A. Theoretical Building Blocks

Macroeconomic theory traditionally modeled investment according to the notion that there are convex adjustment costs to investment, leading to a flexible accelerator model of investment where investments follow a path that results in the actual capital stock adjusting to some desired capital stock over time.⁴ Tobin (1969) introduced “Q”, the ratio of the market value of a firm to the replacement cost of its capital stock, as a measure of its incentive to invest in capital. The link between convex costs of adjustment and the Q-theory of investment was made explicit by Mussa (1977) and Abel (1983), both focusing on marginal q—the ratio of the value of an additional unit of capital to its acquisition cost—and showing that q is a sufficient statistic for investment.

In the corporate finance literature, a large class of theories arose on how asymmetric information or agency problems lead external finance to be costly, and thus affect firm investment. See Stein (2003) for a detailed survey. A prominent model that can roughly be placed in this group is Myers’s (1977) theory of debt overhang, which shows how outstanding debt can distort a firm’s investment incentives and thus lead to underinvestment. With high debt, decision makers that seek to maximize the value of their firm’s equity (shareholders or managers compensated with equity) may be reluctant to invest in positive net present value projects because part of the return will accrue to the debt holders, compared to a situation where the face value of the firm’s debt is lower. Steigum (1983) tied these two literatures together by modeling a specific financial friction—where a firm’s cost of borrowing depends on its debt to equity ratio—and showing that the optimal investment plan was approximated by the flexible accelerator model.

More recent theories of investment focus on: (i) the structure of firm financing; and (ii) the composition of investment. An implication of Myers (1977) is that debt of shorter maturity would lessen the underinvestment problem, as debt that matures before an investment decision cannot cause overhang. Diamond and He (2014) study the effect of debt maturity on investment in a dynamic model, and find that shorter maturities can both increase or decrease overhang effects. For immediate investment, shorter-term debt typically imposes lower overhang. Shorter-term debt leads to more volatile future debt overhang, however, making future investment incentives volatile. Regarding the composition of investment, Hennessy (2004) shows that debt overhang distorts both the level and composition of investment, with underinvestment being more severe for long-lived assets.

B. Empirical Literature

A large empirical literature on the determinants of firm investment exists, much of it focused on estimating the relationship between investment and Tobin’s Q, alongside many studies that establish that establish that firms with more cash and less debt invest more,⁵ countering the implication of Modigliani-Miller (1958) that a firm’s investment should depend only on the profitability of its investment opportunities as measured, for example, by Q.

More recent papers revisit the Tobin’s Q hypothesis, focusing on the measurement error of Q in previous work. Since marginal q is unobservable, Tobin’s average Q, the ratio of equity plus debt value to replacement cost of the capital stock, is commonly used as an empirical proxy.⁶ However, the focus on Tobin’s Q to explain firm-level investment is receding. Eberly et al. (2012), for example, show that the best predictor of current investment at the firm level is lagged investment, working better than Tobin’s Q and cash flow combined on the usual sample of U.S. firms. Gala and Gomes (2013) question the use of Q-ratios in empirical studies of investment, showing that Q bears only a weak relation to optimal investment under all but very stringent assumptions about the nature of markets and technologies, and find that it is better to use the underlying state variables, such as firm size and sales, directly. Using measures of firm fundamentals directly also avoids the measurement error concerns induced by potential stock market misvaluations (Blanchard, Rhee and Summers, 1993), which is even more relevant when studying investment in countries with a lower share of publicly-listed firms and lower stock-market participation.

Empirically, investment expenditures are often lumpy, irregular and infrequent (Doms and Dunne 1998) as opposed to the rather smooth increments to the capital stock implied by theory, and big investment projects typically are irreversible, or at least nearly so. Cooper, Haltiwanger, and Power (1995) find that the probability of an investment spike for a plant increases with the time that has elapsed since the previous spike, lending additional support to the view of a microeconomic environment characterized by nonconvexities in the adjustment technology.

The only paper, to our knowledge, that has focused on the composition of investment is Aghion et al (2010), who model firms that engage in two types of investment: short-term and long-term, where long-term investment contributes more to productivity growth. They show that the share of long-term investment in total investment is countercyclical when financial markets are perfect; but that this share may turn pro-cyclical when firms face tight credit constraints due to the higher liquidity risk of long-term investment. Thus, through its effect on the cyclical composition of investment, tighter credit can lead to both higher volatility and lower mean growth.

