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
Energy prices and business cycles: Lessons from a simulated small open economy model Torsten Schmidt, RWI Essen* Tobias Zimmermann, RWI Essen* Preliminary Version, October 2008 Abstract Despite energy price hikes in recent years growth rates turned out to be high in most industrialised countries. Given the well-known adverse effects of energy price shocks in the 1970s and 1980s, this may be puzzling. This study investigates if a reduction in the energy cost share or different sources of energy price hikes can explain this puzzle. By adding an exogenous two variable VAR to a new open economy model for Germany, it is considered that energy prices and the global economy are independent from domestic variables but influence each other. We show that calibrating the energy cost share to data averages is not sensible. Instead long run fluctuations in important observable structural parameters and VAR coefficients change the calibration of the model each period. By shocking the model by an increase in energy prices (global output), the effects of supply-driven (demand-driven) energy price increase are simulated. The results point out that the effects of recent energy price hikes have only been different, because they were demand-driven. Therefore, supply-driven energy price increases are still an important source of business cycle fluctuations. JEL Classification: E31, E32, F41 Keywords: Energy prices, new Keynesian open economy model * * Hohenzollernstraße 1-3, 45128 Essen. E-Mail: [email protected]. Hohenzollernstraße 1-3, 45128 Essen. E-Mail: [email protected]. 2 1. Introduction Energy prices have risen dramatically since 2004 while economic growth remains relatively high in most energy importing countries. This suggests that the current effects of energy price hikes are different from the seventies and early eighties. In general, there are two categories of explanations. First, the difference can be caused by a change in the economic structure, in particular a reduction of the energy cost share. In this case, the effects of supply-driven energy price shocks were reduced permanently. Second, recent energy price increases are likely to be demand-driven to a large extent, i.e. triggered by a soaring world economy. Thus, the favourable stance of the world economy might have compensated the adverse effects of more costly energy. By simulating a new open economy (NOE) model for the German economy, we analyse these two explanations more elaborately. In the model the size of the energy cost share determines the strength of adverse supply effects of energy price hikes. Due to the importance of this variable for our simulations we extract the underlying trend of this and other important variables and use them to change the calibration of the model for each period. Moreover, energy price and global economy shocks are supposed to be related. We account for this co-movement by adding an two variable VAR for energy prices and global GDP to the NOE model. Changes in the persistency and the relation between shocks are captured by varying the relevant parameters each period according to 40 period rolling window estimates. Two different types of energy price shocks are performed each period: supplydriven energy price shocks – simulated by an innovation to energy prices, and demand-driven energy price movements – simulated by an innovation to global output. The results of our simulations suggest that, energy prices are still an important source for business cycles in Germany. Except for the nineties, energy price hikes still have substantial negative effects if they were supply-driven; hence, the source of the shock matters a lot. Then, only the simulated responses to a global economy shock resemble the evolution of the energy price and important domestic variables during the recent energy price shock episode. Consequently, the effects of recent energy price hikes have only been moderate, because they were demand-driven to a large extent. Our results contribute to the ongoing discussion whether the effects of oil price shocks have changed since the early seventies. In favour of non-fundamental changes Hamilton (1996) stresses that the effects of energy price increases are still the same as in the seventies if one concentrates on strong oil price increases. In line with our approach, Kilian (2006) claims that the energy price shocks in the 3 seventies and early eighties were exogenous supply shocks caused by political events while the recent energy price hike was caused by excess demand, in particular related to increasing economic activity in transition economies like China and India. In contrast, other authors argue that the