DRAFT EFFECT OF GENERAL UNCERTAINTY ON EARLY AND LATE VENTURE- CAPITAL INVESTMENTS: A CROSS-COUNTRY STUDY Rajeev K. Goel* Illinois State University Iftekhar Hasan New Jersey Institute of Technology and Rati Ram Illinois State University Abstract We use a simple linear model to study the effect of general economic uncertainty on venturecapital investments in 19 OECD countries over the period 1986-1995. The estimates indicate that venture-capital investment, like other investments, is adversely affected by such uncertainty. Following the Dixit-Pindyck (1994) argument, we also estimate the effect of general uncertainty on venture-capital investments at early and late stages, since the former appears less irreversible than the latter. The estimates do suggest the predicted pattern, and late venturecapital investments are apparently more adversely affected by such uncertainty than early-stage venture investments. The pattern is broadly similar for the subsample of European countries. June 2001 * Corresponding author: Economics Department, Illinois State University, Normal, IL, 61790-4200, U.S.A. E-mail: rkgoel@ilstu.edu; Fax: 1-309-438-5228; Tel: 1-309-438-2360 Research assistance of Musliyati Mustaffa is appreciated.
Goel et al. 1 EFFECT OF GENERAL UNCERTAINTY ON VENTURE-CAPITAL INVESTMENTS: A CROSS-COUNTRY STUDY Introduction Venture capital (VC) investments have increased significantly in the past decade and have become a major instrument for financing new business ventures. Researchers have argued that VC investments are complementary, rather than substitute, for other means of financing such as banks (Gompers, 1995, p.1467)). VC investments have also been found to have a positive relation with innovation rates (Kortum and Lerner (2000)). Since most VC investment is at the infancy of business ventures, investment in new ventures is characterized by a high level of uncertainty (Ruhnka and Young (1991, p.115)). For example, there are risks associated with the feasibility of new projects and with returns on investments. In this paper we examine the effect of uncertainty of VC investment. Changes in uncertainties might be due to changes in external or internal factors (Ruhnka and Young, 1991, p.126). Further, the effect of uncertainty might be different across various stages of venture financing (Gompers (1995)). For instance, in later stages of VC investment, the specificity of assets increases and thus the response to uncertainty might be different from early stages when assets are less firm-specific. VC investment is also broken up into early VC investment (including early stage seed and startup investment from both private and government sources) and later stage (expansion investment) to examine whether uncertainty has a different effect of various stages of VC investment. The importance of venture capital as a funding instrument has been increasing in Europe. In our sample of nineteen developed countries, the share of fourteen European countries in total VC
Goel et al. 2 investment increased from twelve percent of total in 1986 to about thirty-one percent in 1995. Analysis is performed for a sample of developed countries and a subsample of European countries to see if there are significant differences between the response of Europe and the full sample. Understanding of these responses will help in determining VC investments in various countries. This will also help explain why VC investments in a particular country might vary over time. If certain types of venture investments respond adversely to greater uncertainty, then other means of financing would have to be considered. To our knowledge this paper is the first attempt to empirically examine the effect of uncertainty on VC investment. The Model, Data, and Main Results The following equation is estimated that relates the share of VC investment in GDP as a linear function of uncertainty, real interest rate and change in aggregate economic activity (see Pindyck and Solimano (1993) and Goel and Ram (1999)): (VC/GDP) it = a + b 1 (UNCERTAINTY) it + b 2 (RLR) it + b 3 (GRY) it-1 + u it (1) Equation (1) shows the venture capital investment divided by the GDP (VCY) in country i during year t as a function of uncertainty, real lending rate (RLR) and the lagged growth rate of GDP (GRY). u is an error term, and a is a constant. We employ use the ordinary least squares (OLS) estimation procedure using the SAS statistical package. Country dummy variables were added to equation (1) to account for country-specific effects. There is some evidence that the nature of VC investment varies across different investment stages (Gompers (1995). For instance, there might be more general and intangible investments in
Goel et al. 3 early stages, while later stages might involve investments in firm-specific assets. R&D-intensive firms are likely to have greater firm- and industry-specific investments (Gompers (1995, p. 1477); also see Goel and Ram (2001)). The specificity of assets leads to greater irreversibility of investments and hence greater (negative) response to uncertainty (Pindyck (1991); also see Dixit and Pindyck (1994)). Ruhnka and Young (1991) report that internal risks in a new venture are greater in its early stages, while external risks are higher in later stages. Venture capital investment is broken up into early venture capital investment (EVCY) and late venture capital investment (LVCY) to determine whether uncertainty has different effects on various stages of VC investment. We employ two measures of uncertainty: a three-year moving standard deviation of inflation (STDINF) and a three-year moving average of inflation (MINF). STDINF reflects inflation uncertainty and MINF is included because of the positive relation between the level of inflation and uncertainty. Further, Pindyck and Solimano (1993) found average inflation to be a major correlate of economic volatility. Note that while these measures are unable to capture all uncertainties that a new venture might be exposed to, they provide a reasonable and quantifiable way to capture external uncertainties. The coefficient of uncertainty on VC investment is expected to be negative. However, the effect of uncertainty on various stages of VC is unclear a priori. Do early and late venture investments respond similarly to changes in uncertainty? A higher lending rate is also expected to have an adverse effect on investment. The effect of the level of general economic activity on VC investment is generally positive as investors have more resources to invest and are optimistic in good economic times. However, this effect could be negative when bad economic times increase
