The impact of R&D expenditure on capital structure of high-tech firms

The Impact of R&D Expenditure on

Capital Structure of High-tech Firms

Abstract This paper investigates the impact of R&D expenditure on leverage level of high-tech firms. A regression model is constructed to study the above impact, and also the difference between the impact of R&D expenditure on computer and software industry and the impact of R&D expenditure on medical technology and surgical equipment industry. Results show, for high-tech firms, R&D expenditure is negatively correlated to leverage level, and there is no significant difference between the two industries. Moreover, the change in leverage level is not significantly affected by the change in R&D expenditure. Key words: leverage level, R&D expenditure, high-tech firms. Name: Jiawen Zhang Student Number: 10761160 Program: Economics and Business Track: Finance and Organization Supervisor: Rola-Janicka Magdalena Date: 30/01/2017

Statement of Originality

This document is written by Student Jiawen Zhang who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

1. Introduction

The choice of research and development (R&D) investments is crucial to a firm. Schilling and Hill (1998) indicate that firms need to continuously invest into R&D to achieve competitive advantages. In order to be competitive and successful in today’s unpredicted and dynamic market, R&D investments are especially important for firms competing on the basis of innovation (O’Brien, 2003; Kor, 2006). However, R&D projects are risky investments as Baysinger et al. (1991) stated. Because of the nature of R&D, several problems may arise. First, R&D projects are hard to monitor and their results are difficult to forecast, which can contribute to conflicts of interest between shareholders and debt holders. Shareholders may want to over-invest in risky R&D projects which do not generate positive NPVs (Jensen and Meckling, 1976; Garvey and Mawani, 2005), while debt holders would require additional risk premium to compensate additional risking taking of underinvestment (Bah and Dumontier, 2001; Ho et al., 2006). Second, managers commonly have more details about R&D projects than debt holders. Due to asymmetric information problems, debt holders will overestimate the risk and require higher risk premium (Singh and Faircloth, 2005). Third, R&D projects create intangible assets, which cannot generally be used as collaterals for debt financing, thus limiting a firm’s ability to access external funds (Long and Malitz, 1985). Above problems increase a firm’s governance costs and increase the transaction costs, thus imply that firms with significant R&D expenditure prefer equity financing to debt financing. Moreover, if a firm is competing on the basis of innovation, a capital structure that can continuously support R&D projects is necessary. O’Brien (2003) states that firms with considerable R&D opportunities are advised to maintain financial slack, to protect R&D projects in volatile markets. Consequently, with considerable R&D expenditure, firms tend to finance R&D projects with equity, rather than debt. Thus, high R&D expenditure may induce low leverage level within a firm. This paper intends to investigate the impact of R&D expenditure on leverage level of high-tech firms. How does R&D expenditure impact on capital structure of high-tech firms? Is the influence of R&D expenditure on software and computer companies different from that of medical technology and surgical instrument companies? Numerous studies find a negative impact of R&D expenditure on leverage ratio, and high-tech firms have lower leverage level (e.g. Bradley at el., 1984; Titman and Wessels, 1988; Hovakimain, 2001). For

high-tech firms, R&D expenditure is expected to be negatively correlated to leverage level. What’s more, even though computer and software industry and medical technology and surgical equipment industry both belong to high-tech industries, the two industrial structures may be different. Computer and software firms may grow faster, since they develop new products and accelerate the upgrade of products to maintain their competiveness in short period. They may make more profits and face lower possibility of future financial constraint. Whereas, medical technology and surgical equipment industry may focus on breakthroughs on advanced medical technology and precise surgical equipment, which provide people with better health care. However, achieving breakthroughs takes a lot of time and requires a great deal of R&D expenditure. Medical technology and surgical equipment firms may not make profits as fast as computer and software firms. Thus, comparing to computer and software firms, medical technology and surgical equipment firms may have more need for financial slack. The impact of R&D expenditure on leverage level is expected to be different between the two industries. In order to investigate the impact of R&D expenditure on leverage level a regression model is constructed. The R&D ratio is used as main independent variable, and leverage ratio is used as dependent variable. The model includes a dummy industry variable to distinguish different industries, and an interaction variable to measure the simultaneous effect of R&D expenditure and industry. Three control variables (firm size, market-to-book ratio, and profitability) are included to avoid omitted variable bias. The negative effect is first examined using data from 2014 and 2015. Next, the effect of a change of R&D expenditure on the change of leverage level is measured. Results show that there is a negative effect of R&D expenditure on leverage level. There is no significant difference in the impact of R&D expenditure between the two industries. Finally, a change of R&D expenditure results in insignificant impact on the change of leverage level. Results in empirical study make contributions to the existing literature. Previous studies indicate the negative effect of R&D expenditure on leverage level, this empirical study is an expansion which not only find a negative effect of R&D expenditure on leverage level, but also find that this effect is insignificant among the two high-tech industries. The effect of a change in R&D expenditure is insignificant on the change in leverage level.