energy intensity of production has been reduced by industrial countries since the seventies (e.g. Blanchard, Gali 2007; Sanchez 2008). With the experience of the energy price shocks, firms increased their energy efficiency of production. Moreover, a trend towards more flexible real wages lead to smaller responses of real output and inflation after an energy price shock. Finally, stagflation in the U.S. during the seventies and early eighties was mainly caused by monetary policy (Barsky, Kilian 2001; Bernake, Gertler, Watson 1996). In this case, the political events in the Middle East only amplified the economic downturn. The outline of the paper is as follows: in the next section we describe some stylized facts concerning important economic variables. In section three we describe the model, the calibration and the solution methodology. Section four presents the simulation results. Section five summarizes and concludes. 2. Basic facts In this section we identify important energy price shocks and characterise the evolution of important macroeconomic variables during these episodes. It is shown that real variables evolve quite differently, whereas the magnitudes of the price increases are quite similar. Figure 1 visualises the developments of nominal and real prices of imported energy sources. The price series is the price index of imported energy provided by the Federal Statistical Office. Since 1970 there are four big jumps in real energy prices. The first jump in the energy price index occurred in 1974 from 40 to 140. The second increase from 80 to 160 started in 1979. The increase of energy prices after 1999, also proceeded in two steps. In the first step energy prices jumped from 40 to 100 and moderated to 90 in 2002. The second increase started in 2004 and reached in 2007 the level of 1981. 4 Figure 1. Energy shock episodes 240 200 160 120 80 40 0 1970 1975 1980 1985 1990 1995 2000 2005 Real pric e of im ported energy Nom inal pric e of im ported energy Notes: To get the real series the nominal price index of imported energy (2000 = 100) is deflated by the GDP-deflator. – Source: Federal Statistical Office. The movements of the price for imported energy sources during the four previously defined episodes are characterized in table 1. The last two columns show the change of the nominal and the real energy price index, respectively. During the first energy shock episode the real energy price increased by more than 130% and nearly doubled again between 1978:Q4 and 1981:Q3. Though the recent shock episodes show slight differences according to these numbers, it is unlikely that these differences are sufficient to explain the different effects on energy importing economies. Table 1. Energy shock episodes run-up period change in % (nominal) change in % (real) Shock 1 1973:Q3-1974:Q1 135 132 Shock 2 1978:Q4-1981:Q3 107 94 Shock 3 1999:Q1-2000:Q4 105 106 Shock 4 2004:Q1-2006:Q3 50 48 5 Table 2 reports the development of important macroeconomic variables during the episodes of substantial energy price increases. Here, growth rates are added eight quarters subsequently to the quarter in which the energy price starts to rise. Then the cumulated growth rates of the previous eight quarters are deducted. Inflation and interest rates are averaged over these periods. It is shown that the first two energy price shock episodes were accompanied by a substantial loss in GDP, while in the third and fourth episode there was a gain instead. Except for the first episode, when there was a loss, exports increased faster in periods of energy price shocks. However, the gains in exports were substantially higher in the recent two episodes than previously. There is also a slight change in the effects on inflation. During the first two energy price shocks inflation increased about 2.6 percentage points, while the inflation rate has increased moderately (by 1.4 percentage points) during the recent energy price shocks. It is therefore not surprising that the interest rate increased to a lesser extent during the more recent energy price shocks. However, table 2 shows that monetary policy has even lowered interest rates during the first energy shock episode and was all the more restrictive afterwards. The reaction with regard to the recent shocks seems to be more consistent. Table 2: Changes in selected macroeconomic variables during energy shock episodes Shock 1 Shock 2 Shock 3 Shock 4 GDP -7.4 -1.2 2.9 0.7 Exports -10.6 3.4 31.3 8.9 Inflation -0.3 5.5 1.3 1.5 Interest rate -2.5 1.5 0.7 0.5 3. Model simulations To investigate the reasons why similar increases in energy prices have quite different effects on the German economy, a new open economy- (NOE-) model in style of McCallum and