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
A joint initiative of Ludwig-Maximilians University’s Center for Economic Studies and the Ifo Institute CESifo Conference Centre, Munich Area Conferences 2013 CESifo Area Conference on Applied Microeconomics 8 – 9 March The Impact of Uncertainty on Small Businesses: Evidence from Employment and the Number of Businesses Vivek Ghosal and Yang Ye CESifo GmbH · Poschingerstr. 5 · 81679 Munich, Germany Tel.: +49 (0) 89 92 24 - 14 10 · Fax: +49 (0) 89 92 24 - 14 09 E-mail: [email protected] · www.CESifo.org The Impact of Uncertainty on Small Businesses: Evidence from Employment and the Number of Businesses Vivek Ghosal1 and Yang Ye Version: December 2012 Work-in-progress Abstract Within the broad context of theoretical models related to financing-constraints and real-options, we examine the effects of economic uncertainty on small business employment and the number of small businesses. We construct measures of economic uncertainty related to GDP growth, inflation, the S&P500 stock price index, and fuel prices. Our findings are that economic uncertainty has a negative impact on employment and the number of businesses, and the effects are primarily felt by the relatively smaller businesses. The impact on large businesses are generally non-existent or quantitatively smaller. We note implications of our findings and comment on potential policies towards small businesses. JEL codes: L11, D80, O30, G10, L40. Keywords: uncertainty; small businesses; employment; number of firms; financing constraints; credit constraints; real options; economic stabilization policy. 1 Contact author: Vivek Ghosal, Professor, School of Economics, Georgia Institute of Technology, Atlanta, GA USA. Email: [email protected] Web: http://www.econ.gatech.edu/people/faculty/ghosal Yang Ye: Graduate student, School of Economics, Georgia Institute of Technology, Atlanta, GA USA. Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 2 1. Introduction In this paper we use data on U.S. small businesses to examine the effects of economic uncertainty on employment, as well as the total number of small businesses. While there is a significant literature that has examined the impact of uncertainty on various aspects of firms’ decision-making and key choice variables, we are not aware of any systematic study which has examined the uncertainty-employment link and for small businesses in particular. As we note below, given the economic importance of the small businesses, examining the effects of uncertainty on employment and number of small businesses may have important implications for the study of economic cycles as well as framing of policy. As we discuss in section 2, there are several channels identified in theory via which uncertainty may have important effects on firms’ key decision variables, such as physical capital investment outlays, entry and exit, employment and output. The broad classes of theories include: (1) information asymmetries between borrowers and lenders which may affect the flow of credit (e.g., Greenwald and Stiglitz, 1990); (2) real-options and the irreversibility of capital expenditures (e.g., Dixit and Pindyck, 1994); (3) firms’ attitudes towards risk (e.g., Appelbaum and Katz, 1986; Hartman, 1976); and (4) the convexity of the marginal product of capital (e.g., Abel, 1983). The first three channels tend to predict a negative effect of uncertainty on firms’ key decision variables. The fourth channel predicts positive effect. While there are a large number of papers that have presented the core models and many refinements, in this paper we focus on the credit-market and the real-options channels as these two have come to dominate the literature in terms of gaining insights into firms’ decision-making under uncertainty. While the theoretical channels present relatively clear predictions, empirical analysis, using firm, industry or macroeconomic data, reveals that while uncertainty generally appears to depress economic activity, the estimated quantitative effects and qualitative inferences vary considerably across the studies and the particular decision variable under consideration (investment, entry, output, among others). We take a quick look at the empirical literature in section 2. Our focus on small businesses is motivated by several factors. First, small businesses play an important role in the economy in several dimensions. The U.S. Small Business Administration (SBA) defines a small business as an independent business having fewer than 500 employees. According to the SBA, small businesses represent about 99 percent of all 3 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) employer firms, and employ about one-half of all private sector employees. Small businesses pay about 43 percent of total U.S. private payroll and have generated 65 percent of net new jobs over the past 17 years. Small businesses create more than half of the non-farm private GDP and made up 97 percent of all identified exporters and produced 31 percent of export value in 2008. Small businesses are also important to U.S. technology improvement. According to the SBA, small businesses hire over 40 percent of high tech workers (scientists, engineers, computer programmers, and others). Finally, small businesses generate patents more efficiently than large firms: they produce 16 times more patents per employee than large patenting firms. Irrespective of the specific dimension under consideration, small businesses play a vital role in the economy. The importance of small businesses is a global phenomenon. As noted in ACCA (2010) and di Giovanni et al. (2010), for example, the vast majority of businesses globally are very small. They note that small businesses are special because they have several important characteristics: they contribute disproportionately to local economic growth by improving innovation ability; increase job opportunities; and contribute significant tax revenues. Given these attributes, it has become commonplace in the aftermath of the current global economic crisis to refer to SMEs as the backbone of the global economy. Second, while the empirical literature on uncertainty is quite extensive, there have been relatively few studies that have distinctly emphasized the effects of uncertainty on the economic activity of small businesses. Examining this link is important, given the structural significance of small business in the economy, and during economic expansions and contractions. Third, those papers that have focused on the uncertainty-small business effects, have tended to be related to investment (e.g., Ghosal and Loungani, 2000). Our primary focus is on the uncertaintyemployment link, along with the impact on the number of businesses, and the potential distinction between the effects on small versus large businesses. Examining the effects of economic uncertainty on employment and creation and survival of businesses is important as it has taken on a larger than life significance following the 2007 onwards economic crisis, as well as past crises, where employment effects have taken center stage in policy debates on how to improve economic conditions. Against this backdrop, we examine the effects of economic uncertainty on small businesses. Our data on small businesses obtained from the SBA are US-wide, including all small, and other, businesses in the US and their employment. Since the small business data 4 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) portray a broader economy-wide picture, our uncertainty measures are constructed using economy-wide variables related to GDP, inflation, stock prices, and fuel prices. In section 2 we briefly note the central results from theory, present a cursory overview of the empirical evidence, and note the key hypotheses. Section 3 discusses our data. In section 4 we outline construction of our measures of economic uncertainty. Sections 5 and 6 describe the empirical specification we estimate, and present the results. Our findings are that economic uncertainty negatively affects employment and the number of businesses, and the impact is primarily concentrated in the relatively smaller business category. The impact on large businesses is either non-existent or quantitatively smaller. Section 7 concludes with a discussion of our findings and some policy issues. 2. Review of literature on the impact of uncertainty The literature on the impact of uncertainty on firms decisions is quite extensive. Below we briefly review selected papers and results in the theoretical and empirical literatures with an eye towards spelling out the core hypotheses for our empirical analysis. 2.1 Review of theoretical results In this section we briefly review some of the predictions from the theoretical literature and highlight the implications for our empirical analysis. There are two strands of literature which offer predictions on the impact on uncertainty on a range of firms’ decision variables. A large portion of this literature has posed the problem as one of firms’ investment decisions under uncertainty. But several researchers have examined the complementary issues related to production, employment, R&D, among other variables. In our discussion below we note it as an investment problem, and then note the implications for employment and number of businesses. 2.1.1 Uncertainty and financing constraints Information asymmetry between borrowers and lenders may tend to constrain the flow of credit. In an influential paper, Greenwald and Stiglitz (1990) model firms as maximizing expected equity minus expected cost of bankruptcy and examine scenarios where firms may be equity or borrowing constrained. A key result is that greater uncertainty exacerbates information asymmetries, tightens financing constraints and lowers capital outlays. Since uncertainty increases the risk of bankruptcy, firms cannot issue equity to absorb the risk. Those relying primarily on credit and 5 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) operating outside of equity markets, find the flow of credit constrained. Delli Gatti et al. (2003) develop a model in which the financial conditions of businesses affect capital accumulation and entry and exit which affect the distribution of firms differentiated by the equity ratio. In their model flow of exiting firms is endogenously determined through bankruptcy while the flow of entering firms is affected by stochastic factors. Lensink, Bo and Sterken (2001) provide a lucid discussion of financing constraints in the general context of investment behavior, including the roles played by uncertainty and sunk costs. Overall, this literature shows that periods of greater uncertainty widens the information asymmetry, increases the likelihood of bankruptcy and exacerbates financing constraints. Incumbent businesses who are more dependent on borrowing and adversely affected by tighter credit will have greater likelihood of exit. Similarly, entry is likely to be impeded for potential entrants who are more adversely affected by tighter credit conditions. Thus, periods of greater uncertainty are expected to accelerate exits and retard entry leading to negative net entry – that is, a decline in the number of firms in an industry. The results are similar if we examine physical capital investment or employment. Considering employment, for example, periods of greater uncertainty, by constraining credit flows, limit the ability of existing firms to grow. Since new entry is dampened, this negatively affects employment. As the likelihood of exit by businesses is greater, this also negatively affects employment. Overall, periods of greater uncertainty are likely to lead to lower employment. The literature points to not all businesses being equally affected by uncertainty and credit market conditions. In different strands of this literature, the papers by Evans and Jovanovic (1989) and Fazzari, Hubbard and Petersen (1988), for example, offer insights on small versus larger firm effects. Gertler and Gilchrist (1994), for example, strongly emphasize the negative impact of information asymmetric and credit constraints on smaller firms. Tying the two literatures a prediction that emerges from is that periods of greater uncertainty, via the credit market channel discussed above, is likely to have a dampening effect on employment and the number of businesses. Given that the effects of financing constraints, at the margin, are more likely to be borne by the smaller businesses, the prediction is that the negative effect of uncertainty will be likely be greater on smaller businesses. 2.1.2 Uncertainty and real-options An alternative channel examines the effects of uncertainty on firms’ decisions via the realoptions effects. The results in Caballero and Pindyck (1996), Dixit (1989) and Dixit and Pindyck Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 6 (1994) are sufficient to generate the essential predictions for our empirical analysis, and we briefly discuss the results. Dixit (1989) shows that uncertainty and sunk costs imply an option-value of waiting and this raises the entry trigger price and lowers the exit trigger price.2 For prices below the entry trigger, the potential entrant holds on to its option to enter, and an incumbent firm does not exit at prices above the exit trigger. Another way of looking at this is that to enter during periods of greater uncertainty, firms would require a premium over the conventional Marshallian entry price. And incumbent firms would wait longer to exit (i.e., let prices fall below average variable cost before they exit) as they know that to re-enter the market they would have to re-incur the sunk entry costs. The key results related to entry and exit can be summarized as follows: (1) numerical results in Dixit (p.632-33) show that even small amounts of uncertainty are sufficient to generate significant changes in entry and exit patterns; and (2) the numerical simulations in Dixit (1989, p.632-33) and Dixit and Pindyck (Ch. 7, p.224-228; and Ch. 8.) show that an increase in uncertainty results in the entry trigger price increasing by more than the decrease in the exit trigger. These results reveal that entry is affected more than exit leading to negative net entry. In other words, periods of greater uncertainty lead to a decrease in entry but exits, while lower, continue at a closer to normal pace resulting in the industry experiencing a decrease in the number of firms. The results are similar if we examine physical capital investment or employment. Dixit and Pindyck (1994) present an extensive analysis of the detrimental effects of uncertainty on investment. The effects on employment will be qualitatively similar. If greater uncertainty retards entry and investment, it will also be expected to have a negative effect on employment. Assessing the implications for our empirical analysis, we note the following. One of our interests is to evaluate whether these effects may be different for small versus large firms. Previous studies have shown that the vast majority of the entry and exit churning is for the small businesses (e.g., see Audretsch (1995), Sutton (1997) and Caves (1998)). The literature shows that: (1) entrants are typically small compared to incumbents and have high failure rates; (2) the typical exiting firm is small and young; and (3) larger firms are older with higher survival rates. This implies that, via the real-options channel, most of the negative effects of uncertainty on the number of businesses and employment will likely be observed for the smaller businesses. 2 Dixit and Pindyck (1994) outline the theoretical framework for studying firms’ decision-making regarding entry and exit under uncertainty and sunk entry costs. Hopenhayn (1992) and Pakes and Ericsson (1998), for example, study firm dynamics with firm-specific uncertainty. These models, however, are best subjected to empirical tests using micro-datasets as in Pakes and Ericsson, and is beyond the scope of our paper. 7 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 2.3 Overview of empirical findings The literature examining the effects of uncertainty is quite extensive. Studies vary considerably in terms of the underlying variables they use to measure uncertainty, such as profits, input prices, stock prices, inflation, GDP growth, economic policy news, survey of businesses or forecasters, among others. Equally, there is a large variety in terms of the specific constructs to measure uncertainty, such as conditional or unconditional variance (or standard deviation), among others. Finally, some studies use micro-data at the firm level, while others use data at the industry or macroeconomic level. The literature is quite expansive and it is difficult to review the full spectrum of this literature. To offer a perspective in a convenient format, we present a table in Appendix A which summarizes selected papers. These are not meant to be comprehensive, but merely display the variety of constructs used to measure uncertainty (GDP, inflation, prices, energy prices, stock prices, among others), the level of aggregation of the studies (firm or industry level, economywide, or cross-country), and the estimated quantitative and qualitative effects. Regarding the effects of uncertainty on small business, we briefly note some of the findings in the empirical literature useful for our analysis. Using industry-level data, Ghosal and Loungani (2000) find investment-uncertainty relationship is negative and this negative impact is greater in industries dominated by small firms. Koetse and Vlist (2006) find that there are differential effects of uncertainty on input and output variables. Bianco, Bontempi, Golinelli and Parigi (2012) find family firms’ investments are significantly more sensitive to uncertainty than non-family firms. 2.4 Hypotheses Based on our discussion of theory and the summary of the results in sections 2.1 and 2.2, the following hypotheses emerge: H1: Greater uncertainty is expected to negatively affect employment, and this effect may be more pronounced in smaller businesses. H2: Greater uncertainty is expected to negatively affect the number of businesses, and this effect may be more pronounced for smaller businesses. 