ln (CPI)it)2 + β3 Growthit-1 + β4 ∆ ln (HCONPC)it + β5 ∆ ln (FD)it + β6 ∆ ln (GOVCON)it + β7 ∆ ln (OPEN)it + β8 ∆ ln (AGR)it + β9 DUMOILt + β10 DUMCOMPRt + εit (5.2) where, Growth is measured by two alternative ways, ∆ ln (real GDP) and ∆ ln (real GDP per capita). CPI refers to consumer price index and HCONPC, FD, GOVCON, OPEN, AGR refer to real household consumption per capita, broad money-GDP ratio, government consumption expenditure as a per cent of GDP, trade openness as defined above, agriculture share of GDP respectively. DUMOIL and DUMCOMPR are the dummy variables for oil price growth and commodity price growth respectively, as explained above. The rest of the notations are also explained in the previous section. The inclusion of time fixed effects requires some explanations. We use time fixed effects in our model to make the estimation more robust. As Kumar and Woo (2010) explain that this is to account for the possibility of any structural changes over the sample period, including changes in trends in global growth or global risk factors. Besides, the autocorrelation tests and robust standard errors reported later in the chapter, we assume absence of autocorrelation across countries in the error term. Thus including time dummies increases the chances of this assumption to hold (see Kathavate & Mallik 2012; Roodman 2006). In all our estimations the time fixed effects are found to be significant. One could, however, find contrasting evidence on this. For instance, a recent similar study by Rao and Hassan (2011) ignores these effects since they are found to be insignificant. We should, however, note that Rao and Hassan’s (2011) study uses five year averages as opposed to annual data used by us. We have also used robust standard errors in all our estimations to take care of the heteroscedasticity or unequal variance unlike Rao and Hassan (2011). We have preferred the use of robust standard errors since in practice we usually do not know the structure of heteroscedasticity. Thus, it is safe to use robust standard errors especially when one has a large sample size. Even if there is no heteroscedasticity, the robust standard errors will become just conventional OLS standard errors. Thus, the robust standard errors are believed to be appropriate even under homoscedasticity. 130 5.3 Empirical Results As a starting exercise, we estimate the impact of inflation on economic growth by ordinary least squares (OLS). We are interested in testing whether the marginal impact of inflation on growth is statistically significant. To begin with, Table 5.3 reports the OLS results following five time periods: 1960-2010, 1970-2010, 19802010, 1990-2010, and 2000-2010. Annual data for all variables are used and cases with inflation rate above 40 per cent are avoided. This allows us to cover the long run (fifty and forty year time horizons) as well as the medium term (twenty and ten year time horizons). Such differing estimation periods provide us with an understanding of any changes in inflation-growth nexus over time. However, OLS estimates cannot be taken seriously in order to interpret the causality as it suffers from omitted variable bias or heterogeneity bias resulting from possible correlation between country-specific fixed effects and the regressors. Any significant correlation between unobserved country specific factors and the right hand side regressors renders the OLS estimations inconsistent (Hansen & Tarp 2001). The second is the endogeneity problem due to potential correlation between the regressors and the error term. Besides, this dynamic pooled OLS model (as a result of including the lagged dependent term as a regressor) is likely to suffer from coefficient bias since the lagged dependent variable may be correlated with time invariant country specific effects (see Bond 2002). Table 5.3: Impact of Inflation on GDP Growth, OLS Estimations Dependent variable is growth measured by real GDP growth (1) (2) (3) (4) (5) 1960-2010 1970-2010 1980-2010 1990-2010 2000-2010 0.00184 0.0308 0.00971 0.0396 0.129 (0.0677) (0.0757) (0.0873) (0.105) (0.143) -0.0569 -0.124 -0.0255 -0.0757 -0.0419 (0.236) (0.258) (0.312) (0.339) (0.345) Turning point 1.616872 12.41935 19.03922 26.15588 153.9379 Growth (-1) 0.203*** 0.226*** 0.24*** 0.222** 0.146 (0.0743) (0.0773) (0.0855) (0.0904) (0.19) Inflation Inflation2 131 ∆ ln HCONPC 0.281*** 0.262*** 0.241*** 0.241*** 0.195*** (0.0483) (0.0461) (0.0456) (0.0524) (0.0593) -0.0439 -0.0394 -0.0418 -0.0377 -0.0407 (0.0306) (0.0316) (0.0339) (0.0373) (0.0787) 0.031 0.0284 0.0305 0.0537 0.0413 (0.0299) (0.0304) (0.0328) (0.0362) (0.0432) 0.0128 0.0233 0.0324 0.0264 0.0258 (0.0191) (0.0196) (0.0214) (0.024) (0.0314) -0.0492 -0.0519 -0.0494 -0.056 -0.0864 (0.0407) (0.0424) (0.0443) (0.0487) (0.0561) -0.00231 -0.00177 0.00348 0.00326 -0.00106 (0.0043) (0.0043) (0.0048) (0.0051) (0.0058) 0.00395 0.00394 0.00088 0.00173 0.00294 (0.0086) (0.0086) (0.0102) (0.0106) (0.0112) 0.0363*** 0.0328*** 0.0322*** 0.0308*** 0.0324*** (0.0048) (0.0052) (0.0056) (0.0071) (0.01) 380 348 295 224 129 R-squared 0.278 0.28 0.29 0.261 0.181 Adjusted R-squared 0.259 0.259 0.265 0.226 0.111 ∆ ln FD ∆ ln GOVCON ∆ ln OPEN ∆ ln AGR DUMOIL DUMCOMPR Constant Observations F Countries (N =14) 9.22 10.1 9.56 7.89 5.26 Bangladesh, Cambodia, India, Indonesia, Lao PDR, Malaysia, Pakistan, The Philippines, Thailand, Vietnam, Kazakhstan, Kyrgyz Republic, Papua New Guinea, and Tajikistan, Notes: 1. Estimation is based on annual observations and all cases of inflation greater than 40 per cent are excluded to avoid outlier effect of inflation. 2. All standard errors are robust and reported below coefficient estimates. 3. ***, **, and *, denote significance at 1%, 5%, and 10% respectively. 4. Turning point of inflation = [(Coefficient of linear term) / (2×coefficient of squared term)] × 100. 5. Growth (-1) refers to economic growth lagged by one period. In all five cases reported in Table 5.3, the signs of the coefficients of linear and squared terms of inflation confirm a nonlinear (inverted u-shaped curve) relationship between inflation and growth. However, none of the coefficients is significant, applying the robust standard error estimates. Lagged growth rate seems to have a strong and significant positive effect on growth in the medium and long run but the positive effect in the short run (2000-2010) is insignificant. Growth in real household consumption per capita (HCONPC) also has a strong positive effect on growth across 132 all five time periods. This result is expected since aggregate demand is supposed to have a strong positive impact on growth. Apart from these two variables none of the other variables seem to have a significant impact on GDP growth. Growth in financial deepening shows a negative impact on GDP growth which is not uncommon in the findings of growth literature (see, for instance, Rajan & Subramanian 2008, Table: 9, p. 658). Government expenditure does not affect GDP growth negatively both in the short run and long run, as the conventional wisdom suggests. Openness seems to have a growth enhancing effect. Agriculture share of GDP has a negative impact on GDP growth. The possible rationale for this is explained earlier – as the economy moves away from the primary sector, becoming more industrialised, it moves to a higher growth path. As we mentioned above, the results from Table 5.3 merely provides us with an initial view of the story and we must be cautious interpreting these results. Figure 5.7 shows a scatter plot between inflation and unexplained growth derived from the OLS estimation in column 1 of Table 5.3. It does not show any clear evidence of a strong negative relationship between the two. However, as we have said that these results are based on some preliminary estimations. A more rigorous approach follows in the following sections. 133 .2 Figure 5.7: Conditional Correlation between Inflation and Growth, 1961-2010 0 -.1 ind TJK TJK ind KAZ TJK PNG mys PNG KGZ tha KGZkhm KGZ tha khm tha idn mys mys vnm TJK mys mys tha mys PNG idn ind ind lao idn ind PNG phl khm tha mys khm KAZ tha vnm tha KAZ phlidn tha ind idn mys khm KAZ ind khm idn idn tha ind mys tha PNG mys khm idn ind indPNG mys mys mys mys vnm mys lao tha ind ind tha khm tha mys lao idn pak tha mys phl indbgd vnm khm mys PNG vnm mys tha phl vnm pak ind tha PNG idn mys ind idn ind ind TJK phl mys KAZ PNG KAZ bgd mys pak mys tha phl vnm vnm bgd lao ind mys idn bgd phl ind ind tha idn KAZ pak tha tha vnm pak idn ind phl KGZ khm bgd phl idn bgd phl idn mys phl bgd pak bgd phl ind phl vnm vnm bgd phl lao idn idn tha pak bgd KAZ idn phl tha bgd phl bgd KAZ tha phl bgd tha bgd tha KAZ phl mys tha ind pak phl bgd lao vnm idn pak tha phl bgd mys bgd ind PNG vnm idn tha mys ind bgd ind pak mys ind phl ind idn mys bgd tha khm tha phl phl pak bgd idn pak bgd mys phl tha mys khm ind bgd KAZ idn phl idn idnphllaoind mys tha pak phl phl ind tha vnmkhm tha mys bgd KGZ pak PNG idn phl bgd PNG tha phl pak tha phl bgd mys phl phl phl bgd idn phl ind PNG idn idn phl tha mys phl bgd KAZ ind lao ind pak ind KGZ ind khm ind pak ind PNG idn mys pak tha phl PNG mys KGZ ind tha idn pakPNG mys phl phl tha idn tha ind PNG bgd TJK phl phl thaindmys pakpak PNG ind KAZ pak idn PNG indPNG PNG PNG KGZ phl KGZ PNG mys tha mys PNG thamys khm phl idnPNG ind mys PNG phl PNGindPNG tha PNG KAZ KGZ -.2 Residuals of growth .1 PNG TJK idn KAZ ind phl KGZ TJK -.1 0 .1 .2 .3 Inflation conditional upon other covariates Residuals .4 Fitted values Source: Graph generated from the estimation done by the author in column 1 of Table 5.3. Note: All cases of inflation greater than 40 per cent are excluded in the estimation to avoid the outlier effect of inflation. 5.3.1 Static panel estimation results This section presents results based on Fixed Effects (FE) and Random Effects (RE) models the two most common models used in panel data estimations. Such panel estimations are superior to simple OLS as they can capture the country heterogeneity more efficiently. We do not use the lagged dependent variable on the right hand side of the regression as it may cause bias in the estimators. Kumar and Woo (2010, p. 12) state that “in the dynamic panel setting, the within transformation in the estimation process of FE introduces a correlation between transformed lagged dependent variable and transformed error,…[making] FE inconsistent.” Therefore all estimations performed in this section are within the static framework. In the case of RE models, it is assumed that the observable regressors are uncorrelated with the unobservable characteristics in both country specific effects and error term. This 134 assumption is somewhat relaxed in the case of FE model which does not impose that country specific effects and observable regressors are uncorrelated. Using the two estimation techniques several regressions are run. Tables 5.4.1, 5.4.2, and 5.4.3 record the best possible results from various estimations. To begin with, Table 5.4.1 provides results where regressors FD, GOVCON, OPEN, and AGR are in their level form. In both sets of estimations (using two measures of the dependent variable) the Hausman test shows that the FE model is preferable to the RE one. The FE estimates in Table 5.4.1 provides turning points of inflation at around 15 and 17 per cent for real GDP growth and real GDP per capita growth as dependent variables, respectively. Both coefficients of linear and squared of inflation are significant. In terms of taking care of the outliers of inflation, we consider case 1 of inflation, ignoring all observations when inflation rate is greater than 40 per cent. In this instance, apart from growth in HCONPC no other explanatory variables are found to be significant.91 Table 5.4.1: Fixed and Random Effects Panel Regressions (Case 1) Dependent variable Inflation 2 Inflation Turning point ∆ ln HCONPC ln FD ln GOVCON ln OPEN Real GDP Growth Real GDP Per Capita Growth (1) (2) (3) (4) FE1 RE1 FE2 RE2 0.177* 0.0612 0.155* 0.0182 (0.0882) (0.106) (0.0846) (0.106) -0.563** -0.212 -0.433** -0.02 (0.247) (0.268) (0.187) (0.23) 15.71 14.43 17.89 45.50 0.234** 0.274*** 0.23*** 0.281*** (0.0797) (0.0825) (0.073) (0.0814) -0.0116 -0.00445 -0.0136* -0.00874* (0.0077) (0.0051) (0.0069) (0.0051) 0.0167 -0.0107 0.0214 -0.0102 (0.0162) (0.0085) (0.0178) (0.0097) 0.0227 0.00681 0.0206 0.00473 (0.0135) (0.0044) (0.0126) (0.0053) 91 We also made an attempt to estimate the equations, in the static and dynamic frameworks (results of the latter are presented in the following sections), with a conditionality variable (country × DUMOIL) and (country × DUMCOMPR) – the multiplicative dummies – to test if oil and commodity price shocks have a growth effect in countries with oil reserves and self-sufficiency in food production. These versions of estimates were found to be insignificant and therefore are not reported. 135 ln AGR 0.00607 -0.00297 -0.00164 -0.012 (0.0111) (0.0076) (0.0115) (0.0087) 0.011 -0.00331 -0.00226 0.0151 (0.0232) (0.0178) (0.0212) (0.018) -0.0269 -0.0111 -0.0192 0.00427 (0.033) (0.0181) (0.0323) (0.0203) 388 388 388 388 14 14 14 14 R-squared overall 0.20 0.35 0.20 0.38 Within 0.34 0.31 .0.35 0.31 DUMOIL DUMCOMPR Observations Number of Countries Between Hausman test: Chi-square/(p-value) Countries 0.08 0.81 0.15 0.83 36.84 54.58 (0.003) (0.00) Bangladesh, Cambodia, India, Indonesia, Lao PDR, Malaysia, Pakistan, The Philippines, Thailand, Vietnam, Kazakhstan, Kyrgyz Republic, Papua New Guinea, and Tajikistan. Notes: 1. Estimation is based on annual observations and all cases of inflation greater than 40 per cent are excluded to avoid the outlier effect of inflation. 2. All standard errors are robust and reported below coefficient estimates. 3. ***, **, and *, denote significance at 1%, 5%, and 10% respectively. 4. The turning point of inflation is estimated by examining the partial derivatives of GDP growth with respect to inflation – the coefficient of the linear term of inflation divided by 2 times the coefficient of the squared term of inflation. 5. In Hausman test the null hypothesis is that Random Effect model is preferable to Fixed Effect model. 6. Constant term and time dummies are not reported to conserve space. Since we argued that our preferred form of the control variables is in first differenced form (growth rates92) Table 5.4.2 presents results incorporating the first differenced forms. Here again FE is more suitable than RE according to Hausman test. In FE models only the squared term of the inflation is significant but the linear one is not. All cases of inflation rate greater than 40 per cent are excluded. The turning points in this case (from the preferred FE models) seem to be more conservative than the previous ones, around 13 and 14 per cent for two measures of dependent variables, respectively. Growth in HCONPC shows strong positive impact on economic growth 92 First difference of natural logarithm of a variable is a close approximation of the growth rate of that variable. 136 as the previous models presented in Table 5.4.1. In the case of other control variables, growths in FD and AGR have significant negative impact on economic growth. In terms of shocks, both oil and commodity prices are significant, the former having a negative and the latter having a positive impact on growth. We explored rationale behind such possibilities in the earlier section. Table 5.4.2: Fixed and Random Effects Panel Regressions (Case 2) Dependent variable Inflation 2 Inflation Turning point Real GDP Growth Real GDP Per Capita Growth (1) (2) (3) (4) FE3 RE3 FE4 RE4 0.163 0.0751 0.139 0.0633 (0.093) (0.0759) (0.086) (0.0829) -0.622** -0.267 -0.481** -0.0996 (0.246) (0.245) (0.187) (0.211) 13.10 14.06 14.44 31.77 0.232*** 0.282*** 0.227*** 0.294*** (0.074) (0.0729) (0.0668) (0.0701) -0.0835** -0.0238 -0.0817** -0.00975 (0.0364) (0.0353) (0.0325) (0.034) 0.0188 0.0217 0.0107 0.0141 (0.0266) (0.0317) (0.0275) (0.0338) ∆ ln OPEN 0.00338 -0.00388 -0.00513 -0.0117 (0.016) (0.0181) (0.0146) (0.0184) ∆ ln AGR -0.0649* -0.0869** -0.0867*** -0.117*** (0.0335) (0.0382) (0.0164) (0.0326) -0.0303** 0.0217 -0.0323** 0.0353** (0.0104) (0.0202) (0.0109) (0.0156) 0.0637** 0.0144 0.0764*** 0.0245 (0.0218) (0.0163) (0.0185) (0.0182) 380 380 380 380 14 14 14 14 R-squared overall 0.31 0.35 0.32 0.38 Within .0.36 0.33 .0.38 0.34 ∆ ln HCONPC ∆ ln FD ∆ ln GOVCON DUMOIL DUMCOMPR Observations Number of countries Between Hausman test: Chi-square/(p-value) Countries 0.004 0.61 0.03 0.77 45.07 63.68 (0.00) (0.00) Bangladesh, Cambodia, India, Indonesia, Lao PDR, Malaysia, Pakistan, The Philippines, Thailand, Vietnam, Kazakhstan, Kyrgyz Republic, Papua New Guinea, and Tajikistan. Notes: 137 1. Estimation is based on annual observations and all cases of inflation greater than 40 per cent are excluded to avoid the outlier effect of inflation. 2. All standard errors are robust and reported below coefficient estimates. 3. ***, **, and *, denote significance at 1%, 5%, and 10% respectively. 4. The turning point of inflation is estimated by examining the partial derivatives of GDP growth with respect to inflation – the coefficient of the linear term of inflation divided by 2 times the coefficient of the squared term of inflation. 5. In Hausman test the null hypothesis is that Random Effect model is preferable to Fixed Effect model. 6. Constant terms and time dummies are not reported to conserve space. Table 5.4.3 shows the results using the case 3 of inflation where inflation is transformed as π/(1+ π). In terms of the turning points of inflation, the results are more conservative than the ones presented in previous tables. According to the FE estimates, as this is preferred over RE, the turning points are around 8 per cent with only squared term of inflation being significant. The only other result that is significantly different from the previous cases is the impact of oil price shocks. In this instance, it has a significant positive impact on growth. Table 5.4.3: Fixed and Random Effects Panel Regressions (Case 3) Dependent Variable Inflation Inflation2 Turning point ∆ ln HCONPC ∆ ln FD ∆ ln GOVCON ∆ ln OPEN ∆ ln AGR DUMOIL Real GDP Growth Real GDP Per Capita Growth (1) (2) (3) (4) FE5 RE5 FE6 RE6 0.0951 0.0768 0.0745 0.0741 0.0814 0.069 0.0862 0.0785 -0.609*** -0.503*** -0.456** -0.339** 0.158 0.139 0.152 0.139 8.46 8.26 8.89 12.27 0.229*** 0.278*** 0.222*** 0.288*** 0.0755 0.0725 0.0678 0.0692 -0.0784* -0.0198 -0.0773** -0.00579 0.0389 0.0368 0.0343 0.0356 0.0263 0.0283 0.0205 0.0234 0.027 0.0311 0.027 0.0325 -0.0141 -0.0206 -0.0253** -0.0319** (0.011) (0.0129) (0.0115) (0.0139) -0.0788** -0.101** -0.101*** -0.132*** (0.0359) (0.0399) (0.0216) (0.0336) 0.0431* 0.0274 0.0548*** 0.0416** 138 0.022 0.0218 0.0182 0.0172 0.0761** 0.018 0.0906*** 0.0285 (0.0253) (0.0165) (0.0217) (0.0181) 384 384 384 384 14 14 14 14 R-squared overall 0.39 0.43 0.37 0.43 Within .0.44 0.42 0.44 0.40 DUMCOMPR Observations Number of countries Between Hausman test: Chi-square/(p-value) Countries 0.00 0.65 0.03 0.78 42.75 62.17 (0.00) (0.00) Bangladesh, Cambodia, India, Indonesia, Lao PDR, Malaysia, Pakistan, The Philippines, Thailand, Vietnam, Kazakhstan, Kyrgyz Republic, Papua New Guinea, and Tajikistan. Notes: 1. Estimation is based on annual observations and inflation, π, in the table is transformed as ( ) to avoid the outlier effect of inflation. 2. All standard errors are robust and reported below coefficient estimates. 3. ***, **, and *, denote significance at 1%, 5%, and 10% respectively. 4. The turning point of inflation is estimated by examining the partial derivatives of GDP growth with respect to inflation. It is calculated as β1 / (2 β2 + β1) where β1 and β2 are the coefficients of linear and squared terms of inflation respectively. 5. In Hausman test the null hypothesis is that Random Effect model is preferable to Fixed Effect model. 