C. Investment in India

Tokuoka (2012) took a first look at the investment slowdown in India and found that macroeconomic factors could partially explain corporate investment, but could not fully account for the recent weak investment, and that corporate investment was weaker in cities and states with more difficult business environments. Anand and Tulin (2014) also found that standard macro-financial variables could not fully explain the slowdown in investment at the aggregate level, and that heightened policy uncertainty played a key role. Bahal, Raissi, and Tulin (2016) study the relationship between public-capital accumulation and private investment at the aggregate level India. Their results suggest that while public-capital accumulation crowds out private investment in India over 1950–2012 the opposite is true after 1980, which they attribute to the policy reforms which started during early 1980s and gained momentum after the 1991 Indian balance of payments crisis.

A. Benchmark Specification, Firm-Level Investment Rates

The benchmark econometric specification is as follows:

where Inv_it is investment rate of firm i in year t. This is measured as the expenditure on fixed assets in period t, scaled by the estimate of capital stock at the beginning of the period. That is, by net fixed assets at the end of period t-1. The main explanatory variable of interest is the leverage ratio, Lev, calculated as the debt to equity ratio (book values) of the firm, and X is a vector of control variables including:

• ▪ the cash flow to (beginning of period) capital stock ratio,

• ▪ the sales to (beginning of period) capital stock ratio,

• ▪ the size of the firm (log of beginning of period capital stock).

We include a firm-specific effect, μ_i, year dummies, v_t, to capture aggregate shocks and ɛ_it is an error term uncorrelated with current or lagged values of X. We first employ a panel fixed effects estimator, as our benchmark for the subsequent regressions. The hypothesis is the investment rate is lower for firms that are more highly levered, i.e. that β < 0 .

B. Growth Opportunities and External Demand

In the next specification we explore further the relationship between expectations of future profitability and investment. First, with exporting firms being more vulnerable to the global trade slowdown following the global financial crisis, we study how investment was affected at firms in industries where export demand fell, and in particular at exporting firms. The corresponding specification is as follows:

where:

• ▪ Exp is indicator variable for exporting firms (in specifications without fixed effects) or export intensity, measured as the export to sales ratio;

• ▪ WorldImportGrowth_jt–1 is the growth in the volume of imports in industry j at time t-1 of India’s trading partners;⁸

• ▪ the specification includes an interaction term between the exporting firm indicator variable, or export intensity, and the growth in the volume of world imports by industry;

• ▪ and certain estimations also include an interaction terms between the exporting firm indicator variable, or export intensity, and a post-global financial crisis dummy variable, which is equal to one from 2009 onwards.

C. Financial Frictions

To further test the hypothesis that financial frictions are a factor in decreasing firm investment, we next estimate an interaction terms between leverage and several measures of financial constraints. The three proxies for financial constraints are:

• 1) Firm age, as older firms have developed long-term lending relationships and are less subject to asymmetric information problems (Hadlock and Pierce 2010).

• 2) Whether a firm is a member of a business group. Firms that belong to business groups have access to intra-group loans from the affiliated group firms (Gopalan et al. 2014) and should thus be less financially constrained compared to other firms.

• 3) The Rajan and Zingales (1998) measure of dependence on external finance, which captures an industry’s technological need for external financing.

If leverage affects firm investment by making borrowing more difficult or costly, the negative relationship between investment and leverage should be stronger for firms that are more financially constrained.

Additionally, as financial constraints may differ between bank-dependent firms, firms that rely more on capital markets, and, in the case of India, firms that rely more on informal sources of finance, we test whether leverage with a greater bank-debt component affects investment. Unlike in samples of U.S. firms, where the distinction is between bank borrowing and capital market borrowing, and bank-dependent firms are more financially constrained (see Chen et al. 2017), in India many firms, particularly smaller ones, rely on alternative sources of finance.⁹ This includes financing from all nonbank, nonmarket sources, and is generally backed by non-legal mechanisms. Bank loans, in turn, are the second most important external financing source for firms, while financial markets have played a limited role in financing the investment and operations of Indian firms (Allen et al. 2012). We find that Indian firms with a greater share of bank debt in fact have higher investment. However, the greater the share of bank debt that is secured debt, the lower investment, and the higher the share of short-term bank debt, the lower investment.