8 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 3. Data description For our econometric analysis, we use data from multiple sources. Below we provide details of the datasets we use. First, is the U.S. Small Business Administration (SBA) database. From the SBA database we use annual data on the number of firms according to the size of firms from 1988 to 2011. (Data prior to 1988 are not available.) The SBA database also provides employment data for businesses by the size of firms. Second, U.S. macroeconomic data on real GDP and GDP implicit price deflator are from the Federal Reserve Economic Data. The data on S&P 500 stock price index is from Yahoo Finance. Finally, we obtain data on fuels and related products and power PPI from the U.S. Bureau of Labor Statistics. To consider size classes for our employment and number of businesses, we use the following taxonomy to parse out the small business related effects versus other. The overall sample we examine are: 1. ‘All’ businesses; 2. ‘Large’ businesses – these are businesses with ≥500 employees; 3. ‘Small’ businesses – these are businesses with <500 employees; and 4. ‘Smaller’ businesses – these are businesses with <20 employees. The 500 employee cutoff is the one used by the U.S. SBA, and we use this as the baseline. We consider an additional cutoff of <20 employees for the following reasons: (a) a 500 employee firm is relatively large, so we wanted to consider an alternate cutoff for defining small; (b) data with the <20 employee cutoff was available consistently for both our variables (employment and number of businesses); and (c) a large percentage of the truly small business fall in this category. In our estimation of employment and number of businesses specifications, we will present estimates for each of the four groupings (1-4) noted above. In Figures 1 and 2 we display the time paths for the employment levels and number of businesses by size class. Table 1 presents the summary statistics for the employment and number of businesses by size class. 4. Measures of uncertainty To examine the impact of economic uncertainty on small businesses, we construct alternate measures. As we discussed earlier, our objective is to examine the effects of economic Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 9 uncertainty on small businesses. Our data on small businesses are US-wide and include all small and larger businesses in the US and their employment. Since the data are designed to portray a broader picture, our uncertainty measures are created using economy-wide indicators. The four measures of uncertainty we construct relate to uncertainty about: 1. real GDP growth. This indicates the overall state of the economy capturing demand and supply effects;3 2. inflation rate. We measure inflation as the annual percentage growth of the GDP deflator. Inflation uncertainty captures effects related to input and product prices, as well as affecting firms’ real borrowing rates;4 3. stock prices. We use the S&P500 stock price index. As with real GDP, this is an indicator of the overall state of the economy, and forward looking indicator of investor and business confidence;5 and 4. real fuel price growth. Here we use the BLS price index of a range of commonly used fuels by businesses. The nominal fuel price index is converted to real values after deflating by the implicit GDP deflator. This variable serves to proxy a critical input – fuels and energy – price for businesses.6 To create our measures of uncertainty, we use the following procedure. We assume that firms use a forecasting equation to predict future values of the relevant variable, an economywide indicator in our case. As noted earlier, our objective is to examine the effects of economic uncertainty on small businesses. Our data on small businesses is US-wide (i.e., all small businesses in the US and their employment). Our uncertainty measures, therefore, are created using economy-wide variables related to GDP, inflation, stock prices, and fuel prices. As our baseline, we use an autoregressive, AR(2), specification as the forecasting model. AR(n) models are based on Box and Jenkins (1970) formulation for forecasting economic variables, and historically they have performed well in forecasting exercises (e.g., Meese and Geweke (1984) and Marcellino, Stock and Watson (2003)). Given this aspect, and our annual 3 GDP uncertainty measures, in different forms, have been used by, for example, Driver et al. (2005), Asteriou et al. (2005) and Bloom (2009). 4 Inflation has been used to construct uncertainty measures by, for example, Huizinga (1993), Fountas et al. (2006) and Elder (2004). 5 Stock prices have been used to measure uncertainty by, for example, Bloom (2009), Chen et al. (2011), Bloom et al. (2007), Greasley et al. (2006), and Stein et al. (2010). 6 Fuel and energy prices have been used to construct uncertainty measures by, for example, Koetse et al. (2006), Kilian (2008) and Guo et al. (2005). Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 10 time-series data for the small business variables, we use this standard methodology to forecast the relevant state variables. The forecasting specification is: where Z is either: (i) real GDP growth; or (ii) inflation rate; or (iii) S&P500 growth; or (iv) real fuel price growth. From specification (1), the predicted values represent the forecastable component. The residuals: represent the unsystematic, or unforecastable, component. Since values, we use the squared value of can take positive or negative , interpreted as the conditional-variance, as our measure of uncertainty about the relevant variable. Using the residuals from forecasting equations and conditional-variance to construct uncertainty measures is common: see, for example, the insightful review in Lensink, Bo and Sterken (2001), and the references cited in the above footnotes. If the forecasting specification is for GDP (i.e., Z is real GDP growth), then we denote the uncertainty measure as: Using this procedure for our four underlying economy-wide variables (Z), we obtain four measures of uncertainty denoted by: (i) ; (ii) ; (iii) ; and (iv) .7 To estimate the forecasting specification (1), we use available data on Z from 1960 to 2010. The rationale is as follows. The objective is obtain a good forecasting specification, which is better done with a longer time-series ensuring that the parameters of the equation are estimated precisely. While data on GDP, inflation and S&P 500 are available for earlier periods, the BLS fuel price indices are available starting 1960. So we start the time-period for the equation in 1960 for which all of our variables (that we consider for Z) are available. The terminal period, 2010, is 7 While we report the results for an AR(.) specification, our results were robust to experiments with lag lengths. Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 11 the same as the last period for which the small business data are available. This implies that the generated time-series in our four uncertainty measures , , and are over 1962-2010 (two initial observations are lost due to use of lagged values). A final remark we make is that while our AR(.) forecasting equation is fairly standard, the length and frequency (annual from 1988-2010) of the available small business variables restricts us from considering alternate procedures, such as using ARCH models, to construct measures of uncertainty. While this may be viewed as a limitation, the SBA data on employment and number of businesses are simply not available at a higher frequency. Since we need to harmonize the estimation periods and frequency, we are restricted to using annual data to forecast our specification (1). 5. Estimating Specification To examine the hypotheses noted in section 2.4, we estimate autoregressive-distributed lag AD(n, m) specification of the form: where is a measure of small business activity (in our case, employment or number of businesses), growth, and or or is the measure of uncertainty measured in natural logarithms, is real GDP is the error term. As described earlier, the four measures of uncertainty are or . In specification (4), the autoregressive component has one lag, and the distributed lag (for uncertainty and GDP) also contain one lag implying an AD(1,1) model. Since we are interested in the shorter-term effects of uncertainty on small business employment or number of businesses, and that the data displayed in Figures 1 and 2 show a clear trend, we measure in annual logarithmic first-differences (annual percentage changes). is included in the specification to capture any persistence in the dependent variable. The two real GDP growth variables, and , are included to control for overall economic conditions. As noted in the data section, the available data on Small Businesses are annual and cover the period 1988-2010. We, therefore, estimate specification (4) over 1988-2010. While, as noted earlier, our four generated uncertainty measures , , and are over 1962- Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 12 2010, we only use their values over 1988-2010 in estimating specification (4). Estimating (4) informs us about the impact of uncertainty on employment and number of businesses, after controlling for overall economic activity (GDP growth) and the dynamic lagged structure for the included variables.8 In our examination of the effects of uncertainty on employment and number of businesses, we will present estimates for four groups noted in our data section 3: 1. ‘All’ businesses; 2. ‘Large’ businesses – these are businesses with ≥500 employees; 3. ‘Small’ businesses – these are businesses with <500 employees; and 4. ‘Smaller’ businesses – these are businesses with <20 employees. 6. Estimation results << Preliminary >> The results for the employment specifications are presented in tables 2.1 and 2.2. Table 2.1 presents estimates for ‘All’ and ‘Large’ businesses, and table 2.2 presents estimates for the ‘Small’ and ‘Smaller’ size classes. A similar format is followed for presenting the results on the number of businesses in tables 3.1 and 3.2. In table 2.1, the are generally in the 80% range indicating good fit of the specification. The estimates of the first-order autocorrelation coefficient, ρ, are relatively low. The estimates show that uncertainty related to GDP, Inflation and Fuel prices have a negative effect on ‘All’ business employment, but the quantitative effects vary. Turning to the ‘Large’ businesses, only Inflation and Fuel price uncertainty dampen employment, and the estimated quantitative effects are smaller for the Large business group. In the small business groups in table 2.2, the vary in the 63% to 78% range. The estimates of the first-order autocorrelation coefficient, ρ, are is relatively low for the ‘Smaller’ group, and somewhat higher for the ‘Small’ group. The estimates show that uncertainty related to GDP, Inflation and Fuel prices have a negative effect on employment for the ‘Smaller’ businesses. None of the uncertainty effects are significant for the broader ‘Small’ business group. Examining the broad inferences from the estimates in tables 2.1 and 2.2, uncertainty appears to dampen employment, and this effects is primarily concentrated in the smaller businesses. 8 Our experiments using longer lag lengths did not provide additional insights into the effects of uncertainty. Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 13 Next we turn to the specifications for the number of businesses in tables 3.1 and 3.2. In table 3.1, the for ‘All’ businesses specifications are in the 73% to 82% range, while they are only in the 13% to 29% range for the ‘Large’ businesses, indicating a rather low explanatory power for the latter group. The estimates of the first-order autocorrelation coefficient, ρ, are relatively low. The estimates show that uncertainty related to GDP and Inflation have a negative effect on the number of ‘All’ business, but the quantitative effects vary. Fuel price or S&P500 uncertainty does not appear to affect the number of ‘All’ businesses. Turning to the ‘Large’ businesses, only Fuel price uncertainty appears to decrease the number of these businesses. For the smaller businesses specifications in table 3.2, the are in the 60% range for the ‘Smaller’ group, and in the 90% range for the ‘Small’ group. The estimates of the first-order autocorrelation coefficient, ρ, are uniformly low. The estimates show that uncertainty related to GDP and have a negative effect on the number of ‘Smaller’ business, but Fuel price and S&P500 uncertainty does not appear to affect this group. Turning to the ‘Small’ group, GDP and Inflation uncertainty have a negative effect on the number of businesses. Based on the estimates in tables 3.1 and 3.2, uncertainty has a negative impact on the number of businesses, and this effect appears to be generally concentrated in the smaller business groups. The overall inferences we draw from the set of estimates presented in tables 2 and 3 are that economic uncertainty dampens employment and the number of businesses, and the effects appear to be concentrated in the Smaller (<20 employees) and Small (<500 employees) groups. At broad brush, our findings appear supportive of the results from the theoretical models we discussed in sections 2.1 and 2.2. Before closing this section we note that the estimated specifications include real GDP growth, and model the dynamics of the included variables via lagged effects. Real GDP growth is probably the single most important control variable in either employment or number of businesses specifications. With increasing GDP growth, business opportunities are expected to expand allowing for creation of more jobs as well as new businesses. The fact that we find effects related to uncertainty even after controlling for GDP growth and the lagged dynamics, makes our finding even more noteworthy. << Additional results and checks of robustness to be added >> Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 14 7. Discussion of results and conclusions Our findings on the negative impact of economic uncertainty on employment in small businesses as well as the number of small businesses are particularly revealing as these results appear even after controlling for GDP growth and appropriate controls for lagged dynamics of the included variables. The underlying theories we reviewed related to potential financingconstraints and the real-options channels. Given that the data we use from the U.S. Small Business Administration is relatively aggregated, it is difficult to disentangle which of these two channels may be playing the more dominant role. It is fair to assume that both channels are important in determining the outcomes. As was noted earlier, there exists a significant literature on financing-constraints. This literature has noted important differences between smaller and larger businesses, and point to smaller firms being relatively credit-constrained (e.g., Fazzari, Hubbard and Petersen, 1988; Gertler and Gilchrist, 1994; Evans and Jovanovic, 1989; Lensink, Bo and Sterken, 2001; Ghosal and Loungani, 2000; Himmelberg and Petersen, 1994; Winker, 1999; and Audretsch and Elston, 1997). Viewing our estimation results and the above information on financing constraints collectively, part of our results for the smaller business is undoubtedly emerging from the financing-constraints channel. If this is true, then important policy implications emerge, primarily in the form of initiatives and instruments designed to partly ease the credit-constraints faced by smaller businesses. To the question as to why governments might pay special attention to small businesses, there are several responses. First, we noted earlier in the paper that a large fraction of employment and businesses fall into the smaller categories. Second, a number of emerging structural factors – such as those related to globalization and banking sector consolidation – are likely to favor large businesses relative to the smaller ones. These considerations alone provide important economic policy justification. As with many governments worldwide, the U.S. has recently implemented policies and programs to help small businesses bridge the capital and market gap and encouraged publicprivate partnerships to support small business and entrepreneurship by, for example: (a) supporting more than $53 billion in SBA loan guarantees to more than 113,000 small businesses; (b) awarding more than $221 billion in Federal contracts to small businesses (FY 2009 through Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 15 April 30, 2011); and (c) awarding more than $4.5 billion in research funding through the Small Business Innovation and Research Program during FY 2009 and FY 2010.9 Such initiatives, along with appropriate lending policies, can help ease some of the financing-constraints faced by smaller businesses.10 By doing so, and in the context of this paper, such policies may also help alleviate some of the negative impact of uncertainty on smaller businesses. 9 The National Economic Council (2011) and Sheets and Sockin (2012) provide extensive discussion on the importance of small businesses and policy. 10 The papers by Audretsch and Elston (1997, 2002), for example, provide important insights in this dimension from German policy initiatives. 16 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) References Abel, Andrew. “Optimal Investment under Uncertainty,” American Economic Review, 1983, 228–233. Appelbaum, Elie, and Eliakim Katz. “Measures of Risk Aversion and Comparative Statics of Industry Equilibrium,” American Economic Review, 1986, 524–529. Association of Chartered Certified Accountants (ACCA). “Small Business: A Global Agenda,” ACCA Accountants for business Report, 2010. Asteriou, Dimitrios, and Simon Price. “Uncertainty, Investment and Economic Growth: Evidence from a Dynamic Panel,” Review of Development Economics, 2005, 277–288. Audretsch, David. Innovation and Industry Evolution. Cambridge: MIT Press, 1995. Audretsch, David, and Julie Elston. “Does Firm Size Matter? Evidence on the Impact of Liquidity Constraints on Firm Investment Behavior in Germany,” International Journal of Industrial Organization, 2002, 1–17. Audretsch, David, and Julie Elston. “Financing the German Mittelstand,” Small Business Economics, 1997, 97-110. Baker, Scott, Nicholas Bloom and Steven Davis. “Has Economic Policy Uncertainty Hampered the Recovery?” Chicago Booth Paper, No. 12-06, 2012. Baker, Scott, Nicholas Bloom and Steven Davis. “Measuring Economic Policy Uncertainty,” University of Chicago and Stanford University, 2012. Bianco, Magda, Maria Elena Bontempi, Roberto Golinelli and Giuseppe Parigi. “Family Firms’ Investments, Uncertainty and Opacity,” Small Business Economics, 2012, 1-24. Bloom, Nicholas. “The Impact of Uncertainty Shocks,” Econometrica, 2009, 623-685. Bloom, Nicholas. “The Impact of Uncertainty Shocks: Firm Level Estimation and a 9/11 Simulation,” CEP Discussion Paper, No. 718, 2006. Bloom, Nick, Stephen Bond and John van Reenen. “Uncertainty and Investment Dynamics,” Review of Economic Studies, 2007, 391-415. Bo, Hong, and Elmer Sterken. “Volatility of the Interest Rate, Debt and Firm Investment: Dutch Evidence,” Journal of Corporate Finance, 2002, 179–193. Box, G.E.P., and , G.M. Jenkins. Time Series Analysis, Forecasting and Control. San Francisco: Holden Day, 1970. 