This paper is organized as follows. Section 2 presents related previous studies. Section 3 shows hypothesis. Section 4 describes the data and explains the econometric model. Section 5 discusses empirical results. Section 6 gives the discussion and conclusion.

2. Theoretical Background

Capital Structure The “irrelevance” proposition proposed by Modigliani and Miller (1958) implies that under a perfect capital market (no transaction cost, no income taxes and no bankruptcy costs), the value of a firm is independent of its capital structure in equilibrium. By taking corporate income tax into account, interest expense reduces the amount of corporate taxes, and thus, increases firm value, which provides an incentive to use debt financing. However, the costs of debt financing may offset the tax benefits in the real world. If the firm has difficulty covering its debt outstanding, the firm is in financial distress, and may go bankrupt in the nearly future. Moreover, bankruptcy and financial distress are not costless. The firm needs to hire outside professionals to assist in the handling of the firm. It also faces indirect costs associated with loss of customers, suppliers and employees and fire sale of assets. In addition, the costs of debt financing may also arise from agency problems. For example, the firm has excessive existing amount of debt which causes the firm under-invest in positive NPV projects. In addition, the pecking order theory proposed by Myers and Majluf (1984) implies that managers prefer internal financing (retained earnings) to external financing, and prefer debt to equity when retained earnings are not sufficient. They state that under the circumstance of asymmetric information between outside investors and managers, outside investors do not know the actual type of the firm, and invest as their expectations. Thus, outside investors may overestimate the investment risk and pay less. If the firm uses external funds to finance its new projects, it will suffer a decline in its value. So issuing new shares is not a good choice, and if the firm has internal surplus, it should first finance new projects use retained earnings. But if the firm’s only choice is to rely on external funds, the firm should issue bonds which are under symmetric information, which will not reduce the firm’s value. In this situation, debt financing takes priority over equity financing.

R&D intensity and capital structure The choice of R&D investments can influence a firm’s capital structure. Studies have noted significant differences in leverage across sectors, that regulated utilities and real estate firms tend to have particularly high debt level, but technology-oriented industries tend to maintain very low debt level (Person and Titman, 2009). Several empirical researches propose that R&D-intensive firms prefer equity financing and have lower leverage (e.g. Bradley et al., 1984; Titman and Wessels, 1988). Hovakimain (2001) also supports the negative relationship between R&D intensity and leverage. Furthermore, several studies find that firms with considerable R&D opportunities prefer equity financing to debt financing (e.g. Myers, 1977; Jensen, 1986; Balakrishnan and Fox, 1993; Ho et al., 2004). Kale and Sahrur (2007) propose four aspects of reason to explain the negative relationship between R&D expenditure and leverage level, there are agency cost problems, asymmetric information problems, asset specificity and financial slack. 1. Agency cost problems If the firm finance R&D projects use debt financing, conflict of interests between debt holders and firms incurs agency costs and thus makes equity financing more attractive. On the one hand, because of substantial amount of debt, the firm will under-invest in positive NPV R&D projects (Jensen and Meckling, 1976; Myers, 1977). Since the firm pays back debt holders when it makes profits, and underinvestment means the amount of repayment to debt holders is lower than the amount of repayment if the firm invests sufficiently. On the other hand, due to protection by limited liability, shareholders would like to over-invest in risky R&D projects which do not generate positive NPV but with large payoffs in good states (Frank and Goyal, 2007; Jensen and Meckling, 1976; Garvey and Mawani, 2005). Shareholders make themselves better-off, but decrease the firm’s value. Debt holders are worse-off with facing the risk of firm’s bankruptcy and no repayment. Thus, debt holders will not invest or invest a small amount of investment at first, which makes equity financing more attractive. What’s more, R&D projects are difficult to monitor and the results are hard to forecast. Patience and substantial amount of R&D investments is necessary, since no one knows when will get results. Specialized knowledge is required in order to assess the future