6. Constant term and time dummies are not reported to conserve space. In summary, perhaps the estimates of turning point between 15 and 17 per cent according to FE1 and FE2 presented in Table 5.4.1 are better than the others since both the coefficients of inflation are significant in this instance. In short, the turning point does not seem to be 5 per cent which calls for caution for policy makers when targeting inflation below 5 per cent. 5.3.2 Dynamic panel estimation results In this section, we perform dynamic panel estimation techniques since FE and RE provide biased results incorporating lagged dependent variable as a regressor. It is difficult to find an appropriate external instrument for inflation and other economic variables to overcome the endogeneity problem. We use System Generalised Method of Moment (SGMM) approach arising from the works of Arellano and Bover (1995) 139 and Blundell and Bond (1998) to take care of this endogeneity issue. This approach uses suitable lagged levels and lagged first differences of the regressors as their instruments. It has recently gained popularity and is extensively used in applied economic research (see Kumar & Woo 2010). This estimator helps eliminate any endogeneity that may arise because of the correlation of country-specific, timeinvariant, factors and the right hand side variables. In addition, in this type of regressions, lagged values of the regressors are used to prevent simultaneity or reverse causality. We also use the one-step estimator as opposed to two-step estimator since the latter does not produce any meaningful results. Table 5.5.1 provides initial results by regressing real GDP growth on different regressors. Adding different regressors changes the threshold levels of inflation. It indicates that the relationship between economic growth and inflation is not simple and it is necessary to take into account the effects of other variables into this relationship. The results showing a threshold level of 12.94 per cent, in column 3 and column 4, seems to be a good estimate as both the inflation coefficients are significant at 1 per cent level. The results are robust since post-regression diagnostic tests are satisfactory showing that the underlying assumptions of the model are valid. For instance, AR (1) and AR (2) tests are performed to test first and second order serial correlation in the disturbances. One should reject the null hypothesis of the absence of first order serial correlation and not reject the absence of second order serial correlation (see Baltagi, Demetriades & Law 2009). Our tests satisfy these conditions. The other diagnostic test, Sargan test does not reject the overidentification restrictions. In fact these conditions relating to post-regression diagnostic tests are met in all SGMM estimations that follow. Table 5.5.1: Impact of Inflation on Growth, Dynamic Panel Estimations Dependent variable is real GDP growth Growth (-1) Growth (-2) (1) (2) (3) (4) SGMM1 SGMM2 SGMM3 SGMM4 0.131 0.152** 0.15** 0.15** (0.0857) (0.0719) (0.0733) (0.0733) -0.141*** -0.0967* -0.0483 -0.0483 (0.05) (0.0534) (0.0736) (0.0736) 140 Inflation 2 Inflation Turning point 0.0158 0.0504 0.167*** 0.167*** (0.0465) (0.042) (0.0562) (0.0562) -0.0345 -0.0488* -0.645*** -0.645*** (0.0336) (0.0267) (0.115) (0.115) 22.89 51.63 12.94 12.94 0.208*** 0.237*** 0.237*** (0.0552) (0.0673) (0.0673) -0.0596*** -0.0596*** (0.0203) (0.0203) 0.0524*** 0.0524*** (0.0168) (0.0168) 0.0128 0.0128 (0.0195) (0.0195) -0.0267** -0.0267** (0.0124) (0.0124) ∆ ln HCONPC ∆ ln FD ∆ ln FD (-1) ∆ ln GOVCON ∆ ln GOVCON (-1) ∆ ln OPEN ∆ ln OPEN (-1) ∆ ln AGR ∆ ln AGR (-1) 0.0154 0.0154 (0.0189) (0.0189) 0.075* 0.075* (0.0397) (0.0397) -0.0565* -0.0565* (0.0319) (0.0319) -0.089** -0.089** (0.0349) (0.0349) DUMOIL 0.0202*** (0.0078) DUMCOMPR 0.057*** (0.0155) Observation Number of countries SGMM estimation method AR (1) test (p-value) AR (2) test (p-value) Sargan test (p-value) Countries 435 396 370 370 14 14 14 14 One step One step One step One step -2.98 -2.83 -2.9 -2.9 (0.00) (0.00) (0.00) (0.00) 0.24 -0.58 0.31 0.31 (0.80) (0.55) (0.75) (0.75) 353.18 432.82 537.81 537.81 (0.14) (0.14) (0.87) (0.87) Bangladesh, Cambodia, India, Indonesia, Lao PDR, Malaysia, Pakistan, The Philippines, Thailand, Vietnam, Kazakhstan, Kyrgyz Republic, Papua New Guinea, and Tajikistan. Notes: 1. Estimation is based on annual observations and all cases of inflation greater than 40 per cent are excluded to avoid the outlier effects of inflation. 141 2. All standard errors are robust and reported below coefficient estimates. 3. ***, **, and *, denote significance at 1%, 5%, and 10% respectively. 4. Regressions use the Blundell and Bond (1998) system GMM (SGMM) estimator. 5. The turning point of inflation is estimated by examining the partial derivatives of GDP growth with respect to inflation – the coefficient of the linear term of inflation divided by 2 times the coefficient of the squared term of inflation. 6. AR (1) and AR (2) tests are Arellano-Bond first and second order serial correlation tests respectively. The null hypothesis is that residuals show no serial correlation. 7. Sargan test is for over identifying restrictions. The null hypothesis is that over-identifying restrictions are valid. 8. Time dummies are not reported to conserve space. 9. Variables lagged in one period are represented by -1 within parentheses after the variable. Table 5.5.2 and 5.5.3 have produced GMM estimation results in the case of dependent variable real GDP growth and real GDP per capita growth, respectively, using three different cases of inflation conditioned upon other variables. In the first case, the threshold figure varies from 10 to 15 per cent, while in the second case, the threshold varies from 11 to 14 per cent. The results of threshold inflation depending on measures of economic growth (GDP vs. GDP per capita) do not vary much. Table 5.5.2: System GMM Estimations (Case 1) Dependent variable is real GDP growth Growth (-1) (1) (2) (3) SGMM5 SGMM6 SGMM7 0.15** 0.154** 0.153** (0.0733) (0.0733) (0.0731) -0.0483 -0.0446 -0.0455 (0.0736) (0.0736) (0.0739) Inflation 0.167*** 0.248*** 0.315*** (0.0562) (0.0607) (0.0623) Inflation2 -0.645*** -1.12*** -1.64*** (0.115) (0.139) (0.175) 12.94 14.89 10.62 0.237*** 0.235*** 0.234*** (0.0673) (0.0659) (0.0654) -0.0596*** -0.0592*** -0.059*** (0.0203) (0.0198) (0.0197) 0.0524*** 0.0531*** 0.0523*** (0.0168) (0.0165) (0.0164) Growth (-2) Turning point ∆ ln HCONPC ∆ ln FD ∆ ln FD (-1) 142 ∆ ln GOVCON ∆ ln GOVCON (-1) ∆ ln OPEN ∆ ln OPEN (-1) ∆ ln AGR ∆ ln AGR (-1) DUMOIL DUMCOMPR Observations 0.0128 0.0124 0.0142 (0.0195) (0.0192) (0.0196) -0.0267** -0.025** -0.0252** (0.0124) (0.0121) (0.012) 0.0154 0.0125 0.0106 (0.0189) (0.0165) (0.016) 0.075* 0.0788** 0.0806** (0.0397) (0.04) (0.04) -0.0565* -0.0647** -0.0675** (0.0319) (0.0315) (0.0317) -0.089** -0.0931*** -0.0939*** (0.0349) (0.0345) (0.0342) 0.0202*** 0.0194** 0.0188** (0.0078) (0.0081) (0.0082) 0.057*** 0.0133 0.0156* (0.0155) (0.0094) (0.0092) 370 373 373 Number of countries SGMM estimation method 14 14 14 One step One step One step AR (1) test (p-value) -2.9 -2.9 -2.9 (0.00) (0.00) (0.00) AR (2) test (p-value) 0.31 0.32 0.28 (0.75) (0.74) (0.77) Sargan test (p-value) 537.81 544.33 547.88 Countries (0.87) (0.85) (0.82) Bangladesh, Cambodia, India, Indonesia, Lao PDR, Malaysia, Pakistan, The Philippines, Thailand, Vietnam, Kazakhstan, Kyrgyz Republic, Papua New Guinea, and Tajikistan. Notes: 1. Estimation is based on annual observations. To avoid outlier effect of inflation three cases are considered. In column 1 all cases of inflation greater than 40 per cent are excluded. In column 2 inflation, π, is transformed as ln (1 + π) while in column 3 it is transformed as ( ). 2. All standard errors are robust and reported below coefficient estimates. 3. ***, **, and *, denote significance at 1%, 5%, and 10% respectively. 4. Regressions use the Blundell and Bond (1998) system GMM (SGMM) estimator. 5. The turning point of inflation is estimated by examining the partial derivatives of GDP growth with respect to inflation. In column 1, 2, and 3 it is calculated as (β 1 / 2β2), ( – 1), and β1 / (2 β2 + β1) where β1 and β2 are the coefficients of linear and squared terms of inflation respectively. 143 6. AR (1) and AR (2) tests are Arellano-Bond first and second order serial correlation tests respectively. The null hypothesis is that residuals show no serial correlation. 7. Sargan test is for over identifying restrictions. The null hypothesis is that over-identifying restrictions are valid. 8. Time dummies are not reported to conserve space. 9. Variables lagged in one period are represented by -1 within parentheses after the variable. Table 5.5.3: System GMM Estimations (Case 2) Dependent variable real GDP per capita growth Growth (-1) Growth (-2) Inflation 2 Inflation Turning point ∆ ln HCONPC ∆ ln FD ∆ ln FD (-1) ∆ ln GOVCON ∆ ln GOVCON (-1) ∆ ln OPEN ∆ ln OPEN (-1) ∆ ln AGR ∆ ln AGR (-1) DUMOIL DUMCOMPR (1) (2) (3) SGMM8 SGMM9 SGMM10 0.188** 0.193** 0.193** (0.0841) (0.0834) (0.0827) 0.00421 0.00193 0.00057 (0.0499) (0.0498) (0.0507) 0.157*** 0.248*** 0.309*** (0.0539) (0.063) (0.0667) -0.555*** -1.07*** -1.54*** (0.102) (0.156) (0.214) 14.17 14.18 11.15 0.234*** 0.23*** 0.229*** (0.0555) (0.0532) (0.0527) -0.0598*** -0.0606*** -0.06*** (0.0228) (0.0228) (0.0227) 0.053*** 0.0545*** 0.0537*** (0.0128) (0.0116) (0.0117) 0.00632 0.00653 0.00825 (0.0215) (0.0211) (0.0215) -0.0097 -0.00814 -0.00802 (0.0145) (0.0137) (0.0137) -0.00412 -0.0105 -0.0121 (0.0173) (0.0163) (0.0156) 0.0762* 0.0792** 0.081** (0.0392) (0.0397) (0.0398) -0.0917*** -0.0991*** -0.101*** (0.0242) (0.0237) (0.0241) -0.0708* -0.0744* -0.075* (0.0409) (0.0393) (0.0391) -0.00773 -0.00818 -0.0087 (0.0075) (0.0079) (0.0081) 0.0344*** 0.0158 0.0178* (0.0131) (0.0101) (0.0101) 144 Observations 370 373 373 Number of countries SGMM estimation method 14 14 14 One step One step One step AR (1) test (p-value) -2.58 -2.57 -2.57 (0.00) (0.01) (0.01) AR (2) test (p-value) 1.28 1.32 1.3 (0.19) (0.18) (0.19) Sargan test (p-value) 545.72 554.65 558.17 Countries (0.81) (0.76) (0.73) Bangladesh, Cambodia, India, Indonesia, Lao PDR, Malaysia, Pakistan, The Philippines, Thailand, Vietnam, Kazakhstan, Kyrgyz Republic, Papua New Guinea, and Tajikistan. Notes: 1. Estimation is based on annual observations. To avoid outlier effect of inflation three cases are considered. In column 1 all cases of inflation greater than 40 per cent are excluded. In column 2 inflation, π, is transformed as ln (1 + π) while in column 3 it is transformed as ( ). 2. All standard errors are robust and reported below coefficient estimates. 3. ***, **, and *, denote significance at 1%, 5%, and 10% respectively. 4. Regressions use the Blundell and Bond (1998) system GMM (SGMM) estimator. 5. The turning point of inflation is estimated by examining the partial derivatives of GDP growth with respect to inflation. In column 1, 2, and 3 it is calculated as (β 1 / 2β2), ( – 1), and β1 / (2 β2 + β1) where β1 and β2 are the coefficients of linear and squared terms of inflation respectively. 6. AR (1) and AR (2) tests are Arellano-Bond first and second order serial correlation tests respectively. The null hypothesis is that residuals show no serial correlation. 7. Sargan test is for over identifying restrictions. The null hypothesis is that over-identifying restrictions are valid. 8. Time dummies are not reported to conserve space. 9. Variables lagged by one period are represented by -1 within parentheses after the variable. 5.3.3 Summary of the results from static and dynamic panel estimations A summary of the results from static and dynamic panel estimations are presented in Tables 5.6.1 and 5.6.2, considering the dependent variable real GDP growth and real GDP per capita growth respectively. In the case of static estimations only FE estimates are reported as they are superior to RE estimates (revealed by Hausman 145 test). The results on threshold levels broadly vary from 8 to 15 per cent. Columns 1 and 2 in Table 5.6.1 show FE estimates on turning points based on case 1 (inflation less than 40 per cent) and case 2 (inflation transformed into (π/1+π)) of inflationtreatment, respectively. It is interesting to note that case 1 provides a higher threshold, at 13 per cent, than case 2, at around 8 per cent. Perhaps the explanation behind this is that in the former estimation, we dropped the cases of inflation higher than 40 per cent altogether. This ignores the unusual cases of growth as well (for instance severe decline in growth rates as a result of very high inflation). In the latter case, however, only inflation is transformed to avoid the outlier effect on growth. The unusual cases of growth and other control variables were not dropped. We notice a similar situation in columns 1 and 2 of FE estimates in Table 5.6.2. The turning points are slightly higher, around 14 and 9 per cent, respectively. In the case of dynamic panel estimation, using SGMM, we notice variations in turning points depending on the way in which we treat the dependent variable and inflation cases. The estimated thresholds broadly vary between 10 and 15 per cent. In short, we observe differences in inflation threshold due to different modelling techniques and estimation methods. But none of these estimates justifies a case for keeping inflation at 5 per cent or below. Table 5.6.1: Summary of Results from Static and Dynamic Panel Estimations (Case 1) Dependent variable is real GDP growth Static Models Dynamic Models (1) (2) (3) (4) (5) FE3 FE5 SGMM5 SGMM6 SGMM7 13.10 8.46 12.94 14.89 10.62 Growth (-1) NA NA (+)** (+)** (+)** Growth (-2) NA NA (-) (-) (-) ∆ ln HCONPC (+)*** (+)*** (+)*** (+)*** (+)*** ∆ ln FD (-)** (-)* (-)*** (-)*** (-)*** ∆ ln FD (-1) NA NA (+)*** (+)*** (+)*** ∆ ln GOVCON (+) (+) (+) (+) (+) ∆ ln GOVCON (-1) NA NA (-)** (-)** (-)** Turning point of inflation 146 ∆ ln OPEN (+) (-) (+) (+) (+) ∆ ln OPEN (-1) NA NA (+)* (+)** (+)** ∆ ln AGR (-)* (-)** (-)* (-)** (-)** ∆ ln AGR (-1) NA NA (-)** (-)*** (-)*** DUMOIL (-)** (+)* (+)*** (+)** (+)** DUMCOMPR (+)** (+)** (+)*** (+) (+)* Table 5.6.2: Summary of Results from Static and Dynamic Panel Estimations (Case 2) Dependent variable is real GDP per capita growth Static Models Dynamic Models (1) (2) (3) (4) (5) FE4 FE6 SGMM8 SGMM9 SGMM10 14.44 8.89 14.17 14.18 11.15 Growth (-1) NA NA (+)** (+)** (+)** Growth (-2) NA NA (+) (+) (+) ∆ ln HCONPC (+)*** (+)*** (+)*** (+)*** (+)*** ∆ ln FD (-)** (-)** (-)*** (-)*** (-)*** ∆ ln FD (-1) NA NA (+)*** (+)*** (+)*** ∆ ln GOVCON (+) (+) (+) (+) (+) ∆ ln GOVCON (-1) NA NA (-) (-) (-) ∆ ln OPEN (-) (-)** (-) (-) (-) ∆ ln OPEN (-1) NA NA (+)* (+)** (+)** ∆ ln AGR (-)*** (-)*** (-)*** (-)*** (-)*** ∆ ln AGR (-1) NA NA (-)* (-)* (-)* DUMOIL (-)** (+)*** (-) (-) (-) DUMCOMPR (+)*** (+)*** (+)*** (+) (+)* Turning point of inflation In the following section, we perform some robustness checks by taking subsamples and using our preferred estimation technique SGMM. 5.3.4 Robustness checks and comparing the differences in the threshold levels of inflation Table 5.7.1 presents the results on turning points based on different time frames. The findings broadly provide us with estimated threshold levels within the range of 8-15 147 per cent, close to what we have estimated in the above sections. All estimates show threshold levels above 10 per cent, in other words, double digit figures. Table 5.7.1: Inflation Turning Points in Different Time Frames Dependent variable is real GDP growth Inflation 2 Inflation Turning point (1) (2) (3) (4) (5) (6) (8) SGMM11 19602010 SGMM12 19702010 SGMM13 19802010 SGMM14 19902010 SGMM15 20002010 SGMM16 19601989 SGMM17 19601979 0.167*** 0.177*** 0.149*** 0.225*** 0.248** 0.161** 0.19*** -0.645*** -0.679*** -0.631*** -0.723*** -0.84*** -0.726*** -0.703*** 12.94 13.03 11.8 15.56 14.76 11.08 13.51 Notes: 1. Estimation is based on annual observations and all cases of inflation greater than 40 per cent are excluded to avoid the outlier effect of inflation. 2. ***, **, and *, denote significance at 1%, 5%, and 10% respectively. 3. Regressions use the Blundell and Bond (1998) system GMM (SGMM) estimator. 4. The turning point of inflation is estimated by examining the partial derivatives of GDP growth with respect to inflation – the coefficient of the linear term of inflation divided by 2 times the coefficient of the squared term of inflation. 5. Estimates of column (1) are reported from column (4) of Table 5.5.1. 6. Control variables, time dummies, and post estimation tests (AR test and Sargan test) are not reported to conserve space. Table 5.7.2 shows the findings from subsamples based on different regions and income groups. There is strong evidence from the results that poorer countries have a higher threshold level of inflation. For instance, in column 1, when we drop relatively richer East Asian countries, we find the threshold at around 14 per cent. The threshold reduces, in column 2 and 3, as the mean real GDP per capita and real household consumption per capita increases. This result convincingly proves one of our research issues that threshold varies according to economic circumstances of groups of countries. The result on threshold also shows that our estimates from earlier sections are robust, since the range falls within 8-15 per cent. 148 Table 5.7.2: Impact of Inflation on Growth, the Regional Effect Dependent variable is real GDP growth (1) (2) (3) SGMM18 Excluding East Asia SGMM19 Excluding South Asia SGMM20 Excluding Former Command Economies Inflation 2 Inflation Turning point Number of countries Countries dropped from the sample of 14 countries 0.216** 0.132** 0.094* -0.756*** -0.596*** -0.571*** 14.28 11.07 8.31 10 11 Indonesia, Malaysia, The Philippines, and Thailand Mean real GDP per capita Mean real household consumption per capita Bangladesh, India, and Pakistan 8 Cambodia, Lao PDR, Vietnam, Kazakhstan, Kyrgyz Republic, and Tajikistan 534.24 1099.92 1268.83 385.9 682.1 724.15 Notes: 1. Estimation is based on annual observations and all cases of inflation greater than 40 per cent are excluded to avoid the outlier effect of inflation. 2. ***, **, and *, denote significance at 1%, 5%, and 10% respectively. 3. Regressions use the Blundell and Bond (1998) system GMM (SGMM) estimator. 4. The turning point of inflation is estimated by examining the partial derivatives of GDP growth with respect to inflation – the coefficient of the linear term of inflation divided by 2 times the coefficient of the squared term of inflation. 5. Control variables, time dummies, and post estimation tests (AR test and Sargan test) are not reported to conserve space. 6. Unweighted mean values of real GDP per capita and real household consumption per capita within each group over the period 1990-2010 are reported. Next, we are interested to look at how the threshold of inflation might vary according to changes in agricultural dependence, financial deepening, and trade openness. First, Table 5.7.3 groups the countries according to above and below sample mean based on these three criteria. 149 Table 5.7.3: Averages of Selected Variables in the Sample Countries, 1990-2010 Agriculture share of GDP: Sample mean 24.63 Countries below sample mean Countries above sample mean Country Country Mean Mean Kazakhstan 10.28 Lao PDR 47.01 Thailand 10.44 Cambodia 38.91 Malaysia 11.1 Kyrgyz Republic 35.88 Indonesia 16.18 Papua New Guinea 34.74 The Philippines 16.28 Tajikistan 27.55 India 23.28 Vietnam 25.85 Bangladesh 23.89 Pakistan 24.07 Financial deepening: Sample mean 45.25 Countries above sample mean Countries below sample mean Country Country Mean Mean Malaysia 114.37 Tajikistan 9.9 Thailand 97 Lao PDR 14.29 The Philippines 52.34 Cambodia 15.52 India 52.03 Kyrgyz Republic 15.66 Kazakhstan 19.83 Vietnam Trade openness: Sample mean 47.6 Papua New Guinea 34.4 Bangladesh 37.3 Pakistan 41.65 Indonesia 42.98 88.08 Countries above sample mean Countries below sample mean Country Country Mean Mean Malaysia 188.57 India 30.23 Papua New Guinea 113.91 Bangladesh 33.23 Vietnam 113.57 Pakistan 34.61 Thailand 111.71 Indonesia 58.04 Tajikistan 108.1 Lao PDR 67.33 Cambodia 104.72 The Philippines 86.09 Kyrgyz Republic 96.13 Kazakhstan 89.37 Table 5.7.4 presents the regression results based on the above criteria. The results are consistent with what we have argued so far. The threshold varies according to the levels of economic development, and less developed countries tend to have a higher 150 threshold level of inflation. For instance, in the case of agro dependence, countries more dependent on agriculture appears to have a higher turning point, around 13 per cent, compared to that of countries relatively less dependent on agriculture (turning point at around 10 per cent; see columns 1 and 2 in Table 5.7.4). We find similar evidence in the case of financial deepening and trade openness. The findings on threshold vary, but remain within the range between 8 and 15 per cent. This again proves that our estimates are robust. Table 5.7.4: The Impact of Inflation on Growth Based on Structural Characteristics of the Economy Dependent variable is real GDP growth Inflation 2 Inflation Turning point Number of countries Countries (1) (2) (3) (4) (5) (6) SGMM21 SGMM22 SGMM23 SGMM24 SGMM25 SGMM26 Agro dependence Financial deepening Trade openness Least Most Most Most Least Least 0.0764* 0.635*** 0.0903* 0.651*** 0.114 0.171*** -0.357*** -2.35*** -0.576*** -2.34*** -0.719** -0.763*** 10.7 13.51 7.83 13.91 7.92 11.2 4 Kazakhstan, Thailand, Malaysia, and Indonesia 4 Lao PDR, Cambodia, Kyrgyz Republic, and Papua New Guinea 4 Malaysia, Thailand, The Philippines, and India 4 Tajikistan, Lao PDR, Cambodia, and Kyrgyz Republic 4 Malaysia, Papua New Guinea, Vietnam, and Thailand 4 India, Bangladesh, Pakistan, and Indonesia Notes: 1. Estimation is based on annual observations and all cases of inflation greater than 40 per cent are excluded to avoid the outlier effect of inflation. 2. ***, **, and *, denote significance at 1%, 5%, and 10% respectively. 3. Regressions use the Blundell and Bond (1998) system GMM (SGMM) estimator. 4. The turning point of inflation is estimated by examining the partial derivatives of GDP growth with respect to inflation – the coefficient of the linear term of inflation divided by 2 times the coefficient of the squared term of inflation. 5. P-value for the linear term of inflation in column (5) is 0.153 and therefore the estimate is significant at 16% significance level. 6. Control variables, time dummies, and post estimation tests (AR test and Sargan test) are not reported to conserve space. 151 7. 4 least and most agro dependent, financial deepening, and trade open countries are selected from Table 5.7.4 for each group of estimation. 5.4 Conclusion The empirical evidence based on panel estimations finds no support to keep inflation at a very low level in the selected developing countries of Asia. The results are based on the estimation techniques which take care of endogeneity problem, an issue many past studies have failed to address. Our technique to find the threshold level of inflation is similar to that of Pollin and Zhu (2006). The findings are broadly similar to those of Pollin and Zhu’s (2006) study as well. Their study covers a wider range of countries (80 middle and low income countries). The findings of this chapter are based on a much smaller sample size of 14 countries. While Pollin and Zhu (2006) suggest a threshold level of inflation between 15 and 18 per cent, our estimates are slightly more conservative, between 8 and 15 per cent. But the range is still close to the findings from other cross-country studies reviewed in Chapter 3. In short, our results find no justification for targeting inflation at 5 per cent or below as advised by the IMF. The next chapter seeks to investigate this empirical issue in a countryspecific context. 