17 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) Bulan, Laarni. “Real Options, Irreversible Investment and Firm Uncertainty: New Evidence from U.S. Firms,” Review of Financial Economics, 2005, 255-279. Caballero, Ricardo, and Robert Pindyck. “Investment, Uncertainty and Industry Evolution,” International Economic Review, 1996, 641-662. Caves, Richard. “Industrial Organization and New Findings on the Turnover and Mobility of Firms,” Journal of Economic Literature, 1998, 1947-1982. Crain, Nicole, and W. Mark Crain. “The Impact of Regulatory Costs on Small Firms,” Small Business Administration, 2010. Delli Gatti, Domenico, Mauro Gallegati, Gianfranco Giulioni and Antonio Palestrini. “Financial Fragility, Patterns of Firms’ Entry and Exit and Aggregate Dynamics,” Journal of Economic Behavior and Organization 51, 2003, 79-97. Di Giovanni, J., Levchenko, A. A. and Rancière, R. “Power Laws in Firm Size and Openness to Trade: Measurement and Implications,” IMF Working Paper No. 10/109. Driver, Ciaran, Paul Temple and Giovanni Urga. “Profitability, Capacity, and Uncertainty: A Model of UK Manufacturing Investment,” Oxford Economic Papers, 2005, 120–141. Driver, Ciaran, and Brendan Whelan. “The Effect of Business Risk on Manufacturing Investment,” Journal of Economic Behavior and Organization 44, 2001, 403-412. Evans, David and Boyan Jovanovic. “An Estimated Model of Entrepreneurial Choice under Liquidity Constraints,” Journal of Political Economy, 1989, 808-827. Fazzari, Steven, Glenn Hubbard and Bruce Petersen, “Financing Constraints and Corporate Investment.” Brookings Papers on Economic Activity, 1988, 141-195. Folta, Timothy, and Jonathan O’Brien. "Entry in the Presence of Dueling Options," Strategic Management Journal, 2004, 121-138. Folta, Timothy. “Governance and Uncertainty: The Trade-off Between Administrative Control and Commitment,” Strategic Management Journal, 1998, 1007-1028. Fuss, Catherine, and Philip Vermeulen. “Firms’ Investment Decisions in Response to Demand and Price Uncertainty,” Applied Economics, 2008, 2337-2351. Gertler, Mark, and Simon Gilchrist. “Monetary Policy, Business Cycles, and the Behavior of Small Manufacturing Firms,” Quarterly Journal of Economics, 1994, 309-340. Ghosal, Vivek, and Prakash Loungani. “The Differential Impact of Uncertainty on Investment in Small and Large Businesses,” Review of Economics and Statistics, 2000, 338-343. 18 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) Ghosal, Vivek, and Prakash Loungani. “Product Market Competition and the Impact of Price Uncertainty on Investment: Some Evidence From US Manufacturing Industries,” Journal of Industrial Economics, 1996, 217-228. Ghosal, Vivek. “The Effects of Uncertainty and Sunk Costs on Firms’ Decision-Making: Evidence from Net Entry, Industry Structure and Investment Dynamics.” In The Economics of Imperfect Markets, Giorgio Calcagnini and Enrico Saltari (Eds.), Contributions to Economics, Springer: New York, 2009, Chapter 9, 167 – 182. Ghosal, Vivek. “Does Uncertainty Influence the Number of Firms in an Industry?” Economics Letters 50, 1996, 229 – 237. Ghosal, Vivek. “Demand Uncertainty and the Capital-Labor Ratio: Evidence from the U.S. Manufacturing Sector,” Review of Economics and Statistics 73, 1991, 157 – 162. Greasley, David, and Jakob Madsen. “Investment and Uncertainty: Precipitating the Great Depression in the United States,” Economica, 2006, 393–412. Greenwald, Bruce, and Joseph Stiglitz. “Macroeconomic Models with Equity and Credit Rationing,” in Hubbard, R. Glenn. ed., Asymmetric Information, Corporate Finance, and Investment. Chicago: University of Chicago Press, 1990, 15-42. Guo, Hui, and Kevin Kliesen. “Oil Price Volatility and U.S. Macroeconomic Activity,” Federal Reserve Bank of St. Louis Review, 2005, 669-83. Hartman, Richard. “Factor Demand with Output Price Uncertainty,” American Economic Review, 1976, 675–681. Himmelberg, Charles, and Bruce Petersen. “R & D and Internal Finance: A Panel Study of Small Firms in High-Tech Industries,” Review of Economics and Statistics, 1994, 38-51. Hopenhayn, Hugo. “Entry, Exit and Firm Dynamics in Long Run Equilibrium,” Econometrica, 1992, 1127-1150. Huizinga, John. “Inflation Uncertainty, Relative Price Uncertainty, and Investment in U.S. Manufacturing,” Journal of Money, Credit and Banking, 1993, 521-549. Kilian, Lutz. “Exogenous Oil Supply Shocks: How Big Are They and How Much Do They Matter for the U.S. Economy?” Review of Economics and Statistics. 2008, 216-240. Koetse, Mark J., Arno J. van der Vlist and Henri L. F. de Groot, “The Impact of Perceived Expectations and Uncertainty on Firm Investment,” Small Business Economics, 2006, 365-376. Lensink, Robert, Hong Bo and Elmer Sterken Investment. Capital Market Imperfections, and Uncertainty: Theory and Empirical Results. Edgar Elgar, 2001. 19 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) Lensink, Robert, Paul van Steen and Elmer Sterken. “Uncertainty and Growth of the Firm,” Small Business Economics, 2005, 381-391. Marcellino, Massimiliano, James Stock and Mark Watson. “Macroeconomic Forecasting in the Euro Area: Country Specific versus Euro Wide Information,” European Economic Review, 2003, 1-18. Meese, Richard, and John Geweke. “A Comparison of Autoregressive Univariate Forecasting Procedures for Macroeconomic Time Series,” Journal of Business and Economic Statistics, 1984, 191-200. National Economic Council. “The Small Business Agenda: Growing America’s Small Businesses to Win the Future,” 2011. Oriani, Raffaele, and Maurizio Sobrero. “Uncertainty and the Market Valuation of R&D within a Real Options Logic,” Strategic Management Journal, 2008, 343-361. Pakes, Ariel, and Richard Ericson. “Empirical Implications of Alternate Models of Firm Dynamics,” Journal of Economic Theory, 1998, 1-45. Saemundsson, Rognvaldur, and Asa Lindholm Dahlstrand. “How Business Opportunities Constrain Young Technology-Based Firms from Growing into Medium-Sized Firms,” Small Business Economics, 2005, 113-129. Schmitt, John, and Nathan Lane “An International Comparison of Small Business Employment,” Center for Economic and Policy Research (CEPR) Reports and Issue Briefs, 2009. Sheets, Nathan, and Robert Sockin. “Does Size Really Matter?: The Evolving Role of Small Firms in the U.S. Economy,” Empirical & Thematic Perspectives, Citi Research, 2012. Stein, Luke, and Elizabeth Stone. “The Effect of Uncertainty on Investment: Evidence from Options,” Stanford University Working Paper, 2010. Sutton, John. “Gibrat’s Legacy,” Journal of Economic Literature, 1997(a), 40-59. U.S. Small Business Administration. Advocacy: the voice of small business in government. “Frequently Asked Questions about Small Business Finance,” http://www.sba.gov/sites/default/files/files/Finance%20FAQ%2082511%20FINAL%20for%20w eb.pdf Winker, Peter. “Causes and Effects of Financing Constraints at the Firm Level,” Small Business Economics, 1999, 169-181. Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 20 Figures: employment levels (fig. 1.1-1.4) and number of businesses (fig. 2.1-2.4) by size class Figure 1.1. Total employment 120 100 80 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Figure 1.2. Employment in large (≥500 employees) businesses 65 55 45 35 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Figure 1.3. Employment in small (<500 employees) businesses 65 60 55 50 45 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Figure 1.4. Employment in small (<20 employees) businesses 22 21 20 19 18 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 21 Figure 2.1. Total number of businesses 6.5 6.0 5.5 5.0 4.5 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Figure 2.2. Number of large (≥500 employees) businesses 0.020 0.018 0.016 0.014 0.012 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Figure 2.3. Number of small (<500 employees) businesses 6.5 6.0 5.5 5.0 4.5 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Figure 2.4. Number of small (<20 employees) businesses 5.5 5.0 4.5 4.0 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 22 Tables Table 1. Summary Statistics Variable By size class Employment EMP Size: All EMP Size: Large w ≥500 employees EMP Size: Small w <20 employees EMP Size: Small w <500 employees Number of businesses FIRMS Size: All FIRMS Size: Large w ≥500 employees FIRMS Size: Small w <20 employees FIRMS Size: Small w <500 employees Mean Std. Dev. 106.540 10.665 51.999 6.797 20.134 1.048 54.563 3.900 5.539 0.347 0.016 0.002 4.956 0.305 5.523 0.345 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 23 Tables: Estimates << Preliminary estimates – additional results to be added >> Table 2.1. Employment Specifications Dependent variable: Size Group: All Businesses Const S1 -0.0251* (0.060) -0.3356* (0.031) 0.6972* (0.001) 0.7932* (0.001) 0.0031* (0.048) -0.0032* (0.011) - Size Group: Large Businesses S2 -0.0584* (0.006) -0.3338* (0.052) 0.5971* (0.001) 0.8326* (0.001) - S3 -0.0295* (0.001) -0.1424 (0.279) 0.5680* (0.001) 0.7371* (0.001) - S4 -0.0485* (0.001) -0.4121* (0.018) 0.4462* (0.001) 0.9239* (0.001) - - - - - - - - - - -0.0005 (0.694) -0.0027* (0.022) - -0.0059 (0.739) -0.0307 (0.838) 0.7417* (0.001) 0.7056* (0.001) 0.0041 (0.130) -0.0024 (0.167) - - - - - - -0.0005 (0.427) -0.0011 (0.174) - - - - 0.837 0.814 0.798 - S1 S2 -0.0497* (0.035) 0.0802 (0.360) 0.5890* (0.001) 0.6046* (0.003) - S3 -0.0240* (0.011) 0.1824* (0.008) 0.6098* (0.001) 0.5634* (0.002) - S4 -0.0382* (0.001) 0.0212 (0.832) 0.5119* (0.001) 0.6823* (0.001) - - - - - - - - - -0.0010 (0.550) -0.0019* (0.096) - - - - - -0.0039* (0.002) -0.0017 (0.177) 0.869 - - 0.0001 (0.920) -0.0012 (0.145) - - - - 0.802 0.774 0.767 -0.0026* (0.079) -0.0017 (0.159) 0.795 Ρ 0.110 0.126 -0.061 -0.224 0.083 0.183 0.078 -0.069 Notes: 1. p-values (two-tailed test), based on efficient standard errors, are reported in parentheses. An * denotes significance at least at the 10% level. All specifications are estimated using data over 1988-2010. 2. The first-order autocorrelation coefficient is denoted by ρ. 3. The variable definitions are as follows. (As noted in section 5, the uncertainty terms are measured in logarithms.) = annual percentage change in employment. = annual percentage change in real GDP. = gdp uncertainty. = inflation uncertainty. = S&P 500 uncertainty. = fuel price uncertainty. Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 24 Table 2.2. Employment Specifications Dependent variable: Size Group: Small<20 Employees Businesses Const S1 -0.0406* (0.001) -0.0536 (0.773) 0.4056* (0.001) 0.0986 (0.495) -0.0012 (0.325) -0.0023* (0.011) - Size Group: Small<500 Employees Businesses S2 -0.0356* (0.043) 0.1118 (0.617) 0.4386* (0.001) 0.0359 (0.817) - S3 -0.0154* (0.001) 0.1600 (0.413) 0.4435* (0.001) 0.0826 (0.580) - S4 -0.0237* (0.011) -0.0202 (0.913) 0.3880* (0.001) 0.1542 (0.192) - - - - - - - - - - -0.0019* (0.064) -0.0006 (0.620) - S1 -0.0532* (0.008) -0.1196 (0.701) 0.6390* (0.003) 0.3619 * (0.024) -0.0006 (0.765) -0.0030 (0.106) - - - - - - -0.0004 (0.427) -0.0008 (0.246) - - - - 0.727 0.632 0.615 - S2 -0.0560* (0.045) 0.0753 (0.775) 0.5855* (0.001) 0.3721* (0.010) - S3 0.0344* (0.001) -0.3362* (0.065) 0.5429 * (0.001) 0.7774 * (0.001) - S4 -0.0332 * (0.099) -0.0785 (0.869) 0.5628* (0.001) 0.4896 * (0.001) - - - - - - - - - -0.0025 (0.139) -0.0008 (0.640) - - - - - -0.0028* (0.002) -0.0001 (0.912) 0.715 - - -0.0003 (0.655) -0.0015 (0.103) - - - - 0.684 0.658 0.718 -0.0031 (0.174) 0.0006 (0.773) 0.683 ρ 0.146 -0.015 0.019 0.085 -0.272 -0.414 -0.046 -0.332 Notes: 1. p-values (two-tailed test), based on efficient standard errors, are reported in parentheses. An * denotes significance at least at the 10% level. All specifications are estimated using data over 1988-2010. 2. The first-order autocorrelation is denoted by ρ. 3. The variable definitions are as follows. (As noted in section 5, the uncertainty terms are measured in logarithms.) = annual percentage change in employment. = annual percentage change in real GDP. = gdp uncertainty. = inflation uncertainty. = S&P 500 uncertainty. = fuel price uncertainty. Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 25 Table 3.1. Number of Businesses Specifications Dependent variable: Size Group: All Businesses S1 Const S2 Size Group: Large Businesses -0.0186 (0.207) 0.6352* (0.002) 0.3736* (0.001) -0.2613* (0.008) - S3 -0.0091* (0.049) 0.5691* (0.001) 0.3950* (0.001) -0.1813* (0.031) - - - - - - - - - - -0.0021* (0.013) 0.0005 (0.600) - S1 -0.0065 (0.846) -0.4450 (0.013) 0.3575 (0.318) 0.6893* (0.086) 0.0035 (0.332) -0.0030 (0.387) - - - - - - -0.0006 (0.215) -0.0005 (0.385) - - - - 0.729 0.732 0.703 -0.0201* (0.059) 0.4916* (0.004) 0.3278* (0.001) -0.1249 (0.160) -0.0021* (0.018) 0.0002 (0.868) - S4 S2 -0.0133 (0.813) -0.3788* (0.045) 0.1198 (0.682) 0.8189 * (0.063) - S3 -0.0295* (0.062) -0.3725* (0.018) 0.1125 (0.734) 0.7453* (0.030) - S4 -0.0488* (0.064) -0.4430* (0.015) 0.0684 (0.830) 0.6541* (0.027) - - - - - - - - - -0.0013 (0.776) 0.0009 (0.831) - - - - - -0.0007 (0.368) 0.0005 (0.630) 0.682 - - -0.0017 (0.387) 0.0028 (0.244) - - - - 0.158 0.132 0.218 -0.0046 (0.384) 0.5705* (0.007) 0.3985* (0.001) -0.1686* (0.098) - -0.0065* (0.029) -0.0025 (0.499) 0.290 ρ 0.120 -0.006 -0.054 0.035 -0.228 -0.231 -0.344 -0.346 Notes: 1. p-values (two-tailed test), based on efficient standard errors, are reported in parentheses. An * denotes significance at least at the 10% level. All specifications are estimated using data over 1988-2010. 2. The first-order autocorrelation coefficient is denoted by ρ. 3. The variable definitions are as follows. (As noted in section 5, the uncertainty terms are measured in logarithms.) = annual percentage change in number of businesses. = annual percentage change in real GDP. = gdp uncertainty. = inflation uncertainty. = S&P 500 uncertainty. = fuel price uncertainty. Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 26 Table 3.2. Number of Businesses Specifications Dependent variable: Size Group: Small<20 Employees Businesses S1 Const S2 -0.0210 (0.240) 0.5124* (0.024) 0.3447* (0.007) -0.2439* (0.013) - -0.0068 (0.198) 0.4854* (0.011) 0.3604* (0.001) -0.1699* (0.051) - -0.0052 (0.266) 0.4700* (0.024) 0.3683* (0.001) -0.1658* (0.066) - - - - - - - - - - -0.0020* (0.032) 0.0001 (0.885) - S1 -0.0185* (0.001) 0.6303* (0.001) 0.1954* (0.003) 0.0917* (0.070) -0.0003 (0.638) -0.0010 * (0.034) - - - - - - -0.0004 (0.396) -0.0005 (0.453) - - - - 0.618 0.603 0.564 -0.0197 (0.140) 0.3991* (0.058) 0.2806 * (0.004) -0.1099 (0.251) -0.0024* (0.021) 0.0003 (0.793) - S3 Size Group: Small<500 Employees Businesses S4 S2 -0.0180* (0.043) 0.6370* (0.001) 0.1966* (0.002) 0.1088 * (0.029) - S3 -0.0098* (0.001) 0.6498* (0.001) 0.1805* (0.005) 0.1456* (0.001) - S4 -0.0106* (0.009) 0.6553* (0.001) 0.1631* (0.003) 0.1425 * (0.001) - - - - - - - - - -0.0006 (0.213) -0.0004 (0.521) - - - - - -0.0007 (0.462) 0.00002 (0.987) 0.545 - - -0.0003 (0.188) -0.0002 (0.448) - - - - 0.919 0.901 0.899 -0.0011* (0.027) 0.0003 (0.521) 0.918 ρ 0.024 -0.101 -0.035 -0.004 -0.175 -0.224 -0.187 -0.273 Notes: 1. p-values (two-tailed test), based on efficient standard errors, are reported in parentheses. An * denotes significance at least at the 10% level. All specifications are estimated using data over 1988-2010. 2. The first-order autocorrelation is denoted by ρ. 3. The variable definitions are as follows. (As noted in section 5, the uncertainty terms are measured in logarithms.) = annual percentage change in number of businesses. = annual percentage change in real GDP. = gdp uncertainty. = inflation uncertainty. = S&P 500 uncertainty. = fuel price uncertainty. Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 27 Appendix A: Selected empirical findings on the impact of uncertainty The table below highlights the variables which are used to measure uncertainty, specific constructs for uncertainty, and empirical findings. The papers included below are not meant to represent a comprehensive survey, but merely to show the diversity of research in this area. Table A.1. Selected papers examining the effects of uncertainty Paper Data Estimation method State variables Uncertainty measure Estimation results Lensink, Robert, Paul van Steen and Elmer Sterken. “Uncertainty and Growth of the Firm,” Small Business Economics, 2005, 381-391. Survey of 1,097 Dutch firms in 1999. Sales. Return on Investment. Logit model. Coefficient of variation. Uncertainty has a negative impact on the size of investment, no matter what the type of investment is used. In the level-level regression, one unit increase in measured uncertainty is estimated to reduce the investment (which is scaled by sales) by 0.288. In the logit model, for one unit increase in uncertainty, the log odds of investment decreased by 0.878. Koetse, Mark J., Arno J. van der Vlist and Henri L.F. de Groot. “The Impact of Perceived Expectations and Uncertainty on Firm Investment,” Small Business Economics, 2006, 365-376. Survey of 135 plant locations in Netherlands in 1998. Wages. Energy prices. Output prices. Tobit model. Survey based. Bo, Hong, and Elmer Sterken. “Volatility of the interest rate, debt and firm investment: Dutch evidence,” Journal of Corporate Finance, 2002, 179–193. Data for 41 Dutch listed firms from 1984 to 1995. Interest rate. Driver, Ciaran, and Brendan Whelan. “The Effect of Business Risk on Manufacturing Investment,” Journal of Economic Behavior and Organization, 2001, 403-412. Disaggregated survey data of Ireland in 1995. Conditional variance. Panel Data, ARCH model, Fixed effect estimation. The paper analyzes the channel how uncertainty affects firms’ investment by comparing the percentage of different respondents in the survey questions. Future demand and future price Future unit input cost Capacity Delay risk Subjective descriptions For small firms, that there are important differences between the effects of uncertainty about input and output variables. The regression results show that one point increase in wage uncertainty is associated with 0.06 unit decrease in predicted value of investment to sales; while the increasing of output price uncertainty by one point is estimated to increase the predicted value of ratio of aggregate firm investment to sales by 2.44% point. Using fixed effect estimation, they find that an increase in the interest rate of 1 standard deviation leads to an expected decrease in the investment to capital ratio by 0.27 standard deviations holding other explanatory variables constant. One standard deviation increase in the interaction between debt and the volatility of the interest rate leads to an increase in the investment to capital ratio by 0.07 standard deviations. No strong effect of risk due to convexities, though where price-taking behavior dominated, as in the Food sector, there seemed a weaker negative effect on investment from demand or price uncertainty. Disequilibrium effect recorded in the survey could cause an upward investment bias for those firms worried about capital shortage. However, in practice firms may not invest more than under certainty because the effect is blunted by risk aversion. Risk did affect the timing of investment for between a quarter and a third of the sample. The greatest 28 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) Oriani, Raffaele, and Maurizio Sobrero. “Uncertainty and the Market Valuation of R&D within a Real Options Logic,” Strategic Management Journal, 