profitability and progress of these R&D projects. Therefore, impatient managers may be stressed with substantial amounts of debt and under-invest in ongoing R&D projects, shareholders may be selfish and over-invest in risky R&D projects without carefully assessing their future profitability. Debt holders would require higher risk premium to compensate their excessive risk taking, which increases the costs of debt financing (Bah and Dumontier, 2001; Ho et al., 2006). As a consequence, agency cost problems make debt financing costlier, and firms may use equity financing to fund R&D projects and have lower debt levels. 2. Asymmetric information problems Because of asymmetric information between debt holders and firms, firms with high R&D expenditure tend to maintain low debt level. Singh and Faircloth (2005) argue that debt holders may overestimate the risk of R&D projects and require a premium for bearing the risk, and thus increase the costs of debt financing. Managers generally get hold of much more details of R&D projects than creditors, and it is hard to evaluate these innovative projects accurately without professional knowledge. Because of highly uncertain outcomes of R&D projects, debt holders will require extra compensation for bearing excessive risk, which rises transaction costs (Ou and Haynes, 2006). 3. Asset specificity The characteristics of assets affect a firm’s leverage. Firms with a large stock of specific and intangible assets are unable to borrow as much as others, and consequently choose equity financing. Liu and Wong (2009) state that if a firm has a substantial amount of tangible assets, it is easier for the firm to access external funds and liquidate the assets. Kochhar (1996) also argues that firms with less intangible assets have lower transaction costs and generally prefer debt financing. R&D investment is risky and creates intangible assets, which cannot commonly be used as collaterals for debt financing (Long and Malitz, 1985). Therefore, intangible assets limit a firm’s capable of borrowing. Firms with plenty of R&D projects may use equity funds. 4. Financial Slack Firms with innovation as core competitiveness may choose capital structures that provide sufficient financial slack. In order to be competitive in the market, a continuous rate

of R&D investment is crucial for firms competing on the basis of innovation (O’Brien, 2003; Kor, 2006). Since firms with intangible assets are more likely to face financial constraint, O’Brien (2003) states that financial slack can provide protection against future cash flow fluctuation, and ensure firms have sufficient financial resources to create new products and support firms’ expansion. This suggests, that innovative firms that want to invest in R&Ds’ will choose a capital structure that provides them financial slack. Thus, such firms can be expected to have low leverage levels.

3. Literature review

The pecking order theory proposes that managers prefer internal financing (retained earnings) to external financing (Myers and Majluf, 1984), which implies firms tend to use equity financing and have low leverage levels. Numerous empirical studies find the negative relationship between R&D expenditure and leverage level (e.g. Bradley et al., 1984; Titman and Wessels, 1988; Hovakimain, 2001). Thus, the negative correlation between R&D expenditure and leverage level is expected in high-tech firms. Hypothesis 1: There is a negative relationship between R&D expenditure and leverage level in high-tech firms. As discussed in previous section, O’Brien (2003) proposes that financial slack can provide protection against future cash flow fluctuation, and ensure firms have sufficient financial resources to create new products and support firms’ expansion. High-tech firms with substantial amounts of intangible assets are expected to maintain financial slack, and these firms tend to have lower possibility to finance its R&D projects with leverage. Thus, for high-tech firms, a change in R&D expenditure is expected to have no impact on the change in leverage level. Hypothesis 2: There is no impact of a change in R&D expenditure on the change in leverage level in high-tech firms.