152 Appendix to Chapter 5 Table A.5.1: List of Variables and Their Sources Mnemonic Variable description Source Indicator code RGDP GDP (constant 2000 USD). GDP, gross domestic product, at purchaser's prices is the sum of gross value added by all resident producers in the economy plus any product taxes and minus any subsidies not included in the value of the products. It is calculated without making deductions for depreciation of fabricated assets or for depletion and degradation of natural resources. Data are in constant 2000 U.S. dollars. Dollar figures for GDP are converted from domestic currencies using 2000 official exchange rates. For a few countries where the official exchange rate does not reflect the rate effectively applied to actual foreign exchange transactions, an alternative conversion factor is used. WDI NY.GDP.MKTP.KD RGDPCAP GDP per capita (constant 2000 USD). GDP per capita is gross domestic product divided by midyear population. Data are in constant U.S. dollars. WDI NY.GDP.PCAP.KD CPI Consumer price index (2005 = 100). Consumer price index reflects changes in the cost to the average consumer of acquiring a basket of goods and services that may be fixed or changed at specified intervals, such as yearly. The Laspeyres formula is generally used. WDI FP.CPI.TOTL INFCPI Inflation, consumer prices (annual %). Inflation is measured by the consumer price index. WDI, IFS, WEO, SDBS FP.CPI.TOTL.ZG M2GDP (FD) Broad money (% of GDP); M2 Nominal GDP ratio. Broad money is the sum of currency outside banks; demand deposits other than those of the central government; the time, savings, and foreign WDI FM.LBL.BMNY.GD.ZS 153 currency deposits of resident sectors other than the central government; bank and traveller’s checks; and other securities such as certificates of deposit and commercial paper. GOVCON General government final consumption expenditure (% of GDP). General government final consumption expenditure (formerly general government consumption) includes all government current expenditures for purchases of goods and services (including compensation of employees). It also includes most expenditure on national defence and security, but excludes government military expenditures that are part of government capital formation. WDI NE.CON.GOVT.ZS HCONPC Household final consumption expenditure per capita (constant 2000 USD). Household final consumption expenditure per capita (private consumption per capita) is calculated using private consumption in constant 2000 prices and World Bank population estimates. Household final consumption expenditure is the market value of all goods and services, including durable products (such as cars, washing machines, and home computers), purchased by households. It excludes purchases of dwellings but includes imputed rent for owner-occupied dwellings. It also includes payments and fees to governments to obtain permits and licenses. Here, household consumption expenditure includes the expenditures of nonprofit institutions serving households, even when reported separately by the country. Data are in constant 2000 U.S. dollars. WDI NE.CON.PRVT.PC.KD OPEN Trade (% of GDP) Trade is the sum of exports and imports of goods and services measured as a share of gross domestic product. WDI NE.TRD.GNFS.ZS 154 AGR OIL COMPR Agriculture value added as a % of GDP UK Brent crude oil index (2005=100) WDI Non-energy commodity price index by the World Bank for low and middle income countries (2005=100) IFS NV.AGR.TOTL.ZS IFS 155 Chapter 6 The Relationship between Inflation and Growth: The Bangladesh Experience 6.1 Introduction This chapter investigates the research questions relating to growth-inflation nexus in the context of Bangladesh. The purpose of this chapter is to find out if a countryspecific study provides similar evidence to that of cross-country panel study. The panel study carried out in Chapter 5 includes 14 developing countries of Asia where relative dissimilarity amongst the countries exists. Although panel estimation models used in Chapter 5 are said to be efficient in taking care of the persistent unobserved heterogeneity across countries, studies, such as Temple (2000), mention that the impact of inflation might vary due to differences in the cross-country setting. In addition, a country-specific time series analysis allows greater freedom in using control variables relevant to that particular country. More importantly we can allow for the specific historical and institutional characteristics of the country in our analysis. The threshold estimates of inflation in Chapter 5 vary approximately between 8 and 15 per cent, depending on variations in structural issues in the samples and different modelling and estimation techniques used. We reiterate that none of the estimates of threshold is as low as 5 per cent and it tends to increase for groups of countries with lower levels of development. For estimations using subsamples comprised of Bangladesh, the threshold figures appear to be within double digits. The task of this chapter’s investigation is to provide a more specific estimate of inflation threshold for one country, Bangladesh. Considering Bangladesh’s level of development, we expect the threshold figure would be more than 10 per cent. This expected figure substantially differs from two recent important studies on Bangladesh reviewed in Chapter 4. The first one by Ahmed and Mortaza (2005) suggests that inflation threshold for Bangladesh is at 6 per cent. The other study by Hayat and Kalirajan (2009) does not confirm any threshold level, but suggests that increase in inflation from any level would hurt economic growth in Bangladesh. 156 The chapter is organised as follows. Section 6.2 provides an overview of the countryspecific case. It attempts to provide a rationale for a close examination of Bangladesh – an interesting country for development discourse.93 The stylised facts revealed from trends in economic growth and inflation in the country is discussed in Section 6.3. The next two sections carry out an econometric exercise to determine the growth-inflation nexus and the inflation threshold in Bangladesh, respectively. Section 6.4 examines the output-price level relationship in a cointegration framework using annual data in order to understand the short run dynamics, between economic growth and inflation, and the long run equilibrium relationship, between output and price level. Section 6.5 estimates the threshold level of inflation in a non-linear framework using control variables pertinent to the Bangladesh economy. Finally, Section 6.6 presents some concluding remarks. 6.2 An Overview of the Bangladesh Perspective The country Bangladesh provides an interesting example of a country-specific case study. The Economist concludes that Bangladesh is “one of the most intriguing puzzles in development” because “it has done better than most countries in improving the basic standard of living of its people.”94 Table 6.1 provides some of the key demographic and geographic characteristics of Bangladesh. It is a small but densely populated developing country in South Asia, with 31.5 per cent people living 93 During the first few years of its independence, Bangladesh experienced high inflation and low growth. The then US Secretary of State Henry Kissinger described Bangladesh as an “international basket case” and Kevin Rafferty of the Financial Times characterised it as “the end of the great development dream”. Two World Bank economists, Just Faaland and Jack Parkinson (1976), termed Bangladesh as “the test case of development” – meaning that if development could happen in Bangladesh, it could happen anywhere. But in less than three decades, Bangladesh became an example of success. One leading economic development textbook (Todaro & Smith 2006, p. 89) observed, “And without question, from the viewpoint of the broader meaning of development, Bangladesh seemed to be forging ahead in the early years of the twenty-first century. As a result, there are good reasons to expect Bangladesh to soon move into the lead in income as well.” The UNDP Human Development Report (2000) summarised Bangladesh’s achievements in the following words: “Bangladesh’s case stood out as a potentially instructive story of development. This was, indeed, seen as one of those rare cases of graduation on an international scale, from the status of the poorest of the poor to the place of lead performer among LDCs, and in some respects, even placing itself among the better performers in the entire developing world. It was like a re-birth for Bangladesh, having moved from the status of being at the margin of history in the whirlpool of history, from the status of a mere testing site to that of a learning site from where even master development discourse can benefit.” 94 The Economist, Out of the Basket, 2 November 2012. Available at http://www.economist.com/blogs/feastandfamine/2012/11/bangladesh 157 below the national poverty line.95 Although it is the 8th most populous country of the world (see Table 6.1), it is expected to experience a demographic dividend in the near future. This is because it is passing through the third phase of demographic transition with declining birth and death rates, leading to a growing working age population (see World Bank 2012, p. 2). This attribute is expected to payoff, for instance, by allowing opportunities for greater remittances from abroad and domestic demand-led growth. Table 6.1: Key Demographic and Geographic Features of Bangladesh Indicators Facts and figures Population 149 million (approximately)* World ranking in terms of population 8th most populous nation Density of population 800 people per square kilometre (approximately) Percentage of young population 70% under the age of 25 Religious composition 88% Muslims and 11% Hindus (approximately) Climate Subtropical Terrain Mostly flat low-lying plains, hilly in the southeast Key natural resources Natural gas, arable land, timber, and coal Source: Central Intelligence Agency, The United States, The World Fact Book, available at https://www.cia.gov/library/publications/the-world-factbook/geos/bg.html; and Bangladesh Bank, Bangladesh in brief, available at http://www.bangladesh-bank.org. Note: * Figure reported in January 2011 by the Bangladesh Bureau of Statistics, a government agency of the country. However the figure stands at 162 million, approximately, according to the World Bank, World Development Indicators. The country’s long border with its neighbour India has always been of political and economic significance. On the economic side, trade with India affects some of the major economic indicators such as trade deficits and inflation. In recent years, several possible opportunities have put the country into a greater geographic and strategic importance in the region. These include potential oil reserves and the establishment of a deep sea port in the Bay of Bengal and the construction of the Asian Highway which would connect several Asian countries to Bangladesh. Such 95 The figure 31.5 per cent is poverty headcount ratio at national poverty line (% of population) in 2010, according to the World Bank, available at http://data.worldbank.org/country/bangladesh. 158 prospects, among others, would better connect the country with the two emerging economic powers, China and India, both of which would benefit from using Bangladesh’s ports and infrastructure. The country is prone to supply shocks such as flooding and cyclones.96 However, the regular flooding, along with the country’s large rivers, makes the soil fertile providing a natural ground for agricultural activities. This perhaps explains why agriculture is considered the mainstay of the economy. Besides supply shocks resulting from natural disasters, the country is also exposed to external shocks due to its high dependence on imports as well as exports and remittances. For instance, in a report by the UNDP (2007) Bangladesh is identified as one of the high oil price vulnerable countries. However, one must also bear in mind that governments in Bangladesh, using subsidy, do not always pass through international oil price changes to the domestic economy immediately. In recent years, an additional contributing factor by which the economy has become more likely to experience the supply shocks is the adoption of a flexible exchange rate regime in 2003. On the political side, a great part of the relatively modern political history of the nation can be characterised as subjugation under different colonial rules and a struggle for attaining political autonomy. After about 200 years of British colonial rule and being a part of Pakistan for around 25 years, Bangladesh gained independence in 1971. In recent decades, however, the nation seems to be experiencing a struggle to put the self-governance into a smooth path and use the political emancipation to achieve socio-economic development. Still it is interesting that the detrimental force of political turmoil has not limited the country’s economic progress. Studies on Bangladesh at different times record that the country has made considerable economic progress particularly in alleviating poverty (ADB 2009; Hossain, M & Sen 1992; Khan 1990; Osmani 1990; Rahman, HZ & Hossain 1995; Sen 2003; World Bank 2012). Not only has the rate of growth risen steadily, agencies such as the World Bank has given the country high marks for poverty reduction (Khan 2007, p. 1). There seems to be a resilience developed by the nation 96 The average loss of rice production per annum due to normal flooding is roughly 4 per cent of total rice production; it is higher during the years of severe floods (Paul and Rashid 1993 cited in Hayat and Kalirajan 2009, p. 6). 159 amidst the political setbacks and natural disasters, helping the country moving forward from its reputation of being ‘an international basket case’. This is echoed by Shand (1996, p. ix) who states that the country’s “belief in its own capacity to do better is firm” and it has made undeniable social and economic progress since its independence.97 Nevertheless, Bangladesh still has large development deficits. It is a least developed country with low level of food security and high exposure to external shocks and climate risks. Infrastructure shortage is a critical obstacle to its growth. The above discussion sets the background for the importance of policies to unlock Bangladesh’s geopolitical and demographic potentials. Broad-based policies targeting economic growth and developmental issues would help the country overcome its key problems such as poverty, food security, and climate change mitigation and adaptation. A recent report by the World Bank (2012, p. 1) comments that “[y]et after 40 years of independence, Bangladesh remains a low income country with…its economic growth potential under-exploited [emphasis added].” The report also concludes that Bangladesh requires 8 per cent growth to reach the middle income country threshold (ibid. p. 4). This appears to be a necessary, if not sufficient, condition in the face of poverty still at 31.5 per cent with 50 million people still impoverished.98 Failure to reach a higher growth path thereby generating more income and reducing poverty may have dire consequences, for instance embracing political extremism leading to conditions of a failed state.99 Despite the fact that Bangladesh needs policies with broader development perspective, a greater emphasis on attaining higher economic growth, the IMF urges the country to pursue much more narrowly-based macroeconomic policies mainly to 97 A number of events can be noted to record the initial plight of the country – the devastating 9month war in 1971 and the cyclone a year before, the difficulty in dealing with international politics at the time of the Cold War, and the famine in 1974 which was aggravated due to food aid stopped by the United States primarily in retaliation to the accord made by Bangladesh with Cuba. Several papers state that the War of Independence in 1971 and its aftermath have dealt a severe blow to the economic structure of Bangladesh (Islam 1974, p. 3, cited in Alauddin and Hasan 1996, p. xiii). In addition, after the war as Robinson (1978, p. 370) states “[t]he new government had come to power with…no clearly defined economic policies, beyond general aspiration towards socialism”. Despite the adversities, Bangladesh managed to advance to a near self-sufficiency in food production by the end of 1990s (see Talukder 2005, p. 38). 98 See World Bank (2012). 99 Rashid (2008, p. 1) states that although the State Department of the United States praised Bangladesh as a ‘moderate Muslim democracy’, if the projected poverty and population growth continues, the country will be a fertile ground for extremism. 160 control inflation. The idea rests on the belief that instead of paying attention to multiple tasks to manage macroeconomic conditions, it is better to narrow down the focus and put emphasis on maintaining price stability to achieve sustained economic growth. This is evident from the conditions of a recent IMF loan, equivalent to US $ 1 billion. Table 6.2 reveals that the key message of the conditions is to employ contractionary monetary and fiscal policies.100 In line with the conditions, the central bank increased the policy rates four times in 2011 and in early 2012 it increased the repo rate from 7.25 per cent to 7.50 per cent and raised the reverse repo rate to 5.75 per cent from 5.25 per cent.101 Along with the rising interest rate, the other conditions such as removal of interest rate ceiling and higher energy prices (see conditions 6 and 7 in Table 6.2) are expected to raise operational costs of business. This is contradictory to the goal of achieving higher growth by encouraging investment. The reason is that it may conflict with the need for credit in a low growth economy such as Bangladesh.102 In fact the President of the Federation of Bangladesh Chambers of Commerce and Industry (FBCCI) points out that the higher interest rate is one of the key impediments for the domestic entrepreneurs to undertake the much needed investment to achieve the 8 per cent GDP growth.103 Table 6.2: Key Conditions of IMF’s US $ 1 billion Loan to Bangladesh, 2010 1. Pursuing a contractionary monetary policy in order to check inflationary pressure. 2. Further liberalization of tariff level 3. More liberal exchange regime by moving into floating exchange rate regime. 4. Privatization of the loss making state owned enterprises (SOE’s) in order to create more efficiency and more competencies. 5. Placing new VAT and income tax in order to achieve revenue targets. 6. Phasing out bank lending rate ceilings (for bringing in greater flexibility in the lending 100 In fact the set of conditions imposed by the IMF can be aptly summarised by “the triple commandments” as coined by Rodrik (2004, p. 1) which are “stabilise, liberalise, and privatise.” The government is asked to achieve macroeconomic “stability” by reducing inflation and fiscal deficits (see conditions 1 and 8 in Table 6.2), “liberalise” trade by relaxing trade barriers (condition 2) and entering into flexible exchange rate regime (condition 3), and “privatise” allowing greater ownership of private sector on resources to gain efficiency (condition 4). 101 Source: The Daily Star, BB raises policy rates, 6 January 2012. The repo rate, equivalent to bank rate in the United States, is an interest rate at which the commercial banks borrow money from the central bank. The reverse repo rate is the return commercial banks earn on excess funds kept with the central bank. 102 This is also the reason why setting the main aim of monetary policy to controlling inflation only is imprudent. 103 Source: The Daily Star, 23 November 2012. 161 regime). 7. Raising CNG and furnace oil prices. 8. Introducing a debt management strategy to reduce budget deficit. Source: Adapted from Unnayan Onneshan (2011, p. 3).104 It is also important to point out internal inconsistencies of the IMF’s conditions. One major contributory factor for higher inflation is the removal of subsidies and the upward adjustment of administrative prices, designed to reduce government’s budget deficits and debt. Tighter monetary policy certainly cannot lower inflation coming from this source. It reduces inflation by slowing down overall economic activities as evident from the concerns expressed by the President of the FBCCI. The policy becomes self-defeating as slower economic activities reduce government’s revenue and hence puts more pressure on the government budget. In sum, the importance of a higher and sustained growth requirement for Bangladesh is recognised by the studies carried out both inside and outside the IMF and the World Bank. Still policy prescriptions suggested by the IMF in practice appears to be heavily tilted towards reining in inflation at a low single digit level in the belief that it will spur investment and growth.105 The fear that inflation might accelerate drives the government to pursue constractionary monetary policy even at the cost of hurting growth. Not only is this self-defeating, there is no conclusive evidence that growth in money supply alone causes inflation in Bangladesh. For instance, in an earlier study, Ahmed (1984) concludes that the monetarist hypothesis that inflation is caused by an excessive growth in money supply is “irrelevant for explaining Bangladesh’s inflation.” The study also shows that the demand management policies would reduce inflation but at the cost of a slowdown in income growth. He argues that supply management policies help inflation to go down without adversely affecting output growth. On the other hand, in a recent study, Hossain (2010) argues that supply shocks are unlikely to be a “regular source of inflation” (p. 572) and “broad money growth causes inflation in the short run” (p. 574). In contrast, CPD (2011, p. 24) points out that growth in money supply and inflation moved along the same direction 104 The study by Unnayan Onneshan (translates ‘in search of development’, a multidisciplinary research centre in Bangladesh), funded by the Christian Aid, attempt to evaluate IMF’s loan and its implications on Bangladesh economy. 