2008, 343-361. Data for 290 manufacturing firms in UK from 1989 to 1998. Bianco, Magda, Maria Elena Bontempi, Roberto Golinelli and Giuseppe Parigi. “Family Firms’ Investments, Uncertainty and Opacity,” Small Business Economics, 2012, 1-24. Data for 2,959 Italian private companies from 1996 to 2007. Bloom, Nick, Stephen Bond and John Van Reenen. “Uncertainty and Investment Dynamics,” Review of Economic Studies, 2007, 391-415. Data for 672 UK manufacturing firms from 1972 to 1991. Ghosal, Vivek, and Prakash Loungani. “Product Market Competition and the Impact of Price Uncertainty on Investment: Some Evidence From US Manufacturing Industries,” Journal of Industrial Economics, 1996, 217-228. Data for 254 US 4-digit SIC manufacturing industries from 1958 to 1989. Fuss, Catherine, and Philip Vermeulen. “Firms' Investment Decisions in Response to Demand and Price Uncertainty,” Applied Economics, 2008, 2337-2351. Survey of 279 firms from 19872000, and another survey of 319 firms from 1987-1999. Bulan, Laarni. “Real Options, Irreversible Investment and Firm Uncertainty: New Evidence from U.S. Firms,” Review of Financial Economics, 2005, 255-279. Data for 2,901 US firms from 1964 to 1999. Panel Data, Hedonic model Industry output. Patents. Absolute percentage difference. Inverse of the median age. Sales. Coefficient of variation. Panel data, GMM Panel data, GMM Panel data, fixed effect – IVE. Stock returns. Std. deviation of daily stock returns. Product price. Rolling regression based conditional std. deviation. Expectations of future demand and prices. Theil index. Panel data, GMM. Equity returns. Standard deviation. Panel data, 2SLS. caution in respect of timing was in the Hi-tech sector which was also the sector with the greatest damage from delay. They find that increased market uncertainty reduces the value of R&D investment until a certain level of market uncertainty is reached; whereas beyond that level it augments the value. To be specific, this is a U-shaped relationship between Market Uncertainty and Value of investment. Also, they find an inverted U-shaped relationship between technological uncertainty and the value of R&D capital. Family firms’ investments are significantly more sensitive to uncertainty than nonfamily firms and the greater sensitivity to uncertainty is basically due to the greater opacity of family firms and to their higher risk aversion, rather than to the degree of sunk fixed capital as is typical in the literature on investment decisions. A reduction in the degree of uncertainty from the third to the first quartile of its sample distribution induces an increase in planned investments of 1.69% for the median family firm, while investment plans of non-families are nearly unchanged; also, a 0.27-0.36% increase in investment plans associated with an equal uncertainty reduction. “Cautionary effects” of uncertainty are large – going from the lower quartile to the upper quartile of the uncertainty distribution typically halves the first year investment response to demand shocks. A negative relationship between investment and price uncertainty only exists in competitive industries. One percentage increases in price uncertainty is estimated to cause the ratio of gross industry investment (I/K) decrease by 0.358 for most competitive industries. Demand uncertainty at the time of planning depresses planned and subsequent realized investment. One standard deviation increases in demand uncertainty is estimated to reduce 6% of the average investment ratio. Increased industry uncertainty and firm-specific uncertainty display a pronounced negative effect on firm investment consistent with real options behavior. A one standard deviation increase in industry uncertainty reduces a firm’s investment-to-capital ratio by 6.4%. A one standard deviation increase in firm- 29 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) specific uncertainty decreases firm investment by 19.3%. In contrast, a one standard deviation increase in Tobin’s q increases investment by only 5%. Ghosal, Vivek, and Prakash Loungani. “The Differential Impact of Uncertainty on Investment in Small and Large Businesses,” Review of Economics and Statistics, 2000, 338343. Data for 330 US SIC 4-digit manufacturing industries from 1958 to 1991. Profits. Panel data, IVE. Rolling regression based conditional std. deviation.. Huizinga, John. “Inflation Uncertainty, Relative Price Uncertainty, and Investment in U.S. Manufacturing,” Journal of Money, Credit and Banking, 1993, 521-549. Data for 450 U.S. SIC 4-digit manufacturing industries from 1954 to 1989. Real wage. Output price. Real materials price. Bivariate ARCH model. Conditional std. deviation. Stein, Luke C.D., and Elizabeth C. Stone. “The Effect of Uncertainty on Investment: Evidence from Options,” Stanford University Working Paper, 2010. Data for 2,230 US manufacturing firms from 1996 to 2009. Stock price. Folta, Timothy, and Jonathan P. O’Brien. “Entry in the Presence of Dueling Options,” Strategic Management Journal, 2004, 121-138. Data for 2,230 US manufacturing firms from 1996 to 2009. Expected volatility. Panel data, 2SLS. Panel data, 2SLS. Data on 17,897 firms from 1980 to 1999. GARCH model. Industry’s contribution to GDP. Square root of conditional variance. Investment-uncertainty relationship is negative and this negative impact is greater in industries dominated by small firms. The estimated profit uncertainty in small firms is in -0.80 ranges, while that for large firms are around -0.14. Therefore, it appears that as large-firm dominance increases, the impact of uncertainty on investment gets quantitatively weaker. A one-standard-deviation increase in estimated conditional standard deviation of the real wage from is estimated to reduce the predicted value of ratio of capital spending to output, by 0.72. A one-standard-deviation increase in estimated conditional standard deviation of output price would eventually lower the expected value of ratio of capital spending to output, by 2.9. A one-standard-deviation increase in estimated conditional standard deviation of profit rate is estimated to immediately increase predicted value of ratio of capital spending to output, by 0.74. and if sustained would, ceteris paribus, increase the predicted value by 1.99. They find a negative and statistically significant relationship between uncertainty and investment that is robust across a variety of specifications. The coefficient estimates are larger in magnitude after addressing the endogeneity of the uncertainty measure, suggesting potential reverse causation that biases the OLS estimates towards zero. Given a standard deviation of implied volatility of 0.24 in data sample, a one standard deviation increase in uncertainty is associated with a 0.7% decline in the quarterly investment rate. After adding Tobin’s q into the model, a one standard deviation increase in market-wide uncertainty is associated with a 0.1% decline in a firm’s investment rate. They find the effect of uncertainty on entry is non-monotonic and U-shaped, for the coefficient for Uncertainty2 is positive (1.00), while the coefficient for Uncertainty is negative (-4.16). When there are potentially strong early mover advantages (i.e., Scale Advantages at 95th percentile), the effect of Uncertainty on entry turns positive at about the 54th percentile of Uncertainty. In contrast, the effect of 30 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) Uncertainty does not turn positive until around its 94th percentile for most industries. Baker, Scott, Nick Bloom and Steven J. Davis. “Has Economic Policy Uncertainty Hampered the Recovery?” Chicago Booth Paper, No. 12-06, 2012. Index of economic policy uncertainty, news-based proxy, government purchases data, disagreement about future indexes from 1985 to 2011. And tax code expiration data is from 1991 to 2011. Economic policy uncertainty, news-based proxy, government purchases data, disagreement about future indexes, and merge them into a new proxy. VAR model Aggregating the above Components to Obtain an Index of Economic Policy Uncertainty Baker, Scott, Nicholas Bloom and Steven Davis. “Measuring Economic Policy Uncertainty,” University of Chicago and Stanford University, 2012. Constructing index data about political uncertainty from 19852009. VAR model Newspaper coverage of policy-related economic uncertainty. Number of federal tax code provisions set to expire in future years. Disagreement among economic forecasters. Constructing an index from three types of underlying components. Bloom, Nicholas. “The Impact of Uncertainty Shocks,” Econometrica, 2009, 623-685. Firm-level data from 2548 firms U.S. firms from 1981 to 2000. Panel data, VAR model, SMM estimation Driver, Ciaran, Paul Temple and Giovanni Urga. “Profitability, capacity, and uncertainty: a model of UK manufacturing investment,” Oxford Economic Papers, 2005, 120– 141. Aggregate data of UK manufacturing on two capital assets (machinery and building) from 1972 to 1999. Asteriou, Dimitrios and Simon Price. “Uncertainty, investment and economic growth: evidence from a dynamic panel,” Review of Development Economics, 2005, 277– 288. Data of 59 industrial and developing countries from 1966 to 1992. Profits growth, stock returns, TFP growth, GDP forecasts. Standard deviation. Output growth. Time-series conditional volatility. GARCH model Panel data, PMG, MG, GARCH GDP per capita growth. Conditional variance. Their analysis indicates that the historically high levels of policy uncertainty in 2010 and 2011 mainly reflect concerns about tax and monetary policy and secondarily a broader range of other policy-related concerns. Policy-related concerns now account for a large share of overall economic uncertainty. A rise in policy uncertainty, similar in magnitude to the actual change since 2006, is associated with substantially lower levels of output and employment over the following 36 months. Policy related uncertainty played a role in the slow growth and fitful recovery of recent years, and they invite further research into the effects of policyrelated uncertainty on economic performance. VAR estimates show that a policy uncertainty shock equal in size to actual increase in the index value from 2006 to 2011 foreshadows drops in private investment of 16 percent within 3 quarters, industrial production drops of 4 percent after 16 months, and aggregate employment reductions of 2.3 million within two years. Uncertainty appears to dramatically increase after major economic and political shocks like the Cuban missile crisis, the assassination of JFK, the OPEC I oil-price shock, and the 9/11 terrorist attacks. Uncertainty shocks appear to have large real effects, based on the assumption that firms have non-convex adjustment costs: the uncertainty component alone generates a 1% drop and rebound in employment and output over the following 6 months, and a milder long-run overshoot in these economic shocks. The GARCH model shows uncertainty variable for the full sample are estimated to be negatively significant at the 5% level for machinery, and negative but not significant for building. Using PMG estimation, they find the estimated coefficients were -0.360 for the industrial and -0.053 for the developing countries, which showed a negative relationship between growth and uncertainty in both cases. For the effect of uncertainty on investment, the PMG results revealed a significant negative relationship for both subgroups, with higher negative magnitudes for the industrial countries (-3.280) than that of developing 31 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) countries (-0.081). Bloom, Nick. “The Impact of Uncertainty Shocks: Firm Level Estimation and a 9/11 Simulation,” CEP Discussion Paper, No. 718, 2006. Data of 579 US manufacturing firms from 1991 to 2000. Greasley, David, and Jakob B. Madsen. “Investment and Uncertainty: Precipitating the Great Depression in the United States,” Economica, 2006, 393–412. Data of investment information in US from 1920 to 1938. Kilian, Lutz. “Exogenous Oil Supply Shocks: How Big Are They and How Much Do They Matter for the U.S. Economy?” Review of Economics and Statistics. 2008, 216-240. Monthly production data for all OPEC countries and for aggregate non-OPEC oil production since 1973. Monthly share returns. Standard deviations. Panel data, time-varying second moment model Tobin’s q Real stock prices. Squared monthly proportional change. Oil production. Exogenous variation. OLS Guo, Hui, and Kevin L. Kliesen. “Oil Price Volatility and U.S. Macroeconomic Activity,” Federal Reserve Bank of St. Louis Review, 2005, 669-83. Data of daily price of U.S. 1month futures and 12- month futures contracts from 1983 to 2004. Oil prices. Realized variance series. Forecasting regression John Elder. “Another Perspective on the Effects of Inflation Uncertainty,” Journal of Money, Credit & Banking, 2004, 911-928. Data of U.S. output growth rate, inflation, CPI, commodity price from 1966 to 2000. Cross-section data, MGARCH-M VAR model Inflation Conditional variance Monthly neighborhood returns Conditional variance Stilianos Fountas, Menelaos Karanasos and Jinki Kim. “Inflation Uncertainty, Output Growth Uncertainty and Macroeconomic Performance,” Oxford Bulletin of Economics and Statistics. 2006, 319- Monthly data for the G7 on PI and the IPI. Inflation Output growth Bivariate GARCH model, VAR model Conditional variance Uncertainty appears to vary strongly over time, temporarily rising by up to 200% around major shocks like the Cuban Missile crisis, the assassination of JFK and 9/11. Based on a model with a time varying second moment, it finds temporary impact of a second moment shock is different from the typically persistent impact of a first moment shock. While the second moment effect has its biggest drop in month 1 and has completely rebounded by month 5, a persistent first moment shock will generate a drop in activity lasting several quarters. For investment, the results here show that the effects of heightened uncertainty surrounding the expected marginal profitability of capital, measured by share price volatility, can explain around 80% of the actual fall in the business fixed investment ratio in 1930. Thus, to a large extent, the investment slump led the declines in income. Using this approach and the new exogenous oil supply shock measure, it finds statistically significant evidence of a sharp drop in real GDP growth five quarters after an exogenous oil supply shock and of a spike in CPI inflation three quarters after the shock. And it is shown that exogenous oil supply shocks made remarkably little difference overall for the evolution of U.S. real GDP growth and CPI inflation since the 1970s. A volatility measure constructed using daily crude oil futures prices has a negative and significant effect on future GDP growth over the period 19842004. Moreover, the effect becomes more significant after oil price changes are also included in the regression to control for the symmetric effect. The main empirical result is that uncertainty about inflation has significantly reduced real economic activity over the post-1982 period, with the effect concentrated after a twomonth lag. Using MGARCH-M VAR model, they find that one standard deviation increase in inflation uncertainty has tended to reduce real economic activity over three months by about 22 basis points in the post-1982 period. First, inflation does cause negative welfare effects, directly and indirectly. Secondly, in some countries, e.g. Canada and the UK, more inflation uncertainty provides an incentive to Central Banks to surprise the public by Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 32 343. raising inflation unexpectedly. Thirdly, in contrast to the assumptions of some macroeconomic models, business cycle variability and the rate of economic growth are related. More variability in the business cycle leads to more output growth. The Impact of Uncertainty on Small Businesses: Evidence from Employment and the Number of Businesses Vivek Ghosal1 and Yang Ye Version: February 18 2013 Work-in-progress Abstract Within the broad context of theoretical models related to real-options and financing-constraints, we examine the effects of economic uncertainty on small business employment and the number of small businesses. We construct a range of measures of economic uncertainty derived from survey forecasts, as well as those constructed from our estimation of forecasting models for GDP growth, inflation, S&P500 stock price index, and fuel prices. Our findings are that economic uncertainty has a negative impact on employment and the number of businesses, and the effects are primarily felt by the relatively smaller businesses. The impact on large businesses are generally non-existent or weaker. We note the linkages of our findings to the underlying theories, and comment on potential policy implications. JEL codes: L11, D80, O30, G10, L40. Keywords: uncertainty; small businesses; employment; number of firms; financing constraints; credit constraints; real options; economic stabilization policy. 1 Contact author: Vivek Ghosal, Professor, School of Economics, Georgia Institute of Technology, Atlanta, GA USA. Email: [email protected] Web: http://www.econ.gatech.edu/people/faculty/ghosal Yang Ye: Graduate student, School of Economics, Georgia Institute of Technology, Atlanta, GA USA. Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 2 1. Introduction In this paper we use data on U.S. small businesses to examine the effects of economic uncertainty on employment, as well as the total number of small businesses. While there is a significant literature that has examined the impact of uncertainty on various aspects of firms’ decision-making and key choice variables, we are not aware of any systematic study which has examined the uncertainty-employment link and for small businesses in particular. As we note below, given the economic importance of the small businesses, examining the effects of uncertainty on employment and number of small businesses may have important implications for the study of economic cycles as well as framing of policy. As we discuss in section 2, there are several channels identified in theory via which uncertainty may have important effects on firms’ key decision variables, such as physical capital investment outlays, entry and exit, employment and output. The broad classes of theories include: (1) information asymmetries between borrowers and lenders which may affect the flow of credit (e.g., Greenwald and Stiglitz, 1990); (2) real-options and the irreversibility of capital expenditures (e.g., Dixit and Pindyck, 1994); (3) firms’ attitudes towards risk (e.g., Appelbaum and Katz, 1986; Hartman, 1976); and (4) the convexity of the marginal product of capital (e.g., Abel, 1983). The first three channels tend to predict a negative effect of uncertainty on firms’ key decision variables. The fourth channel predicts positive effect. While there are a large number of papers that have presented the core models and many refinements, in this paper we focus on the credit-market and the real-options channels as these two have come to dominate the literature in terms of gaining insights into firms’ decision-making under uncertainty. While the theoretical channels present relatively clear predictions, empirical analysis, using firm, industry or macroeconomic data, reveals that while uncertainty generally appears to depress economic activity, the estimated quantitative effects and qualitative inferences vary considerably across the studies and the particular decision variable under consideration (investment, entry, output, among others). We take a quick look at the empirical literature in section 2. Our focus on small businesses is motivated by several factors. First, small businesses play an important role in the economy in several dimensions. The U.S. Small Business Administration (SBA) defines a small business as an independent business having fewer than 500 employees. According to the SBA, small businesses represent about 99 percent of all 3 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) employer firms, and employ about one-half of all private sector employees. Small businesses pay about 43 percent of total U.S. private payroll and have generated 65 percent of net new jobs over the past 17 years. Small businesses create more than half of the non-farm private GDP and made up 97 percent of all identified exporters and produced 31 percent of export value in 2008. Small businesses are also important to U.S. technology improvement. According to the SBA, small businesses hire over 40 percent of high tech workers (scientists, engineers, computer programmers, and others). Finally, small businesses generate patents more efficiently than large firms: they produce 16 times more patents per employee than large patenting firms. Irrespective of the specific dimension under consideration, small businesses play a vital role in the economy. The importance of small businesses is a global phenomenon. As noted in ACCA (2010) and di Giovanni et al. (2010), for example, the vast majority of businesses globally are very small. They note that small businesses are special because they have several important characteristics: they contribute disproportionately to local economic growth by improving innovation ability; increase job opportunities; and contribute significant tax revenues. Given these attributes, it has become commonplace in the aftermath of the current global economic crisis to refer to SMEs as the backbone of the global economy. Second, while the empirical literature on uncertainty is quite extensive, there have been relatively few studies that have distinctly emphasized the effects of uncertainty on the economic activity of small businesses. Examining this link is important, given the structural significance of small business in the economy, and during economic expansions and contractions. Third, those papers that have focused on the uncertainty-small business effects, have tended to be related to investment (e.g., Ghosal and Loungani, 2000). Our primary focus is on the uncertaintyemployment link, along with the impact on the number of businesses, and the potential distinction between the effects on small versus large businesses. Examining the effects of economic uncertainty on employment and creation and survival of businesses is important as it has taken on a larger than life significance following the 2007 onwards economic crisis, as well as past crises, where employment effects have taken center stage in policy debates on how to improve economic conditions. Against this backdrop, we examine the effects of economic uncertainty on small businesses. Our data on small businesses obtained from the SBA are US-wide, including all small, and other, businesses in the US and their employment. Since the small business data 4 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) portray a broader economy-wide picture, our uncertainty measures are constructed using economy-wide variables related to GDP, inflation, stock prices, and fuel prices. In section 2 we briefly note the central results from theory, present a cursory overview of the empirical evidence, and note the key hypotheses. Section 3 discusses our data. In section 4 we outline construction of our measures of economic uncertainty. Sections 5 and 6 describe the empirical specification we estimate, and present the results. Our findings are that economic uncertainty negatively affects employment and the number of businesses, and the impact is primarily concentrated in the relatively smaller business category. The impact on large businesses is either non-existent or quantitatively smaller. Section 7 concludes with a discussion of our findings and some policy issues. 2. Review of literature on the impact of uncertainty The literature on the impact of uncertainty on firms decisions is quite extensive. Below we briefly review selected papers and results in the theoretical and empirical literatures with an eye towards spelling out the core hypotheses for our empirical analysis. 2.1 Review of theoretical results In this section we briefly review some of the predictions from the theoretical literature and highlight the implications for our empirical analysis. There are two strands of literature which offer predictions on the impact on uncertainty on a range of firms’ decision variables. A large portion of this literature has posed the problem as one of firms’ investment decisions under uncertainty. But several researchers have examined the complementary issues related to production, employment, R&D, among other variables. In our discussion below we note it as an investment problem, and then note the implications for employment and number of businesses. 2.1.1 Uncertainty and financing constraints Information asymmetry between borrowers and lenders may tend to constrain the flow of credit. In an influential paper, Greenwald and Stiglitz (1990) model firms as maximizing expected equity minus expected cost of bankruptcy and examine scenarios where firms may be equity or borrowing constrained. A key result is that greater uncertainty exacerbates information asymmetries, tightens financing constraints and lowers capital outlays. Since uncertainty 5 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) increases the risk of bankruptcy, firms cannot issue equity to absorb the risk. Those relying primarily on credit and operating outside of equity markets, find the flow of credit constrained. Delli Gatti et al. (2003) develop a model in which the financial conditions of businesses affect capital accumulation and entry and exit which affect the distribution of firms differentiated by the equity ratio. In their model flow of exiting firms is endogenously determined through bankruptcy while the flow of entering firms is affected by stochastic factors. Lensink, Bo and Sterken (2001) provide a lucid discussion of financing constraints in the general context of investment behavior, including the roles played by uncertainty and sunk costs. Overall, this literature shows that periods of greater uncertainty widens the information asymmetry, increases the likelihood of bankruptcy and exacerbates financing constraints. Incumbent businesses who are more dependent on borrowing and adversely affected by tighter credit will have greater likelihood of exit. Similarly, entry is likely to be impeded for potential entrants who are more adversely affected by tighter credit conditions. Thus, periods of greater uncertainty are expected to accelerate exits and retard entry leading to negative net entry – that is, a decline in the number of firms in an industry. The results are similar if we examine physical capital investment or employment. Considering employment, for example, periods of greater uncertainty, by constraining credit flows, limit the ability of existing firms to grow. Since new entry is dampened, this negatively affects employment. As the likelihood of exit by businesses is greater, this also negatively affects employment. Overall, periods of greater uncertainty are likely to lead to lower employment. The literature points to not all businesses being equally affected by uncertainty and credit market conditions. In different strands of this literature, the papers by Evans and Jovanovic (1989) and Fazzari, Hubbard and Petersen (1988), for example, offer insights on small versus larger firm effects. Gertler and Gilchrist (1994), for example, strongly emphasize the negative impact of information asymmetric and credit constraints on smaller firms. Tying the two literatures a prediction that emerges from is that periods of greater uncertainty, via the credit market channel discussed above, is likely to have a dampening effect on employment and the number of businesses. Given that the effects of financing constraints, at the margin, are more likely to be borne by the smaller businesses, the prediction is that the negative effect of uncertainty will be likely be greater on smaller businesses. Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 6 2.1.2 Uncertainty and real-options An alternative channel examines the effects of uncertainty on firms’ decisions via the real-options effects. The results in Caballero and Pindyck (1996), Dixit (1989) and Dixit and Pindyck (1994) are sufficient to generate the essential predictions for our empirical analysis, and we briefly discuss the results. Dixit (1989) shows that uncertainty and sunk costs imply an option-value of waiting and this raises the entry trigger price and lowers the exit trigger price.2 For prices below the entry trigger, the potential entrant holds on to its option to enter, and an incumbent firm does not exit at prices above the exit trigger. Another way of looking at this is that to enter during periods of greater uncertainty, firms would require a premium over the conventional Marshallian entry price. And incumbent firms would wait longer to exit (i.e., let prices fall below average variable cost before they exit) as they know that to re-enter the market they would have to re-incur the sunk entry costs. The key results related to entry and exit can be summarized as follows: (1) numerical results in Dixit (p.632-33) show that even small amounts of uncertainty are sufficient to generate significant changes in entry and exit patterns; and (2) the numerical simulations in Dixit (1989, p.632-33) and Dixit and Pindyck (Ch. 7, p.224-228; and Ch. 8.) show that an increase in uncertainty results in the entry trigger price increasing by more than the decrease in the exit trigger. These results reveal that entry is affected more than exit leading to negative net entry. In other words, periods of greater uncertainty lead to a decrease in entry but exits, while lower, continue at a closer to normal pace resulting in the industry experiencing a decrease in the number of firms. The results are similar if we examine physical capital investment or employment. Dixit and Pindyck (1994) present an extensive analysis of the detrimental effects of uncertainty on investment. The effects on employment will be qualitatively similar. If greater uncertainty retards entry and investment, it will also be expected to have a negative effect on employment. Assessing the implications for our empirical analysis, we note the following. One of our interests is to evaluate whether these effects may be different for small versus large firms. 2 Dixit and Pindyck (1994) outline the theoretical framework for studying firms’ decision-making regarding entry and exit under uncertainty and sunk entry costs. Hopenhayn (1992) and Pakes and Ericsson (1998), for example, study firm dynamics with firm-specific uncertainty. These models, however, are best subjected to empirical tests using micro-datasets as in Pakes and Ericsson, and is beyond the scope of our paper. 7 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) Previous studies have shown that the vast majority of the entry and exit churning is for the small businesses (e.g., see Audretsch (1995), Sutton (1997) and Caves (1998)). The literature shows that: (1) entrants are typically small compared to incumbents and have high failure rates; (2) the typical exiting firm is small and young; and (3) larger firms are older with higher survival rates. This implies that, via the real-options channel, most of the negative effects of uncertainty on the number of businesses and employment will likely be observed for the smaller businesses. 2.3 Overview of empirical findings The literature examining the effects of uncertainty is quite extensive. Studies vary considerably in terms of the underlying variables they use to measure uncertainty, such as profits, input prices, stock prices, inflation, GDP growth, economic policy news, survey of businesses or forecasters, among others. Equally, there is a large variety in terms of the specific constructs to measure uncertainty, such as conditional or unconditional variance (or standard deviation), among others. Finally, some studies use micro-data at the firm level, while others use data at the industry or macroeconomic level. The literature is quite expansive and it is difficult to review the full spectrum of this literature. To offer a perspective in a convenient format, we present a table in Appendix A which summarizes selected papers. These are not meant to be comprehensive, but merely display the range of variables used to measure uncertainty (GDP, inflation, prices, energy prices, stock prices, among others), the specific statistical constructs to capture uncertainty (unconditional variance, conditional variance derived from regression estimates, survey measures), the level of aggregation of the studies (firm or industry level, economy-wide, or cross-country), and the estimated quantitative and qualitative effects. Regarding the effects of uncertainty on small business, we briefly note some of the findings in the empirical literature useful for our analysis. Using industry-level data, Ghosal and Loungani (2000) find investment-uncertainty relationship is negative and this negative impact is greater in industries dominated by small firms. Koetse and Vlist (2006) find that there are differential effects of uncertainty on input and output variables. Bianco, Bontempi, Golinelli and Parigi (2012) find family firms’ investments are significantly more sensitive to uncertainty than non-family firms. 8 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 2.4 Hypotheses Based on our discussion of theory and the summary of the results in sections 2.1 and 2.2, the following hypotheses emerge: H1: Greater uncertainty is expected to negatively affect employment, and this effect may be more pronounced in smaller businesses. H2: Greater uncertainty is expected to negatively affect the number of businesses, and this effect may be more pronounced for smaller businesses. 3. Data description For our econometric analysis, we use data from multiple sources. Below we provide details of the datasets we use. First, is the U.S. Small Business Administration (SBA) database. From the SBA database we use annual data on the number of firms according to the size of firms from 1988 to 2011. (Data prior to 1988 are not available.) The SBA database also provides employment data for businesses by the size of firms. Second, U.S. macroeconomic data on real GDP and GDP implicit price deflator are from the Federal Reserve Economic Data. The data on S&P 500 stock price index is from Yahoo Finance. Finally, we obtain data on fuels and related products and power PPI from the U.S. Bureau of Labor Statistics. To consider size classes for our employment and number of businesses, we use the following taxonomy to parse out the small business related effects versus other. The overall sample we examine are: 1. ‘All’ businesses; 2. ‘Large’ businesses – these are businesses with ≥500 employees; 3. ‘Small’ businesses – these are businesses with <500 employees; and 4. ‘Smaller’ businesses – these are businesses with <20 employees. The 500 employee cutoff is the one used by the U.S. SBA, and we use this as the baseline. We consider an additional cutoff of <20 employees for the following reasons: (a) a 500 employee firm is relatively large, so we wanted to consider an alternate cutoff for defining small; (b) data with the <20 employee cutoff was available consistently for both our variables (employment and number of businesses); and (c) a large percentage of the truly small business fall in this category. Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 9 In our estimation of employment and number of businesses specifications, we will present estimates for each of the four groupings (1-4) noted above. In Figures 1 and 2 we display the time paths for the employment levels and number of businesses by size class. Table 1 presents the summary statistics for the employment and number of businesses by size class. 4. Measures of uncertainty As noted earlier, Appendix A summarizes selected papers and displays the range of variables used to measure uncertainty (GDP, inflation, prices, energy prices, stock prices, survey opinions, among others), and the specific statistical constructs to capture uncertainty (unconditional variance, conditional variance derived from regression estimates, survey forecast variance). The broader literature reveals a wide range of variables and methods to capture uncertainty. In the spirit of this literature, we construct six alternate measures to examine the impact of economic uncertainty on small businesses. As we discussed earlier, our objective is to examine the effects of economic uncertainty on small businesses. Our data on small businesses are US-wide and include all small and larger businesses in the US and their employment, as well as the total number of such businesses. Since the data are designed to portray a broader picture, our uncertainty measures are created using economy-wide indicators. First, we use the Forecasts for the Survey of Professional Forecasters provided by the Federal Reserve Bank of Philadelphia. From the survey forecasts, we use index of industrial production and forecasts for the level of real GDP to construct the two measures of uncertainty. Our measures of uncertainty are the within-year variance of survey forecasts. These two measures of uncertainty are labeled and , with sf denoting survey forecast. Second, we use forecasting specifications to construct four additional measures related to uncertainty about: 1. real GDP growth. This indicates the overall state of the economy capturing demand and supply effects;3 3 GDP uncertainty measures, in different forms, have been used by, for example, Driver et al. (2005), Asteriou et al. (2005) and Bloom (2009). Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 10 2. inflation rate. We measure inflation as the annual percentage growth of the GDP deflator. Inflation uncertainty captures effects related to input and product prices, as well as affecting firms’ real borrowing rates;4 3. stock prices. We use the S&P500 stock price index. As with real GDP, this is an indicator of the overall state of the economy, and forward looking indicator of investor and business confidence;5 and 4. real fuel price growth. Here we use the BLS price index of a range of commonly used fuels by businesses. The nominal fuel price index is converted to real values after deflating by the implicit GDP deflator. This variable serves to proxy a critical input – fuels and energy – price for businesses.6 To create our measures of uncertainty, we use the following procedure. We assume that firms use a forecasting equation to predict future values of the relevant variable, an economywide indicator in our case. As noted earlier, our objective is to examine the effects of economic uncertainty on small businesses. Our data on small businesses is US-wide (i.e., all small businesses in the US and their employment). Our uncertainty measures, therefore, are created using economy-wide variables related to GDP, inflation, stock prices, and fuel prices. As our baseline, we use an autoregressive, AR(2), specification as the forecasting model. AR(n) models are based on Box and Jenkins (1970) formulation for forecasting economic variables, and historically they have performed well in forecasting exercises (e.g., Meese and Geweke (1984) and Marcellino, Stock and Watson (2003)). Given this aspect, and our annual time-series data for the small business variables, we use this standard methodology to forecast the relevant state variables. The forecasting specification is: 4 Inflation has been used to construct uncertainty measures by, for example, Huizinga (1993), Fountas et al. (2006) and Elder (2004). 5 Stock prices have been used to measure uncertainty by, for example, Bloom (2009), Chen et al. (2011), Bloom et al. (2007), Greasley et al. (2006), and Stein et al. (2010). 6 Fuel and energy prices have been used to construct uncertainty measures by, for example, Koetse et al. (2006), Kilian (2008) and Guo et al. (2005). Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 11 where Z is either: (i) real GDP growth; or (ii) inflation rate; or (iii) S&P500 growth; or (iv) real fuel price growth. From specification (1), the predicted values represent the forecastable component. The residuals: represent the unsystematic, or unforecastable, component. Since values, we use the squared value of can take positive or negative , interpreted as the conditional-variance, as our measure of uncertainty about the relevant variable. Using the residuals from forecasting equations and conditional-variance to construct uncertainty measures is common: see, for example, the insightful review in Lensink, Bo and Sterken (2001), and the references cited in the above footnotes. If the forecasting specification is for GDP (i.e., Z is real GDP growth), then we denote the uncertainty measure as: Using this procedure for our four underlying economy-wide variables (Z), we obtain four measures of uncertainty denoted by: (i) ; (ii) ; (iii) ; and (iv) .7 To estimate the forecasting specification (1), we use available data on Z from 1960 to 2010. The rationale is as follows. The objective is obtain a good forecasting specification, which is better done with a longer time-series ensuring that the parameters of the equation are estimated precisely. While data on GDP, inflation and S&P 500 are available for earlier periods, the BLS fuel price indices are available starting 1960. So we start the time-period for the equation in 1960 for which all of our variables (that we consider for Z) are available. The terminal period, 2010, is the same as the last period for which the small business data are available. This implies that the generated time-series in our four uncertainty measures , , and are over 1962-2010 (two initial observations are lost due to use of lagged values). A final remark we make is that while our AR(.) forecasting equation is fairly standard, the length and frequency (annual from 1988-2010) of the available small business variables 7 While we report the results for an AR(.) specification, our results were robust to experiments with lag lengths. Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 12 restricts us from considering alternate procedures, such as using ARCH models, to construct measures of uncertainty. While this may be viewed as a limitation, the SBA data on employment and number of businesses are simply not available at a higher frequency. Since we need to harmonize the estimation periods and frequency, we are restricted to using annual data to forecast our specification (1). 5. Estimating Specification To examine the hypotheses noted in section 2.4, we estimate autoregressive-distributed lag AD(n, m) specification of the form: where is a measure of small business activity (in our case, employment or number of businesses), growth, and or or is the measure of uncertainty measured in natural logarithms, is real GDP is the error term. As described earlier, the four measures of uncertainty are or . In specification (4), in our estimation we found that for several specifications a second autoregressive lag was significant, whereas in others only one lag was significant. Rather than alternating the lag length, we parsimoniously used two autoregressive lags for all specifications. Since we are interested in the shorter-term effects of uncertainty on small business employment or number of businesses, and that the underlying data on employment and the number of businesses contained a clear trend, we measure differences (annual percentage changes). in annual logarithmic first- and are included in the specification to capture any persistence in the dependent variable and the variable’s own dynamics. The two real GDP growth variables, and , control for overall economic conditions. As noted in the data section, the available data on Small Businesses are annual and cover the period 1988-2010. We, therefore, estimate specification (4) over 1988-2010. While, as noted earlier, our four generated uncertainty measures , , and are over 1962- 2010, we only use their values over 1988-2010 in estimating specification (4). Estimating (4) informs us about the impact of uncertainty on employment and number of businesses, after Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 13 controlling for overall economic activity (GDP growth) and the dynamic lagged structure for the included variables.8 In our examination of the effects of uncertainty on employment and number of businesses, we will present estimates for four groups noted in our data section 3: 1. ‘All’ businesses; 2. ‘Large’ businesses – these are businesses with ≥500 employees; 3. ‘Small’ businesses – these are businesses with <500 employees; and 4. ‘Smaller’ businesses – these are businesses with <20 employees. 6. Estimation results << Preliminary >> The results for the employment specifications are presented in tables 2.1 and 2.2. Table 2.1 presents estimates for ‘All’ and ‘Large’ businesses, and table 2.2 presents estimates for the ‘Small’ and ‘Smaller’ size classes. A similar format is followed for presenting the results on the number of businesses in tables 3.1 and 3.2. In table 2.1, the are in the 75% to 87% range indicating good fit of the specification. The estimates of the first-order autocorrelation coefficient, ρ, are relatively low on average. The estimates show that uncertainty related to GDP, Inflation, S&P500 and Fuel prices have a negative effect on ‘All’ business employment, but the timing (contemporaneous versus lagged) and quantitative effects and vary. Turning to the ‘Large’ businesses, only Inflation and Fuel price uncertainty dampen employment, and the estimated quantitative effects are a bit smaller for the Large business group. For the small business groups in table 2.2, the are in the 62% to 78% range. The estimates of the first-order autocorrelation coefficient, ρ, are generally quite low. The estimates show that aside from the mixed inferences from the Inflation and S&P500 based measures, uncertainty related to the two survey-based measures, and GDP, Inflation and Fuel prices have a negative effect on employment for the ‘Small’ and ‘Smaller’ businesses. Examining the broad inferences from the estimates in tables 2.1 and 2.2, uncertainty appears to dampen employment, and the effects are primarily concentrated in the smaller businesses. Next we turn to the specifications for the number of businesses in tables 3.1 and 3.2. 8 Our experiments using longer lag lengths did not provide additional insights into the effects of uncertainty. 14 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) In table 3.1, the for ‘All’ businesses specifications are in the 70% to 75% range, while they are only in the 40% to 55% range for the ‘Large’ businesses, indicating lower explanatory power for the latter group. Aside from a couple of specifications, the estimates of the first-order autocorrelation coefficient, ρ, are relatively low. The estimates show that uncertainty related to the survey-based measures, and GDP and S&P500 have a negative effect on the number of ‘All’ business, but the quantitative effects vary. Fuel price or Inflation uncertainty does not appear to affect the number of ‘All’ businesses. Turning to the ‘Large’ businesses, we see that four of the uncertainty measures appears to decrease the number of these businesses. For the smaller businesses specifications in table 3.2, the are in the 50% to 60% range for the ‘Smaller’ group, and in the 70% range for the ‘Small’ group. The estimates of the first-order autocorrelation coefficient, ρ, are uniformly low. The estimates show that uncertainty related to both the survey-based measures, and GDP and Inflation have a negative effect on the number of ‘Smaller’ business, but Fuel price and S&P500 uncertainty does not appear to affect this group. Turning to the ‘Small’ group, the two survey-based measures, and GDP and Inflation uncertainty have a negative effect on the number of businesses. Based on the estimates in tables 3.1 and 3.2, uncertainty has a negative impact on the number of businesses, and this effect appears to be more mixed across the various size classes. The overall inferences we draw from the set of estimates presented in tables 2 and 3 are that economic uncertainty dampens employment, and the effects appear to be concentrated in the Smaller (<20 employees) and Small (<500 employees) groups. At broad brush, our findings appear supportive of the results from the theoretical models we discussed in sections 2.1 and 2.2. The effects on the number of businesses appears to be more mixed and need further investigation. Before closing this section we note that the estimated specifications include real GDP growth, and model the dynamics of the included variables via lagged effects. Real GDP growth is probably the single most important control variable in either employment or number of businesses specifications. With increasing GDP growth, business opportunities are expected to expand allowing for creation of more jobs as well as new businesses. The fact that we find effects related to uncertainty even after controlling for GDP growth and the lagged dynamics, makes our finding even more noteworthy. << Additional results and checks of robustness to be added >> Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 15 7. Discussion of results and conclusions Our findings on the negative impact of economic uncertainty on employment in small businesses as well as the number of small businesses are particularly revealing as these results appear even after controlling for GDP growth and appropriate controls for lagged dynamics of the included variables. The underlying theories we reviewed related to potential financingconstraints and the real-options channels. Given that the data we use from the U.S. Small Business Administration is relatively aggregated, it is difficult to disentangle which of these two channels may be playing the more dominant role. It is fair to assume that both channels are important in determining the outcomes. As was noted earlier, there exists a significant literature on financing-constraints. This literature has noted important differences between smaller and larger businesses, and point to smaller firms being relatively credit-constrained (e.g., Fazzari, Hubbard and Petersen, 1988; Gertler and Gilchrist, 1994; Evans and Jovanovic, 1989; Lensink, Bo and Sterken, 2001; Ghosal and Loungani, 2000; Himmelberg and Petersen, 1994; Winker, 1999; and Audretsch and Elston, 1997). Viewing our estimation results and the above information on financing constraints collectively, part of our results for the smaller business is undoubtedly emerging from the financing-constraints channel. If this is true, then important policy implications emerge, primarily in the form of initiatives and instruments designed to partly ease the credit-constraints faced by smaller businesses. To the question as to why governments might pay special attention to small businesses, there are several responses. First, we noted earlier in the paper that a large fraction of employment and businesses fall into the smaller categories. Second, a number of emerging structural factors – such as those related to globalization and banking sector consolidation – are likely to favor large businesses relative to the smaller ones. These considerations alone provide important economic policy justification. As with many governments worldwide, the U.S. has recently implemented policies and programs to help small businesses bridge the capital and market gap and encouraged publicprivate partnerships to support small business and entrepreneurship by, for example: (a) supporting more than $53 billion in SBA loan guarantees to more than 113,000 small businesses; Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 16 (b) awarding more than $221 billion in Federal contracts to small businesses (FY 2009 through April 30, 2011); and (c) awarding more than $4.5 billion in research funding through the Small Business Innovation and Research Program during FY 2009 and FY 2010.9 Such initiatives, along with appropriate lending policies, can help ease some of the financing-constraints faced by smaller businesses.10 By doing so, and in the context of this paper, such policies may also help alleviate some of the negative impact of uncertainty on smaller businesses. 9 The National Economic Council (2011) and Sheets and Sockin (2012) provide extensive discussion on the importance of small businesses and policy. 10 The papers by Audretsch and Elston (1997, 2002), for example, provide important insights in this dimension from German policy initiatives. Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 17 Appendix A: Selected empirical findings on the impact of uncertainty The papers included below are not meant to be comprehensive, but display the range of variables used to measure uncertainty (GDP, inflation, prices, energy prices, stock prices, among others), the specific statistical constructs to capture uncertainty (unconditional variance, conditional variance derived from regression estimates, survey measures), the level of aggregation of the studies (firm or industry level, economy-wide, or cross-country), and the estimated quantitative and qualitative effects. Table A.1. Selected papers examining the effects of uncertainty Paper Data Estimation method State variables Uncertainty measure Estimation results Lensink, Robert, Paul van Steen and Elmer Sterken. “Uncertainty and Growth of the Firm,” Small Business Economics, 2005, 381-391. Survey of 1,097 Dutch firms in 1999. Sales. Return on Investment. Logit model. Conditional variance over the conditional mean. Koetse, Mark J., Arno J. van der Vlist and Henri L.F. de Groot. “The Impact of Perceived Expectations and Uncertainty on Firm Investment,” Small Business Economics, 2006, 365-376. Survey of 135 plant locations in Netherlands in 1998. Wages. Energy prices. Output prices. Tobit model. Survey based. Bo, Hong, and Elmer Sterken. “Volatility of the interest rate, debt and firm investment: Dutch evidence,” Journal of Corporate Finance, 2002, 179–193. Driver, Ciaran, and Brendan Whelan. “The Effect of Business Risk on Manufacturing Investment,” Journal of Economic Behavior and Organization, 2001, 403-412. Data for 41 Dutch listed firms from 1984 to 1995. Interest rate. Uncertainty has a negative impact on the size of investment, no matter what the type of investment is used. Smaller firms have a lower probability to invest if uncertainty increases. In logit model, for one unit increase in uncertainty, the log odds of investment decreased by 0.878. Uncertainty has a larger influence on decision making in small firms than in large firms specifically for investment in energy-saving technologies. In small firms, input uncertainty and output uncertainty have a differential impact on both aggregate and energy-saving investments. Cross-effect of the interest rate volatility and debt on investment is positive. This effect is more important for highly indebted firms than for lessindebted firms. No strong effect of risk due to convexities. Risk did affect the timing of investment for between a quarter and a third of the sample. The greatest caution in respect of timing was in the Hi-tech sector which was also the sector with the greatest damage from delay. They find a U-shaped relationship between market uncertainty and value of investment. Also, they find an inverted U-shaped relationship between technological uncertainty and the value of R&D capital. Family firms’ investments are significantly more sensitive to uncertainty than nonfamily firms and the greater sensitivity to uncertainty is basically due to the greater opacity of family firms and to their higher risk aversion, rather than to the degree of sunk fixed capital. “Cautionary effects” of uncertainty are large – going from the lower quartile to the upper quartile of the uncertainty distribution typically halves the first year investment response to demand shocks. A negative relationship between investment and price uncertainty only exists in competitive industries. One percentage increases in price Conditional variance. Panel Data, ARCH model, Fixed effect estimation. Disaggregated survey data of Ireland in 1995. Comparing the percentage of different respondents in the survey questions. Future demand and future price Future unit input cost Capacity Delay risk Subjective descriptions Oriani, Raffaele, and Maurizio Sobrero. “Uncertainty and the Market Valuation of R&D within a Real Options Logic,” Strategic Management Journal, 2008, 343-361. Data for 290 manufacturing firms in UK from 1989 to 1998. Bianco, Magda, Maria Elena Bontempi, Roberto Golinelli and Giuseppe Parigi. “Family Firms’ Investments, Uncertainty and Opacity,” Small Business Economics, 2012, 1-24. Data for 2,959 Italian private companies from 1996 to 2007. Bloom, Nick, Stephen Bond and John Van Reenen. “Uncertainty