The relationship between leverage and investment has been well researched in corporate finance literature. Under the neoclassical view as contained in the original Modigliani-Miller (1958) irrelevance propositions where the capital market is perfect and complete and external financing can be secured without incurring any extra costs, leverage and investment are considered to be unrelated. According to this school, a firm's investment policy should depend solely on the fundamental determinants of profitability, cash flow, and net worth. Where profitable investment opportunities are available, a firm can always secure funding regardless of the nature of its current financial status.
However, subsequent studies on capital structure have argued against this position and posit that financing considerations significantly complicate real investment decisions, introducing important determinants beyond neoclassical fundamentals. In the real world where there are missing or incomplete markets due to transaction costs, asymmetric information and agency problems (arising from interactions between shareholders, debt holders, and management), it make conditions of suboptimal investment decisions or overinvestment incentives possible, creating a strong relation between leverage and investment (Aivazian et al, 2005).
Empirical Literature on the Relation between Leverage and Investment
In his study, Myers (1977) for example, posits that capital structure plays an important role in firms' investment policies. He argued that as a result of agency problem, when firm have sufficiently high leverage, it reduces the alignments of managers and shareholders to control firm resources and be able to exploit available positive net present value investment opportunities. This compels them to make suboptimal investment decisions since the benefits at least partially accrue to bondholders. Therefore, high-leverage firms are less likely to take advantage of valuable growth opportunities compared to firms with low level of leverage, a situation referred to as underinvestment problem of debt financing and this has a negative effect on firm value.
Jensen (1986) studied the agency problem associated with financing decisions and the conflict between managers and shareholders. He argue that since managers are rewarded for expanding the scale of the firm, they therefore have the incentive and the opportunity (given their discretion to apply the available internal fund as they deem fit) to do so even if they have to undertake poor, negative net present value projects- overinvestment. He observe that when firms have more free cash flow than available positive net present value investment opportunities, the presence of debt in the firm's capital structure pre-commits managers to service interest and principal obligation and forces them to pay out funds that might otherwise have been invested in negative net present value projects. Leverage therefore may overcome overinvestment. This suggests that corporate value is negatively correlated with leverage for firms with strong growths options (high Tobin's q) and positively correlate with leverage for firms with low growth opportunities (low Tobin's q)
Stulz (1990), theoretically similar to Jensen (1986), predict a negative relation between leverage and investment emphasizing the beneficial effect of this effect for shareholders of low-growth firms. He argues that high leverage in low growth firms is used to discourage management from undertaking non-profitable investments. Here, debt pre-commits firms to pay cash as interest and principal and such commitments in low growth firms can reduce managerial discretion over free cash flows that may have otherwise been allocated to negative NPV projects. In other words, the banks and other debt-holders perform a beneficial monitoring and disciplinary role in low growth firms where a high level of debt can limit the overinvestment bias caused by managerial agency problems
Aivazian et al, (2005) examines the impact of financial leverage on the firms' investment decisions using information on Canadian publicly traded companies. Their result demonstrates that leverage is negatively related to investment and that this negative effect is significantly stronger for firms with low growth opportunities than those with high growth opportunities. The results also provide support to agency theories of leverage, and especially the theory that leverage has a disciplining role for firms with low growth opportunities
The work of McConnell and Servaes (1995) provide support for both the overinvestment and the underinvestment theories. They examine a large sample of US firms separating them into two groups of firms with strong growth opportunities and those with weak growth opportunities. Their result show that firm value is negatively correlated with leverage for firms with strong growth opportunities (indicated by high Tobin's q), and positively correlated with leverage for firms with weak growth opportunities (or low Tobin's q).
Lang et al (1996), in their related underinvestment theory centers on a liquidity effect and suggest that firms with large debt commitments invest less irrespective of the nature of their growth opportunities. Their study demonstrates a negative relation between leverage and future growth in a broad sample of firms. They report that negative relation between leverage and investment exists only for low q firms. This implies that leverage does not constrain investment in those firms in which the market recognizes profitable growth opportunities.