105 This is also evident from the discussion based on the IMF Article IV Consultation report on Bangladesh in Chapter 3. 162 over the major part of the period between 2001 and 2010, and that no causal relationship could be established between the two in the short run. Given this setup, the next section looks at the time series data on growth and inflation to reveal some stylised facts on their relationship. 6.3 Trends in Growth and Inflation in Bangladesh The data on real GDP growth and inflation rates in Bangladesh for the period 19712009 provide some interesting observations into the relationship between the two indicators. Figure 6.1 depicts growth and inflation rates between 1971 and 2009 while Table 6.3 provides decadal summary statistics. It shows that growth has never reached a double digit figure during this period (Figure 6.1). Each decade has experienced about one percentage point increase in growth since the 1980s (Table 6.3). Average inflation on the other hand shows a declining trend according to the statistics. Another important fact to note, from Table 6.3, is that volatility in growth, measured in terms of coefficient of variation (CV), is much less compared to its inflation counterpart since the 1990s (Table 6.3). In short, the average growth rate has steadily picked up pace and become less volatile after the transition to democracy, with an associated declining trend but greater variability in inflation.106 This, however, does not provide us with the information relating to causality between the two variables. 106 After about a 9-year military rule Bangladesh experienced a transition to democracy in the late 1990 (see Mahmud 2008, p. 93). 163 Figure 6.1: Real GDP Growth and CPI Inflation in Bangladesh, 1971 - 2009 60 50 40 30 20 10 2008 2006 2004 2002 2000 1998 1996 1994 1992 1990 1988 1986 1984 1982 1980 1978 1976 1974 1972 -10 1970 0 -20 Real GDP Growth CPI Inflation Source: World Bank, World Development Indicators; and IMF, World Economic Outlook. Table 6.3: Decadal Averages of Growth and Inflation in Bangladesh, 1971 – 2009 Real GDP growth rates Inflation rates (CPI) Average SD CV Average SD CV 1970s 0.80 7.67 9.53 22.09 20.99 0.95 1980s 3.15 1.22 0.38 11.26 2.24 0.19 1990s 4.68 0.69 0.15 6.39 2.92 0.46 2000s 5.64 0.63 0.11 5.68 2.45 0.43 Source: Author’s calculations based on data available from World Bank, World Development Indicators and IMF, World Economic Outlook. Note: SD and CV stand for standard deviation and coefficient of variation (SD/Average) respectively. The correlation statistics between growth and inflation rates also shows some interesting stylised facts. For the entire period between 1971 and 2009, the correlation (-0.27) shows a moderate negative relationship. However, if we divide the period into halves, 1971-1989 and 1990-2009, what we notice is a negative relationship (-0.13) for the former period but a positive relationship for the latter (0.21). This perhaps reflects the changing pattern of growth-inflation dynamics over time as the economy undergoes important structural changes. 164 The above descriptive analysis provides mixed experience about the growth-inflation nexus over the sample period, 1971-2009. In order to employ policy to sustain growth and manage inflation, an understanding of the sources of inflation is also important. Before using policy instruments it is essential to identify whether inflation is predominantly demand-driven or a supply-side phenomenon. This is because, contractionary monetary policy attempting to stabilise inflation in the event of a supply shock may cause further decline and greater variability in growth. 107 It is particularly relevant for Bangladesh which is susceptible to supply shocks, discussed in the previous section. It is clear from Figure 6.1 that the incidence of inflation in the early period of 1970s was discrete and it reached its highest in the 40-year history of Bangladesh.108 This is no surprise as inflation typically shoots up after a crisis such as war (see Blanchard & Sheen 2009, Chapter 24). Not only was it inevitable due to the war, the two international oil price shocks of the 1970s also triggered the high bouts of inflation in the economy. Therefore, studies (for example, Bose 1973; Rahim 1973; Siddique 1975) which claim that a prime cause of the inflation during this period is excess monetary expansion to finance budget deficits in a war-ravaged economy miss the point and fail to see the causes that triggered inflation in the first place. Mahmud and Osmani (1980) identify remittances, a demand-side cause of inflation. They argue that remittances are mostly used for consumption and unproductive investment such as real estate thus exacerbating the already explosive inflationary situation. The process of inflation in the economy is also examined by Taslim (1982) in the light of structuralist-monetarist debate and he concludes that both structural bottlenecks and monetary aggregates are relevant for the inflationary process in Bangladesh. The above studies attempt to analyse inflationary situation during an unusual period of the country – the 1970s which was dogged by a series of major international and domestic problems as discussed earlier. Besides, these studies have some empirical shortcomings in terms of incorporating the advances in time series analysis such as 107 As discussed in Chapter 3, the research by Selassie et al. (2006) carried out at the IMF also recognise that procyclical monetary policy, when inflation is cause by a supply shock, would have unintended consequence on growth in the developing countries. 108 As Chowdhury and Hossain (1996, p. 11) note, the annual inflation rate during 1972-1975 was the highest since the 1943 Bengal Famine. 165 addressing the non-stationarity issue resulting in spurious relationship between variables. Therefore, a look at recent studies is more appropriate. Mortaza (2006) supports the IMF view and claims that money supply – a demand-driven cause of inflation – and exchange rate have significant positive impact on inflation during the fiscal period 1990-2006. His finding of negative relationship between deposit interest rate and inflation also lends support to fighting inflation by raising interest rates. Contrary to this, Rahman et al. (2008) find that external shocks such as the upsurge in international commodity prices and structural issues such as supply-side constraints and distortions in domestic supply chain are mainly responsible for inflation in 2007-2008. Another study by Osmani (2007), around this time, argues that demand-driven factors are not important in explaining inflation in recent years. The author points out that the extra demand resulting from higher income and growth should not exert pressure on inflation since higher national income also entails higher national output.109 Therefore, additional demand should be matched by additional output. His argument is also based on the fact that rising income leads to falling relative price of food vis-à-vis luxury or manufactured products.110 The above discussion on growth-inflation relationship in Bangladesh based on descriptive statistics and findings from existing literature does not provide a convincing argument in favour of the IMF view. Fighting inflation without paying attention to its causes and keeping it at a low single digit level at all times may not ensure high long term sustained growth. The discussion paves the way for an empirical exercise which is carried out in the following two sections to further substantiate this view. 6.4 A Time Series Analysis of the Growth-Inflation Relationship This section attempts to look at whether there exists a long run equilibrium relationship between output and price level and if so, what kind of short run 109 In accordance with National Income Accounting Identity. Income elasticity of demand for food, according to Engel’s Law, is said to be inelastic and therefore as income rises, even if expenditure on food rises, the proportion of income spent on food decreases. This is expected to lower the relative price of food. However, the behaviour of the relative price of food as incomes and hence the demand for food increase will also depend on how fast the supply of food grows. Structuralists have long argued that if the agricultural sector lags behind other sectors in the process of economic growth, it may lead to an increase in the relative price of food, thereby triggering an inflationary spiral. 110 166 dynamics exists between growth and inflation. The annual data for Bangladesh is gathered from the World Development Indicators of the World Bank; and International Financial Statistics and World Economic Outlook of the IMF for econometric analysis.111 Considering the turbulent phase of the early half of the 1970s in Bangladesh, we feel a sample period between 1977 and 2009 is more appropriate to carry out this empirical investigation. Our first step, as is standard in any time series analysis, is to test for non-stationarity in the variables to avoid any spurious relationship. Empirical studies on unit root provide a whole battery of tests to identify nonstationarity. We rely on Augmented Dickey Fuller (ADF) test (Dickey & Fuller 1979, 1981) and Phillips Perron (PP) (Phillips & Perron 1988) test, most widely used in the literature. PP test performs better in the event of low frequency data as suggested by Choi and Chung (1995). The annual data used in this chapter fit into that category. Because PP test is supposedly superior to ADF test, if differences arise in terms of the results from the two unit root tests, we draw conclusion based on the PP test. Table 6.4 presents the unit root test results on real GDP (output level), consumer price index (price level), real GDP growth, and inflation (first difference of CPI). The findings show expected order of integration of the variables. Both output and price levels possess unit root and integrated of order one, in other words, they are nonstationary. On the other hand, their first differences – growth and inflation respectively – are found to be stationary. This is consistent with the idea that in a finite sample first differenced variables are expected to be stationary (since mean reversion process should take place in a small-finite sample). 111 The empirical work of this chapter started in 2010 when the latest information available was up to 2009. 167 Table 6.4: Unit Root tests with Real GDP, CPI, Growth, and Inflation, 1977–2009 Variables Augmented Dickey Fuller (ADF) Phillip Perron (PP) Tests Process Tests Statistics p-value Unit root Statistics p-value Unit root Intercept 4.77 (8) 1.00 Yes 6.85 1.00 Yes I (1) Intercept 1.61 (8) 1.00 Yes 1.22 0.99 Yes I (1) Intercept -3.19 (8) 0.02 No -3.20 0.02 No I (0) Intercept -1.66 (8) 0.74 Yes -2.65 0.25 Yes I (1) Intercept -1.12 (8) 0.69 Yes -4.54 0.00 No I (0) Intercept -4.57 (8) 0.00 No -6.71 0.00 No I (0) Intercept -2.80 (8) 0.06 Yes -5.04 0.00 No I (0) Intercept -2.25 (8) 0.44 Yes -5.86 0.00 No I (0) Real GDP with trend CPI with trend Real GDP growth with trend CPI inflation with trend Notes: 1. Both real GDP and CPI are in natural logarithm form. 2. The null hypothesis of ADF and PP tests is that the series has a unit root (in other words nonstationary). 3. Figures within parentheses in the ADF statistics indicate the number of lags taken to perform the test. 4. PP test is carried out by employing Bartlett Kernel estimation method with Newey-West Bandwith selector. 5. I(1), integrated of order one, and I(0), integrated of order zero, refer to non-stationary and stationary variables respectively. 6. Tests are carried out using EViews version 7. Since both real GDP and price level possess unit roots and they are integrated of the same order, I(0), it provides an opportunity to test for cointegration to find out if there is a long run equilibrium relationship between real GDP and price level. We employ Johansen Maximum Likelihood Cointegration Test (Johansen 1991) the most 168 widely used technique in recent literature in this regard.112 The bivariate cointegration test also involves the use of a dummy, DUM92 as an exogenous variable. The purpose of incorporating a dummy variable is to take into account the structural break in the growth rate, since the economy’s transition to democracy in the early 1990s.113 As discussed earlier, this transition is seen to have a significant impact on the economy. The election took place in early 1991 and because policies tend to affect the economy with lags we expect that the structural break takes place in 1992. Chow test (Chow 1960) for known dates confirms existence of a structural break in real GDP growth in the year 1992 (see Table 6.5).114 In other words, diagnostic test supports the structural break date based on historical evidence. So a shift dummy, DUM92, with value ‘0’ before 1992 and ‘1’ since 1992, is constructed and used as an exogenous variable in the cointegration analysis. Table 6.5: Structural Breakpoint Test for Real GDP Growth in 1992 Variable Test F-statistics P-value Real GDP growth Chow Test for known dates 8.347 0.001 Notes: The null hypothesis – no break in the specified breakpoint – is rejected at 5 per cent significance level according to the p-value. Test is carried out using EViews version 7. The cointegration test results are reported in Table 6.6. Based on Trace and maximum eigenvalue tests, cointegration results indicate the presence of no cointegrating vector at 5 per cent level of significance for the sample period 19772009. In both cases, the value of test statistics is less than the critical value at 5 per cent significance level and therefore we cannot reject the null hypothesis of no cointegrating vector. It means that we do not find a stable long run equilibrium relationship between real GDP and price level in Bangladesh, once the structural change in the economy is taken care of. It contradicts with the findings of Ahmed and Mortaza (2005) which did not take the structural break issue into account. The finding casts doubt about the notion advocated by the IMF that stable price will 112 Johansen approach performs better than Engle-Granger two-step approach (Engle & Granger 1987) in identifying cointegration. For a detail discussion on the comparative advantage of the Johansen approach see Asteriou and Hall (2007, Chapter 17). 113 Inference on tests of cointegration may be affected by structural change (Maddala & Kim 2000, p. 410). Out analysis is unique in this sense since most of the studies in the context of Bangladesh do not take into account the issue of structural break. 114 Chow test is for stationary variables and a single break (Maddala & Kim 2000, p. 390). 169 ensure higher output and growth in the long run as they do not seem to move together over a long period of time. The test of cointegration also helps us determine the modelling issue whether to use a restricted model such as error correction model (when variables are cointegrated) or an unrestricted model (in the absence of cointegration) in the subsequent analyses. Table 6.6: Cointegration Test Results Null Alternative hypothesis hypothesis Test statistics P-value 5 per cent Conclusion critical value Trace test r=0 r=1 18.92 0.28 25.87 No r≤1 r=2 6.82 0.36 12.51 cointegrating equation Maximum eigenvalue test r=0 r=1 12.09 0.40 19.38 No r≤1 r=2 6.82 0.36 12.51 cointegrating equation Notes: Results reported above are based on the assumptions of a constant and a linear trend in the data with lag length 3. According to AIC and SIC a lag length of 1 is sufficient. A dummy variable, DUM92 explained in the text is used as an exogenous variable. MacKinnon-Haug-Michelis (1999) pvalues are used. In short, the analysis of this section does not provide strong evidence of a long run relationship between output and price level in Bangladesh. The apparent lack of a long-run relationship between inflation and growth may be due to incorrect modelling. That is, the actual relationship between the two variables may be nonlinear, while the cointegration analysis uses a linear equation. Therefore, the next section examines the possibility of existence of a non-linear relationship and if it does, at what level inflation might turn harmful for economic growth, assuming a nonlinear relationship between the two. 6.5 Identifying the Threshold Level of Inflation in Bangladesh In this section we try to estimate the threshold level of inflation using a nonlinear model. It is reasonable to assume a nonlinear relationship between growth and 170 inflation since existing literature (see Chapter 4) and the descriptive statistics (particularly correlation) on Bangladesh data provide support for this. Inflation may have a positive or a negative association with growth. The previous chapter using a panel study also essentially uses this nonlinear framework to estimate the threshold levels. So, we model economic growth as a function of inflation in a quadratic specification similar to Chowdhury and Siregar (2004). However, the bivariate relationship used in their estimation poses serious limitations. This is because growth-inflation relationship is conditioned upon other variables. So taking this into account, we consider some other covariates, pertinent to the Bangladesh economy, in determining the impact of inflation on growth. The specification of the model is given below: yt = α + β1 π + β2 π2 + ∑ i Zt + εt (6.1) where, yt is economic growth measured by the first difference of natural logarithm of real GDP (measured at constant 2000 US dollar); π is the inflation rate measured in terms of first difference of natural logarithm of the consumer price index (CPI); Zt includes the control variables; α is a constant term; and εt is the classical error term. The squared term of inflation captures the nonlinearity into this relationship and since we assume it to be inverted U shaped we expect β1 > 0 and β2 <0. Applying the first order condition, the turning point is measured as -β1/2 β2. To avoid any spurious relationship, the control variables (Z) include variables in their first differenced form. Our unit root test results in the previous section already prove that both growth and inflation are stationary variables. Z considers a number variables, seen as appropriate to influence the growth-inflation relationship and some of which are used in the panel study (Chapter 5) as well. The variables in their growth form are real household consumption per capita (HCONPC), broad money supply (M2), government consumption expenditure as a percentage of GDP (GOVCON), trade openness (OPEN) which the summation of export and import as a percentage of GDP; agriculture value added at constant 2000 US dollar (AGR), oil price index (OILPR) (UK Dated Brent Crude Oil Index, 2005=100), commodity price index (COMPR) (World Bank Non-Energy Commodities for Lower Middle Income Countries Index, 2005=100), remittances as a percentage of GDP 171 (REMGDP)115, and a dummy variable DUM92 to take into account the structural break in the dependent variable as discussed above. In addition a lagged dependent variable is also considered as a regressor in the growth equation to allow for any long term impacts. We employ OLS technique to regress growth on inflation, conditioned upon other covariates. Table 6.7 records the results of the OLS estimations using equation 6.1. The estimation is carried out in two steps, following a general-to-specific approach. First, growth is regressed on all the variables discussed above, findings of which are presented in Model 1 in Table 6.7. Next, we identify the conditioned variables which are significant at least at 10 per cent level in Model 1. We again regress growth on inflation and these conditioned variables in Model 2. In other words, the insignificant conditioned variables are omitted to create a more parsimonious model to determine the impact of inflation on growth. Table 6.7: Impact of Inflation on Growth, Conditioned upon Other Covariates, 19772009 Dependent variable is real GDP growth Model 1 Model 2 Variable Coefficient SE P-value Coefficient SE P-value VIF Inflation 0.272 0.099 0.012 0.295 0.073 0.001 12.641 Inflation -0.772 0.648 0.248 -0.934 0.482 0.064 14.269 Turning point 17.628 Growth (-1) 0.048 0.077 0.541 ∆ ln HCONPC 0.113 0.039 0.009 0.129 0.032 0.000 1.655 ∆ ln M2 0.017 0.013 0.197 ∆ ln GOVCON 0.024 0.011 0.051 0.022 0.010 0.035 1.219 ∆ ln OPEN -0.001 0.010 0.934 ∆ ln AGR 0.256 0.042 0.000 0.265 0.038 0.000 1.545 ∆ ln OILPR 0.007 0.003 0.053 0.006 0.003 0.043 1.060 ∆ ln COMPR -0.005 0.008 0.557 ∆ ln REMGDP -0.007 0.006 0.272 DUM92 0.016 0.002 0.000 0.015 0.002 0.000 1.608 Constant 0.009 0.005 0.105 0.012 0.003 0.000 2 15.775 115 The remittance-variable, not considered in the growth regressions of Chapter 5, is particularly important in the context of Bangladesh. Bangladesh is one of the leading remittance recipient countries, with remittances equivalent to 12 per cent of the GDP (Chowdhury, MB 2011, p. 2600). 