چرا حساسیت جریان سرمایه گذاری پول نقد، به شدت کاهش یافته است؟ افزایش R & D و تحولات بازار سهام

The study of the investment-cash flow (ICF) sensitivity constitutes one of the largest literatures in corporate finance, yet little is known about changes in the ICF relationship over time, and the literature has largely ignored how rising R&D investment and developments in equity markets have impacted ICF sensitivity estimates. We show that for the time period 1970–2006, the ICF sensitivity: (i) largely disappears for physical investment, (ii) remains comparatively strong for R&D, and (iii) declines, but does not disappear, for total investment. We argue that these findings can largely be explained by the changing composition of investment and the rising importance of public equity as a source of funds, particularly for firms with persistent negative cash flows.

The study of the investment-cash flow (ICF) sensitivity constitutes one of the largest empirical literatures in corporate finance. Many studies find that firms which are a priori more likely to confront binding financing constraints display a greater sensitivity of investment to cash flow. Although there is disagreement on how to interpret the findings in ICF studies, ICF regressions continue to be used extensively as a tool to study a variety of issues in corporate finance. But despite the literature’s prominence, little is known about the stability of the ICF relationship over time and the R&D-cash flow relationship has been largely ignored. In particular, the literature has not explored how the rising importance of R&D or continued improvements in equity markets may have affected measures of the ICF sensitivity. There are several reasons to suspect that the ICF sensitivity has declined significantly. Perhaps the most important reason is developments in US equity markets over the last three decades. One major improvement was the creation of the Nasdaq – launched in 1971 and repeatedly improved thereafter – which likely gave young firms access to a much more efficient stock exchange than was available to them for most of the 20th century. In the last few decades, there has been a sharp increase in the use of public equity finance by young firms, suggesting that stock issues may have become a closer substitute for internal finance. A second, closely related reason for a declining ICF sensitivity is the sharp increase in the fraction of publicly traded firms that report persistent negative cash flows. Since these firms often make very heavy use of public equity to expand investment when cash flow is particularly low, failure to account for external finance in ICF regressions can result in a downward omitted variable bias in the estimated cash flow coefficient. Third, there has been a sharp change in the composition of total investment: the absolute and relative importance of physical investment has declined substantially and R&D intensity has risen dramatically for the typical publicly traded manufacturing firm. Because almost all ICF studies focus on physical investment, the declining relative importance of physical investment can, by itself, lead to a decline in the conventionally measured ICF sensitivity. This paper makes three main contributions. First, we provide a systematic documentation of what has happened to the ICF sensitivity over time. The only other studies to examine the ICF sensitivity over relatively long periods of time are Allayannis and Mozumdar, 2004 and Agca and Mozumdar, 2008, both of which show a substantial decline in the ICF sensitivity for physical investment over time. We expand on these studies by considering R&D and total investment in addition to physical investment. Comparatively few studies have examined the ICF sensitivity for R&D and no other studies have explored changes in the R&D-cash flow sensitivity over time. Second, we examine the role of external finance in ICF regressions by estimating dynamic investment models that include measures of stock and debt issues. We argue that these are potentially important omitted variables in most ICF studies, and their inclusion helps address some concerns that have been raised about interpreting ICF sensitivities. Finally, we explore why the estimated ICF sensitivity has changed over time, focusing on the impact of both capital market developments and the changing composition of investment. We explore changes in the ICF sensitivity between 1970 and 2006 using Compustat data for manufacturing firms, divided into three subperiods: 1970–1981, 1982–1993 and 1994–2006. We also split firms into two categories, young and mature, where young firms have stock prices for fewer than ten years before the start of a given subperiod. We expect improvements in equity markets to have the greatest impact on young firms, since they are the most likely to be “equity-dependent.” Our summary statistics show that, over time, there has been: (i) a very large decline in the physical investment share of total investment, (ii) a dramatic rise in the R&D-to-assets ratio, particularly for young firms, (iii) a very sharp rise in the proportion of negative cash flow observations, (iv) a substantial decline in the median cash flow ratio, particularly for young firms, and (v) a pronounced rise in the share of young firm finance accounted for by new stock issues. We estimate the sensitivity of physical investment, R&D, and total investment (physical investment plus R&D) to cash flow with the standard OLS fixed effects model used in the ICF literature. Our main results, however, are based on dynamic investment regressions using GMM where cash flow and other financial variables are treated as endogenous. The other financial variables include both new stock issues and debt finance, variables that potentially matter a great deal for investment but are rarely included in ICF studies. A consistent pattern of results emerges from the OLS and GMM regressions. For physical investment, our OLS estimates show that, even after controlling for negative cash flows, there is a dramatic decline in the ICF sensitivity over time. Similarly, GMM regressions that control for negative cash flow and include measures of external finance show a decline in the ICF sensitivity of at least 70% between 1970–1981 and 1994–2006. We argue that much of this decline is due to the sharp fall in the physical investment share of total investment. For R&D investment, on the other hand, there is no decline in the ICF relationship over time in OLS regressions (that control for negative cash flow) or GMM regressions (that control for external finance). We will emphasize, however, that the cash flow coefficients for R&D, absent improvements in equity markets, should have risen a great deal because of the sharp rise in the R&D share of total investment that occurred during the period we study. Finally, the ICF pattern for total investment reflects a blend of the ICF pattern for physical and R&D investment and shows that the overall ICF sensitivity declined substantially over the time period of our study, but it did not disappear. Our regression findings also shed light on the different roles of debt and stock issues as well as the rise in importance of the US stock market. First, in the physical investment regressions (where investment presumably has collateral value), debt coefficients are substantial but there is little or no evidence of stock effects. Second, for R&D (arguably the equity-dependent type of investment), stock issues play a more important role than debt, especially in the final period. Third, stock issues appear especially important for young firms, particularly young firms with negative cash flows (arguably the most equity-dependent type of firm). Fourth, in the R&D regressions, estimated coefficients on stock issues rise from near zero to large values by the final period for young firms, consistent with improvements in equity markets. Finally, our findings show that young firms with persistent negative cash flows rely heavily on stock issues to finance R&D and failure to account for this appears to cause a downward bias (e.g., negative cash flow coefficients) in the estimated R&D-cash flow sensitivity. To summarize, the ICF sensitivity for physical investment has fallen dramatically, the ICF sensitivity for R&D remains comparatively strong, and the ICF sensitivity for total investment has fallen substantially. Absent improvements in equity markets, however, the ICF sensitivity of R&D should have increased a great deal, given the sharp change in the composition of investment. The bottom line is that the overall ICF sensitivity has declined noticeably in recent decades and improvements in equity markets were likely a significant contributing factor.