Firth et al, (2008) in their study investigate the relations among investment, leverage, growth, and performance for China's listed firms using data from 1991 to 2004. Their analysis yields results consistent with existing literature. The establish a negative relation between leverage and investment, consistent with the existence of a debt overhang problem even where banks are state-owned rather than privately owned, the negative relation between leverage and investment is weaker in firms with low growth opportunities and poor operating performance than in firms with high growth opportunities and good performance. This contradicts evidence obtained in the U.S. where low growth firms find it difficult to borrow money to finance expansion. Their findings also predict a negative relation between leverage and investment as weaker in firms with a higher level of state shareholding than in firms with a lower level of state shareholding. This is in line with the hypothesis that the state-owned banks in China impose fewer constraints on capital spending by low growth and poorly performing firms and firms with greater state ownership creating an overinvestment bias in these firms.
Data
Panel data set from COMPUSTAT industrial database merged with CRSP database containing financial/accounting profile (operating summaries, annual balance sheet and income statements, sources and uses of fund, growth rates, financial ratios summary stock data etc) and market information on publicly quoted US companies covering the period 1996 to 2007. Firm in the financial industry (with SIC code 6000-6999), utility firms and those in other regulated industries (with SIC code 4000-4999 and 9000-9999) will be excluded from the data.
Model Specification
I will estimate a reduced form investment equation to examine the impact of leverage on investment. The specification is similar to Aivazian et al., (2005) and Lang et al., (1996) but extended to a panel setting. Specifically, the following equation will be estimated:
Ii,t/Ki,t-1 = α + λt + β(CFi,t/Ki,t-1) + δQi,t-1 + ηLEVERAGEi,t-1 + ïª(SALEi,t-1/Ki,t-1) + μi + εi,t
Where Ii,t is the net investment of firm i at time t; Ki,t-1 is the lagged net fixed assets; CFi,t is cash flow of firm I at time t; Qi,t-1 is lagged Tobin's q; LEVERAGEi,t-1 is lagged leverage; SALEi,t-1 represents lagged net sales of firm I, α is the intercept; λt is a set of time dummy controlling for possible differences in the macroeconomics environment of each year; μi is the individual effect of firm I, and εi,t is the residual error term.
Variables
Investment: This is the acquisition of capital expenditure such as plant and machinery to create output, the higher the level of investment in a firm, the higher the expected growth based on such factors as the rate of interest, the availability of profitable opportunities and the general climate of business confidence. Specifically, I will investigate investment as the dependent variable and how it is impacted by the level of debt financing employed by firms. In this paper and in line with the model proposed in Aivazian et al. 2005, investment will be measured as the ratio of firm's net investment at time t to lagged net fixed assets. Net investment is measured as capital expenditures at time t less depreciation. This is consistent with the measure adopted by Lang et al. (2005) and Firth et al. (2008).
Leverage (Lagged): Two alternative measures used in the literature are adopted. As proposed in Aivazian et al. (2005), the first measure of leverage is the ratio of book value of total liabilities to the book value of total assets (taking the totality of short term and long term debt) while the second measure is the ratio of the book value of long-term debt to the total assets (emphasizing the central role of long-term debt as a determinant of investment). The measure is also consistent with that employed in Firth et al. (2008).
Tobin's Q (Lagged): This is defined as the ratio of the market value of total assets of the firm to the book value of assets and proxies for growth opportunities. Market value of the firm is calculated as the sum of total liabilities, the value of the common stocks, and the estimated value of preferred stocks. Adam and Goyal (2008) in the study on investment opportunities tested a number of measures of proxies of investment opportunities including the market-to-book equity ratio, the earnings-price ratio and the ratio of capital expenditures over the net book value of plant, property, and equipment. The results show that, on a relative scale, the market-to-book assets ratio, (or Tobin's q) has the highest information content with respect to investment opportunities. Although both the market-to-book equity and the earnings-price ratios are related to investment opportunities, they do not contain information that is not already included in the market-to-book assets ratio and does not improve on the performance of the market-to-book assets ratio. This measure was also adopted in Aivazian et al., (2005) and Firth et al., (2008)
Cash flow: This is measured as the sum of earnings before extraordinary items and depreciation deflated by lagged net fixed assets. This follow prior studies (Aivazian et al., 2005 and Firth et al., 2008) to control for firms' cash flow conditions and it is a measure of a firm's profitability.
Sale (Lagged): is defined as lagged net sales deflated by lagged net fixed assets. This is to control for sales condition and also measures the profitability of the firm.