Next, as an alternative measure to financial leverage, we proxy firm financial distress by its interest coverage ratio (ICR). An ICR of less than one is an often used indicator of corporate vulnerability, however, some firms have a low ICR due to shocks to revenues or costs, or because they are new firms that have not yet realized their revenue potential. To account for this, Chung and Ratnovski (2016) use a more persistent measure of ICR: firms that have had and ICR less that one for several consecutive years. We find, as expected, that firms with higher ICRs have higher investment rates, and that firms with ICR < 1 in the previous year, or for several consecutive years, have lower investment rates.

D. New Investment Variable and Investment Project Length

In the next specification, we study a new investment variable built up from the project-level data: the probability that a firm announces a new investment project in a particular year. The predictions of theory apply more directly to this variable than cash expenditure on investment, which is the variable primarily studied in the previous empirical literature, and which has been shown to depend strongly on past investment (Eberly et al. 2012).

where NewProject_sit = 1 if firm i announces it is beginning new project s in year t. Since firms with high leverage may also have to abandon investment projects due to financing difficulties, we also estimate the relationship between financial leverage and probability of completing an investment project. Finally, we get at the composition of investment by analyzing the factors that affect project length:

E. Robustness

Several additional specifications were estimated as robustness exercises. To allow for the possibility that high leverage is less of an issue for firms with:

• ▪ higher cash flow or sales, who may have better growth opportunities, or

• ▪ for larger firms who may have more established relationship with lenders,

an interaction term of leverage with each of cash flow, sales, and size variables was included one by one in equation (1). In each case, the estimated coefficient on the interaction term was not statistically significant. This suggests that that issue of leverage negatively affecting investment is widespread across Indian firms, and not mitigated for firms with higher sales or for larger firms.

For specifications with the firm investment rate as the dependent variable, the investment rate in the previous period was included as an explanatory variable and two-stage least squares estimation, with the second and third lags of the investment rate as instruments, as well as Arellano and Bond’s (1991) consistent GMM estimator, were used given the potential endogeneity of the lagged dependent variable. The results presented in Tables 4–8 are not sensitive to the inclusion of the lagged investment rate.

Table 4.

Firm Investment—Benchmark Specification

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1Notes: Size is the log of the beginning of period capital stock and other variables are as in Table 2.

Table 4.

Firm Investment—Benchmark Specification

Dependent variable: Investment/K
	(1)	(2)	(3)
	OLS	FE	FE
	Industry Effects
Debt/Equity (t-1)	−0.944***	−1.056***	−1.025***
	(0.099)	(0.157)	(0.159)
Cashflow/K (t-1)	0.001	0.001	0.002*
	(0.001)	(0.001)	(0.001)
Sales/K (t-1)	0.011***	0.014***
	(0.000)	(0.001)
Sales growth (t-1)			0.000
			(0.000)
Size (t-1)	−2.977***	−26.010***	−29.530***
	(0.172)	(0.875)	(0.828)
Constant	57.217***	77.054***	87.712***
	(2.210)	(2.117)	(1.800)
Observations	74,205	74,205	74,205
No. Firms	10974	10974	10974
Rsquared	0.078	0.149	0.137

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1Notes: Size is the log of the beginning of period capital stock and other variables are as in Table 2.

View Table

Table 4.

Firm Investment—Benchmark Specification

Dependent variable: Investment/K
	(1)	(2)	(3)
	OLS	FE	FE
	Industry Effects
Debt/Equity (t-1)	−0.944***	−1.056***	−1.025***
	(0.099)	(0.157)	(0.159)
Cashflow/K (t-1)	0.001	0.001	0.002*
	(0.001)	(0.001)	(0.001)
Sales/K (t-1)	0.011***	0.014***
	(0.000)	(0.001)
Sales growth (t-1)			0.000
			(0.000)
Size (t-1)	−2.977***	−26.010***	−29.530***
	(0.172)	(0.875)	(0.828)
Constant	57.217***	77.054***	87.712***
	(2.210)	(2.117)	(1.800)
Observations	74,205	74,205	74,205
No. Firms	10974	10974	10974
Rsquared	0.078	0.149	0.137

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1Notes: Size is the log of the beginning of period capital stock and other variables are as in Table 2.