Although computer and software industry and medical technology and surgical equipment industry both belong to high-tech industries, medical technology and surgical equipment industry is expected to have more need for financial slack. Medical technology and surgical equipment industry may focus more on breakthroughs on R&Ds’, in order to provide people with more advanced medical technology and better health care. However, it takes a lot of time and cost substantial amounts of investment to achieve successful outcomes. These firms may be more likely to face financial constraint if there is no successful result after substantial amounts of investment. On the contrary, computer and software firms may be less likely to face financial constraint. These firms tend to accelerate the upgrade of products to maintain their competiveness in short period, profits are expected to generate and cover R&D expenditure. With financial slack, R&D expenditure may less likely to impact on leverage level in computer and software firms. Therefore, the impact of R&D expenditure on leverage level in computer and software industry is expected to be different from the impact of R&D expenditure on leverage level in medical technology and surgical equipment industry. Hypothesis 3: The impact of R&D expenditure on leverage level in computer and software industry is different from the impact of R&D expenditure on leverage level in medical technology and surgical equipment industry.

4. Econometric Model Setup

Based on literature studies, the following regression is estimated: !" = $% + $'(&*"+ $+,-./012!" + $3(&*"×,-./012!" + $56728 :7;<"+ $=>?2@<1/ BCC@_" + $_DE2CF71?G7H71!_" + I_" . Leverage ratio is the depend variable in this estimation, one of main independent variables is R&D ratio. Since Person and Titman (2009) state that industry effect is one of determinants of leverage level, a dummy variable industry is used to distinguish between medical technology and surgical equipment industry and computer and software industry. An interaction term is included to measure the effect of simultaneous influence of R&D ratio and Industry.

In order to avoid the estimated effect of interest to be influenced by omitted variable bias, this model includes three control variables: firm size, market-to-book ratio, and profitability. First, the positive relationship between firm size and leverage ratio is verified by numerous studies (e.g. Graham et al., 1998). Smaller firms are short of diversification, which limit the ability to minimize unsystematic risk. Therefore, they are more likely to suffer strict financial distress costs and go bankruptcy (Bhagat and Welch, 1995). In order to maintain R&D investments, smaller firms prefer not to borrow large amounts of debt. Therefore, firm size influences leverage ratio, and a positive correlation is expected. Second, the market-to-book ratio is the most commonly used proxy for growth opportunities (Frank and Goyal, 2009). The fact that firms with substantial growth opportunities prefer equity financing is proposed by several studies (e.g. Myers, 1977, Jensen, 1986). Hence, M/B ratio is expected to have negative relationship with leverage ratio. Third, numerous studies find the negative correlation between profitability and leverage ratio (e.g. Titman and Wessels, 1988, Rajan and Zingales, 1995). When a firm is profitable, paying out profits through dividends or share repurchase to shareholders incurs taxes, while repaying debt does not. Thus, firms tend to use less debt financing when they are profitable, and a negative correlation is expected. On the basis of literature studies, the data source is Compustat files. The sample will focus on public firms that are actively traded on the U.S. stock market. Observations will be selected by SIC industry codes. According to Liu and Wong (2009), any firms with SIC industry codes falling into 2833-2836 (medical technology), 3661-3669 (communications), 3671-3679 (semi-conductors), 3841-3845 (surgical equipment), 3570-3579 (computer equipment) and 7370-7377 (software)have intensive intellectual capital and are classified as high- tech firms. In order to investigate whether R&D impact on capital structure is different between high-tech industries, the sample will include firms in medical technology and surgical equipment industry (2833-2836, 3841-3845), and computer and software industry (3570-3579, 7370-7377). Total 2986 observations from 2014 and 2015 are collected. Any firms with missing data will be dropped. Extreme variable values will also be dropped to mitigate the impact of outliers. After screening, 880 observations for year 2014, and 930 observations for year 2015 are included. The impact of R&D on leverage will be estimate through regressions for each year. To measure the effect of a R&D change, panel data is used with 827 firms and two-year span.