172 Sample period 1978-2009 1977-2009 No of observation 32 33 Estimation method Ordinary Least Squares Ordinary Least Squares R-squared 0.943 0.928 Adjusted R-squared 0.907 0.907 S.E. of regression 0.004 0.004 Log likelihood 138.253 134.636 F-statistic 26.088 44.368 Prob(F-statistic) 0.000 0.000 Akaike criterion -7.828 -7.915 Schwarz criterion -7.233 -7.548 DW statistics 1.892 2.276 JB Normality test 0.581 0.503 p-value BG Serial correlation (1-4) LM test 0.747 0.777 4.927 7.322 p-value BPG heteroscedasticity test 0.294 0.119 6.695 9.116 p-value 0.877 0.244 ARCH (1-4) test 2.174 5.063 p-value 0.703 0.28 Stable Stable Diagnostic tests Stability test CUSUM Test CUSUMSQ Test Stable Source: Author’s estimations using EViews version 7. Stable Notes: 1. Estimations are performed using the model specified in equation 6.1. 2. CPI inflation is used as a measure of inflation. 3. Turning points of inflation are calculated by examining the partial derivatives of real GDP growth with respect to inflation – the coefficient of the linear term of inflation divided by 2 times the coefficient of the squared term of inflation. 4. Growth (-1) refers to real GDP growth lagged in one period. 5. ∆ and ln indicate first difference operator and natural logarithm of the variables respectively. 6. The acronyms of the variables are explained in the text. 7. SE stands for standard error of the coefficient and the respective p-value indicates the level of significance. For instance a p-value less than 0.05 indicates that the coefficient is significant at 5 per cent significance level. 173 8. Null hypotheses of JB normality test, BG serial correlation LM test, and BPG heteroscedasticity test are that the residuals are normally distributed, no residual autocorrelation, and the residuals are homoscedastic respectively. 9. Null hypothesis of ARCH (autoregressive conditional heteroscedasticity) test states no ARCH effect in the residuals. In Model 1, although the linear term of inflation is significant at 5 per cent significance level, the squared term is not. Both terms show the expected signs, positive and negative, respectively, confirming the assumption of an inverted U shaped relationship. The turning point in this case is around 17 per cent. Amongst the conditioned variables, growth rates in HCONPC, GOVCON, AGR, OILPR, and dummy variable DUM92 are found to be significant at 10 per cent significance level. On the other hand, growth lagged in one period and growth rates in M2, OPEN, COMPR, and REMGDP are found to be insignificant.116 All the significant conditioned variables (growth rates in HCONPC, GOVCON, AGR, and OILPR) have a positive impact on growth. In a developing country such as Bangladesh, household and government consumptions are likely to have a strong influence on aggregate demand, thereby enhancing economic growth. The positive effect of agriculture on growth is also no surprise since we know the sector is still the mainstay of the Bangladesh economy. What is perhaps not expected is the positive sign of the coefficient of growth in oil price index. Rise in Oil price is supposed to have a negative shock on growth. However, we know that governments in Bangladesh, through the use of subsidy do not always pass on the changes in oil price into the domestic economy immediately. Therefore, economic growth may not be affected by changes in oil prices. Moreover, because natural gas is in many respects a substitute for oil, increases in oil prices lead to increases in the demand for natural gas which could be expected to boost economic activity in Bangladesh. We pay greater attention to the findings from Model 2 which is more parsimonious. First, both linear and squared terms of inflation are significant at 10 per cent and 116 The coefficient of remittances is found to be negative and insignificant. It is at odds with the idea that the flow of remittances is supposed to influence the growth of the economy positively. Perhaps we should bear in mind that the flow of remittance in Bangladesh takes place irrespective of financial development or higher investment opportunities of the country. Families of migrant workers, 90-95 per cent, are mostly poor living in villages (see Chowdhury, MB 2011, p. 2605). They are highly dependent on the remittance income which may not have productive use and therefore having a positive impact on growth. 174 have signs consistent with the assumption. The turning point calculated from the coefficients is at 15 per cent, approximately. This result matches our expectation of a double digit figure for the threshold level of inflation in Bangladesh. The figure is higher than the single digit level of 6 per cent suggested by Ahmed and Mortaza (2005). In a developing country such as Bangladesh, low productivity and structural bottlenecks for example poor infrastructure and energy shortage are key impediments for growth. Our finding suggests that inflation up to 15 per cent should not impose a threat to economic growth. This is consistent with all the major studies done in the context of developing countries – the threshold broadly remaining between 11 and 18 per cent. The coefficients of other conditioned variables in Model 2, growth rates in HCONPC, GOVCON, AGR, OILPR, and DUM92 provide a similar result to that of Model 1. It is interesting to note that government consumption is not found to have a detrimental effect on growth in any of the models. Again this contradicts the conventional idea advocated by the IMF that government’s intervention essentially creates more harm than good to the economy. It is also important to note that higher household consumption, through reducing poverty and raising productivity and aggregate demand, has a growth enhancing effect. The importance of agriculture, the production side of the economy, is also evident from the findings. The overall evidence suggests that broader issues such as higher domestic consumption leading to poverty reduction, greater role of government in boosting aggregate demand, and rising agricultural production to attain self-sufficiency in food production lead to a higher growth rate. These issues, therefore, are more important for economic growth than mere control of money supply growth in keeping inflation at a low single digit level. To check the reliability of the above estimates a number of diagnostic tests are performed. Both models pass comfortably in terms of all diagnostic tests. The adjusted R2 is high in both cases suggesting that the models have a good fit. However, Model 2 has higher F statistics and has more significant explanatory variables than Model 1. Values from information criteria, such as Akaike Information Criterion (Akaike 1974) and Schwarz criterion (Schwarz 1978) reported here, also indicate a preference towards Model 2 since it has lower values of all those 175 statistics (for model selection based on information criteria see Asteriou & Hall 2007, p. 66). The Durbin-Watson (DW) test statistic for Model 2 is close to the value 2 suggesting no residual autocorrelation in the regression model. We should be cautious about interpreting the DW test statistic for Model 1 since this test statistic should not be used when a lagged dependent variable is used as a regressor in the model (see Vogelvang 2005, Chapter 6, p. 125). In terms of the standard residual tests to determine normality, serial correlation, and heteroscedasticity, both models perform well. P-values for these tests are greater than 0.05 and therefore we cannot reject the respective null hypotheses at 5 per cent significance level (see notes 8 and 9 in Table 6.7). This means that the residuals are normally distributed and do not suffer from problems such as autocorrelation and heteroscedasticity. We also check the stability of the estimates over time using the CUSUM and the CUSUM squares tests proposed by Brown Durbin and Evans (1975). No significant break is observed in both models. The test statistics derived from the regressions remain within the 5 per cent confidence bands. This indicates that the estimates from the corresponding regressions are stable over time.117 Finally, since the estimates of Model 2 are more important we check for the multicollinearity problem in this case (reported in the last column of Table 6.7). We use the formal procedure to detect multicollinearity variance inflation factors (VIFs) of the coefficients. A rule of thumb is that a VIF of 10 and greater is a problem (see for instance Verbeek 2012, Chapter 2, p. 45). The VIFs of all explanatory variables, except inflation terms, in Model 2 are well below 10 suggesting that the estimates to do not suffer from multicollinearity problem. We ignore the VIF values for inflation terms as it is obvious that linear dependency would exist among them.118 Some of the post-estimation facts from Model 2 are illustrated in the following figures. Figure 6.2 shows the residuals from the regressions while Figure 6.3 depicts a scatter plot between residual growth and inflation controlled for the other 117 Since we are more interested in the results from Model 2 only diagrams of the parameter stability tests for this model are shown in Figure 6.4 and 6.5. 118 Our variable transformation techniques also help ensure the remedies for problems such heteroscadasticity and multicollinearity. These are standard procedures employed in empirical work. For instance, in the case of time series data, log-transformation and taking first differences often take care of heteroscadasticity and multicollinearity problems (see Maddala & Lahiri 2009, Chapter 5, p. 224 and Chapter 7, p. 300, respectively). 176 covariates. Observation from the scatter plot does not provide a clear evidence of a negative relationship between growth and inflation. The parameter stability tests in the case of Model 2 are shown in Figures 6.4 and 6.5, interpretations of which are discussed above. Figure 6.2: Growth Rate, Actual and Fitted, 1977 – 2009 .08 .06 .04 .008 .02 .004 .00 .000 -.004 -.008 78 80 82 84 86 88 90 Residual 92 94 96 Actual 98 00 02 04 06 08 Fitted Note: The graph is drawn based on the results of Model 2 presented in Table 6.7. 177 Figure 6.3: Scatter Plot between Unexplained Growth and Inflation, 1977 – 2009 .008 .006 Residual of growth .004 .002 .000 -.002 -.004 -.006 -.008 .00 .04 .08 .12 .16 .20 Inflation conditioned upon other covariates Note: The scatter plot is drawn based on the results of Model 2 presented in Table 6.7. Figure 6.4: Parameter Stability Test using CUSUM Test 12 8 4 0 -4 -8 -12 1994 1996 1998 2000 CUSUM 2002 2004 2006 2008 5% Significance Note: Parameter stability test for Model 2 detailed results of which are presented in Table 6.7. 178 Figure 6.5: Parameter Stability Test using CUSUMSQ Test 1.6 1.2 0.8 0.4 0.0 -0.4 1994 1996 1998 2000 CUSUM of Squares 2002 2004 2006 2008 5% Significance Note: Parameter stability test for Model 2, results of which are presented in Table 6.7. In empirical time series analysis, it is customary to see the response of one variable to an impulse (exogenous shock or innovation) in another variable in a system. Our final task is to carry out an impulse response analysis between growth and inflation in a vector autoregressive (VAR) framework. In this regard, Lutkepohl (2005, p. 62) points out a major limitation that arises because of the “potential incompleteness” in the system. This incompleteness may arise as a result of omitting important variables from the system thus leading to major distortions in the responses. Since we realise that the growth-inflation relationship is conditioned upon other important variables, we use the control variables of Model 2 to avoid this problem. Table 6.6 illustrates these impulse responses. 179 Figure 6.6: Impulse Response Analysis for Growth and Inflation Response to Cholesky One S.D. Innovations ± 2 S.E. Response of growth to growth Response of growth to inflation .0100 .0100 .0075 .0075 .0050 .0050 .0025 .0025 .0000 .0000 -.0025 -.0025 -.0050 -.0050 1 2 3 4 5 6 7 8 9 10 1 2 Response of inflation to growth 3 4 5 6 7 8 9 10 9 10 Response of inflation to inflation .04 .04 .03 .03 .02 .02 .01 .01 .00 .00 -.01 -.01 -.02 -.02 -.03 -.03 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 Notes: 1. Impulse response between growth and inflation conditioned upon variables used in Model 2 in Table 6.7. 2. A VAR lag length of 3 based on information criteria is used. 3. The graphs are generated using EViews version 7. In the four graphs of Figure 6.6 we are more interested to see the cross-responses than own-responses. The upper right graph shows the response of growth for a shock in inflation. The response of growth to an increase in inflation is insignificant as the confidence band touches zero. The response is positive but very weak from year 1 to about year 2. After that inflation seems to have a negative effect on growth which starts to fade away after year 5. On the other hand, in the lower left graph, an increase in growth positively affects inflation after one year but this effect becomes insignificant after about 6 months as the confidence band touches zero.119 119 This also tells us that current growth does not seem to affect current inflation causing a possible reverse causality problem in estimating equation 6.1. According to the responses, growth only starts to affect inflation after 1 year and the effect continues for about 6 months. 180 These findings from the multiplier effects clearly find support from existing theories explaining the growth-inflation relationship discussed in the opening chapters. The initial positive influence of inflation on growth, albeit statistically insignificant and weak, finds support from the Kalecki-Keynesian schools arguing in line with the proposition of transferring resources through forced savings and inflation tax. One can also explain the positive growth effect using Kaldor’s argument of transferring income to profit earners with higher propensity to save and Tobin’s argument of higher capital intensity as real return to financial investment declines. The negative impact of inflation on growth after about 2 years, continued for the next 3 years approximately, can be supported by the classical-mainstream view. This classical view, in brief, brings in the idea that inflation retards growth through its distortionary effects on investment.120 We must note that these dynamic influences of inflation on growth are interesting – complex and surely not linear – but unfortunately the results are not statistically significant. On the other hand, the positive effect of growth on inflation, after 1 year continued to be significant for the next 6 months, shows that inflation is an inevitable companion to growth. Again this is explained by the Structuralist School which says that inflationary pressure is caused by the process of growth due to various structural rigidities within the economy. Together the findings clearly reject the views expressed by Hayat and Kalirajan (2009) who advocate that growth-inflation relationship is unidirectional and linear and decline in inflation from any level is beneficial for growth. To summarise the findings of this section, we can point out two important facts. First, there is no evidence that inflation beyond 5 per cent level will start to have a detrimental effect on growth in Bangladesh. Inflation threshold is found at around 15 per cent for the country. Second, in an underdeveloped country such as Bangladesh, inflation does not seem to be the key factor causing low economic growth. The dynamic impulse response analysis shows that inflation starts to pick up when the economic growth is taking place. This inevitable consequence should not alarm the government immediately. The choice of anti-inflationary measure should be 120 A relatively older study by Mallik and Chowdhury (2001) comprising four south Asian countries including Bangladesh provides some empirical support to our findings. According to the authors, moderate inflation is found to be beneficial for growth. However, as discussed in Chapter 2, they did not provide any estimate of this moderate level of inflation. 181 employed with caution as in doing so a low inflation target may be achieved but at the cost of hurting growth.121 6.6 Conclusion The key estimation result presented in this chapter provides strong evidence that the inflation threshold for Bangladesh is not as low as 5 per cent. Based on annual data between 1977 and 2009 we find the threshold level is around 15 per cent. This evidence is consistent with the findings of major cross-country panel studies on developing countries, discussed in Chapter 2 and Chapter 5, but differs from the two recent country-specific studies on Bangladesh by Ahmed and Mortaza (2005) and Hayat and Kalirajan (2009) advocating very low inflation thresholds. Similar to the findings of panel study in Chapter 5, the threshold estimate of this chapter in a country specific case also questions the conventional wisdom of keeping inflation at a low single digit level to attain high economic growth. Bangladesh has shown promising and steady growth since its transition to democracy in the early 1990s. This has taken place in spite of several bouts of inflation and greater inflation volatility during this time. This steady growth performance, however, is not enough in the face of poverty rate still around 31.5 per cent. The country needs at least a growth rate of 8 per cent for further poverty reduction. Institutions such as the World Bank have acknowledged it (see World Bank 2012). The World Bank also admits that Bangladesh’s potentials are underutilised. Yet our discussion in this chapter shows that the IMF imposes conditions, while sanctioning loans to Bangladesh, with a view to fighting inflation first. This policy prescription is similar to what we have seen in Chapter 3 in the content analysis of Article IV Country Consultation Report by the IMF staff. That report, in the aftermath of the Great Recession, also gives priority to rein in inflation instead of outlining strategies for economic growth and rapid poverty reduction. The country has passed a couple of decades with a priority to keep inflation at a low single digit level, but failed to achieve the required level of growth. Perhaps growth fails to reach a higher trajectory because of strict anti- 121 As former World Bank economist Surjit Bhalla says “Let’s kill GDP, inflation will fall” (The Financial Express, 15 October, 2011; available via the internet at http://www.financialexpress.com/news/Column---Let-s-kill-GDP--inflation-will-fall/860148. 182 inflationary measures such as interest rate hikes. The domestic entrepreneurs feel strongly that by raising operational costs this acts as a major obstacle to grow at 8 per cent rate. On the other hand, the acceptance of a moderate inflation rate would have widened the government’s fiscal and policy space, needed to tackle the country’s infrastructure bottlenecks. 183 Chapter 7 Conclusion This thesis provides an extensive study of the growth-inflation relationship in selected developing countries of Asia. The study questions the conventional wisdom of macroeconomic policies targeting low inflation in developing countries. Following several bouts of high inflation together with growth stagnation during the 1970s and 1980s, macroeconomic policies in developing countries have started to place too much emphasis on attaining price stability by keeping inflation low. The IMF also strongly advises the developing countries to maintain inflation within 5 per cent. Because of the conditions attached as part of the loan-agreement, developing countries often do not have much option but to pursue this low inflation targeting policy. This policy, as the conventional wisdom suggests, is expected to stop acceleration in inflation and bring benefits in terms of sustained long term economic growth. There is a sense of apprehension that inflation accelerates when it goes beyond 5 per cent, thus leading to a crisis. It is also believed that once inflation accelerates, it is quite difficult to control it and often becomes too costly in terms of high unemployment and lost output. Therefore, even at the cost of immediate adverse effects on growth, the policy of fighting inflation first, keeping it as low as 5 per cent, is advised. Moreover, the IMF seems to believe that this one policy fits all works the best. This is reflected in its policy prescription contained in the Article IV consultation irrespective of country-specific circumstances. There is little evidence that this perception has changed in the aftermath of global crisis of 2007-2009. Leading economists, even within the Bretton Woods institutions, have argued about the limitations of this conventional idea after the crisis. They have also raised concerns about deteriorating condition of poverty throughout the developing world despite the policy adherence towards low inflation. In spite of the need for a change in this policy, suggested by the leading economists, the IMF seems to continue with its pre-crisis policy prescription. Therefore, keeping a watchful eye on inflation still receives priority in macroeconomic policy management in developing economies. 