We conclude with a summary of our main findings and a brief discussion of the main implications. We have shown that the ICF sensitivity for physical investment has fallen dramatically. Even after controlling for negative cash flow firms and the use of external finance (Table 4, Panel B), cash flow coefficients for physical investment have fallen by over 70% since the 1970–1981 period for both young and mature firms. A likely explanation for much of this decline is the sharp fall in the physical investment share of total investment, which we denoted earlier in the paper as βCAP. The summary statistics show that for positive cash flow young firms, median βCAP declined from 0.82 in 1970–1981 to 0.33 in 1994–2006, a decline of around 60%. Thus, based on Prediction 1, we would expect a very large fall in the physical investment sensitivity simply because physical investment has become a much smaller use of funds. Superficially, the R&D-cash flow sensitivity (Table 5) displays a very different pattern: Cash flow coefficients do not decline over time. But recall that the R&D share of total investment (βRD) rose dramatically over time. For example, for all young firms ( Table 1), the median value of βRD rose from 0.18 in 1970–1981 to 0.75, a four-fold increase. Based on Prediction 1, without improvements in capital markets, there should have been a very sharp rise in the R&D ICF sensitivity. As summarized below, a likely reason for the absence of a rise in the R&D ICF sensitivity is improvements in equity markets. Because of compositional issues, the clearest evidence of a decline in the ICF sensitivity can be found in the GMM regressions for total investment (Table 6). For all firms (Panel A), over the period of our study, cash flow coefficients fell 60% for young firms and 41% for mature firms. This decline is considerably smaller than for physical investment ( Table 4), consistent with Prediction 4. For positive cash flow firms ( Table 6, Panel B), cash flow coefficients declined by approximately 45% for young firms and 51% for mature firms. Among these positive cash flow firms, the cash flow coefficient for young firms is still substantial in the final period (0.240). Because of the changing composition of investment, future ICF studies should consider total investment, not just physical investment. Several findings in our study suggest that public equity markets improved substantially in recent decades and contributed to the decline in the ICF sensitivity. Most obviously, stock issues by young firms rose from near zero to very high levels in the final period. (In contrast, debt finance rises only modestly.) In addition, we have argued (Prediction 5) that R&D is likely the equity-dependent investment and young firms are the equity-dependent type of firm, and we in fact find strong stock effects primarily for R&D, and only for young firms. Furthermore, stock coefficients for young firm R&D rise from near zero in the early period to large and significant values in the final period, which we argue (Section 2.4.5) is consistent with improvements in equity markets. Finally, we point to the impact that the dramatic increase in R&D’s share of total investment (βRD) had on the R&D-cash flow sensitivity. Absent improvements in equity markets, this change in the composition of investment should have led to a sharp rise in the R&D-cash flow sensitivity. Our results indicate that this did not happen. Given the large size of stock issues and the large stock coefficients in the R&D regression, a likely explanation for the static R&D-cash flow sensitivity is that public equity finance became a much closer substitute for internal equity over this time period. The sharply declining total investment-cash flow sensitivity, together with the rising stock effects in total investment regressions, is also consistent with public equity becoming a better substitute for internal funds over the period we study. The above findings have implications for an important literature on the role of the stock market for corporate investment. Well-known studies such as Morck et al., 1990 and Blanchard et al., 1993 conclude that the stock market plays only a limited role in firm investment decisions, and may even be a “sideshow.” An important recent study by Baker et al. (2003), however, ranks firms according to “equity dependence” and finds that the total investment of equity-dependent firms over the period 1980–1999 is more sensitive to Tobin’s Q, supporting the existence of an “equity finance channel.” While our approach differs from their study – we focus on stock issues rather than stock prices – our findings support their conclusion concerning the importance of the equity finance channel. Our findings strongly indicate, however, that the equity finance channel is not uniform across types of investment: We find strong stock effects only for R&D. Furthermore, we do not detect substantial stock effects for R&D until the 1980s, suggesting that the stock market may have largely been a sideshow until the last few decades.