Goel et al. 4 the incentives to invest in new ventures. Our sample covers the period 1986-1995 and includes 19 developed countries. These countries are Australia, Belgium, Canada, Denmark, Finland, France, Germany, Ireland, Italy, Japan, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, UK and USA. Note that all except Australia, Canada, Japan, New Zealand and USA are European countries. Data on venture capital investments are from Jeng and Wells (2000; pp. 244, 249-50, 260) and include investment in seed, startup and expansion of new ventures. 1 Early venture capital excludes expansion investment. The data on inflation rates, lending rates and GDP growth are from the World Bank (1999). Several noteworthy points can be noted from our regression results reported in Table 1. The overall fit of regressions is quite good as shown by the adjusted R 2 and F values. We also performed a RESET test to check for the goodness of fit (see Ramsey and Schmidt (1976) for background on the RESET test) and the test generally reported absence of any significant specification errors. 2 Total VC investments (VCY) appear to be negatively and significantly influenced by both measures of uncertainty. As expected, increases in uncertainty have a dampening effect on VC investment. The effect of lending rate is negative, but statistically insignificant. Higher GDP growth does not seem to have a major impact on VC investments. An 1 Jeng and Wells also report venture capital data for Austria. However, Austria had to be dropped from our sample due to missing data for other variables. 2 A simple version of the RESET test deals with inclusion of the squared predicted value as an additional regressor. Statistical insignificance of the coefficient of the squared predicted value signifies absence any major specification error. Specific details are available from the authors upon request.
Goel et al. 5 important implication of this is that in cases of high or increasing uncertainty, venture capital might not be a suitable funding vehicle and other sources of financing might have to be tapped. When total VC investment is divided into different stages, early VC investments (EVCY), including capital for startups and seed money, is not significantly influenced by uncertainty. While the statistical significance of the estimated parameters in this case is relatively low, the signs are in the right direction. One reason for the lack of strong results in this case might be that the nature of early stage VC investments is quite different from other stages and includes investment in intangible assets (Gompers (1995)). Therefore, separate models might have to be developed to study early stage investments. The results for late VC investments (LVCY), including expansion investments that are likely to be more asset- and firm-specific, are generally similar to total VC investment and starkly different from the early VC. In particular, the effect of uncertainty in this case is negative and significant. This suggests that different stages of VC investment respond differently to changes in uncertainty. While our data are aggregated, it might be the case that we are picking up external risks with our measures of uncertainty that are reported to be higher in later stages of a venture (Ruhnka and Young (1991)). To learn about cross-continent differences, we reran the regression with only 14 European countries. The European subsample consists of a more coherent group of countries than the full sample and the importance of venture capital as a financing vehicle has been increasing in Europe. The results were similar for the European subsample to the full
Goel et al. 6 sample signifying a negative impact of uncertainty on overall VC investment and more pronounced negative impact on late VC investment than early VC investment. The pattern of results is quite similar whether uncertainty is measured as a three-year moving standard-deviation of inflation or a three-year moving average of inflation. Increases in both measures of uncertainty lead to reductions in overall and later stage VC investments. Conclusions We used a simple linear model to determine the effect of uncertainty on VC investments for a broad cross-section of developed countries. Specifically, using data over 1986-1995 for 19 countries, investments in a new venture are related in a simple linear model to uncertainty, lending rate and a measure of the overall economic activity. Analysis is performed for the full sample and the European subsample. The results show that VC investments are adversely affected by increases in uncertainty. This is true whether uncertainty is measured as a three-year moving standard deviation of inflation or as a three-year moving average of inflation. When VC investments are broken up into different stages, it turns out that late venture capital, due to greater irreversibility of investments associated with asset specificity, is more (adversely) affected by increase in uncertainty. Early venture investments are not significantly influenced by changes in uncertainty. The findings are broadly consistent for the European subsample. An important caveat is that a new venture might be exposed to various types of uncertainties: both external and internal (see Cochrane (2001), Ruhnka and Young (1991)). While our
Goel et al. 7 framework is unable to capture all (or even most) of these uncertainties, the results do provide an empirical insight into the effect of uncertainty on VC investment in a large sample of developing countries. It is hoped that with the availability of appropriate data and modeling techniques, the effects of the other types uncertainties can also be empirically determined.