Dummy Variables: These are included in the model to control for firm and time fixed effects, time dummy to capture the potential differences in the macroeconomic environment over time and firm dummy to control for endogeneity problem in the data and eliminate non-observed time-invariant firm effect. In Lang et al. (1996), unobservable individual effect is assumed to be zero and therefore used a pooling regression to estimate the investment equation. As noted by Aivazian et al., (2005) however, the assumption of zero unobservable individual effect may be too strong considering possibility of large heterogeneity across industries and across firms within the same industry. As modeled by Aivazian (2003), a random effect as well as a fixed effect will be used.
To identify the most suitable methodology in the options- pooling, random effect or fixed effects, two statistical tests will be performed. The Lagrangian Multiplier (LM) test for the random effect model and the Hausman specification test will compare the fixed effect and the random effect models.
There may be econometric problems which may affect the estimation of the fixed effect model. There is a possibility of high correlation among the independent variables, for instance, Tobin's q, which represents growth opportunities may affect the decision on leverage. An improvement in the firm's growth opportunities generated from increased demand or its marginal efficiency of investment may result in increased investment. This improvement may also affect other economic fundamental which may improve its market standing and access to external funding. In estimating the relationship between investment and financing, its important to identify meaningful measures correlated with either demand or supply. Tobin's q also represents firm value and may also be affected by leverage. High correlation among the variables may affect the efficiency of the estimated coefficients.
The variance of the error terms may also not be constant across firms causing Heteroscedasticity. This may be because the variance of error term correlating with firm size. Since the variables have been deflated by lagged net fixed assets, the problem has been alleviated; White's correction for heteroscedasticity will be used to obtain consistent standard errors.
Another econometrics problem to be avoided is the correlation of error terms across periods. Controlling for auto-correlation by assuming first order correlation where the errors are independently and identically distributed will resolve this problem.
High and Low Growth Opportunities and The Effect of leverage
To investigate the impact of leverage on firms with high growth opportunities versus low growth opportunities, the baseline model will be modified to include a dummy variable which will split firms into high and low growth categories using the following specification
Ii,t/Ki,t-1 = α + λt + β(CFi,t/Ki,t-1) + δQi,t-1 + ηLEVERAGEi,t-1 + ï§Di,t-1 x LEVERAGEi,t-1 + ïª(SALEi,t-1/Ki,t-1) + μi + εi,t
Where D is the dummy variable which equals to 1 if Tobin's q>1 and 0 if q<1
It is also important in the leverage-investment relationship to test for endogeneity problem in the model. While leverage may be affected by expected investment opportunities, and this has been controlled for by the inclusion of Tobin's q, it may not be adequate because Tobin's q only reflects public information while the firm's choice of leverage may reflect inside information. To deal with this endogeneity issue, a test for differences in the effect of leverage in the firm's core business and investment decisions in its non-core business is important since leverage in the firm's non-core business should be less affected by growth opportunities than that in its core business. To the extent if leverage proxies for investment opportunities, there should not be a strong relationship between the firm's leverage and its noncore investment. However, it is hard to classify core and non-core segments appropriately to ensure sufficiently different growth opportunities. There may also be sample bias since information may only cover a partial sample of firms and provide limited details about segment investments and assets.
We adopt an instrumental variable approach to deal with the endogeneity problem in respect of the relationship between leverage and investment. The instrumental variable for leverage to be employed is the proportion of the value of tangible assets to total assets.
Using tangibility as an instrumental variable is sensible because bankruptcy costs are an important determinant of the firm's leverage level and tangible assets tend to reduce bankruptcy costs and increase the use of leverage. Therefore, the tangibility of assets should be highly correlated with the firm's leverage level. Also, tangibility of assets is not highly correlated with the firm's investment opportunities. Tangibility of assets is measured by the value of property, plant, and equipment, plus the value of inventory divided by total assets.
Robustness Test
As a further test of the robustness of the estimates, industry effects will be controlled for and data will be restricted to manufacturing firms. There may be significant heterogeneity in investment behavior across different industries and hence it is important to control for industry effects. In the panel model, industry effects were subsumed under individual firm effects. To check the robustness, industry median will be used to adjust the variables in the investment equation. Each variable is adjusted by deducting the median of the industry to which the firm belongs; this allows for control of industry heterogeneity in terms of average scales.