184 A number of studies in recent years have argued against this wisdom of macroeconomic policies targeting low inflation, both empirically and analytically. They have revealed that too low an inflation is inimical to growth and how policies aiming to achieve this can create an inflation trap, ultimately having an adverse impact on poverty. However, a systematic empirical research on Asian developing countries, from Asia-Pacific region in particular, is limited. In this thesis, we make an attempt to fill this gap. This thesis investigates the nature of growth-inflation relationship in developing countries of Asia, questioning the validity of the widely held belief that low inflation is beneficial for economic growth. This research demonstrates that inflation beyond 5 per cent is not necessarily detrimental for economic growth in our sample of Asian developing countries. Moreover, the inflation threshold is not fixed at a particular level and it varies according to levels of economic development. The contribution of the thesis, as such, is twofold. First is to provide a unique contribution to the literature pertaining to growth-inflation nexus. Second, the findings of our study have profound implications for macroeconomic policy making in countries that still follow low inflation targeting policies. As mentioned above, despite the dominance of low inflation policy, there is much evidence that the condition of poverty in the developing countries has not improved significantly over the last few decades. In fact, it has deteriorated in certain cases. There is also historical evidence that developing countries can grow under relatively moderate inflationary environment. On the other hand, as mentioned above, very low inflation can inhibit economic growth in developing countries. Attempts to keep inflation at low single digit levels (for example, below 5 per cent) may lead to a stabilisation trap – low inflation and inadequate growth for reducing poverty, essential for developing countries. Motivated by these facts, we ask three important questions: (a) is there a threshold level of inflation at 5 per cent for developing countries of Asia beyond which inflation is harmful for economic growth?; (b) does the threshold level vary depending on the level of development within the developing countries of Asia with poorer countries tend to have higher threshold?; and (c) is there a consistency between the need for a change in the low inflation policy suggested by the leading economists and country policy reports by the IMF? 185 The thesis is organised in seven chapters. Chapter 1 introduces the issues and sets the research agenda while Chapter 7 provides the concluding remarks. The rest of the chapters are devoted to research investigation. Our investigation and analyses to address the issues took place within two broad frameworks. In the first case, we provided a critical review of existing literature, including IMF Article IV country consultation reports, in Chapters 2, 3, and 4. Chapter 2 mainly reported an in-depth content analysis on the IMF country consultation reports with a view to examining whether there has been a change in the perception of policy advice by the IMF in the wake of the global crisis of 2007-2009. Chapter 3 critically reviewed the recent cross-country studies relating to inflation-growth nexus in developing countries to find broad evidence on inflation threshold beyond which inflation is harmful for economic growth. Chapter 4 sought to explain the nature of growth-inflation nexus by reviewing country-specific literature on selected developing countries of Asia. This review of existing studies revealed evidence from selected individual countries of Asia-Pacific region in line with our research interests. In the second part of our investigation, we provided an empirical analysis to find evidence on the research questions. This has been dealt with in Chapters 5 and 6. Chapter 5 employed crosscountry panel estimation techniques to show empirical evidence from selected Asian developing countries. Chapter 6, on the other hand, examined a country-specific case in the context of Bangladesh. Findings from both critical reviews and econometric analyses tell us that the empirical justification to keep inflation as low as 5 per cent is rather weak. Our empirical evidence suggests that inflation thresholds for these developing countries range between 8 and 15 per cent, well above the level, 5 per cent, advocated by the IMF. Besides, poorer countries tend to have a higher threshold, providing convincing evidence against one size fits all policy and that country-specific circumstances matter. We also find little evidence of a change in policy prescription by the IMF in the post-crisis era. Our content analysis using the country reports provides strong support on this. The results from our analyses broadly support earlier findings that threshold level of inflation is well above 5 per cent for developing countries and the level also differs with higher inflation tolerance observed for poorer countries. In the following paragraphs, we summarise the issues discussed in each chapter. 186 In Chapter 1, we introduced the issues relating to our investigation. We provided an overview of the IMF policy supporting low inflationary environment in the developing countries and the likely consequent danger these countries might face. The argument put forward by the IMF, in short, is that inflation acts as a tax for the poor and also raises uncertainty which, in turn, is bad for investment and growth. This persuades the IMF to advocate for keeping inflation at 5 per cent or below in developing countries. An inflation rate above this is believed to be bad for growth and hence poverty reduction. Maintaining this nominal target is believed to be safe in terms of preventing the inflation rate from accelerating. In order to achieve this, the countries need to rely heavily on tight monetary policy and reining in fiscal deficits. The likely adverse consequence of pursuing such tight policies is that they restrict the development prospects of these countries. High interest rates raise the cost of borrowing for business and thus are not beneficial for private sector growth. This is an important consideration given the fact that World Bank’s Enterprise Surveys show access to finance is among the top-5 business impediments for 93 per cent of the countries in Asia and the Pacific. Access to finance is critical for small and medium size enterprises (SMEs) as well as for agriculture, which depend solely on the banking sector for external financing. Policies aiming at containing inflation at a very low level also reduce the ability of fiscal policy to perform its developmental role. In many cases, governments reduced fiscal deficits and public debt by cutting vital infrastructure investment, especially in agriculture as noted in the United Nations Economic and Social Survey of Asia and the Pacific 2013. In addition to its direct adverse impact on productivity, insufficient infrastructure adds to the cost of business. As a result, insufficient private sector investment for development is a common trait in the developing countries. Reduced government spending coupled with insufficient investment often resulted in a stabilisation trap, especially in countries that were under IMF’s support programmes. This was particularly evident in the wake of the Asian financial crisis in 1997-98. Malaysia and Thailand recovered quickly by following policies contrary to the IMF advice while Indonesia’s recession worsened due to contractionary policies of the IMF.122 Interestingly, one study (Chowdhury 122 For a review of Indonesia’s policy dilemma after the crisis, see Chowdhury and Siregar (2004). Chowdhury (2004) provides an assessment of monetary policy in selected Asian countries in the wake 187 1999) traces the seed of the Asian crisis that originated in Thailand to its stabilisation programme with the IMF in the late 1980s. Thailand achieved programme conditionality by drastically reducing public infrastructure investment, which affected adversely its international competitiveness as manifested in the rise in current account deficits, making its currency ideal for speculative attacks. The classic example of severe development cost of over-zealous restrictive policies aimed at stabilisation is Argentina in the 1990s. In order to combat against the hyperinflation of the 1980s, Argentina had to undergo the IMF stabilisation program in the 1990s. The policies were so stringent that inflation came down to a very low single digit level, at 1.5 per cent annually. But the pursuit of low inflation target had a severe cost in terms of growth, unemployment, and poverty. The tight macroeconomic policies that the country continued to implement led to a “deflationary spiral”. It led the economy enter into a recession, resulting in unemployment rate to reach at a staggering rate of around 17 per cent by the mid1990s. Poverty rate (head-count ratio) showed a 13 percentage point rise, reaching at 34.3 per cent in 2002. Restrictive macroeconomic policies also have negative magnified impact on growth in the event of a supply shock. This is particularly true in the case of developing countries which are prone to adverse supply shocks such as social unrests and natural disasters. As a negative supply shock results in high inflation, the IMF usually advises to pursue a contractionary monetary and fiscal policy or not to accommodate inflation. Such a policy stance is believed to help tame inflation expectations and thus acceleration in inflation. However, the strict adherence to a fixed nominal target in all circumstances makes macroeconomic policies pro-cyclical and leads to a contraction of aggregate demand causing output to fall even further. Therefore, tight macroeconomic policies to achieve low inflation can have inadvertent consequences on economic growth in developing countries in a number of ways. Moreover, this perception on inflation by the IMF does not seem to be in agreement with its own mandate. According to the agreement stated in the IMF’s Article IV, member of Asian financial crisis of 1997-98 and finds that monetary policy in most of these countries were restrictive and caused a stabilisation trap. 188 countries are supposed to aim at policies to foster orderly economic growth with reasonable price stability, considering the country-specific circumstances. Chapter 1 also provided some empirical evidence, both in broad and country-specific contexts. A glimpse on growth-inflation relationship from 150 developing countries, using a scatter plot, revealed that the relationship is nonlinear, showing a plateau rather than a sharp cliff edge between 5 and 15 per cent of inflation. This general overview shows that the fear of accelerating inflation and a sharp decline in growth when inflation reaches above 5 per cent is unsubstantiated. Country-specific examples from two countries in the Asia-Pacific region, South Korea and Indonesia, provided significant evidence that they grew rapidly amidst high inflation particularly during their early stage of development. The experience of South Korea, which is no longer a developing country now, tells us that in the 1960s and 1970s South Korea was more interested in growth rather than attaining price stability in a narrow sense. Inflation at that time was high, above 20 per cent. But a strong growth rate of 8 per cent, at the same time, encouraged South Korea to adhere to growth promoting policies by allowing both monetary and fiscal policies to play their developmental role, supported by other policies dealing with exchange rate, foreign direct investment, and industry policy. The consequence was that its growth was durable and inflation did not accelerate. A similar story from Indonesia in the 1970s is also evident. The economy grew at around 7.7 per cent even though inflation was around 17 per cent. In the initial phase of development of these countries, inflation appeared to be an efficient way to mobilise resources for capital accumulation and investment needed for rapid economic progress. The theoretical debate surrounding the growth-inflation relationship and broad empirical evidence from cross-country studies were also discussed in Chapter 1. It recounted the Kalecki-Kaldor-Keynesian perspective that moderate inflation may be beneficial for growth and structural transformation for developing economies. Inflation helps redistribute income conducive for raising savings propensity needed for higher investment; it also transfers resources to the government to allow necessary public investments. The chapter also revisited the neoclassical theories which show that inflation is harmful for growth. This is because both consumption and investment decline as a result of fall in real money balances and increased 189 uncertainty, respectively. In an open economy, inflation also leads to real appreciation of domestic currency, thus making exports less competitive. The issue of Phillips curve in explaining the debate over inflation and employment (or growth) was also discussed in the chapter. It explained the two opposing views (originating in the context of Indian economy) of why expansionary policies may cause inflation in the developing economies. One view is that because the supply in developing countries is inelastic due to structural bottlenecks, an increase in aggregate demand as a result of expansionary policies will only cause inflationary pressure. The opposing view argues that it is not the supply rigidities but the low subsistence real wages of workers that cause inflation as a result of the expansionary policies. Since, real wages in developing countries are already at a very low level and cannot go down any further, increases in prices as a result of expansionary policies have to be compensated by raising the nominal wages. Because real wages do not change, i.e. due to real wage resistance, employment level determined by it does not change either. On the empirical side, a brief discussion tells us that most studies on developing countries find a threshold level of inflation above 5 per cent. These empirical works are carried out both inside and outside the IMF. The chapter also outlined the influential structuralist-monetarist debate, originating in the 1950s in the context of Latin American countries. The empirical studies based on the structuralist argument show that causality between inflation and growth is bidirectional. Theoretically aligned mainly towards Keynesianism, they argue that inflation promotes economic growth and is an inevitable companion to it. The latter fact is because of the structural rigidities in developing economies for which supply cannot keep pace with rising demand (resulting from growth of real incomes). Besides, inflation that results directly from economic expansion may not create any significant barrier to expansion. On the other hand, empirical literature, supporting the monetarist view, show that inflation causes economic growth negatively. Their theoretical argument behind this empirical finding is predicated on neoclassical theories stated above. Finally, Chapter 1 concluded with the discussion on the research questions and outline of the way in which we examined the issues in this thesis. 190 Against the backdrop of global economic crisis of 2008-2009, leading economists, both inside and outside the IMF, called for a need to revisit the traditional macroeconomic tenets, particularly the wisdom of low inflation policies. We wanted to understand the nature of policy advice by the IMF and to what extent it reflected the need for a change in the aftermath of the crisis. To achieve this objective, in Chapter 2, we selected IMF Article IV country consultation reports and provided an in-depth content analysis. Our intention was to reveal the experiences from some of the growing developing countries of Asia-Pacific region. To this end, we selected 12 countries mostly from South and East Asia depending on the availability of the IMF reports issued around the time of the crisis. These countries are Bangladesh, Bhutan, Cambodia, China, India, Indonesia, Lao PDR, Malaysia, Maldives, Nepal, Pakistan, and The Philippines. We noticed enough variations in the levels of economic and social developments amongst these countries. For instance, Bangladesh, a small country, initially followed a socialist path but quickly moved into a more market oriented system which resulted into a rising middle class. On the other hand, Nepal, also a small country, but is land-locked geographically and is still emerging out of its social conflicts. Despite these differences in the country characteristics, policy advice from the reports shows that containing inflation at a low single digit level receives priority in general. In other words, there is little evidence that country-specific circumstances have been taken into consideration by these reports. For example, in Bangladesh, although inflation was below 6 per cent in 2009, the IMF wanted the government to become cautious about an increase in inflation. The policies were considered too accommodative in view of the domestic conditions. This suggestion was made when economic growth was around 5 per cent, well below 8 per cent that is required for the country’s rapid development. The example from Nepal shows that the IMF pays similar attention to inflation situation, giving less priority to economic growth. Nepal is also one of the poorest countries of the world like Bangladesh but its socioeconomic characteristics are quite different from that of Bangladesh as noted above. Inflation in Nepal was around 11 per cent in 2009 and the IMF report stated that loose monetary policy was responsible for this double digit inflation. This cautionary note came when the country was still struggling with a low growth 191 performance at around 4 per cent. In both cases, the perception of keeping inflation below 5 per cent is evident, despite the differences in country-specific characteristics. In addition, historical experience from these countries shows that inflation, in most cases, remains well above 5 per cent without affecting their growth performance. Given the findings from the analysis, we have provided a critique on the continued policy advice by the IMF from three directions. We found inconsistency at all three levels – (a) with the need for a change suggested by the IMF’s own senior economists, (b) with the historical data on inflation and growth, and (c) findings from the literature in general. Our content analysis showed that the IMF policy still tends to follow a pre-set view with predictable conclusions that do not allow for alternative perspectives. The analysis, in this chapter, reaffirmed the need for a change in the narrow policy advice by the IMF and put it firmly on its already stated principle of fostering economic growth with reasonable price stability. To examine if there is strong empirical evidence of an inflation-threshold at 5 per cent for developing countries we looked at the existing literature based on crosscountry panel studies. Chapter 3 provided a critical review of some of these important studies. Here, we also found evidence that the inflation-threshold according to recent research lies well above 5 per cent. Broadly, findings from majority of these studies suggest that the threshold varies between 7 and 19 per cent. Interestingly, even studies carried out inside the IMF support a threshold above 5 per cent for developing countries in general. Some of these results are sufficiently robust as they take into account advances in dynamic panel estimation techniques. Therefore, we can reasonably conclude from the cross-country panel evidence based on the literature that policy advice to keep inflation at 5 per cent is not well founded. We have also identified that there is no panel study using dynamic analysis emphasising particularly on developing countries of Asia. Chapter 5 of this thesis filled this gap. We examined the country-specific literature in Chapter 4 as cross-country studies cannot capture country-specific circumstance. Thus we believe that country-specific analysis provides a better understanding about the growth-inflation nexus in a particular country. We provided a critical review of the literature from the growing 192 developing countries of South and East Asia. We noticed a number of interesting findings from the review. First, inflation and growth may move together in developing countries. The reasons behind this are structural causes such as inevitable fiscal deficits due to the need to finance the effort to close large physical and social structural deficits with low tax base, subsistence agriculture, and various other supply rigidities and bottlenecks. Second, studies which find very low threshold level of inflation do not seem to be consistent with the country’s historical inflation-growth experience. Third, studies which find a relatively high inflation threshold seem to be in line with the historical experience but suffer from methodological problems. We filled this gap in Chapter 6 where we performed our country-specific empirical analysis. The review of country-specific literature also revealed that countries trying to keep inflation at a very low level have to sacrifice economic growth. This is evident from the higher volatility in growth. Both Indonesia and Thailand showed this characteristic when we compared these economies between pre and post Inflation Targeting (IT) regimes. The discussion also showed that monetary policy