Nelson (1999) is used. Like Kamps and Pierdzioch (2002) the model considers that energy is not only used for productive purposes, but is also consumed. According to the model, Germany is treated as a small open economy, i.e. the stance of the global economy and the real price of imported energy sources are modelled as exogenous shocks. By using a structural model, the effects of a lower energy cost share can be identified. However, after presenting the model setup, we show that the energy cost share in Germany exhibits long-run fluctuations, including a trend reversal, which are commonly not classified as 6 cyclical. The same is true for the relation between energy prices and world GDP growth. To capture these trends, a time-varying calibration of the models is proposed. According to this concept, the effects of both shocks have to be calculated separately for each period. 3.1 Model setup In this section we describe the NOE-model. Since the derivation of the model is well documented in the literature, only a brief description of the model is given. In the following lower case letters denote logarithms of the corresponding upper-case variables. Aggregate supply In the model energy is utilized to produce domestic output according to a CESproduction function: 1 υ1 υ Yt = ⎡⎢α1 N tυ1 + (1 − α1 ) ( EtY ) ⎤⎥ 1 . ⎣ ⎦ N t and EtY represent labour input or energy input, respectively. It is assumed that under price flexibility labour input equals one for all t . Therefore, energy input y under price flexibility, et , is determined by potential output, y t , and energy prices, pte : y et = y t − 1 pe . 1 − υ1 t Output under price flexibility is: y t = (1 − α1 ) ( eYSS ) et . υ1 y The parameter eYSS represents the energy cost share of production in the steady y state. By combining both relations, et can be eliminated. (1 − α1 ) ( eYSS ) υ1 yt = − υ (1 − υ1 ) ⎡⎢⎣1 − (1 − α1 ) ( eYSS ) 1 ⎤⎥⎦ p te (1) 7 Output under flexible prices, y t , negatively depends on real energy prices measured in domestic currency. Since the model abstracts from all imports besides energy and the real energy price is directly measured in Euro, the real exchange rate does not appear explicitly in equation (1). Because of certain monopoly power each firm treats the price of its good as a choice variable while the aggregate and foreign price level are taken as given. After setting the profit-maximizing price each firm produces whatever quantity of output is demanded. It is assumed that firms behave according to a price adjustment mechanism similar to the one introduced in Fuhrer and Moore (1995). This approach rationalizes a reasonable degree of inertia in inflation dynamics. More precisely, it claims that inflation, measured as the change of the price index of domestically produced goods, is a function of the output gap, y%t , and of the weighted average of lagged and expected inflation Δptc = 0.5 ( Δptc−1 + Δptc+1 ) + Ψ y%t . (2) Here, Ψ is a parameter depending on the degree of price stickiness in the economy. The stickier prices, the flatter the Phillips curve. The output gap y% t = yt − y t (3) is characterized as the difference between actual output, yt , and the amount of production that would prevail under flexible prices. Aggregate demand and monetary policy In this model energy prices affect total consumption, cxt in two ways since in case of an energy price hike both, the production of domestically produced goods, and the consumption of energy become more expensive. Optimal consumption is derived from households, maximizing their expected lifetime utility with respect to total consumption, real money balances, and domestic and foreign bonds. Preferences include habit formation, using a special case of the functional form proposed by Carrol et al. (1995). By combining the first order conditions with regard to consumption and bonds, the expectational difference equation for the change in consumption ( Δcxt ) with respect to expected future consumption, expected price level ( ptcx+1 ) , and the nominal interest rate ( Rt ) is: β g1 Et Δcxt + 2 + g 2 Et Δcxt +1 + g3 Et Δptcx+1 = g1Δcxt + g3 Rt (4) 8 where β is a discount factor, and g1 to g3 depend on β , risk aversion and habit persistence. 