Table 5.

Firm Investment—Exporters and External Demand

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1Explanatory variables not shown are the cashflow to capital ratio, the sales to capital ratio, and the size of the firm.Notes: Exporter is a dummy variable equal to one if a firm is an exporter, Export Intensity is the exports to sales ratio, post-GFC dummy is equal to one from 2009 onwards, World Import Growth (Industry) is the growth in the volume of imports in industry j of India’s trading partners.

Table 5.

Firm Investment—Exporters and External Demand

Dependent variable: Investment/K
	(1a)	(1b)	(2a)	(2b)	(3)	(4a)	(4b)
	OLS	OLS	FE	FE	FE	FE	FE
	Industry EffectsIndustry Effects
Debt/Equity (t-1)	−0.924***	−0.925***	−1.052***	−1.050***	−1.350***	−1.350***	−1.348***
	(0.100)	(0.100)	(0.157)	(0.157)	(0.168)	(0.168)	(0.168)
Exporter	1.892***	1.981***
	(0.566)	(0.705)
Exporter * post-GFC dummy		−0.228
		(1.007)
Export Intensity (t-1)			0.047*	0.060**
			(0.028)	(0.030)
Export Intensity (t-1) * post-GFC dummy				−0.035
				(0.025)
World Import Growth (Industry) (t-1)					−0.006	−0.017	−0.017
					(0.011)	(0.013)	(0.013)
World Import Growth (t-1) * Export Intensity (t-1)						0.001**	0.001**
						(0.000)	(0.000)
World Import Growth (t-1) * Export							−0.032
Intensity (t-1) * post-GFC dummy							(0.030)
Observations	74,205	74,205	73,252	73,252	45,202	45,202	45,202
No. Firms	10974	10974	10827	10827	5954	5954	5954
Rsquared	0.078	0.078	0.148	0.148	0.161	0.161	0.161

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1Explanatory variables not shown are the cashflow to capital ratio, the sales to capital ratio, and the size of the firm.Notes: Exporter is a dummy variable equal to one if a firm is an exporter, Export Intensity is the exports to sales ratio, post-GFC dummy is equal to one from 2009 onwards, World Import Growth (Industry) is the growth in the volume of imports in industry j of India’s trading partners.

View Table

Table 5.

Firm Investment—Exporters and External Demand

Dependent variable: Investment/K
	(1a)	(1b)	(2a)	(2b)	(3)	(4a)	(4b)
	OLS	OLS	FE	FE	FE	FE	FE
	Industry EffectsIndustry Effects
Debt/Equity (t-1)	−0.924***	−0.925***	−1.052***	−1.050***	−1.350***	−1.350***	−1.348***
	(0.100)	(0.100)	(0.157)	(0.157)	(0.168)	(0.168)	(0.168)
Exporter	1.892***	1.981***
	(0.566)	(0.705)
Exporter * post-GFC dummy		−0.228
		(1.007)
Export Intensity (t-1)			0.047*	0.060**
			(0.028)	(0.030)
Export Intensity (t-1) * post-GFC dummy				−0.035
				(0.025)
World Import Growth (Industry) (t-1)					−0.006	−0.017	−0.017
					(0.011)	(0.013)	(0.013)
World Import Growth (t-1) * Export Intensity (t-1)						0.001**	0.001**
						(0.000)	(0.000)
World Import Growth (t-1) * Export							−0.032
Intensity (t-1) * post-GFC dummy							(0.030)
Observations	74,205	74,205	73,252	73,252	45,202	45,202	45,202
No. Firms	10974	10974	10827	10827	5954	5954	5954
Rsquared	0.078	0.078	0.148	0.148	0.161	0.161	0.161

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1Explanatory variables not shown are the cashflow to capital ratio, the sales to capital ratio, and the size of the firm.Notes: Exporter is a dummy variable equal to one if a firm is an exporter, Export Intensity is the exports to sales ratio, post-GFC dummy is equal to one from 2009 onwards, World Import Growth (Industry) is the growth in the volume of imports in industry j of India’s trading partners.

Table 6.