faces a challenge of controlling inflation and maintaining export-led growth at the same time. Higher domestic price level causes loss in competitiveness in international markets. But so does the tight monetary policy to curb inflation. Tight monetary policy using higher interest rate to fight inflation causes domestic currency to appreciate (as higher interest rate attracts more capital inflow) and the country loses its competitiveness in the international markets. The critical review of the existing literature motivated us to examine the relationship between growth and inflation empirically. We provided our empirical analyses in two chapters. Chapter 5 dealt with cross-country panel study using different panel estimation techniques while Chapter 6 investigated the case of Bangladesh using time series analysis. Our selection of developing countries of Asia for panel study in Chapter 5 was influenced by the availability of data on variables considered in the models estimated. The 14 countries used in the analysis are Bangladesh, Cambodia, India, Indonesia, Lao PDR, Malaysia, Pakistan, The Philippines, Thailand, Vietnam, Kazakhstan, Kyrgyz Republic, Papua New Guinea, and Tajikistan. We have excluded China in this case because we believe it is an outlier considering its size, 193 economic structure, and political-economic institutions. Our analysis includes both static models, such as Fixed Effects and Random Effects models, and dynamic panel estimation models. The key strength of our analysis in Chapter 5 is the estimations using dynamic System Generalised Method of Moments (SGMM) as well as robustness checks using a number of subsamples. The SGMM is regarded as a superior instrumentation technique amongst its counterparts in tackling the endogeneity problem. The results are robust since they pass all the essential post diagnostic tests, confirming that the underlying assumptions of the models hold. The findings from the analyses using 14 countries for the period 1961-2010 revealed that the empirical basis for keeping inflation at 5 per cent or below is rather weak. We found that the threshold level ranges between 8 and 15 per cent. In particular, an important finding is that the threshold levels vary according to the levels of development. There is much evidence that for poorer groups of countries thresholds tend to be higher. In short, these findings substantiate the issues we address in our research questions. First, the inflation threshold in developing countries of Asia is much higher than 5 per cent. Second, poorer countries have higher threshold levels of inflation. The time series analysis on Bangladesh, carried out in Chapter 6, provided strong evidence that inflation-threshold is not as low as 5 per cent. Based on annual observations between 1977 and 2009 the threshold is found around 15 per cent. Evidence also revealed that inflation does not appear to be a main reason behind low economic growth in an underdeveloped country such as Bangladesh. Our analysis also showed that inflation starts to pick up when economic growth is taking place. Since inflation appears to be an inevitable consequence of growth, government should not immediately take up anti-inflationary measure which might hurt growth. The analysis of this thesis provides convincing evidence that the wisdom of macroeconomic policies targeting low inflation in developing countries has serious limitations, causing more harm than good. This particular tenet of mainstream economics which continues to influence IMF’s policy advice should be revised. In the context of developing countries of Asia, this study shows that developing countries tend to have higher inflation tolerance and therefore IMF’s policy 194 suggestions are too narrow and seriously flawed. We have successfully addressed our research questions as well. We draw the following conclusions from our investigation: (a) there is no empirical justification for keeping inflation as low as 5 per cent, (b) the threshold level of inflation varies with the level of development and targeting inflation should be done after carefully considering the country-specific circumstances, and (c) policy suggestions by the IMF in developing countries should incorporate the rethinking about macroeconomics. Macroeconomic policies should be seen from a broader perspective and require a balance between the need for stabilisation and development. Stabilisation is required but not sufficient for development. This research suggests poverty reduction and employment creation should be given priority over targeting a low inflation regime. Our study raises some additional intriguing questions which can be pursued as future research. An unresolved question relating to inflation threshold in both developing and developed countries is whether the threshold is a cliff or a plateau in nature. If the threshold is not too kinked and does not have a sharp cliff-edge, there would be some room for accommodative policies and fiscal space, essential for rapid development. It would also assure policy makers that macroeconomic policies need not be overly cautious about inflation and therefore to keep it at a low single digit level. Additionally, research can be carried out to determine the threshold level of debt-GDP ratio or fiscal deficits to find out how they affect economic growth, especially in light of the debate on fiscal austerity in the wake of the Great Recession of 2008-2009 and the IMF’s recent admission that “there is no single threshold for debt ratios that can delineate the “bad” from the “good.””123 The issue is vitally important because public investment in infrastructure, education, health, and social security leads to a positive impact on economic growth. The findings about the threshold levels of inflation and debt-GDP ratio would also have important implications for exchange rate policy because balance of payments, fiscal deficits, and monetary policy stance are all closely intertwined. Therefore, further investigation can be done to find out the implications on exchange rate policy which is an important instrument for structural change as well as stabilisation. In our 123 IMF (2008, p. 109). 195 research, we have shown the importance of country-specific circumstances and how they influence the growth-inflation dynamics. Similarly, the choice of an exchange rate regime might be determined by specific characteristics of a country. This therefore, would help raise questions about the validity of prescribing a more flexible exchange rate regime and open capital account in developing countries. Moreover, a similar research can be carried out to find the implications of the above issues on wage policy and other aspects of labour markets in developing countries. Finally, out of Asia; that is, to do similar studies for other developing regions in Africa and Latin America. 196 References ADB 2009, Bangladesh 2006-2010, Asian Development Bank. Ahmed, S & Mortaza, MG 2005, 'Inflation and economic growth in Bangladesh: 1981 - 2005', Bangladesh Bank Policy Analysis Unit Working Paper Series WP 0604, pp. i - 19. Ahmed, S 1984, 'Inflation in Bangladesh: causes and consequences', Ph.D thesis, Boston University. Akaike, H 1974, 'A new look at statistical model identification', IEEE Transactions on Automatic Control, vol. 19, pp. 716-23. Akerlof, G & Shiller, RJ 2009, Animal spirit: how human psychology drives the economy, and why it matters for global capitalism, Princeton University Press. Akerlof, G, Dickens, WT & Perry, G 1996, 'The macroeconomics of low inflation', Brookings Papers on Economic Activity, vol. 1996, no. 1, pp. 1 - 76. Alauddin, M & Hasan, S 1996, 'Bangladesh: the first twenty five years: an overview of the contributions', in M Alauddin & S Hasan (eds), Bangladesh: economy, people, and the environment, The University of Queensland, Australia. Amato, JD & Gerlach, S 2002, 'Inflation targeting in emerging market and transition economies: lessons after a decade', European Economic Review, vol. 46, pp. 781-90. Anwar, S & Islam, I 2011, 'Should developing countries target low, single digit inflation to promote growth and employment?', ILO Employment Sector Employment Working Paper, no. 87, pp. i - 35. Arellano, M & Bover, O 1995, 'Another look at the instrumental variables estimation of error-components models', Journal of Econometrics, vol. 68, pp. 29-51. Arestis, P & Sawyer, M 2003, 'Reinventing fiscal policy', Journal of Post Keynesian Economics, vol. 26, no. 1, pp. 3-25. 197 —— 2004, Re-examining Monetary and Fiscal Policy for the 21st Century, Edward Elgar, Cheltenham. —— 2005, 'Aggregate demand, conflict and capacity in the inflationary process', Cambridge Journal of Economics, vol. 29, no. 6, pp. 959 - 74. —— 2009, 'Prices and wage determination and the inflation barrier: moving beyond the Phillips curve', in C Gnos & L Rochon (eds), Monetary Policy and Financial Stability: A Post-Keynesian Agenda, Edward Elgar, Cheltenham. Argitis, G & Dafermos, Y 2011, 'Finance, inflation and employment: a postKeynesian/Kaleckian analysis', Cambridge Journal of Economics, vol. 35, pp. 1015 33. Armijo, LE 2005, 'Mass democracy: the real reason that Brazil ended Inflation?', World Development, vol. 33, no. 12, pp. 2013 - 27. Asteriou, D & Hall, SG 2007, Applied econometrics: a modern approach, Revised edn, Palgrave Macmillan, New York. Baer, W 1967, 'The inflation controversy in Latin America: a survey', Latin America Research Review, vol. 2, no. February, pp. 3 - 25. —— 1987, 'The resurgence of inflation in Brazil, 1974 - 86', World Development, vol. 15, no. 8, pp. 1007 - 34. Ball, LM 1997, 'Disinflation and the NAIRU', in CD Romer & DH Romer (eds), Reducing Inflation: Motivation and Strategy, University of Chicago Press, Chicago, pp. 167 - 85. Baltagi, BH, Demetriades, PO & Law, SH 2009, 'Financial development and openness: evidence from panel data', Journal of Development Economics, vol. 89, pp. 285-96. Bangake, C & Eggoh, JC 2011, 'Further evidence on finance-growth causality: a panel data analysis', Economic Systems, vol. 35, pp. 176-88. Barro, R 1996, 'Inflation and growth', Federal Reserve Bank of St. Louis Review, vol. 78, no. 3, pp. 153-69. 198 Bhatia, R 1960-61, 'Inflation, deflation and economic development', IMF Staff Papers, vol. 8, pp. 101 - 14. Bick, A 2010, 'Threshold effect of inflation on economic growth in developing countries', Economic Letters, vol. 108, pp. 126-9. Blanchard, O & Gali, J 2007, 'Real wage rigidities and the New Keynesian model', Journal of Money, Credit, and Banking, vol. 39, pp. 35 - 65. Blanchard, O & Sheen, J 2009, Macroeconomics: Australasian edition, 3rd edn, Pearson Australia, Frenchs Forest. Blanchard, O & Summers, LH 1988, 'Beyond the Natural Rate Hypothesis', American Economic Review, vol. 78, no. May, pp. 182 - 7. Blanchard, O, Dell'Ariccia, G & Mauro, P 2010, 'Rethinking macroeconomic policy', Journal of Money, Credit and Banking, vol. 42 (supplement), no. 6, pp. 199 - 215. Blundell, R & Bond, S 1998, 'Initial conditions and moment restrictions in dynamic panel data models', Journal of Econometrics, vol. 87, pp. 115-43. Bond, S 2002, 'Dynamic panel data models: a guide to micro data methods and practice', Portuguese Economic Journal, vol. 1, no. 2, pp. 141-62. Bose, SR 1973, 'The price situation in Bangladesh - a preliminary analysis', Bangladesh Development Review, vol. 1, no. 3, pp. 243-68. Brown, RI, Durbin, J & Evans, JM 1975, 'Techniques for testing the constancy of regression relationships over time', Journal of the Royal Statistical Society, vol. 37, pp. 149-63. Bruno, M & Easterly, W 1996, 'Inflation and growth: in search of a stable relationship', Federal Reserve Bank of St. Louis Review, vol. 78, no. 3, pp. 139-46. —— 1998, 'Inflation crises and long-run growth', Journal of Monetary Economics, vol. 41, pp. 3-26. 199 Bruno, M 1992, 'From sharp stabilization to growth: on the political economy of Israel's transition', European Economic Review, vol. 36, pp. 310 - 9. —— 1995, 'Does inflation really lower growth?', Finance and Development, vol. September, pp. 35 - 8. Burdekin, RCK, Denzau, AT, Keil, MW, Sitthiyot, T & Willet, TD 2004, 'When does inflation hurt economic growth? Different nonlinearities for different economies', Journal of Macroeconomics, vol. 26, pp. 519 - 32. Busse, M & Hefeker, C 2007, 'Political risk, institutions and foreign direct investment', European Journal of Political Economy, vol. 23, pp. 397-415. Cagan, P 1956, 'The monetary dynamics of hyperinflation', in M Friedman (ed.), Studies in the Quantity Theory of Money, University of Chicago Press, Chicago. Chenery, HB 1960, 'Patterns of industrial growth', The American Economic Review, vol. 50, no. 4, pp. 624-54. Choi, I & Chung, B 1995, 'Sampling frequency and the power of the tests for a unit root: a simulation study', Economics Letters, vol. 49, pp. 131-6. Chow, GC 1960, 'Tests of equality between sets of coefficients in two linear regressions', Econometrica, vol. 28, pp. 591-605. Chowdhury, A & Ham, R 2009, 'Inflation targeting in Indonesia: searching for a threshold', The Singapore Economic Review, vol. 54, no. 4, pp. 645 - 55. Chowdhury, A & Hossain, A 1996, 'The political economy of macroeconomic management in Bangladesh', in M Alauddin & S Hasan (eds), Bangladesh: economy, people, and the environment, The University of Queensland, Australia. Chowdhury, A & Islam, I 2010, 'Rethinking macroeconomics from the perspective of the Millennium Development Goals in the post-crisis era', A commentary in the VoxEU Debate on the Global Crisis - Development and the Crisis. Chowdhury, A & Siregar, H 2004, 'Indonesia's monetary policy dilemma constraints of inflation targeting', The Journal of Developing Areas, vol. 37, no. 2, pp. 137 - 53. 200 Chowdhury, A 1996, 'Macroeconomic management in east Asian newly industrialising economies', in B Kapur, E Quah & HT Hoon (eds), Development, trade, and the Asia-Pacific, Prentice-Hall, Singapore, pp. 141-63. —— 1999, 'Villain of the Asian crisis-Thailand or the IMF?', Asian Economic Bulletin, vol. 16, no. 2, pp. 168-77. —— 2002, 'Does inflation affect economic growth? The relevance of the debate for Indonesia', Journal of the Asia Pacific Economy, vol. 7, no. 1, pp. 20-34. —— 2004, Thematic summary report: Monetary policy, UNDP Asia-Pacific Regional Programme on the Macroeconomics of Poverty Reduction, Colombo. —— 2006, 'The "stabilisation trap" and poverty reduction-what can monetary policy do?', Indian Development Review, vol. 4, no. 2, pp. 407-32. Chowdhury, MB 2011, 'Remittances flow and financial development in Bangladesh', Economic Modelling, vol. 28, pp. 2600-8. CPD 2011, Analytical review of Bangladesh's macroeconomic performance in fiscal year 2011-12 (first reading), Centre for Policy Dialogue, Dhaka. Cukierman, A, Webb, S & Neyapti, B 1992, 'Measuring the independence of central banks and its effect on policy outcomes', World Bank Economic Review, vol. 6, no. 3, pp. 353-98. Dasgupta, AK 1954, 'Keynesian economics and under-developed countries', Economic and Political Weekly, vol. January 26. Datta, K & Mukhopadhyay, CK 2011, 'Relationship between inflation and economic growth in Malaysia - An econometric review', in International Conference on Economics and Finance Research, IPEDR, Singapore, vol. 4, pp. 415 - 9. de Mendonca, HF & de Guimaraes e Souza, GJ 2011, 'Is inflation targeting a good remedy to control inflation?', Journal of Development Economics, vol. Forthcoming, pp. 1-14. Dickey, D & Fuller, WA 1979, 'Distribution of the estimates for autoregressive time series with a unit root', Journal of the American Statistical Association, vol. 74, pp. 427-31. 201 —— 1981, 'Likelihood ratio statistics for autoregressive time series with a unit root', Econometrica, vol. 49, pp. 1057-72. Dornbusch, R, Fischer, S & Startz, R 2008, Macroeconomics, 10 edn, McGraw-Hill Irwin, New York. Dorrance, GS 1964, 'The effect of inflation on economic development', in W Baer & I Kerstenetzky (eds), Inflation and Growth in Latin America, The Economic Growth Center, Yale University, pp. 37 - 88. Dotsey, M & Sarte, P 2000, 'Inflation uncertainty and growth in a cash-in-advance economy', Journal of Monetary Economics, vol. 45, pp. 631 - 55. Dowling, JM & Valenzuela, MR 2004, Economic development in Asia, Thomson Learning, Singapore. Drukker, D, Gomis-Porqueras & Hernandez-Verme 2005, 'Threshold effects in the relationship between inflation and growth: a new panel-data approach', Working paper, pp. 1 - 18. Engle, R & Granger, CWJ 1987, 'Cointegration and error correction: representation, estimation and testing', Econometrica, vol. 55, pp. 251-76. Epstein, G & Yeldan, E 2008, 'Inflation targeting, employment creation, and economic development: assessing the impacts and policy alternatives', International Review of Applied Economics, vol. 22, no. 2, pp. 131 - 44. Espinoza, R, Leon, H & Prasad, A 2010, 'Estimating the inflation-growth nexus - a smooth transition model', IMF Working Paper, no. WP/10/76, pp. 1-22. —— 2011, 'When should we worry about inflation?', The World Bank Economic Review, vol. 26, no. 1, pp. 100-27. Faaland, J & Parkinson, J 1976, Bangladesh: the test case of development, C. Hurst, London. 202 Fasano-Filho, U 1996, 'Economic policy making in Sub-Saharan Africa and IMF involvement', The Quarterly Review of Economics and Finance, vol. 36, no. Special Issue, pp. 115 - 51. Fischer, S 1987, 'Economic growth and economic policy', in V Corbo, M Goldstein & M Khan (eds), Growth-Oriented Adjustment Programs, The International Monetary Fund and the World Bank, Washington DC. —— 1993, 'The role of macroeconomic factors in growth', Journal of Monetary Economics, vol. 32, no. 3, pp. 485 - 512. Frank, C, Kim, KS & Westphal, L 1975, 'Economic growth in South Korea since World War II', in Foreign regimes and economic development: South Korea, NBER, Cambridge, MA. Friedman, M & Schwartz, A 1963, A monetary history of the United States, 1867 1960, Princeton University Press, Princeton. Friedman, M 1968, 'The role of monetary policy', American Economic Review, vol. 58, no. 1, pp. 1-17. —— 1973, Money and economic development, Lexington Books, Toronto. —— 1977, 'Nobel Lecture: Inflation and Unemployment', Journal of Political Economy, vol. 85, no. 3, pp. 451 - 72. Georgescu-Roegen, N 1970, 'Structural inflation - lock and balanced growth', Economies et Societes, vol. 4, no. 3 (March), pp. 557 - 605. Gerlach, S & Tillmann, P 2012, 'Inflation targeting and inflation persistence in AsiaPacific', Journal of Asian Economics, vol. 23, pp. 360 - 73. Ghosh, A & Phillips, S 1998, 'Warning: inflation may be harmful to your growth', IMF Staff Papers, vol. 45, no. 4, pp. 672-710. Goldsbrough, D, Adovor, E & Elberger, B 2007, 'Inflation Targets in IMF-Supported Programs', Center for Global Development Working Group on IMF Programs and Health Expenditures Background Paper, no. March 2007, pp. 1 - 9. 203 Greene, WH 2003, Econometric Analysis, Fifth edn, Prentice Hall, New Jersey. Grier, K & Perry, M 2000, 'The effects of real and nominal uncertainty on inflation and output growth: some GARCH-M evidence', Journal of Applied Econometrics, vol. 15, pp. 45 - 58. Grier, K & Tullock, G 1989, 'An empirical analysis of cross-national economic growth, 1951 - 80', Journal of Monetary Economics, vol. 24, pp. 259 - 76. Grossman, GM & Helpman, E 1991a, 'Quality ladders in the theory of growth', Review of Economic Studies, vol. 58, pp. 43-61. —— 1991b, 'Endogenous product cycles', Economic Journal, vol. 101, pp. 1229-41. Guiliano, P & Ruiz-Arranz, M 2009, 'Remittances, financial development, and growth', Journal of Development Economics, vol. 90, pp. 144-52. Hansen, H & Tarp, F 2001, 'Aid and growth regressions', Journal of Development Economics, vol. 64, pp. 547-70. Harberger, AC 1963, 'The dynamics of inflation in Chile', in C Christ (ed.), Measurement in Economics: Studies in Mathematical Economics and Econometrics in Memory of Yehuda Grunfeld, Stanford University Press, Stanford, pp. 219 - 50. Harrison, A 1996, 'Openness and growth: a time-series, cross-country analysis for developing countries', Journal of Development Economics, vol. 48, pp. 419-47. Hayat, ZUC & Kalirajan, K 2009, 'Is there a threshold level of inflation for Bangladesh?', Margin: The Journal of Applied Economic Research, vol. 3, no. 1, pp. 1-20. Hemming, R, Kell, M & Mahfouz, S 2002, 'The effectiveness of fiscal policy in stimulating economic activity - a review of the literature', IMF Working Paper, no. 02/208. Hossain, A 2005a, 'The Granger causality between money growth, inflation, currency devaluation and economic growth in Indonesia: 1954 - 2002', International Journal of Applied Econometrics and Quantitative Studies, vol. 2-3, pp. 45 - 68. 