Goel et al. TABLE 1 EFFECT OF UNCERTAINTY ON VENTURE CAPITAL INVESTMENT Full Sample Dep. Var. STDINF MINF RLR GRY -1 adj. R 2 N F VCY -0.020* (-2.54) -0.007 (-0.49) 0.003 (0.36) 0.49 133 6.98 EVCY -0.002 (-0.8) -0.002 (-0.49) 0.004 (1.43) 0.25 133 3.05 LVCY -0.018* (-2.67) -0.005 (-0.40) -0.0004 (-0.05) 0.49 133 7.08 VCY -0.013* (-2.49) -0.007 (-0.51) 0.004 (0.38) 0.49 133 6.95 EVCY -0.001 (-0.79) -0.002 (-0.49) 0.004 (1.43) 0.25 133 3.05 LVCY -0.012* (-2.61) -0.005 (-0.42) -0.0003 (-0.03) 0.49 133 7.05 European Subsample VCY -0.014+ (-1.73) -0.006 (-0.39) -0.005 (-0.45) 0.51 107 7.88 EVCY 0.0003 (0.14) 0.0002 (0.06) 0.002 (0.91) 0.20 107 2.64 LVCY -0.014+ (-1.94) -0.006 (-0.44) -0.007 (-0.72) 0.49 107 7.48 VCY -0.009+ (-1.68) -0.006 (-0.41) -0.005 (-0.44) 0.51 107 7.85 EVCY 0.0002 (0.17) 0.0002 (0.05) 0.002 (0.90) 0.20 107 2.64 LVCY -0.009+ (-1.88) -0.006 (-0.46) -0.007 (-0.72) 0.49 107 7.46
Goel et al. Notes: Our pooled sample covers annual data for the period 1986-1995. The countries included are Australia, Belgium, Canada, Denmark, Finland, France, Germany, Ireland, Italy, Japan, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, UK and USA. The European subsample includes all countries except Australia, Canada, Japan, New Zealand, USA. VCY is the total venture capital investment per million GDP, and EVCY, LVCY stand for early and late VC; STDINF is a three-year moving standard deviation of inflation, and MINF is a three-year moving average of inflation; RLR is the real lending rate and GRY -1 is the lagged growth rate of GDP. Relevant t-statistics are in parentheses, (*) denotes statistical significance at least at the 5% level, + indicates significance at the 10% level. All equations include a constant term and countryspecific dummy variables. The estimates for these are not reported and are available on request.
Goel et al. References Cochrane, J.H., The Risk and Return of Venture Capital, working paper, University of Chicago, 2001, http://gsbwww.uchicago.edu/fac/john.cochrane/research/papers/ Dixit, A.K. and R.S. Pindyck, Investment Under Uncertainty, Princeton, NJ: Princeton Univefrsity Press, 1994. Goel, R.K. and R. Ram, Irreversibility of R&D Investment and the Adverse Effect of Uncertainty: Evidence from the OECD Countries, Economics Letters, 71, 2001, 287-291. Goel, R.K. and R. Ram, Variations in the Effect of Uncertainty on Different Types of Investment: An Empirical Investigation, Australian Economic Papers, 38, 1999, 481-492. Gompers, P.A., Optimal Investment, Monitoring, and the Staging of Venture Capital, Journal of Finance, 50, 1995, 1461-1489. Jeng, L.A. and P.A. Wells, The Determinants of Venture Capital Financing: Evidence Across Countries, Journal of Corporate Finance, 6, 2000, 241-289. Kortum, S. and J. Lerner, Assessing the Contribution of Venture Capital to Innovation, Rand Journal of Economics, 31, 2000, 674-692. Pindyck, R.S., Irreversibility, Uncertainty, and Investment, Journal of Economic Literature, 29, 1991, 1110-1148. Pindyck, R.S. and A. Solimano, Economic Instability and Aggregate Investment, in O.J. Blanchard and S. Fischer (eds.), NBER Macroeconomics Annual 1993, Cambridge, MA:
Goel et al. MIT Press, 259-317.Ruhnka, J.C. and J.E. Young, Some Hypotheses About Risk in Venture Capital Investing, Journal of Business Venturing, 6, 1991, 115-133. World Bank, World Development Indicators, 1999. CD-ROM, http://www.worldbank.org/data/