Firm Investment—Interaction with Financial Constraints

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1Explanatory variables not shown are the cashflow to capital ratio, the sales to capital ratio, and the size of the firm.Notes: Age denotes the age of the firm; BusinessGroup is a dummy variable equal to one if the firm is a member of a business group; ExternalDependence is the Rajan and Zingales (1998) industry-level measure of dependence on external finance; Low, Middle, and High External Dependence denote the bottom, middle, and top tertile of industries in terms of their dependence on external finance.

Table 6.

Firm Investment—Interaction with Financial Constraints

Dependent variable: Investment/K
	(1)	(2)	(3)	(4)
	Age	BusGroup	R&Z	R&Z
Debt/Equity	−1.417***	−1.473***	−1.241***
	(0.289)	(0.211)	(0.227)
Debt/Equity * Age	0.012**
	(0.006)
Debt/Equity * BusinessGroup		0.999***
		(0.315)
Debt/Equity * ExternalDependence			−0.515
			(0.639)
Debt/Equity * Low ExternalDependence				−1.230***
				(0.258)
Debt/Equity * Middle ExternalDependence				−1.285***
				(0.356)
Debt/Equity * High ExternalDependence				−1.642***
				(0.290)
Observations	73,963	74,205	44,096	44,096
No. Firms	10895	10974	5736	5736
Rsquared	0.148	0.149	0.166	0.166

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1Explanatory variables not shown are the cashflow to capital ratio, the sales to capital ratio, and the size of the firm.Notes: Age denotes the age of the firm; BusinessGroup is a dummy variable equal to one if the firm is a member of a business group; ExternalDependence is the Rajan and Zingales (1998) industry-level measure of dependence on external finance; Low, Middle, and High External Dependence denote the bottom, middle, and top tertile of industries in terms of their dependence on external finance.

View Table

Table 6.

Firm Investment—Interaction with Financial Constraints

Dependent variable: Investment/K
	(1)	(2)	(3)	(4)
	Age	BusGroup	R&Z	R&Z
Debt/Equity	−1.417***	−1.473***	−1.241***
	(0.289)	(0.211)	(0.227)
Debt/Equity * Age	0.012**
	(0.006)
Debt/Equity * BusinessGroup		0.999***
		(0.315)
Debt/Equity * ExternalDependence			−0.515
			(0.639)
Debt/Equity * Low ExternalDependence				−1.230***
				(0.258)
Debt/Equity * Middle ExternalDependence				−1.285***
				(0.356)
Debt/Equity * High ExternalDependence				−1.642***
				(0.290)
Observations	73,963	74,205	44,096	44,096
No. Firms	10895	10974	5736	5736
Rsquared	0.148	0.149	0.166	0.166

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1Explanatory variables not shown are the cashflow to capital ratio, the sales to capital ratio, and the size of the firm.Notes: Age denotes the age of the firm; BusinessGroup is a dummy variable equal to one if the firm is a member of a business group; ExternalDependence is the Rajan and Zingales (1998) industry-level measure of dependence on external finance; Low, Middle, and High External Dependence denote the bottom, middle, and top tertile of industries in terms of their dependence on external finance.

Table 7.

Firm Investment—Bank Dependence

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1Explanatory variables not shown are the cashflow to capital ratio, the sales to capital ratio, and the size of the firm.

Table 7.

Firm Investment—Bank Dependence

Dependent variable: Investment/K
	(1)	(2)	(3)	(4)
	FE	FE	FE	FE
Debt/Equity (t-1)	−1.056***	−1.244***	−1.277***
	(0.157)	(0.158)	(0.159)
Bank Debt / Total debt (t-1)		0.034**	0.032**	0.073***
		(0.014)	(0.014)	(0.022)
Secured Bank Debt / Bank debt (t-1)			−0.043*
			(0.024)
Long-term Bank Debt/Equity (t-1)				0.281
				(0.685)
Short-term Bank Debt/Equity (t-1)				−4.900***
				(0.635)
Observations	74,205	61,684	60,796	35,735
No. Firms	10,974	9,627	9,520	6,931
Rsquared	0.147	0.159	0.161	0.199

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1Explanatory variables not shown are the cashflow to capital ratio, the sales to capital ratio, and the size of the firm.

View Table

Table 7.