and Investment Dynamics,” Review of Economic Studies, 2007, 391-415. Data for 672 UK manufacturing firms from 1972 to 1991. Ghosal, Vivek, and Prakash Loungani. “Product Market Competition and the Impact of Price Uncertainty on Investment: Some Data for 254 US 4-digit SIC manufacturing industries from 1958 to 1989. Panel Data, Hedonic model Industry output. Patents. Absolute percentage difference. Inverse of the median age. Sales. Coefficient of variation. Panel data, GMM Panel data, GMM Stock returns. Std. deviation of daily stock returns. Product price. Rolling regression based conditional std. 18 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) Evidence From US Manufacturing Industries,” Journal of Industrial Economics, 1996, 217-228. Panel data, fixed effect – IVE. deviation. Fuss, Catherine, and Philip Vermeulen. “Firms' Investment Decisions in Response to Demand and Price Uncertainty,” Applied Economics, 2008, 2337-2351. Survey of 279 firms from 19872000, and another survey of 319 firms from 1987-1999. Expectations of future demand and prices. Bulan, Laarni. “Real Options, Irreversible Investment and Firm Uncertainty: New Evidence from U.S. Firms,” Review of Financial Economics, 2005, 255-279. Data for 2,901 US firms from 1964 to 1999. Ghosal, Vivek, and Prakash Loungani. “The Differential Impact of Uncertainty on Investment in Small and Large Businesses,” Review of Economics and Statistics, 2000, 338343. Huizinga, John. “Inflation Uncertainty, Relative Price Uncertainty, and Investment in U.S. Manufacturing,” Journal of Money, Credit and Banking, 1993, 521-549. Data for 330 US SIC 4-digit manufacturing industries from 1958 to 1991. Theil index. Panel data, GMM. Equity returns. Standard deviation. Panel data, 2SLS. Profits. Panel data, IVE. Rolling regression based conditional std. deviation. Data for 450 U.S. SIC 4-digit manufacturing industries from 1954 to 1989. Real wage. Output price. Real materials price. Bivariate ARCH model. Conditional std. deviation. Stein, Luke C.D., and Elizabeth C. Stone. “The Effect of Uncertainty on Investment: Evidence from Options,” Stanford University Working Paper, 2010. Data for 2,230 US manufacturing firms from 1996 to 2009. Stock price. Folta, Timothy, and Jonathan P. O’Brien. “Entry in the Presence of Dueling Options,” Strategic Management Journal, 2004, 121-138. Data for 2,230 US manufacturing firms from 1996 to 2009. Expected volatility. Panel data, 2SLS. Panel data, 2SLS. Data on 17,897 firms from 1980 to 1999. Industry’s contribution to GDP. Square root of conditional variance. GARCH model. Baker, Scott, Nick Bloom and Steven J. Davis. “Has Economic Policy Uncertainty Hampered the Recovery?” Chicago Booth Paper, No. 12-06, 2012. Index of economic policy uncertainty, news-based proxy, government purchases data, disagreement about future indexes from 1985 to 2011. And tax code expiration data is from 1991 to 2011. Economic policy uncertainty, news-based proxy, government purchases data, disagreement about future indexes, and merge them into a new proxy. VAR model Aggregating the above Components to Obtain an Index of Economic Policy Uncertainty uncertainty is estimated to cause the ratio of gross industry investment (I/K) decrease by 0.358 for most competitive industries. Demand uncertainty at the time of planning depresses planned and subsequent realized investment. One standard deviation increases in demand uncertainty is estimated to reduce 6% of the average investment ratio. Increased industry uncertainty and firmspecific uncertainty display a pronounced negative effect on firm investment consistent with real options behavior. A one standard deviation increase in industry uncertainty reduces a firm’s investment-to-capital ratio by 6.4%. A one standard deviation increase in firmspecific uncertainty decreases firm investment by 19.3%. Investment-uncertainty relationship is negative and this negative impact is greater in industries dominated by small firms. Increased uncertainty about real wages portends an immediate drop in capital expenditures which is large in economic terms; while increased uncertainty about real output price does not signal an immediate decline in capital expenditures hut a sustained increase may indicate a large reduction eventually. They find a negative and statistically significant relationship between uncertainty and investment. The coefficient estimates are larger in magnitude after addressing the endogeneity of the uncertainty measure, suggesting potential reverse causation that biases the OLS estimates towards zero. They find the effect of uncertainty on entry is non-monotonic and U-shaped. And the turning points are influenced by factors that should theoretically influence options to grow and options to defer. Uncertainty has a potent effect on entry even after controlling for firm resource profiles, including the relatedness to the target industry. Historically high levels of policy uncertainty in 2010 and 2011 mainly reflect concerns about tax and monetary policy and secondarily a broader range of other policy-related concerns. Policyrelated concerns now account for a large share of overall economic uncertainty. A rise in policy uncertainty, similar in magnitude to the actual change since 2006, is associated with substantially lower levels of output and employment over the following 36 months. 19 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) Baker, Scott, Nicholas Bloom and Steven Davis. “Measuring Economic Policy Uncertainty,” University of Chicago and Stanford University, 2012. Constructing index data about political uncertainty from 19852009. VAR model Newspaper coverage of policy-related economic uncertainty. Number of federal tax code provisions set to expire in future years. Disagreement among economic forecasters. Constructing an index from three types of underlying components. Bloom, Nicholas. “The Impact of Uncertainty Shocks,” Econometrica, 2009, 623-685. Firm-level data from 2548 firms U.S. firms from 1981 to 2000. Panel data, VAR model, SMM estimation Driver, Ciaran, Paul Temple and Giovanni Urga. “Profitability, capacity, and uncertainty: a model of UK manufacturing investment,” Oxford Economic Papers, 2005, 120– 141. Asteriou, Dimitrios and Simon Price. “Uncertainty, investment and economic growth: evidence from a dynamic panel,” Review of Development Economics, 2005, 277– 288. Aggregate data of UK manufacturing on two capital assets (machinery and building) from 1972 to 1999. Bloom, Nick. “The Impact of Uncertainty Shocks: Firm Level Estimation and a 9/11 Simulation,” CEP Discussion Paper, No. 718, 2006. Data of 579 US manufacturing firms from 1991 to 2000. Greasley, David, and Jakob B. Madsen. “Investment and Uncertainty: Precipitating the Great Depression in the United States,” Economica, 2006, 393–412. Data of investment information in US from 1920 to 1938. Kilian, Lutz. “Exogenous Oil Supply Shocks: How Big Are They and How Monthly production data for all OPEC countries and for GARCH model Data of 59 industrial and developing countries from 1966 to 1992. Profits growth, stock returns, TFP growth, GDP forecasts. Standard deviation. Output growth. Time-series conditional volatility. GDP per capita growth. Conditional variance. Panel data, PMG, MG, GARCH Monthly share returns. Standard deviations. Panel data, time-varying second moment model Tobin’s q Real stock prices. Squared monthly proportional change. Oil production. Policy related uncertainty played a role in the slow growth and fitful recovery of recent years, and they invite further research into the effects of policyrelated uncertainty on economic performance. VAR estimates show that a policy uncertainty shock equal in size to actual increase in the index value from 2006 to 2011 foreshadows drops in private investment of 16 percent within 3 quarters, industrial production drops of 4 percent after 16 months, and aggregate employment reductions of 2.3 million within two years. Uncertainty appears to dramatically increase after major economic and political shocks like the Cuban missile crisis, the assassination of JFK, the OPEC I oil-price shock, and the 9/11 terrorist attacks. The uncertainty component alone generates a 1% drop and rebound in employment and output over the following 6 months, and a milder long-run overshoot in these economic shocks. The GARCH model shows uncertainty variable for the full sample are estimated to be negatively significant at the 5% level for machinery, and negative but not significant for building. Using PMG estimation, they find the estimated coefficients were -0.360 for the industrial and -0.053 for the developing countries, which showed a negative relationship between growth and uncertainty in both cases. For the effect of uncertainty on investment, the PMG results revealed a significant negative relationship for both subgroups, with higher negative magnitudes for the industrial countries (-3.280) than that of developing countries (-0.081). Uncertainty appears to vary strongly over time, temporarily rising by up to 200% around major shocks like the Cuban Missile crisis, the assassination of JFK and 9/11. It finds temporary impact of a second moment shock is different from the typically persistent impact of a first moment shock. While the second moment effect has its biggest drop in month 1 and has completely rebounded by month 5, a persistent first moment shock will generate a drop in activity lasting several quarters. For investment, the results show that the effects of heightened uncertainty surrounding the expected marginal profitability of capital, measured by share price volatility, can explain around 80% of the actual fall in the business fixed investment ratio in 1930. Thus, to a large extent, the investment slump led the declines in income. Using this approach and the new exogenous oil supply shock measure, it 20 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) Much Do They Matter for the U.S. Economy?” Review of Economics and Statistics. 2008, 216-240. aggregate non-OPEC oil production since 1973. Exogenous variation. OLS Guo, Hui, and Kevin L. Kliesen. “Oil Price Volatility and U.S. Macroeconomic Activity,” Federal Reserve Bank of St. Louis Review, 2005, 669-83. Data of daily price of U.S. 1month futures and 12- month futures contracts from 1983 to 2004. Oil prices. Realized variance series. Forecasting regression Elder, John. “Another Perspective on the Effects of Inflation Uncertainty,” Journal of Money, Credit & Banking, 2004, 911-928. Data of U.S. output growth rate, inflation, CPI, commodity price from 1966 to 2000. Cross-section data, MGARCH-M VAR model Inflation Conditional variance Monthly neighborhood returns Conditional variance Stilianos Fountas, Menelaos Karanasos and Jinki Kim. “Inflation Uncertainty, Output Growth Uncertainty and Macroeconomic Performance,” Oxford Bulletin of Economics and Statistics. 2006, 319343. Monthly data for the G7 on PI and the IPI. Inflation Output growth Bivariate GARCH model, VAR model Conditional variance Ghosal, Vivek. “Demand Uncertainty and Capital-Labor Ration: Evidence from U.S. manufacturing Sector,” The Review of Economics and Statistics, 1991, 157-161. Ghosal, Vivek. “Does uncertainty influence the number of firms in an industry?” Economics Letters, 1996, 229-236. Jonathan, O'Brien, Timothy Folta. "Sunk costs, uncertainty and market exit: A real options perspective." Industrial & Corporate Change. 2009, 807-833. Date for 125 U.S. manufacturing industries from 1968 to 1977. GARCH model Standard deviation Li, Yong. “Duration analysis of venture capital staging: A real options perspective,” Journal of Business Venturing, 2008, 497–512. 46,976 portfolio company-round pairs in U.S. for 1975–2005, involving 3737 venture capital firms and 15,786 portfolio companies. Market price volatility; The stage of development at the time of financing. Fluctuations of shipments Standard deviation OLS Date for 196 U.S. industries from 1973-1986. IV Estimation Data for 16,447 firms operating in 405 industries from 1985 to 2003. GARCH model Industry-specific price Standard deviation of residuals Volatility of stock returns Conditional variance for market uncertainty; Use the three stages: Seed/Startup, Early, and Expansion/Late stages. (The first two represent finds statistically significant evidence of a sharp drop in real GDP growth five quarters after an exogenous oil supply shock and of a spike in CPI inflation three quarters after the shock. And it is shown that exogenous oil supply shocks made remarkably little difference overall for the evolution of U.S. real GDP growth and CPI inflation since the 1970s. A volatility measure constructed using daily crude oil futures prices has a negative and significant effect on future GDP growth over the period 1984-2004. Moreover, the effect becomes more significant after oil price changes are also included in the regression to control for the symmetric effect. Uncertainty about inflation has significantly reduced real economic activity over the post-1982 period, with the effect concentrated after a twomonth lag. Using MGARCH-M VAR model, they find that one standard deviation increase in inflation uncertainty has tended to reduce real economic activity over three months by about 22 basis points in the post-1982 period. First, inflation does cause negative welfare effect. Secondly, in some countries, e.g. Canada and the UK, more inflation uncertainty provides an incentive to Central Banks to surprise the public by raising inflation unexpectedly. Thirdly, business cycle variability and the rate of economic growth are related. More variability in the business cycle leads to more output growth. There is a significant negative relationship between demand uncertainty and the capital-labor ratio, and that an increase in firm size counteracts this negative influence. Price uncertainty has a statistically significant and quantitatively large negative impact on the number of firms in an industry. Uncertainty dissuades firms from exiting an industry, but only when the sunk costs of entering and exiting that industry are sizeable. Moreover, sunk costs can be influenced by the technological intensity of an industry, by the extent to which a firm competes on the basis of innovation, and by the firm’s diversification strategy. Market uncertainty encourages venture capital firms to delay investing at each round of financing, whereas competition, project-specific uncertainty and agency concerns prompt venture capital firms to invest sooner. 21 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) Podoynitsyna, Ksenia, Michael Song, Hans van der Bij and Mathieu Weggeman. “Improving new technology venture performance under direct and indirect network externality conditions,” Journal of Business Venturing, 2013 195–210. A sample of 385 NTVs drawn from the VentureOne 2001 database and the 1995–2000 Inc. 500 list. Li, Yong, Joseph T. Mahoney “When are venture capital projects initiated?” Journal of Business Venturing, 2011, 239–254. Date of venture capital investment of 22,164 ventures made between 1980 and 2007 in the U.S. high uncertainty and list one is low uncertainty.) of development at the time of financing as an indicator of the level of project-specific uncertainty. Five main uncertainty management strategies: (Avoidance, Imitation, Control, Cooperation, Real options) OLS multiple regression Use scales to measure each of the five main uncertainty management strategies: (e.g. for avoidance strategy, they ask firms to select from the three performance: introduce their product with low uncertainty, Postpone a market entry if the market is too uncertain, Grow from small scale to large scale when entering a new market.) Volatility of market returns. The standard error of the regression. Accelerated-failure-time models Li, Dan. “Multilateral R&D alliances by new ventures,” Journal of Business Venturing, 2013, 241–260. Freel, Mark S.. “Perceived Environmental Uncertainty and Innovation in Small Firms,” Small Business Economics, 2005, 49–64. They show that real options reasoning does not always perform better under conditions of higher uncertainty, such as uncertainty due to direct network externalities. 173 new ventures involved in multilateral R&D alliances in high-technology industries during 1990–2005, and 173 matching ventures which are not. Heckman two-stage regressions. Firm-level survey data of UK SMEs in 1996, 1998, 2000. Discriminant functions Mean monthly stock price volatility. Standard deviation. Supply, finance, competitors, trade Asking the firms to report their feeling of uncertainty on a fivepoint scale. Venture capitalists tend to defer new investment projects in target industries with substantial market volatility. This delay effect of market volatility is reduced if the target industry experiences high sales growth or if competition among venture capitalists is intense in the target industry. An inverted U-shaped relationship between market uncertainty and a new venture's likelihood of forming multilateral R&D alliances. Firms engaged in novel product innovation appear to perceive a less hostile, competitive environment. Also, higher levels of innovation in manufacturing firms are associated with higher perceptions of supplier uncertainty, whilst, higher levels of innovation in service firms are associated with higher perceptions of human resource uncertainty. 22 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) References Abel, Andrew. “Optimal Investment under Uncertainty,” American Economic Review, 1983, 228–233. Appelbaum, Elie, and Eliakim Katz. “Measures of Risk Aversion and Comparative Statics of Industry Equilibrium,” American Economic Review, 1986, 524–529. Association of Chartered Certified Accountants (ACCA). “Small Business: A Global Agenda,” ACCA Accountants for business Report, 2010. Asteriou, Dimitrios, and Simon Price. “Uncertainty, Investment and Economic Growth: Evidence from a Dynamic Panel,” Review of Development Economics, 2005, 277–288. Audretsch, David. Innovation and Industry Evolution. Cambridge: MIT Press, 1995. Audretsch, David, and Julie Elston. “Does Firm Size Matter? Evidence on the Impact of Liquidity Constraints on Firm Investment Behavior in Germany,” International Journal of Industrial Organization, 2002, 1–17. Audretsch, David, and Julie Elston. “Financing the German Mittelstand,” Small Business Economics, 1997, 97-110. Baker, Scott, Nicholas Bloom and Steven Davis. “Has Economic Policy Uncertainty Hampered the Recovery?” Chicago Booth Paper, No. 12-06, 2012. Baker, Scott, Nicholas Bloom and Steven Davis. “Measuring Economic Policy Uncertainty,” University of Chicago and Stanford University, 2012. Bianco, Magda, Maria Elena Bontempi, Roberto Golinelli and Giuseppe Parigi. “Family Firms’ Investments, Uncertainty and Opacity,” Small Business Economics, 2012, 1-24. Bloom, Nicholas. “The Impact of Uncertainty Shocks,” Econometrica, 2009, 623-685. Bloom, Nicholas. “The Impact of Uncertainty Shocks: Firm Level Estimation and a 9/11 Simulation,” CEP Discussion Paper, No. 718, 2006. Bloom, Nick, Stephen Bond and John van Reenen. “Uncertainty and Investment Dynamics,” Review of Economic Studies, 2007, 391-415. Bo, Hong, and Elmer Sterken. “Volatility of the Interest Rate, Debt and Firm Investment: Dutch Evidence,” Journal of Corporate Finance, 2002, 179–193. Box, G.E.P., and , G.M. Jenkins. Time Series Analysis, Forecasting and Control. San Francisco: Holden Day, 1970. 