1 Total consumption, cxt , is the sum of domestic consumption, ct , and imported energy, which leads to the following linear approximated identities for the consumer price level, ptcx : ptcx = α ptc + (1 − α ) pte (5) where 1 − α stands for the average share of energy imports in total consumption, ptc is the price level for domestic consumption and pte the price level for imported energy. The following relation determines the demand for domestically produced consumption in dependence of the relative domestic price level ct = cxt − ( ptc − ptcx ) . (6) Monetary policy sets the nominal interest rate, Rt , according to a Taylor rule, which entails a reasonable degree of interest rate smoothing. ( ) Rt = μ0 + (1 − μ3 ) ⎡ Δpt cx + μ1 Δpt cx − π + μ 2 y% t ⎤ + μ3 Rt −1 ⎣ ⎦ (7) Here, π represents the inflation objective of the central bank. Open economy elements and goods market equilibrium The world demand shock enters the model through the export equation: ext = byt* + η qt . (8) Exports, ext , positively depend on global output, yt* , and the real exchange rate, qt . b and η are the income and exchange rate elasticity. The real exchange is the sum of the nominal exchange rate, st , and the difference between foreign and domestic price levels: 1 The exact definitions and calibrated values of all parameters of the model and can be found in the appendix. 9 qt = st + ptcx − ptcx . * (9) Additionally, the uncovered interest parity is assumed to hold: Rt = Rt* + Et Δst +1 + κ t . (10) Here, Rt and Rt* represent the domestic and foreign nominal interest rates which are defined as the sum of real interest rates, rt , and expected inflation ( Rt = rt + E Δpt +1 and Rt* = rt* + E Δpt*+1 ). The variable st stands for the nominal exchange rate. In this model the foreign variables are treated as exogenous. To close the model, we follow Schmitt-Grohe and Uribe (2003) by modelling the risk premium, κ t , as a function of the ratio of the nominal value of foreign bonds ( bt* ) and domestic nominal output ( yt ) : κ t = ϕ ( st + bt* − ptY − yt ) . (11) Since capital accumulation is ignored, total domestic production is spent solely for domestic consumption and exports. yt = (1 − ex SS ) ct + ex SS ext . (12) The parameter ex SS denotes the export share in the steady state. Equations (1) – (12) establish a system of 12 difference equations in the endogenous variables ext , ct , cxt , κ t , ptc , ptcx , qt , Rt , st , yt , y% t , yt . In addition we treat the energy price and all foreign variables as exogenous. 3.2. Time varying calibration In this section the calibration of the model is discussed. The most important measure to assess the supply side effects of energy price hikes is the energy cost share, eYSS , (figure 2). Several papers (e.g. Blanchard, Gali 2007; Schmidt, Zimmermann 2005, 2007; Sanchez 2008) distinguish two sub-periods to investigate the effects of price shocks in more and less energy intensive times. Obviously, the calibration of a low energy intensive era from the end of the eighties up to now seems not to be suitable any more. Even if the amount of energy (in physical units) in relation to GDP has remained lower than in the 70s and 80s, the ongo- 10 ing upward trend in energy prices since 2002 has pushed the energy cost share near to its previous peak in 1981. Figure 2. Energy usage and energy imports in proportion to GDP 120 100 80 6% 5% 60 4% 40 3% 2% 1% 0% 1970 1975 1980 1985 1990 1995 2000 2005 Energy usage (right axis) Energy imports (left axis) Notes: To calculate energy usage in proportion to GDP, nominal energy imports are deflated by the price index for imported energy and divided by real GDP. Since the level of this measure is arbitrary 1970 is set to 100. – Source: Federal Statistical Office. Therefore, a calibration on the basis of sample averages seems not to be convenient to capture cost trends adequately. Instead, the HP-filter is used to identify trends in the energy cost share which changes the calibration of the model for each quarter of our sample. Thus, long-run fluctuations which are commonly not classified as cyclical are employed for calibration purposes while the remaining fluctuations of the energy prices are interpreted as shocks.2 The identified cost trend sug2 To calculate the share of imported energy in total consumption, α , we assume that 70 percent of the imported energy is spent for production while 30 percent are consumed by private households. This 11 gests that the cost burden for the German economy, since 2005 is nearly as large as during the eighties. The same filter is applied to calibrate the export share, ex SS , which is an indicator of openness and therefore crucially determines the response of the domestic economy to world GDP shocks. This parameter shows an upward trend that has become steeper after 1995. 