204 —— 2005b, 'The sources and dynamics of inflation in Indonesia: an ECM model estimation for 1952 - 2002', Applied Econometrics and International Development, vol. 5 - 4, pp. 93 - 116. —— 2010, 'Monetary targeting for price stability in Bangladesh: How stable is its money demand function and the linkage between money supply growth and inflation?', Journal of Asian Economics, vol. 21, pp. 564 - 78. Hossain, M & Sen, B 1992, 'Rural poverty in Bangladesh: trends and determinants', Asian Development Review, vol. 10, no. 1, pp. 1-34. Hussain, M 2005, 'Inflation and growth: estimation of threshold point for Pakistan', State Bank of Pakistan Economic Policy Department Working Paper, pp. 1 - 15. Hussain, S & Malik, S 2011, 'Inflation and economic growth: evidence from Pakistan', International Journal of Economics and Finance, vol. 3, no. 5, pp. 262 76. Hwang, J-T & Wu, M-J 2011, 'Inflation and economic growth in China: an empirical analysis', China & World Economy, vol. 19, no. 5, pp. 67 - 84. IEO 2007, The IMF and aid to Sub-Saharan Africa, Independent Evaluation Office, International Monetary Fund, Washington D.C. IMF 2008, World Economic Outlook, International Monetary Fund, Washington DC. —— 2009a, 'Bhutan: 2009 Article IV Consultation - Staff Report; Staff Supplement; and Public Information Notice on the Executive Board Discussion', IMF Country Report No. 09/334. —— 2009b, 'Indonesia: 2009 Article IV Consultation - Staff Report; Staff Statement; Public Information Notice on the Executive Board Discussion; and Statement by the Executive Director for Indonesia', IMF Country Report No. 09/230. —— 2009c, 'Cambodia: 2009 Article IV Consultation - Staff Report; Staff Supplement; Public Information Notice on the Executive Board Discussion; and Statement by the Executive Director', IMF Country Report No. 09/325. 205 —— 2009d, 'Lao People's Democratic Republic: 2009 Article IV Consultation Staff Report; Staff Supplement; Public Information Notice on the Executive Board Discussion', IMF Country Report No. 09/284. —— 2009e, 'Pakistan: 2009 Article IV Consultation and First Review Under the Stand-By Arrangement - Staff Report; Staff Statement; Public Information Notice and Press Release on the Executive Board Discussions; and Statement by the Executive Director for Pakistan', IMF Country Report No. 09/123, vol. April. —— 2010a, 'Bangladesh: 2009 Article IV Consultation - Staff Report; Staff Supplement; Public Information Notice on the Executive Board Discussion', IMF Country Report No. 10/55. —— 2010b, 'People's Republic of China: 2010 Article IV Consultation - Staff Report; Staff Statement; Public Information Notice on the Executive Board Discussion', IMF Country Report No. 10/238. —— 2010c, 'India: 2009 Article IV Consultation - Staff Report; Staff Statement; Public Information Notice on the Executive Board Discussion; and Statement by the Executive Director for India', IMF Country Report No. 10/73. —— 2010d, 'Malaysia: 2010 Article IV Consultation - Staff Report; Staff Statement; Public Information Notice on the Executive Board Discussion; and Statement by the Executive Director for Malaysia', IMF Country Report No. 10/265. —— 2010e, 'Maldives: 2009 Article IV Consultation - Staff Report; Staff Supplements; Public Information Notice on the Executive Board Discussion; Press Release on the Executive Board Discussion; and Statement by the Executive Director for Maldives', IMF Country Report No. 10/28. —— 2010f, 'Nepal: 2010 Article IV Consultation and Request for Disbursement Under the Rapid Credit Facility - Staff Report; Staff Supplements; Public Information Notice on the Executive Board Discussion; Press Release on the Executive Board Discussion; and Statement by the Alternate Executive Director and Advisor for Nepal', IMF Country Report No. 10/185. —— 2010g, 'Philippines: 2009 Article IV Consultation - Staff Report, Public Information Notice on the Executive Board Discussion; and Statement by the Executive Director for the Philippines', IMF Country Report No. 10/45. 206 Islam, MS 2008, 'Commodity boom and inflation challenges for Bangladesh', Institute of South Asian Studies Working Paper, National University of Singapore, no. 40, pp. 1 - 20. Islam, N 1974, 'The state and prospects of the Bangladesh economy', in EAG Robinson & KB Griffin (eds), The economic development of Bangladesh within a socialist framework, Macmillan, London. Jha, R 2008, 'Inflation targeting in India: issues and prospects', International Review of Applied Economics, vol. 22, no. 2, pp. 259 - 70. Johansen, S 1991, 'Estimation and hypothesis testing of cointegrating vectors in Gaussian vector autoregressive models', Econometrica, vol. 59, pp. 1551-80. Johnson, HG 1969, Essays in Monetary Economics, Harvard University Press, Cambridge. Johnston, BF & Mellor, JW 1961, 'The role of agriculture in economic development', The American Economic Review, vol. 51, no. 4, pp. 566-93. Jongwanich, J & Park, D 2009, 'Inflation in Developing Asia', Journal of Asian Economics, vol. 20, pp. 507 - 18. Jung, W & Marshall, P 1986, 'Inflation and economic growth: some international evidence on structuralist and distortionist positions', Journal of Money, Credit, and Banking, vol. 18, no. 2. Kaldor, N 1955-56, 'Alternative theories of distribution', Review of Economic Studies, vol. 23, pp. 83 - 100. Kapur, BK 1976, 'Alternative stabilization policies for less developed countries', Journal of Political Economy, vol. 84, pp. 777-95. Kathavate, J & Mallik, G 2012, 'The impact of the interaction between institutional quality and aid volatility on growth: theory and evidence', Economic Modelling, vol. 29, pp. 716-24. Khan, AR 1990, 'Poverty in Bangladesh: a consequence of and a constraint on economic growth', The Bangladesh Development Studies, vol. 18, no. 3, pp. 19-34. 207 —— 2007, 'Is Bangladesh poised to be launched on an East-Asian development path? some thoughts on performance, prospects, and puzzles', paper presented to BIDS Golden Jubilee Conference on Development Prospects of Bangladesh: Emerging Challenges, Dhaka, Bangladesh, 2-3 December 2007. Khan, M & Senhadji, A 2001, 'Threshold effects in the relationship between inflation and growth', IMF Staff Papers, vol. 48, no. 1, pp. 1-21. —— 2003, 'Financial development and economic growth: a review and new evidence', Journal of African Economies, vol. 12, no. AERC Supplement 2, pp. ii89ii110. Khan, SU & Saqib, OF 2011, 'Political instability and inflation in Pakistan', Journal of Asian Economics, vol. 22, pp. 540 - 9. Kheir-El-Din, H & Abou-Ali, H 2008, 'Inflation and growth in Egypt: is there a threshold effect?', ECES Working Paper, vol. 135. Kormendi, RC & Meguire, PG 1985, 'Macroeconomic determinants of growth: cross-country evidence', Journal of Monetary Economics, vol. 16, pp. 141 - 64. Kremer, S, Bick, A & Nautz, D 2009, 'Inflation and growth: new evidence from a dynamic panel threshold analysis', Economic Risk, vol. SFB 649, Discussion Paper 036. Kumar, MS & Woo, J 2010, 'Public debt and growth', IMF Working Paper, no. WP/10/174, pp. 1-46. Kwon, G, Mcfarlane, L & Robinson, W 2009, 'Public debt, money supply, and inflation: a cross-country study', IMF Staff Papers, vol. 56, no. 3, pp. 476-515. Lewis, A 1964, 'Closing Remarks', in W Baer & I Kerstenetzky (eds), Inflation and Growth in Latin America, The Economic Growth Center, Yale University, pp. 21 36. Lim, J 2008, 'Central banking in the Philippine: from inflation targeting to financing development', International Review of Applied Economics, vol. 22, no. 2, pp. 271 85. 208 Lipsey, RG 1960, 'The relationship between unemployment and the rate of change of money wage rates in the United Kingdom, 1862 - 1967: A further analysis', Economica, vol. 27, pp. 1 - 31. Lopez-Villavicencio, A & Mignon, V 2011, 'On the impact of inflation on output growth: does the level of inflation matter?', Journal of Macroeconomics, vol. 33, pp. 455-64. Lucas, RE & Sargent, T 1978, 'After Keynesian macroeconomics', paper presented to After the Phillips Curve, Federal Reserve Bank of Boston Conference Series, no. 19, Boston. Lucas, RE 1972, 'Expectations and the neutrality of money', Journal of Economic Theory, vol. 4, pp. 103 - 24. —— 1973, 'Some international evidence on output - inflation trade-offs', American Economic Review, vol. 63, pp. 326 - 34. —— 1988, 'On the mechanisms of economic development', Journal of Monetary Economics, vol. 22, pp. 3-42. Lutkepohl, H 2005, New introduction to multiple time series analysis, First edn, Springer, Berlin. Mackie, JAC 1967, Problems of the Indonesian inflation, Department of Asian Studies Monograph Series, Cornell University. MacKinnon, JG, Haug, A & Michelis, L 1999, 'Numerical distribution functions of likelihood ratio tests for cointegration', Journal of Applied Econometrics, vol. 14, no. 5, pp. 563-77. Maddala, GS & Kim, I-M 2000, Unit roots, cointegration, and structural change, Reprinted edn, Cambridge University Press, Cambridge. Maddala, GS & Lahiri, K 2009, Introduction to econometrics, Fourth edn, John Wiley and Sons Limited, Chichester. 209 Mahadevan, R & Asafu-Adjaye, J 2006, 'Is there a case for low inflation-induced productivity growth in selected Asian economies?', Contemporary Economic Policy, vol. 24, no. 2, pp. 249 - 61. Mahmud, W & Osmani, SR 1980, 'Impact of emigrant workers' remittances on the Bangladesh economy', The Bangladesh Development Studies, vol. VIII, no. 3, pp. 128. Mahmud, W 2008, 'Bangladesh', in A Chowdhury & W Mahmud (eds), Handbook on the South Asian economies, Edward Elgar Publishing Limited, Cheltenham, pp. i245. Mallik, G & Chowdhury, A 2001, 'Inflation and economic growth: evidence from four South Asian countries', Asia Pacific Development Journal, vol. 8, no. 1, pp. 123 - 35. Mankiw, NG & Ball, LM 2011, Macroeconomics and the Financial System, 1st edn, Worth Publishers, New York. Mankiw, NG 2001, 'The inexorable and mysterious trade-off between inflation and unemployment', The Economic Journal, vol. 111, pp. 45 - 61. McKinnon, RI 1973, Money and capital in economic development, Brookings Institution, Washington, DC. Mishra, A & Mishra, V 2009, 'Preconditions for inflation targeting in an emerging economy: the case of India', Monash University Development Research Unit Discussion Paper, no. DEVDP09-08, pp. 1-32. —— 2012a, 'Evaluating inflation targeting as a monetary policy objective for India', Economic Modelling, vol. 29, pp. 1053 - 63. —— 2012b, 'Inflation targeting in India: A comparison with the multiple indicator approach', Journal of Asian Economics, vol. 23, pp. 86 - 98. Montiel, PJ 2011, Macroeconomics in Emerging Markets, 2nd edn, Cambridge University Press, New York. 210 Mortaza, MG 2006, 'Sources of inflation in Bangladesh: recent macroeconomic experience', Bangladesh Bank Policy Analysis Unit Working Paper Series, no. WP 0704, pp. i-27. Mubarik, Y 2005, 'Inflation and growth: an estimate of the threshold level of inflation in Pakistan', State Bank of Pakistan Research Bulletin, vol. 1, no. 1, pp. 3544. Munir, Q, Mansur, K & Furuoka, F 2009, 'Inflation and economic growth in Malaysia: a threshold regression approach', ASEAN Economic Bulletin, vol. 26, no. 2, pp. 180-93. Naranpanawa, A & Bandara, JS 2012, 'Poverty and growth impacts of high oil prices: Evidence from Sri Lanka', Energy Policy, vol. 45, pp. 102 - 11. Odedokun, MO 1991, 'Evidence on inflation-output trade-off in developing and industrialized countries', Applied Economics, vol. 23, pp. 731 - 42. Osmani, SR 1990, 'Notes on some recent estimates of poverty in Bangladesh', The Bangladesh Development Studies, vol. 18, no. 3. —— 2007, 'Interpreting recent inflationary trends in Bangladesh and policy options', paper presented to Interpreting Recent Inflationary Trends in Bangladesh and Policy Options, Dhaka, 8 September 2007. Palley, T 2012, 'The Economics of the Phillips Curve: Formation of Inflation Expectations versus Incorporation of Inflation Expectations', Structural Change and Economic Dynamics, vol. forthcoming. Palley, T, Rochon, L & Vernengo, M 2012, 'Economics and the economic crisis: the case for change', Review of Keynesian Economics, vol. 1, pp. 1-4. Paul, BK & Rashid, H 1993, 'Flood damage to rice crop in Bangladesh', Geographical Review, vol. 83, no. 2, pp. 150-9. Paul, S, Kearney, C & Chowdhury, K 1997, 'Inflation and economic growth: a multicountry empirical analysis', Applied Economics, vol. 29, pp. 1387 - 401. Pazos, F 1972, Chronic inflation in Latin America, Praeger Publisher. 211 Phelps, ES 1967, 'Phillips Curves, inflation expectations, and optimal employment over time', Economica, vol. 34, pp. 254-81. Phillips, AW 1958, 'The relation between unemployment and the rate of change of money wage rates in the United Kingdom, 1861 - 1957', Economica, vol. 25, pp. 283-99. Phillips, PC & Perron, P 1988, 'Testing for a unit root in time series regression', Biometrika, vol. 75, no. 2, pp. 335-46. Phiri, A 2010, 'At what level is inflation least detrimental towards finance-growth activity in South Africa?', Journal of Sustainable Development in Africa, vol. 12, no. 6, pp. 354-64. Pollin, R & Zhu, A 2006, 'Inflation and economic growth: a cross-country nonlinear analysis', Journal of Post Keynesian Economics, vol. 28, no. 4, pp. 593-614. Poole, W 1999, 'Is inflation too low?', Cato Journal, vol. 18, no. 3, pp. 453 - 64. Rahim, AMA 1973, 'Some aspects of inflation theories in the context of Bangladesh', Bangladesh Economic Review, vol. 1, no. 2, pp. 173-85. Rahman, HZ & Hossain, M 1995, Rethinking rural poverty: Bangladesh as a case study, Sage Publications, London. Rahman, M, Bhattacharya, D, Shadat, WB & Deb, U 2008, Recent inflation in Bangladesh: trends, determinants, and impact on poverty, Centre for Policy Dialogue, Dhaka. Rajan, RG & Subramanian, A 2008, 'Aid and growth: what does the cross-country evidence really show?', The Review of Economics and Statistics, vol. 90, no. 4, pp. 643-65. Ramayandi, A & Rosario, A 2010, 'Monetary policy discipline and macroeconomic performance: the case of Indonesia', Asian Development Bank Working Paper Series, no. 238, pp. 1 - 22. 212 Rao, BB & Hassan, GM 2011, 'A panel data analysis of the growth effects of remittances', Economic Modelling, vol. 28, pp. 701-9. Rao, VKRV 1952, 'Investment, income, and the multiplier in an underdeveloped economy', Indian Economic Review, vol. 1, no. 1, pp. 55-67. Rashid, S 2008, 'Bangladesh poverty: the need for a "big push"', in SS Andaleeb (ed.), The Bangladesh economy: diagnoses and prescriptions, The University Press Limited, Dhaka, pp. i-338. RBI 2002, Reserve Bank of India Bulletin, Reserve Bank of India, Mumbai. Rivera-Batiz, LA & Xie, D 1993, 'Integration among unequals', Regional Science and Urban Economics, vol. 23, pp. 337-54. Robinson, A 1978, 'Development planning in Bangladesh: a study in political economy by Nurul Islam', The Economic Journal, vol. 88, no. 350, pp. 370-2. Robinson, J 1952, The rate of interest and other essays, Macmillan, London. Rodrik, D 1999a, The new global economy and developing countries: making openness work, Overseas Development Council, Washington D.C. —— 1999b, 'Where did all the growth go? External shocks, social conflicts, and growth collapses', Journal of Economic Growth, vol. 4, no. December, pp. 385 - 412. —— 2004, 'Rethinking growth policies in the developing world', paper presented to Luca d' Agliano Lecture in Development Economics, Torino, Italy, 8 October. Romer, CD & Romer, DH 1999, 'Monetary policy and the well-being of the poor', Federal Reserve Bank of Kansas City Economic Review, vol. 1st Quarter, pp. 21-49. Roodman, D 2006, 'How to do xtabond2: an introduction to "difference" and "system" GMM in Stata', Center for Global Development Working Paper, no. 103. Rowthorn, R 1995, 'Capital formation and unemployment', Oxford Review of Economic Policy, vol. 11, no. 1, pp. 26 - 39. 213 Roy, R & Ramos, RA 2012, 'IMF Article IV Reports: an analysis of policy recommendations', International Policy Centre for Inclusive Growth, UNDP, Working Paper, no. 86. Salami, D & Kelikume, I 2010, 'An estimation of inflation threshold for Nigeria 1970 - 2008', International Review of Business Research Papers, vol. 6, no. 5, pp. 375 85. Samuelson, PA & Solow, RM 1960, 'Analytical aspects of anti-inflation policy', American Economic Review, vol. 50, pp. 177 - 94. Sarel, M 1996, 'Nonlinear effects of inflation on economic growth', IMF Staff Papers, vol. 43, no. 1, pp. 199 - 215. Sawyer, MC 1985, 'The Economics of Michal Kalecki', Eastern European Economics, vol. 23, no. 3/4, pp. v-x+1-319. —— 2001, 'The NAIRU: a critical appraisal', in P Arestis & MC Sawyer (eds), Money, Finance and Capitalist Development, Edward Elgar, Aldershot. —— 2002, 'The NAIRU, aggregate demand and investment', Metroeconomica, vol. 53, no. 1, pp. 66 - 94. —— 2005, 'Towards a simple macroeconomic model incorporating the key Heterodox Propositions', University of Leeds Working Paper, pp. 1-25. Schwarz, G 1978, 'Estimating the dimension of a model', Annals of Statistics, vol. 6, pp. 461-4. Seers, D 1962, 'A theory of inflation and growth in under-developed economies based on the experience of Latin America', Oxford Economic Papers, vol. 14, pp. 173 - 95. Selassie, AA, Clements, B, Tareq, S, Martijn, JK & Di Bella, G 2006, 'Designing monetary and fiscal policy in low-income countries', IMF Occasional Paper, no. 250, pp. i - 69. Sen, B 2003, 'Drivers of escape and descent: changing household fortunes in rural Bangladesh', World Development, vol. 31, no. 3, pp. 513-34. 214 Sepehri, A & Moshiri, S 2004, 'Inflation-growth profiles across countries: evidence from developing and developed countries', International Review of Applied Economics, vol. 18, no. 2, pp. 191 - 207. Shand, RT 1996, 'Foreward', in M Alauddin & S Hasan (eds), Bangladesh: economy, people, and the environment, The University of Queensland, Australia. Shaw, ES 1973, Financial deepening in economic development, Oxford University Press, New York. Siddique, A 1975, 'Money and inflation in Bangladesh', Bangladesh Development Studies, vol. 3, no. 1, pp. 27-42. Sidrauski, M 1967, 'Rational choice and patterns of growth in a monetary economy', American Economic Review, vol. 57, pp. 534 - 44. Singh, K & Kalirajan, K 2003, 'The inflation-growth nexus in India: an empirical analysis', Journal of Policy Modeling, vol. 25, pp. 337 - 96. Siregar, H & Ward, BD 2002, 'Were aggregate demand shocks important in explaining Indonesian macroeconomic fluctuations?', Journal of the Asia Pacific Economy, vol. 7, no. 1, pp. 35 - 60. Siregar, R & Rajaguru, G 2005, 'Sources of variations between the inflation rates of Korea, Thailand and Indonesia during the post-1997 crisis', Journal of Policy Modeling, vol. 27, pp. 876 - 84. Siregar, RY & Goo, S 2010, 'Effectiveness and commitment to inflation targeting policy: evidence from Indonesia and Thailand', Journal of Asian Economics, vol. 21, pp. 113 - 28. Stockman, AC 1981, 'Anticipated inflation and the capital stock in a cash-in-advance economy', Journal of Monetary Economics, vol. 8, pp. 387 - 93. Storm, S & Naastepad, CWM 2007, 'It is high time to ditch the NAIRU', Journal of Post Keynesian Economics, vol. 29, no. 4, pp. 531 - 54. 215 Talukder, RK 2005, 'Food security, self-sufficiency and nutrition gap in Bangladesh', The Bangladesh Development Studies, vol. XXXI, no. 3 & 4, pp. 35-62. Taslim, M 1982, 'Inflation in Bangladesh: a reconsideration of structuralistmonetarist controversy', The Bangladesh Development Studies, vol. X, no. 1, pp. 2153. Taylor, JB 1993, ‘Discretion versus policy rules in practice’, Carnegie-Rochester Conference Series on Public Policy, vol. 39, pp. 195-214. Taylor, L 1983, Structuralist macroeconomics: applicable models for the Third World, Basic, New York. Temple, J 2000, 'Inflation and growth: stories short and tall', Journal of Economic Surveys, vol. 14, no. 4, pp. 395-426. Thomas, V, Dailami, M, Dhareshwar, A, Kaufmann, D, Kishor, N, Lopez, R & Wang, Y 2000, The quality of growth, Oxford University Press for the World Bank. Tobin, J 1965, 'Money and economic growth', Econometrica, vol. 33, pp. 671 - 84. —— 1971a, 'Phillips Curve algebra', in Essays in Economics, North Holland Press, Amsterdam, vol. 2, pp. 11 - 5. —— 1971b, 'The wage-price mechanism', in Essays in Economics, North Holland Press, Amsterdam, vol. 2, pp. 17 - 32. —— 1972, 'Inflation and unemployment', American Economic Review, vol. 62, no. 1/2, pp. 1-18. —— 1987, Policies for prosperity: essays in a Keynesian mode, 1st edn, Wheatsheaf Books LTD, Brighton, Sussex. Todaro, MP & Smith, SC 2006, Economic development, 9 edn, Pearson Addison Wesley, Harlow, England. 216 Tommassi, M 1994, 'The consequences of price instability on search markets: toward understanding the effects of inflation', American Economic Review, vol. 84, pp. 1385 - 96. Tun Wai, U 1959, 'The relation between inflation and economic development: a statistical inductive study', IMF Staff Papers, vol. 7, pp. 302 - 17. UNDP 2000, Human development report 2000, New York. UNDP 2007, Overcoming vulnerability to rising oil prices: options for Asia and the Pacific, UNDP Regional Office, Bangkok. Unnayan Onneshan 2011, Bangladesh economic update March 2011: IMF's loan and its implications on Bangladesh economy, Unnayan Onneshan, Dhaka. Verbeek, M 2012, A guide to modern econometrics, Fourth edn, John Wiley and Sons Limited, Chichester. Vogel, RC 1974, 'The dynamics of inflation in Latin America, 1950 - 1969', American Economic Review, vol. 64, pp. 102 - 14. Vogelvang, B 2005, Econometrics theory and applications with EViews, First edn, Pearson Addison Wesley, London. Wilkinson, F 2000, 'Inflation and employment: is there a third way?', Cambridge Journal of Economics, vol. 24, pp. 643-70. Woo, WT, Glassburner, B & Nasution, A 1994, Macroeconomic policies, crises, and long-term growth in Indonesia, 1965-1990, The World Bank, Washington DC. World Bank 2012, Bangladesh: towards accelerated, inclusive and sustainable growth-opportunities and challenges, 67991, The World Bank. Yanikkaya, H 2003, 'Trade openness and economic growth: a cross-country empirical investigation', Journal of Development Economics, vol. 72, pp. 57-89. Yilmazkuday, H 2013, 'Inflation thresholds and growth', International Economic Journal, vol. 27, no. 1, pp. 1-10. 217 Young, A 1991, 'Learning by doing and the dynamic effects of international trade', Quarterly Journal of Economics, vol. 106, pp. 369-405. Zhang, J 2000, 'Inflation and growth: pecuniary transaction costs and qualitative equivalence', Journal of Money, Credit, and Banking, vol. 32, pp. 1 - 12. 218