23 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) Bulan, Laarni. “Real Options, Irreversible Investment and Firm Uncertainty: New Evidence from U.S. Firms,” Review of Financial Economics, 2005, 255-279. Caballero, Ricardo, and Robert Pindyck. “Investment, Uncertainty and Industry Evolution,” International Economic Review, 1996, 641-662. Caves, Richard. “Industrial Organization and New Findings on the Turnover and Mobility of Firms,” Journal of Economic Literature, 1998, 1947-1982. Crain, Nicole, and W. Mark Crain. “The Impact of Regulatory Costs on Small Firms,” Small Business Administration, 2010. Delli Gatti, Domenico, Mauro Gallegati, Gianfranco Giulioni and Antonio Palestrini. “Financial Fragility, Patterns of Firms’ Entry and Exit and Aggregate Dynamics,” Journal of Economic Behavior and Organization 51, 2003, 79-97. Di Giovanni, J., Levchenko, A. A. and Rancière, R. “Power Laws in Firm Size and Openness to Trade: Measurement and Implications,” IMF Working Paper No. 10/109. Driver, Ciaran, Paul Temple and Giovanni Urga. “Profitability, Capacity, and Uncertainty: A Model of UK Manufacturing Investment,” Oxford Economic Papers, 2005, 120–141. Driver, Ciaran, and Brendan Whelan. “The Effect of Business Risk on Manufacturing Investment,” Journal of Economic Behavior and Organization 44, 2001, 403-412. Evans, David and Boyan Jovanovic. “An Estimated Model of Entrepreneurial Choice under Liquidity Constraints,” Journal of Political Economy, 1989, 808-827. Fazzari, Steven, Glenn Hubbard and Bruce Petersen, “Financing Constraints and Corporate Investment.” Brookings Papers on Economic Activity, 1988, 141-195. Folta, Timothy, and Jonathan O’Brien. "Entry in the Presence of Dueling Options," Strategic Management Journal, 2004, 121-138. Folta, Timothy. “Governance and Uncertainty: The Trade-off Between Administrative Control and Commitment,” Strategic Management Journal, 1998, 1007-1028. Freel, Mark S.. “Perceived Environmental Uncertainty and Innovation in Small Firms,” Small Business Economics, 2005, 49–64. Fuss, Catherine, and Philip Vermeulen. “Firms’ Investment Decisions in Response to Demand and Price Uncertainty,” Applied Economics, 2008, 2337-2351. Gertler, Mark, and Simon Gilchrist. “Monetary Policy, Business Cycles, and the Behavior of Small Manufacturing Firms,” Quarterly Journal of Economics, 1994, 309-340. 24 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) Ghosal, Vivek, and Prakash Loungani. “The Differential Impact of Uncertainty on Investment in Small and Large Businesses,” Review of Economics and Statistics, 2000, 338-343. Ghosal, Vivek, and Prakash Loungani. “Product Market Competition and the Impact of Price Uncertainty on Investment: Some Evidence From US Manufacturing Industries,” Journal of Industrial Economics, 1996, 217-228. Ghosal, Vivek. “The Effects of Uncertainty and Sunk Costs on Firms’ Decision-Making: Evidence from Net Entry, Industry Structure and Investment Dynamics.” In The Economics of Imperfect Markets, Giorgio Calcagnini and Enrico Saltari (Eds.), Contributions to Economics, Springer: New York, 2009, Chapter 9, 167 – 182. Ghosal, Vivek. “Does Uncertainty Influence the Number of Firms in an Industry?” Economics Letters 50, 1996, 229 – 237. Ghosal, Vivek. “Demand Uncertainty and the Capital-Labor Ratio: Evidence from the U.S. Manufacturing Sector,” Review of Economics and Statistics 73, 1991, 157 – 162. Greasley, David, and Jakob Madsen. “Investment and Uncertainty: Precipitating the Great Depression in the United States,” Economica, 2006, 393–412. Greenwald, Bruce, and Joseph Stiglitz. “Macroeconomic Models with Equity and Credit Rationing,” in Hubbard, R. Glenn. ed., Asymmetric Information, Corporate Finance, and Investment. Chicago: University of Chicago Press, 1990, 15-42. Guo, Hui, and Kevin Kliesen. “Oil Price Volatility and U.S. Macroeconomic Activity,” Federal Reserve Bank of St. Louis Review, 2005, 669-83. Hartman, Richard. “Factor Demand with Output Price Uncertainty,” American Economic Review, 1976, 675–681. Himmelberg, Charles, and Bruce Petersen. “R & D and Internal Finance: A Panel Study of Small Firms in High-Tech Industries,” Review of Economics and Statistics, 1994, 38-51. Hopenhayn, Hugo. “Entry, Exit and Firm Dynamics in Long Run Equilibrium,” Econometrica, 1992, 1127-1150. Huizinga, John. “Inflation Uncertainty, Relative Price Uncertainty, and Investment in U.S. Manufacturing,” Journal of Money, Credit and Banking, 1993, 521-549. Kilian, Lutz. “Exogenous Oil Supply Shocks: How Big Are They and How Much Do They Matter for the U.S. Economy?” Review of Economics and Statistics. 2008, 216-240. Koetse, Mark J., Arno J. van der Vlist and Henri L. F. de Groot, “The Impact of Perceived Expectations and Uncertainty on Firm Investment,” Small Business Economics, 2006, 365-376. 25 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) Lensink, Robert, Hong Bo and Elmer Sterken Investment. Capital Market Imperfections, and Uncertainty: Theory and Empirical Results. Edgar Elgar, 2001. Lensink, Robert, Paul van Steen and Elmer Sterken. “Uncertainty and Growth of the Firm,” Small Business Economics, 2005, 381-391. Li, Dan. “Multilateral R&D Alliances by New Ventures,” Journal of Business Venturing, 2013, 241–260. Li, Yong. “Duration Analysis of Venture Capital Staging: A Real Options Perspective,” Journal of Business Venturing, 2008, 497–512. Li, Yong, Joseph T. Mahoney “When are venture capital projects initiated?” Journal of Business Venturing, 2011, 239–254. Marcellino, Massimiliano, James Stock and Mark Watson. “Macroeconomic Forecasting in the Euro Area: Country Specific versus Euro Wide Information,” European Economic Review, 2003, 1-18. Meese, Richard, and John Geweke. “A Comparison of Autoregressive Univariate Forecasting Procedures for Macroeconomic Time Series,” Journal of Business and Economic Statistics, 1984, 191-200. National Economic Council. “The Small Business Agenda: Growing America’s Small Businesses to Win the Future,” 2011. Oriani, Raffaele, and Maurizio Sobrero. “Uncertainty and the Market Valuation of R&D within a Real Options Logic,” Strategic Management Journal, 2008, 343-361. Pakes, Ariel, and Richard Ericson. “Empirical Implications of Alternate Models of Firm Dynamics,” Journal of Economic Theory, 1998, 1-45. Podoynitsyna, Ksenia, Michael Song, Hans van der Bij and Mathieu Weggeman. “Improving New Technology Venture Performance under Direct and Indirect Network Externality Conditions,” Journal of Business Venturing, 2013 195–210. Saemundsson, Rognvaldur, and Asa Lindholm Dahlstrand. “How Business Opportunities Constrain Young Technology-Based Firms from Growing into Medium-Sized Firms,” Small Business Economics, 2005, 113-129. Schmitt, John, and Nathan Lane “An International Comparison of Small Business Employment,” Center for Economic and Policy Research (CEPR) Reports and Issue Briefs, 2009. Sheets, Nathan, and Robert Sockin. “Does Size Really Matter?: The Evolving Role of Small Firms in the U.S. Economy,” Empirical & Thematic Perspectives, Citi Research, 2012. 26 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) Stein, Luke, and Elizabeth Stone. “The Effect of Uncertainty on Investment: Evidence from Options,” Stanford University Working Paper, 2010. Sutton, John. “Gibrat’s Legacy,” Journal of Economic Literature, 1997(a), 40-59. U.S. Small Business Administration. Advocacy: the voice of small business in government. “Frequently Asked Questions about Small Business Finance,” http://www.sba.gov/sites/default/files/files/Finance%20FAQ%2082511%20FINAL%20for%20w eb.pdf Winker, Peter. “Causes and Effects of Financing Constraints at the Firm Level,” Small Business Economics, 1999, 169-181. Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 27 Figures 1.1-1.4: Annual percentage change in employment by size class Figure 1.1. Percentage change in total employment 0.06 0.04 0.02 0 -0.02 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 -0.04 -0.06 Figure 1.2. Percentage change in employment in large (≥500 employees) businesses 0.06 0.04 0.02 0 -0.02 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 -0.04 -0.06 Figure 1.3. Percentage change in employment in small (<500 employees) businesses 0.04 0.02 0 -0.02 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 -0.04 -0.06 Figure 1.4. Percentage change in employment in small (<20 employees) businesses 0.04 0.02 0 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 -0.02 -0.04 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 28 Figures 2.1-2.4: Annual percentage change in number of businesses by size class Figure 2.1. Percentage change in total number of businesses 0.04 0.02 0 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 -0.02 -0.04 Figure 2.2. Percentage change in number of large (≥500 employees) businesses 0.08 0.06 0.04 0.02 0 -0.02 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 -0.04 -0.06 Figure 2.3. Percentage change in number of small (<500 employees) businesses 0.04 0.02 0 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 -0.02 -0.04 Figure 2.4. Percentage change in number of small (<20 employees) businesses 0.03 0.02 0.01 0 -0.01 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 -0.02 -0.03 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 29 Tables Table 1. Summary Statistics Variable By Size Class Employment EMP Size: All EMP Size: Large w ≥500 employees EMP Size: Small w <20 employees EMP Size: Small w <500 employees Number of Businesses FIRMS Size: All FIRMS Size: Large w ≥500 employees FIRMS Size: Small w <20 employees FIRMS Size: Small w <500 employees Mean Std. Dev. 106.540 10.665 51.999 6.797 20.134 1.048 54.563 3.900 5.539 0.347 0.016 0.002 4.956 0.305 5.523 0.345 30 Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) Estimation tables. Employment specifications: Tables 2.1 and 2.2 Number of Businesses specifications: Tables 3.1 and 3.2 Notes to tables: 1. p-values (two-tailed test), based on efficient standard errors, are reported in parentheses. An * denotes significance at least at the 10% level. All specifications are estimated using data over 1988-2010. 2. The first-order autocorrelation coefficient is denoted by ρ. 3. The variable definitions are as follows. (As noted in section 5, the uncertainty terms are measured in logarithms.) = annual percentage change in employment. = annual percentage change in number of businesses. = annual percentage change in real GDP. Survey forecast uncertainty measures = survey forecast variance of gdp = survey forecast variance of industrial (manufacturing) production Estimated uncertainty measures = gdp uncertainty. = inflation uncertainty. = S&P 500 uncertainty. = fuel price uncertainty. Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 31 Table 2.1. Employment Specifications. Dependent variable: Size Group: Large (≥500 Employee) Businesses Size Group: All Businesses Const S1 -0.0269* (0.001) -0.2362 (0.111) 0.0339 (0.706) 0.5673* (0.007) 0.7491* (0.002) 0.0002 (0.936) -0.0025 (0.112) - S2 -0.0272* (0.001) -0.3491* (0.080) -0.0058 (0.949) 0.5104* (0.001) 0.8640* (0.001) - S3 -0.0270* (0.081) -0.3651* (0.038) 0.0596 (0.525) 0.6953* (0.001) 0.8069* (0.002) - S4 -0.0477* (0.022) -0.4119* (0.049) -0.0288 (0.736) 0.5847* (0.001) 0.9379* (0.001) - S5 -0.0359* (0.001) -0.2578* (0.091) 0.1467 (0.141) 0.6067* (0.001) 0.7471* (0.001) - S6 -0.0516* (0.001) -0.5174* (0.009) 0.1003 (0.201) 0.4495* (0.001) 0.9808* (0.001) - - - - - - - - - - - - - - - - -0.0016 (0.225) -0.0017 (0.259) - S1 -0.0189* (0.058) 0.1224 (0.317) 0.0271 (0.834) 0.6167* (0.008) 0.6277* (0.001) 0.0010 (0.795) -0.0008 (0.678) - - - - - - - - - - 0.0029* (0.079) -0.0031* (0.054) - - - - - - - 0.0005 (0.728) -0.0026* (0.027) - - - - - - - - - - 0.792 0.198 - ρ S2 -0.0197* (0.001) 0.1853 (0.107) 0.0020 (0.987) 0.5778* (0.001) 0.5719* (0.003) - S3 -0.0148 (0.518) -0.0435 (0.798) 0.0899 (0.556) 0.7624* (0.003) 0.6561* (0.006) - S4 -0.0570* (0.038) 0.0928 (0.394) 0.0414 (0.725) 0.5902* (0.001) 0.5653* (0.007) - S5 -0.0264* (0.006) 0.1329 (0.129) 0.0668 (0.577) 0.6264* (0.001) 0.5774* (0.001) - S6 -0.0436* (0.001) -0.0807 (0.524 0.1345) (0.188) 0.5152* (0.001) 0.7329* (0.001) - - - - - - - - - - - - - - - -0.0019 (0.235) 0.0009 (0.635) - - - - - - - - - - - - - - 0.0042 (0.137) -0.0034 (0.155) - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 0.812 0.822 0.806 0.816 -0.0040* (0.001) -0.0016 (0.140) 0.878 0.0001 (0.903) -0.0014 (0.160) - - - -0.0004 (0.474) -0.0019* (0.029) - -0.0015 (0.470) -0.0020* (0.081) - 0.732 0.746 0.786 0.753 0.746 -0.0031* (0.026) -0.0017 (0.147) 0.782 0.222 0.177 0.258 0.099 -0.117 0.175 0.115 0.142 0.203 0.147 0.030 - Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 32 Table 2.2. Employment Specifications Dependent variable: Size Group: Small (<500 Employee) Businesses Const S1 -0.0361* (0.001) -0.4275* (0.006) -0.1217 (0.406) 0.4205* (0.047) 0.8137* (0.001) -0.0019 (0.532) -0.0038* (0.013) - -0.0554* (0.001) -0.4867* (0.006) -0.0417 (0.716) 0.5535* (0.003) 0.7747* (0.002) - -0.0686* (0.021) -0.6164* (0.008) -0.2012 (0.227) 0.5509* (0.001) 0.9617* (0.002) - -0.0382* (0.001) -0.3832* (0.045) 0.0554 (0.751) 0.5868* (0.001) 0.7351* (0.003) - -0.0483* (0.001) -0.5633* (0.003) -0.1006 (0.410) 0.4293* (0.001) 0.9505* (0.001) - - - - - - - - - - - - - - - - -0.0028* (0.016) -0.0036* (0.008) - S1 -0.0192* (0.001) 0.0942 (0.593) -0.2843* (0.084) 0.2969* (0.006) 0.1232 (0.314) -0.0028* (0.062) -0.0031* (0.004) - - - - - - - - - - 0.0004 (0.804) -0.0037* (0.011) - - - - - - - -0.0002 (0.932) -0.0035* (0.029) - - - - - - - - - - 0.738 0.084 - ρ S2 -0.0344* (0.001) -0.6403* (0.001) -0.2457 (0.110) 0.4117* (0.002) 0.9656* (0.001) - S3 S4 Size Group: Small (<20 Employee) Businesses S5 S6 S2 -0.0137* (0.001) 0.0228 (0.883) -0.3003 (0.106) 0.3898* (0.001) 0.1082 (0.318) - S3 S4 S5 S6 -0.0353* (0.001) 0.0212 (0.915) -0.1759 (0.277) 0.3849* (0.001) 0.1238 (0.379) - -0.0222 (0.277) 0.1123 (0.604) -0.2511 (0.279) 0.4481* (0.001) 0.1231 (0.351) - -0.0155* (0.007) 0.1421 (0.527) -0.1418 (0.557) 0.4523* (0.001) 0.0931 (0.501) - -0.0220* (0.012) 0.0135 (0.941) -0.2754* (0.013) 0.3781* (0.001) 0.1812 (0.105) - - - - - - - - - - - - - - - -0.0025* (0.001) -0.0017* (0.027) - - - - - - - - - - - - - - -0.0014 (0.237) -0.0016* (0.078) - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 0.815 0.756 0.741 0.737 -0.0038* (0.004) -0.0008 (0.618) 0.785 -0.0004 (0.457) -0.0008 (0.358) - - - -0.0001 (0.824) -0.0022* (0.038) - -0.0005 (0.688) -0.0007 (0.610) - 0.745 0.798 0.718 0.622 0.648 -0.0027* (0.003) -0.0001 (0.899) 0.754 0.064 0.237 0.257 0.110 0.140 0.001 0.089 0.106 0.053 0.090 0.095 - Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 33 Table 3.1. Number of Businesses Specifications. Dependent variable: Size Group: Large (≥500 Employee) Businesses Size Group: All Businesses Const S1 -0.0080 (0.110) 0.6509* (0.001) -0.3361 (0.115) 0.2958* (0.001) -0.0838 (0.384) -0.0027* (0.052) -0.001 (0.846) - S2 -0.0052 (0.127) 0.5933* (0.002) -0.3399 (0.158) 0.3834* (0.001) -0.1614* (0.085) - S3 -0.0131 (0.229) 0.6646* (0.003) -0.2262 (0.260) 0.3013* (0.001) -0.1077 (0.228) - S4 -0.0147 (0.474) 0.6629* (0.011) -0.0519 (0.853) 0.3755* (0.001) -0.2485* (0.025) - S5 -0.0068 (0.127) 0.7189* (0.001) -0.2926* (0.091) 0.3614* (0.001) -0.1501* (0.065) - S6 -0.0059 (0.230) 0.6887* (0.001) -0.3688* (0.075) 0.3713* (0.001) -0.1300 (0.176) - - - - - - - - - - - - - - - - -0.0019* (0.031) -0.0006 (0.409) - S1 0.0092 (0.335) -0.5427* (0.004) -0.5929* (0.003) 0.0102 (0.961) 1.1061* (0.002) 0.0067* (0.073) -0.0027 (0.569) - - - - - - - - - - -0.0023* (0.016) 0.0010 (0.352) - - - - - - - -0.0017 (0.216) 0.0005 (0.562) - - - - - - - - - - 0.729 -0.101 - ρ S2 -0.0028 (0.661) -0.5486* (0.009) -0.5323* (0.019) -0.1991 (0.241) 1.2435* (0.001) - S3 0.0637* (0.068) -0.7865* (0.001) -0.8287* (0.001) 0.1012 (0.679) 1.4109* (0.001) - S4 -0.0031 (0.949) -0.6529* (0.003) -0.6838* (0.001) -0.1944 (0.272) 1.3490* (0.001) - S5 -0.0147 (0.146) -0.6209* (0.001) -0.6059* (0.001) -0.1520 (0.431) 1.1398* (0.001) - S6 -0.0416* (0.010) -0.6841* (0.005) -0.6216* (0.001) -0.3017* (0.067) 1.1185* (0.001) - - - - - - - - - - - - - - - 0.0032 (0.160) -0.0046 (0.161) - - - - - - - - - - - - - - 0.0093* (0.008) -0.0021 (0.552) - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 0.755 0.747 0.706 0.715 -0.0011 (0.107) 0.0002 (0.805) 0.707 0.0001 (0.971) -0.0035* (0.025) - - - -0.0008* (0.085) -0.0002 (0.736) - 0.0018 (0.672) -0.0021 (0.528) - 0.446 0.470 0.556 0.400 0.462 -0.0059* (0.028) -0.0037 (0.254) 0.555 -0.177 -0.130 -0.021 -0.106 -0.174 0.241 0.323 0.135 0.248 0.151 0.100 - Ghosal and Ye. Uncertainty and Small Businesses. (Work-in-progress.) 34 Table 3.2. Number of Businesses Specifications. Dependent variable: Size Group: Small (<500 Employee) Businesses S1 Const S2 -0.0079 (0.112) 0.6645* (0.001) -0.3385 (0.110) 0.2939* (0.001) -0.0881 (0.351) -0.0027* (0.053) -0.0002 (0.854) - S4 S5 S6 S1 S2 -0.0051 (0.128) 0.6055* (0.001) -0.3412 (0.154) 0.3813* (0.001) -0.1651* (0.077) - -0.0127 (0.235) 0.6781* (0.002) -0.2287 (0.249) 0.2988* (0.001) -0.1108 (0.208) - -0.0137 (0.495) 0.6789* (0.008) -0.0721 (0.794) 0.3736* (0.001) -0.2449* (0.025) - -0.0067 (0.127) 0.7356* (0.001) -0.3007* (0.074) 0.3575* (0.001) -0.1529* (0.056) - -0.0057 (0.244) 0.7051* (0.001) -0.3720* (0.070) 0.3685* (0.001) -0.1345 (0.158) - - - - - - - - - - - - - - - - -0.0019* (0.033) -0.0006 (0.401) - -0.0075 (0.175) 0.4901* (0.016) -0.1020 (0.661) 0.2633* (0.003) -0.0859 (0.431) -0.0031* (0.047) 0.0004 (0.716) - - - - - - - - - - -0.0022* (0.016) 0.0010 (0.340) - - - - - - - -0.0016 (0.241) 0.005) (0.576) - - - - - - - - - - - - ρ S3 Size Group: Small (<20 Employee) Businesses S3 S4 S5 S6 -0.0046 (0.248) 0.4257* (0.051) -0.0946 (0.717) 0.3679* (0.004) -0.1891* (0.068) - -0.0153 (0.278) 0.4556* (0.065) 0.0178 (0.939) 0.2646* (0.007) -0.1100 (0.318) - -0.0282 (0.263) 0.4202 (0.112) 0.2864 (0.368) 0.3507* (0.004) -0.3252* (0.037) - -0.0078 (0.195) 0.4713* (0.029) -0.0006 (0.997) 0.3804* (0.004) -0.1895* (0.044) - -0.0060 (0.361) 0.5053* (0.016) -0.1483 (0.605) 0.3606* (0.004) -0.1455 (0.161) - - - - - - - - - - - - - - - -0.0019* (0.067) -0.0004 (0.692) - - - - - - - - - - - - - - -0.0027* (0.014) 0.0012 (0.359) - - - - - - - - - - - - - - - - - - - - - - - -0.0005 (0.413) -0.0007 (0.244) - - - -0.0008* (0.081) -0.0001 (0.762) - -0.0029* (0.094) 0.0004 (0.705) - - - - - - - - 0.581 0.583 0.609 0.584 0.542 -0.0009 (0.375) -0.0001 (0.966) 0.518 -0.087 -0.142 -0.101 0.116 -0.056 -0.095 0.732 0.757 0.750 0.706 0.718 -0.0011 (0.105) 0.0003 (0.781) 0.711 -0.102 -0.178 -0.132 -0.026 -0.107 -0.178 -