Firm Investment—Bank Dependence

Dependent variable: Investment/K
	(1)	(2)	(3)	(4)
	FE	FE	FE	FE
Debt/Equity (t-1)	−1.056***	−1.244***	−1.277***
	(0.157)	(0.158)	(0.159)
Bank Debt / Total debt (t-1)		0.034**	0.032**	0.073***
		(0.014)	(0.014)	(0.022)
Secured Bank Debt / Bank debt (t-1)			−0.043*
			(0.024)
Long-term Bank Debt/Equity (t-1)				0.281
				(0.685)
Short-term Bank Debt/Equity (t-1)				−4.900***
				(0.635)
Observations	74,205	61,684	60,796	35,735
No. Firms	10,974	9,627	9,520	6,931
Rsquared	0.147	0.159	0.161	0.199

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1Explanatory variables not shown are the cashflow to capital ratio, the sales to capital ratio, and the size of the firm.

Table 8.

Firm Investment—Interest Coverage Ratio

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1Explanatory variables not shown are the cashflow to capital ratio, the sales to capital ratio, and the size of the firm.Notes: ICR is the interest coverage ratio (earnings to interest expenses); ICR < 1 is a dummy variable equal to one if the ICR is less than one; ICR < 1 for past five years is a dummy variable equal to one if a firm’s ICR has been last than 1 in each of the past five years.

Table 8.

Firm Investment—Interest Coverage Ratio

Dependent variable: Investment/K
	(1)	(2)	(3)
	FE	FE	FE
Debt/Equity (t-1)	−1.005***	−0.775***	−0.984***
	(0.147)	(0.148)	(0.148)
ICR (t-1)	0.131***
	(0.022)
ICR < 1 in previous year		−7.898***
		(0.588)
ICR < 1 for past five years			−4.884***
			(0.754)
Observations	65,104	65,104	65,104
No. Firms	10,069	10,069	10,069
Rsquared	0.154	0.157	0.154

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1Explanatory variables not shown are the cashflow to capital ratio, the sales to capital ratio, and the size of the firm.Notes: ICR is the interest coverage ratio (earnings to interest expenses); ICR < 1 is a dummy variable equal to one if the ICR is less than one; ICR < 1 for past five years is a dummy variable equal to one if a firm’s ICR has been last than 1 in each of the past five years.

View Table

Table 8.

Firm Investment—Interest Coverage Ratio

Dependent variable: Investment/K
	(1)	(2)	(3)
	FE	FE	FE
Debt/Equity (t-1)	−1.005***	−0.775***	−0.984***
	(0.147)	(0.148)	(0.148)
ICR (t-1)	0.131***
	(0.022)
ICR < 1 in previous year		−7.898***
		(0.588)
ICR < 1 for past five years			−4.884***
			(0.754)
Observations	65,104	65,104	65,104
No. Firms	10,069	10,069	10,069
Rsquared	0.154	0.157	0.154

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1Explanatory variables not shown are the cashflow to capital ratio, the sales to capital ratio, and the size of the firm.Notes: ICR is the interest coverage ratio (earnings to interest expenses); ICR < 1 is a dummy variable equal to one if the ICR is less than one; ICR < 1 for past five years is a dummy variable equal to one if a firm’s ICR has been last than 1 in each of the past five years.

A. Benchmark Specification, Firm-Level Investment Rates

The regression results confirm the expected negative impact of leverage on investment. Specifically, a 1 percentage point increase in company’s debt to equity ratio leads to about a 1 percentage point decline in investment as a share of its capital stock (Table 4). The sales to capital ratio is found to have a statistically significant, positive effect on investment, while the cash flow to capital stock ratio is found to have a positive effect on investment only in some specifications. Larger firms have lower investment rates, which is expected as large firms would be finished with their main expansion or growth phases.

B. Growth Opportunities and External Demand

Table 5 shows the effect of changes in export demand on Indian firms’ investment rates. In column (1a), we see that exporting firms’ investment rates were 1.9 percent higher than other firms on average through the sample period. Column (1b) includes an interaction term between the indicator variable that identifies exporting firms and a post-GFC dummy, which is equal to one from 2009 onwards. The coefficient on the interaction term is negative, at −0.228, but not statistically significant, suggest no significant change in investment at exporting firms after the crisis. Columns (2a) includes a measure of firms’ export intensity, the export to sales ratio, and confirms that exporting firms invest more on average, and column (2b) interacts the export intensity with the post-GFC dummy and we again find a negative but statistically insignificant coefficient on the interaction term.