3 As mentioned above energy prices and the global economic activity are assumed to be not influenced by the German economy but linked. To capture these interactions a VAR(1) which contains the real energy prices and a measure of global output is estimated. E E ⎡ ptIM ⎤ ⎡ p IM ⎤ ⎡ε tIM ⎤ ⎢ * ⎥ = Γ1 ⎢ t *−1 ⎥ + ⎢ Y * ⎥ ⎣⎢ yt −1 ⎦⎥ ⎣⎢ ε t ⎥⎦ ⎣⎢ yt ⎦⎥ E (18) Both time series are HP-filtered before estimation. Thereby, long-run trends in energy prices are already considered by calibration, the remaining fluctuations are interpreted as shocks. Moreover, it is supposed that the relationship between the exogenous variables may have changed. By estimating a rolling window scheme of 40 quarters over the period from 1970:Q1 to 2007:Q4, trends in the relationship between, and the persistency of exogenous shocks are captured. The other parameters of the model are largely set to the same values as in the studies of McCallum and Nelson (1999) and Kamps and Pierdzioch (2002).4 However, by setting the parameter υ1 close to zero a very simple approach is chosen concerning production. Thus, the production function can be interpreted as Cobb Douglas and α1 stands for the non-energy cost share in domestic production. The supply side of our economy resembles the approach that has been taken in Kim and Loungani (1991) as well as Schmidt and Zimmermann (2007). Concerning preferences we also differ from the benchmark calibration in McCallum and Nelson (1999) by assigning σ the conventional value of 1. The coefficient of the risk premium equation, ϕ , is set to -0.02 (Ambler et al. 2004). Well established estimates of the coefficients of a Taylor rule for Germany can be found in the paper of Clarida et al. (1998). The point estimates for the parameters distribution is a very rough estimate based on information from the Arbeitsgemeinschaft Energiebilanzen. 3 Note that the corresponding parameter c SS is not calculated on the basis of the data. Since the model abstracts from saving and capital accumulation all output beside exports is spent for consumption purposes. c SS is therefore simply 1 − ex SS . 4 A table which entails all parameter values can be found in the appendix. 12 μ1 and μ 2 are 0.31 and 0.25, respectively. Faust et al. (2001) find similar values for these coefficients, albeit for slightly diverging periods. All of the mentioned papers suggest a distinctive tendency to smooth interest rates over time. For instance, according to Clarida et al. (1998), the point estimate of the relevant coefficient is equal to 0.91. 4. Simulation Results In this section impulse response functions, showing the reaction of endogenous variables to energy price and global economy shocks, are presented. Due to the time varying calibration proposed above, the responses to shocks are calculated at each point in time. In the first simulation, only long-run movements of the energy cost and the export share are applied whereas shocks remain unchanged. In this case, shocks are assumed to be independent AR(1)-process, that are estimated over the whole sample. The effects of energy price shocks on the German economy are plotted in figure 3. The period at which the initial impulse of a one percent increase in energy prices hits the German economy is depicted on the x-axis. The z-axis represents the number of periods after the initial shock impulse. The vertical axis shows the amount of reaction. Impulse response functions are calculated from 1971:Q1 to 2006:Q4. It is demonstrated that supply side effects of energy price shocks were negligible only in the nineties. At the end of the simulation sample (2006:Q4), the reaction of GDP amounts to over 70% of the maximum effect which is computed for 1982. The same is also true for exports, the price level, and the reaction of the monetary authority. Note that the final simulation starts at the end of 2006 and therefore shows what will happen, when energy prices start to rise from the level of 2006.To answer the question whether energy price shocks have changed it is suggestive to compare the effects at the beginning of the defined energy shock episodes. The magnitude of the reaction in 2004:Q1 is roughly 1/3 larger than in 1973:Q1 and 1/3 lower than in 1980:Q1. The current trend in the energy cost share therefore suggests that the recent energy price shock should have had effects similar to earlier shock episodes. The German economy has not reduced the energy intensity of production or the energy usage for consumption purposes sufficiently to prevent a trend reversal of the energy cost share. Consequently, energy importing countries have not become immune to supply side-driven energy price hikes. 