5. Results

In order to examine the impact of R&D on leverage ratio, table 1 presents regression results. There are four regression models. Model 1 investigates the impact of R&D expenditure on leverage in 2014, and shows moderate explanatory power (58.42%). For R&D ratio and interaction variable (R&D ratio * Industry), two coefficients are both negative but insignificant (-0.000301 and -0.000361, respectively). The coefficient of dummy industry (0.216) is positive but insignificant. Thus, three independent variables show no impact on leverage level. As for control variables, results of firm size and M/B ratio are insignificant, while the coefficient of profitability is negative (-0.731) and statistically significant at 1%. Further, model 3 estimates the effect on the basis of data from 2015. Model 3 shows similar moderate explanatory power at 57.36%. The coefficient of main independent variable R&D ratio is negative (-0.00125) and significant at 10%. The coefficient of the dummy variable and the interaction variable is 0.0811 and -0.000145, respectively. Both are insignificant in this regression. The coefficient of profitability is negative (-0.506) and statistically significant at 5%, which is in line with literature. The other two control variables are insignificant. Since the dummy variable Industry and the interaction variable are estimated to be irrelevant to leverage level, model 2 and 4 are constructed to investigate the impact of R&D expenditure on leverage level without the dummy variable and the interaction variable. Model 2 generates the impact in 2014. The coefficient of R&D expenditure (-0.0003789) shows a negatively significant correlation between R&D expenditure and leverage level, while it is insignificant under model 1. Model 4 is based on data for 2015. A negative and significant coefficient (-0.00131) of R&D expenditure shows a stronger correlation between R&D expenditure and leverage level, comparing to model 3. According to above results, the coefficient of main independent variable R&D ratio is negative and significant in model 2,3 and 4, indicating R&D expenditures of a high-tech firm negatively influences leverage level, which is consistent with hypothesis 1. Further, the insignificant dummy variable industry and interaction variable in model 1 and 3 show that a high-tech firm’s leverage level is irrelevant to which industries it belongs to. After removing the dummy variable and the interaction variable, R&D expenditure shows a stronger impact on leverage level. In addition, the insignificant dummy variable industry shows that there is no significant difference of impact between the two industries. Hypothesis 3 is thus not

confirmed. Only the impact of profitability on leverage level is in line with literature, the impact of firm size and the impact of M/B ratio on leverage level are insignificant in these four regression models. Table 2 presents two regression models to examine whether a change in R&D expenditure affects the change in leverage level from 2014 to 2015. Regression 5 shows good explanatory power (90.5%). As for R&D ratio, dummy Industry, and interaction variable, these three coefficient estimates are all negative but insignificant, with -0.0000378, -0.0788, and -0.000184, respectively. The coefficient of a change of firm size is positive (0.318) and significant at 5%. The coefficient of a change of profitability (-1.012) shows a negative impact on leverage level, and the result is statistically significant at 1%. Conversely, the coefficient of M/B ratio is negative but insignificant. Regression 6 is adjusted without the dummy variable and the interaction variable on the basis of regression 5. The coefficient of the change in R&D expenditure (-0.0000834) is negative but insignificant. The change in firm size is positively significant related to the change in leverage level. The change in profitability is negatively significant related to the change in leverage level. The coefficient of M/B ratio is insignificant, conversely. The results from regression 5 and 6 show no big difference, two regression models both show insignificant correlations between the change in R&D expenditure and the change in leverage level. The analysis indicates that the change in R&D expenditure doesn’t have a significant impact on the change in the level of leverage for high-tech firms, which is consistent with hypothesis 2. High-tech firms with substantial intangible assets are suggested to maintain financial slack to safeguard firms’ R&D projects. These firms may desire a stable capital structure which can provide them financial slack, and they finance R&D projects without leverage. Thus, the change in R&D expenditure insignificantly affect the change in leverage level. Above suggests that the level of R&D expenditure matters more for capital structure than the change of R&D expenditure. As for two control variables, the positive impact of firm size difference and the negative impact of profitability difference on the change of leverage level are verified to be consistent with literature. The change in M/B ratio shows insignificant relationship with the change in leverage level in this regression.

To sum up, hypothesis 1 is partially supported by above results that the impact of R&D expenditure on leverage level is negative. Outcomes are consistent with hypothesis 2 that the change in leverage level is not affected by the change in R&D expenditure. Hypothesis 3 is not verified by above empirical results. The impact of R&D expenditure on leverage level in computer and software industry shows insignificant difference from the impact of R&D expenditure on leverage level in medical technology and surgical equipment industry.