In columns (3) to (4b), we include a measure of external demand – the growth in the volume of world imports at the industry level. The coefficient on WorldImportGrowth in column (3) indicates that that investment is not significantly different at firms in industries with higher external demand. In column (4a), we include an interaction term between WorldImportGrowth and firm’s export intensity and find, as expected, that exporting firms, however, do have higher investment when external demand is higher. Next, in column (4b), the interaction between WorldImportGrowth and a firm’s export intensity is further interacted with the post-GFC dummy variable—to test whether the relationship between the investment rate of exporting firms and external demand changed after the crisis. The negative but statistically insignificant coefficient of −0.032 on this interaction terms indicated that it did not. Overall, the results indicate that, while investment at exporting firms is affected by external demand, the responsiveness of investment to external demand neither increased, as might be expected, or decreased as trade collapsed after the crisis.

C. Financial Frictions

Table 6 shows the results from interacting leverage with three measures of financial constraints: firm age (column 1), whether a firm is a member of a business group (column 2), and the Rajan and Zingales industry-level measure of dependence on external finance (columns 3 and 4). In each case, the relationship between investment and leverage is less negative for firms with lower financial constraints. For example, in column (1), the negative effect of a one percentage point higher leverage ratio on investment for older firms is 0.012 percentage points lower than for the whole sample of firms. The effect of being in a business group is found to be important, with the relationship between investment and leverage being less negative by almost one percentage point for firms belonging to business groups.

In Table 7, we see that firms that are more dependent on bank finance in fact have higher investment than other firms in India. Specifically, firms with a 1 percentage point higher share of bank debt to total debt have marginally higher investment rates, by about 0.034 percentage points (column 2). Interestingly, this positive effect is offset for firms where the bank debt is in the form of secured debt (column 3). That is, when firms are subject to the loss of collateral, firms that have borrowed from banks do not have higher investment rates. Column (4) separates bank debt into its short-term and long-term components,¹⁰ and the estimated coefficients indicate that the negative effect on investment comes from the short-term component. Unlike the underinvestment channel in Diamond and He (2014), where shorter-term debt typically imposes lower overhang, these results suggest that a large repayment burden over the coming year is what prevents near-term investment.

The relationship between firm interest coverage ratios and investment provides further support for the notion of investment being dampened by high financing burdens (Table 8). Firms with one percentage point higher ICRs (greater earnings relative to interest expenses) have higher investment rates, by about 0.13 percent points on average (column 1). Columns (2) and (3) show that firms with ICR<1 in the previous year, or firms that have ICR<1 for five consecutive years, have lower investment rates, by about 5 percentage points.

D. New Investment Variable and Investment Project Length

Turning to the new investment variable (Table 9), we see that leverage has a negative effect on the likelihood that a firm decides to undertake new investment projects (column 1a). Column (1b) shows the corresponding marginal effect: a 1 percentage point increase in the leverage ratio (from the sample mean) leads to a 3.3 percent decline in the probability of undertaking a new investment project. This is a sizeable effect given that, overall, new investment projects are initiated in 9 percent of the firm-year sample. Cash flow and sales do not have economically or statistically significant effects on the probably of undertaking new investment projects. The coefficient on size is positive, unlike in specifications with investment rates as the dependent variable. While large firms would be expected to have lower expenditure on investment relative to their existing capital stock, since they are done with their main expansion or growth phases, this result suggests that large firms are still likely continue to undertake new investments.

Table 9.

New Investment Projects

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1.Notes: NewProject= 1if a firm announces it is beginning new project in year t.

Table 9.

New Investment Projects

Dependent variable: Prob{New Project}
	(1a)	(1b)
	FE Logit	Marginal Effects
Debt/Equity	−0.165***	−0.033***
	(0.011)	(0.002)
Cashflow/K (t-1)	−0.000	−0.000
	(0.000)	(0.000)
Sales/K (t-1)	0.001***	0.000***
	(0.000)	(0.000)
Size (t-1)	0.264***	0.053***
	(0.021)	(0.005)
Observations	46,884
No. Firms	4269
Pseudo Rsquared	0.0815

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1.Notes: NewProject= 1if a firm announces it is beginning new project in year t.