13 Figure 3. Impulse response functions of the NOE-model to a one unit energy price shock, constant shock coefficients x 10 Output -3 x 10 -2 -4 -6 -8 -10 -12 -5 -10 -15 20 15 10 5 1980 1990 2000 20 Consumer price level 15 x 10 10 -3 5 1980 1990 2000 Interest rate 3 0.025 0.02 0.015 0.01 20 Exports -3 2 1 15 10 5 1980 1990 2000 20 15 10 5 1980 1990 2000 So far we have shown that the energy cost share is not able to explain why the effects of energy price shocks where severe in the seventies and early eighties, but have only negligible effects or are even accompanied with exceptional high growth rates nowadays. For a small export-oriented country like Germany the source of the shock matters a lot. On the one hand, the negative effects of higher energy prices can be compensated when they are accompanied or even triggered by a soaring world economy. On the other hand, the negative effects of higher energy prices can be aggravated when they are accompanied by a global recession. In the following simulations the energy prices and global output evolve independently from the German economy, but are allowed be interrelated. Trends in the behaviour of the exogenous shocks are employed in a two variable VAR. Since the coefficients are estimated in a 40 periods rolling window scheme, the first (latest) impulse response functions are available for 1975:1 (2002:Q4). Thus, the simulations can not ultimately answer the question what will happen if energy prices rise again in the current situation. However, simulating shocks at the beginning of 2003 should reasonably explain the evolution of macroeconomic variables during the latest run-up period (2004:Q1-2006:Q3). 14 The setup offers two different explanations for energy price hikes. Firstly, as before an energy price increase is modelled as an exogenous initial innovation. However, contrary to the former case this shock may be aggravated by global economic downswings. Secondly, energy prices are triggered by global output. This variant resembles a demand-driven energy price increase. We present both simulations, in the following. Since the effects on the German economy are quite different, we can decide which of the two variants yields a plausible explanation for different shock episodes. Figure 4 shows what happens to important macroeconomic variables subsequent to a one unit energy price hike when all time varying elements are included. In contrast to the former simulations differences in the impulse functions are not only caused by trends in structural parameter but also by a changing persistency of energy prices themselves and changing effects on the global economy. At first, it becomes evident that the persistency of the energy price itself was exceptionally large in the beginning of the 80s. Then, at the same time the estimated effects on the world economy increased. Thus, the adverse supply effects on the German output and exports were relatively persistent and heavily aggravated by a decrease in demand from abroad. This can be seen by comparing the magnitude of the responses to the ones which are depicted in figure 3. In consequence of the high energy price persistency during this time, the price level effects of energy price shocks are inimitable large from the beginning to the middle of the eighties. These simulations are therefore useful to understand the first energy shock episodes, because they reproduce not only the very large negative impact on domestic and worldwide aggregates, but also a harsh monetary reaction without referring to credibility problems. Key factors are the persistency of the energy price itself and its strong negative impact on the world economy. The last mentioned factor heavily aggravates the pure energy price shock for export intensive countries. 15 Figure 4. Impulse response functions of the NOE-model to a one unit energy price shock, time-varying shock coefficients Energy price 1 0.5 0 -0.5 20 15 10 5 World economic activity 0.1 0 -0.1 -0.2 -0.3 20 15 2000 1990 10 5 1980 Output 0.02 0 -0.02 20 15 10 5 0.04 0.02 0 -0.02 -0.04 20 15 10 5 Exports 0.05 0 -0.05 -0.1 20 15 2000 1990 10 5 1980 Consumer price level 2000 1980 1990 2000 1980 1990 x 10 4 2 0 -2 -4 20 15 2000 1980 1990 -3 Interest rate 10 5 2000 1980 1990 16 Even if the effects of energy prices where strongly aggravated in the seventies and early eighties but not in recent run-up periods, supply-driven energy price shocks in any case cause negative effects on real and positive effects on nominal domestic variables. Thus, these simulations are less appropriate to reproduce the very recent facts, which suggest that energy price shocks have no substantial effects or are even accompanied by unusual high growth rates and moderate inflationary pressure. In a final simulation exercise, the model is shocked by the same innovation to global output in each period. It is shown that these simulations provide a reasonable explanation of the stylized facts of the recent energy price episodes. Especially, after the end-nineties, shocks to global output are accompanied by increasing energy prices. Note that the magnitude of the innovations is chosen to cause approximately a one unit increase in energy prices in recent shock episodes. Moreover, domestic output and exports increase and the effects on the consumer price level are moderate, so that a minor reaction of the monetary authority succeeds. The final simulation, starting in 2002, matches exactly what can be observed in Germany (and in other small energy importing countries) from this point in time on up to now: a booming world economy, rapidly increasing energy prices, an increase in domestic output and exports, moderate inflationary pressure, and a moderate reaction of the central bank.5 The simulation of the end-eighties and nineties suggest that economic booms and energy prices exhibit a weak or even negative relationship, i.e. energy prices were not demand driven during this period. Surprisingly, the simulations, which are conducted for the time before 1980, show that according to the model the demand shock explanation is not completely implausible for previous energy shock episodes. Contrary to recent shock episodes, the negative effects of higher energy prices which succeed economic booms strongly overcompensate the weak positive initial effects on exports and output. As observed in this period the effects on these variables are therefore ultimately negative. However, there is strong evidence that energy supply was shortened previous to the earlier shock episodes. We conclude that the supply shock explanation plays at least a dominant role in the first energy shock episodes. 5 Note, that interest rates would probably be higher, if recent financial markets turbulences have not pushed central banks to keep interest rates constant. 17 Figure 5. Impulse response functions of the NOE-model to a global economy shock, time-varying shock coefficients Energy price 1 0.5 0 20 15 10 5 World economic activity 0.2 0 -0.2 20 15 2000 1990 10 5 1980 Exports Output 0.02 0 -0.02 20 15 0.1 0.05 0 -0.05 10 5 2000 1980 1990 Consumer price level 0.06 0.04 0.02 0 20 15 2000 1980 1990 10 5 2000 1980 1990 20 15 x 10 4 2 0 -2 20 15 10 5 2000 1980 1990 -3 Interest rate 10 5 2000 1980 1990 18 4. Conclusions This paper contributes to the ongoing debate on the changing effects of energy price shocks. We illustrate that during the recent energy price shock episodes the inflation rate and the interest rate increased moderately compared to the seventies and early eighties, while GDP and exports showed unaltered or even higher growth rates following recent energy shock episodes. We show that a permanent reduction of the energy cost share can not be a source for the different effects of energy price hikes on the German economy. While the energy cost share is an important factor for the effects of energy price shocks, it increased substantially since the late nineties. This ongoing upward trend in the energy cost share suggests that energy importing countries like Germany have become again quite vulnerable to energy price shocks in recent times. In addition, due to this upward trend a calibration of the energy cost share to sample averages is not useful to asses why the adverse supply effects of energy price might change over time. Instead we propose to indentify trends in observable structural parameters which change the calibration our model each period. Using a time varying calibration gives us also the opportunity to consider the changing relation between energy prices and the global economy by estimating a rolling window scheme. Simulations with an NOE-model suggest that the source of an energy price shock plays the major role from the perspective of a small open economy. Supply-driven energy price shocks can explain the stylized facts of the first shock episodes very well. Hereby, the succeeding worldwide economic downswing heavily aggravated the pure supply side effects. Moreover, since oil price increases were exceptionally persistent, even the harsh monetary reaction in the early eighties can be explained without referring to credibility problems or other special factors. Surprisingly, also the demand shock view is not completely implausible as an explanation for the earlier shock episodes. During these times the only weak positive effects of shocks to global production on domestic output were strongly overcompensated by the strong negative effects of succeeding energy price increases. On the contrary, the supply shock simulation does not yield a convenient explanation for recent energy price shocks and their consequences while a demand-driven energy price hike can. A world economic boom has no effects from the end of the eighties to the new century, but large energy price increases are the consequence afterwards. Since both shocks compensate each other, the positive reaction of domestic production and exports match the