View Table

Table 9.

New Investment Projects

Dependent variable: Prob{New Project}
	(1a)	(1b)
	FE Logit	Marginal Effects
Debt/Equity	−0.165***	−0.033***
	(0.011)	(0.002)
Cashflow/K (t-1)	−0.000	−0.000
	(0.000)	(0.000)
Sales/K (t-1)	0.001***	0.000***
	(0.000)	(0.000)
Size (t-1)	0.264***	0.053***
	(0.021)	(0.005)
Observations	46,884
No. Firms	4269
Pseudo Rsquared	0.0815

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1.Notes: NewProject= 1if a firm announces it is beginning new project in year t.

Similarly, the probability of completing an investment project, once started, is lower for firms with higher leverage (Table 10). The estimated marginal effect in column (1b) shows that a 1 percentage point increase in the leverage ratio (from the sample mean) leads to an almost 1 percent decline in the probability of completing an investment project.

Table 10.

Investment Project Completion

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1.Notes: ProjectCompletion = 1if the firm completes its investment project in year t.

Table 10.

Investment Project Completion

Dependent variable: Prob{Project Completion}
	(1a)	(1b)
	FE Logit	Marginal Effects
Debt/Equity	−0.046*	−0.008*
	(0.026)	(0.004)
Cashflow/K (t-1)	−0.000	−0.000
	(0.000)	(0.000)
Sales/K (t-1)	−0.000	−0.000
	(0.000)	(0.000)
Size (t-1)	−0.118**	−0.020***
	(0.052)	(0.008)
Observations	11,378
No. Firms	3648
Pseudo Rsquared	0.108

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1.Notes: ProjectCompletion = 1if the firm completes its investment project in year t.

View Table

Table 10.

Investment Project Completion

Dependent variable: Prob{Project Completion}
	(1a)	(1b)
	FE Logit	Marginal Effects
Debt/Equity	−0.046*	−0.008*
	(0.026)	(0.004)
Cashflow/K (t-1)	−0.000	−0.000
	(0.000)	(0.000)
Sales/K (t-1)	−0.000	−0.000
	(0.000)	(0.000)
Size (t-1)	−0.118**	−0.020***
	(0.052)	(0.008)
Observations	11,378
No. Firms	3648
Pseudo Rsquared	0.108

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1.Notes: ProjectCompletion = 1if the firm completes its investment project in year t.

Finally, Table 11 shows the results of estimating the relationship between leverage and the length of an investment project. The coefficient on leverage of −0.02 indicates that firms with 1 percentage point higher leverage are associated with projects that are one year shorter on average in duration.

Table 11.

Investment Project Length

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1.Notes: Project Length is the length of an investment project in years, from beginning to completion.

Table 11.

Investment Project Length

Dependent variable: Project Length
	(1)	(2)
	OLS	OLS
	Industry Effects	Fixed Effects
Debt/Equity	−0.012*	−0.020**
	(0.006)	(0.010)
Cashflow/K (t-1)	−0.000	−0.000
	(0.000)	(0.000)
Sales/K (t-1)	0.000***	0.015
	(0.000)	(0.027)
Size (t-1)	0.045***	0.000
	(0.005)	(0.000)
Observations	11,170	11,170
No. Firms	3815
Pseudo Rsquared	0.208	0.020

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1.Notes: Project Length is the length of an investment project in years, from beginning to completion.

View Table

Table 11.

Investment Project Length

Dependent variable: Project Length
	(1)	(2)
	OLS	OLS
	Industry Effects	Fixed Effects
Debt/Equity	−0.012*	−0.020**
	(0.006)	(0.010)
Cashflow/K (t-1)	−0.000	−0.000
	(0.000)	(0.000)
Sales/K (t-1)	0.000***	0.015
	(0.000)	(0.027)
Size (t-1)	0.045***	0.000
	(0.005)	(0.000)
Observations	11,170	11,170
No. Firms	3815
Pseudo Rsquared	0.208	0.020

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1.Notes: Project Length is the length of an investment project in years, from beginning to completion.