stylized facts. Though, the risk of an energy price induced recession seems to be limited for this time, the outlook is therefore miscellaneous. On the one hand, if energy price movements continue to be demand-driven for the main part, their effects will continue to be negligible. In this case two exogenous shocks will keep on compensating each other. On the other hand, if a new supply-driven energy price hike 19 which causes a worldwide recession takes place, the effects might be even stronger due to the larger openness of the energy importing economies. References Ambler, S., A. Dib and N. Rebei (2003), Nominal Rigidities and Exchange Rate PassThrough in a Structural Model of a Small Open Economy, Bank of Canada Working Paper 2003-29. Barsky, R. B. and L. Kilian (2004), Energy and the Macroeconomy since the 1970s. Journal of Economic Perspectives 18: 115-134. Bernanke, B. S., M. Gertler and M. Watson (1997), Systematic Monetary Policy and the Effects of Energy Price Shocks. Brookings Papers on Economic Activity 1: 91-157. Blanchard, O. J. and J. Gali (2007). The Macroeconomic Effects of Energy Shocks: Why are the 2000s so Different from the 1970s? NBER Working Paper 13368 National Bureau of Economic Research, Inc., Cambridge MA. Carrol, C. D., J. Overland and D. N. Weil (1995), Saving and Growth with Habit Formation, FEDS Working Paper No. 95-42. Federal Reserve Board. Clarida, R., J. Gali and M. Gertler (1998), Monetary Policy Rules in Practice Some International Evidence. European Economic Review 42: 1033-1067. Faust, J., J. H. Rogers and J. H. Wright (2001). An Empirical Comparison of Bundesbank and ECB Monetary Policy Rules, International Finance Discussion Papers No. 705. Board of Governors of the Federal Reserve System. Frondel, M. and Ch. M. Schmidt (2002), The Capital-Energy Controversy: An Artifact of Cost Shares? Energy Journal 23: 53-79. Fuhrer, J. C. and G. R. Moore (1995), Inflation Persistence. Quarterly Journal of Economics 109: 127-159. Hamilton, James D. (1996), This is what happened to the Energy Price-Macroeconomy Relationship. Journal of Monetary Economics 38: 215-220. Kamps, C. and C. Pierdzioch (2002), Monetary Policy Rules and Energy Price Shocks, Kiel Working Paper No. 1090. Kiel Institute of World Economics, Kiel. Kilian, Lutz (2008), Not all Energy Price Shocks are Alike: Disentangling Demand and Supply Shocks in the Crude Energy Market. American Economic Review: forthcoming. Kim, I.-M. and P. Loungani (1992), The Role of Energy in Real Business Cycle Models. Journal of Monetary Economics 29: 173-189. McCallum, B. T. and E. Nelson (1999), Nominal Income Targeting in an Open-Economy Optimizing Model. Journal of Monetary Economics 43: 553-578. Sánchez, M. (2008). Energy Shocks and Endogenous Mark-ups - Results from an Estimated Euro Area DSGE Model, European Central Bank Working Paper No. 860. European Central Bank, Frankfurt. Schmidt, T. and T. Zimmermann (2005), Effects of Energy Price Shocks on German Business Cycles, RWI Discussion Papers 31. Rheinisch-Westfälisches Institut für Wirtschaftsforschung, Essen. 20 Schmidt, T. and T. Zimmermann (2007). Why are the Effects of Recent Energy Price Shocks so Small?, Ruhr Economic Papers #29. Rheinisch-Westfälisches Institut für Wirtschaftsforschung, Ruhr-Universität Bochum, Universität Dortmund, Universität Duisburg-Essen, Essen. Schmitt-Grohe, S. and M. Uribe (2003), Closing Small Open Economy Models. Journal of International Economics 61: 163-185. 21 Appendix A: Parameters Table A.1: Constant parameters υ1 -0.001 σ 1 β 0.99 h 0.6 Ψ 0.02 b 0.33 η 0.33 ϕ -0.02 π 0 μ0 0 μ1 0.3 μ2 0.25 μ3 0.9 g1 = h − σ h g 2 = 1 + β h 2 − σβ h 2 − σβ h g 3 = σ (1 − β h ) 22 Appendix B: Data description GDP: Real Gross Domestic Product. Chain index. Seasonally adjusted by official source. National Accounts. Federal Statistical Office Germany. Consumer price level: GDP, implicit Price Deflator. National Accounts. Federal Statistical Office Germany. Consumption: Real private household consumption. Chain index. Seasonally adjusted by official source. National Accounts. Federal Statistical Office Germany. Exports: Real exports. Chain index. Seasonally adjusted by official source. National Accounts.Federal Statistical Office Germany. Real Energy imports: Nominal energy imports – GDP-deflated. Foreign trade statistics. Statistical Office Germany. Real energy price: Imported energy price index 2000 = 100 – GDP-deflated. Foreign trade price statistics. Federal Statistical Office Germany. Nominal interest rate: three month interbank rate. German Bundesbank. Global GDP: Real Gross Domestic Product of Belgium, Canada, Denmark, France, Italy, Japan, Korea, Mexico, Netherlands, Spain, Sweden, United Kingdom, USA. OECD.