6. Discussion and Conclusion

On the basis of previous studies, a firm’s R&D expenditure influences its leverage level, and a negative impact is discovered (e.g. Bradley et al., 1984; Titman and Wessels, 1988). Thus, this paper investigates how R&D expenditure impacts the capital structure of high-tech firms in the U.S. stock markets, and whether the influence in software and computer industries is different from the influence in medical technology and surgical equipment industries. Two regression models with data collected from 2014 and 2015 are firstly constructed to identify the effect of R&D expenditure on leverage level. Results show that there is a negative effect, and this negative effect is independent with respect to industry. After removing the dummy variable and the interaction variable, the negative effect is more significant. Besides, there is a negative correlation between profitability and leverage level, which is supported by literature. The fifth regression model is created to examine the impact of a change in R&D expenditure on the change in leverage level. Results indicate that the change in R&D expenditure is insignificant related to the change in leverage level. In addition, results show the positive relationship between firm size and leverage level, and the negative correlation between profitability and leverage level, which are consistent with literature. The impact of M/B ratio on leverage ratio is expected to be negative under previous studies (e.g. Myers, 1977; Jensen, 1986), but an insignificant impact is shown in results. According to Frank and Goyal (2009), the M/B ratio is the most commonly used proxy for growth opportunities, which means growth opportunity is insignificant related to leverage level for high-tech firms in this case.

The negative correlation between R&D expenditure and leverage level shown in results illustrates that the decision of R&D expenditure negatively influences a firm’s capital structure. Since the choice of R&D investments is a determinate factor of a firm’s core competitiveness, especially for those technology-based firms, firms will make large amounts of investment to create and maintain their core competitiveness. However, the outcomes of R&D projects are highly uncertain and cannot be used as collaterals. Debt financing is less likely to be used for those technology-based firms. The insignificant impact of a change in R&D expenditure on the change in leverage level is consistent to expectation. O’B rien (2003) proposes financial slack provide protection for R&D projects. High-tech firms with sufficient intangible assets are more likely to face financial constraint. Since intangible assets cannot be used as collaterals, financial slack ensures R&D projects operating as planned. Firms tend to maintain a capital structure that provides them financial slack, and finance R&D projects without leverage. So the change in leverage level is not affected by the change in R&D expenditure. Nevertheless, results find the impact of R&D expenditure on leverage level is indifferent between computer and software industry and medial tech and surgical equipment industry, which is not comply with hypothesis. Although the industrial structures and products from two industries are entirely diverse, the need for financial slack may be different, they both belong to high-tech industries with large amounts of R&D investments. Thus, they may have same industry effect. The purpose of R&D projects is the same for two industries to develop advanced technology, attain core competitiveness and gain significant market shares. The impact of R&D investments on leverage level should then be consistent for computer and software industry and medical technology and surgical equipment industry. To summaries, this paper confirms the negative impact of R&D expenditure on leverage level found in previous studies. This paper also obtains that the impact is indifferent between high-tech industries. However, only less than 1000 observations and two recent fiscal years are examined, the small sample will omit many important details and limit the precision of results. Additionally, only high-tech firms within the U.S. stock market are selected, external validity cannot be guaranteed. Further, the regression model in this paper

is not well constructed. It faces endogenous variable problems. The interaction variable is included as an independent variable to exam whether there is a simultaneous influence of R&D expenditure and industry dummy, and to avoid omitted variable bias between independent variables. However, excluding the dummy variable and the interaction variable, R&D expenditure shows a more significant correlation with leverage level. The dummy variable and the interaction variable weaken the impact of R&D expenditure on leverage level, which is not consistent with expectation. As suggested by Person and Titman (2009), a great deal of determinants will influence leverage level, only firm size, M/B ratio, and profitability is not enough to avoid omitted variable bias. Only industry dummy variable is not sufficient to discovery industry effect, an industry median leverage ratio should also be included as Hovakimian et al. (2001) recommended. For further research, I suggest to focus on the degree of impact and the duration of impact, with observations worldwide. In order to ensure external validity, a big sample with observations worldwide is necessary, and further research is advised to focus not only on high-tech firms, but also on low-tech firms. Since this paper only investigate the negative impact of R&D expenditure on leverage level, and gives possible reasons for the negative correlation. Further research is recommended to figure out the reason why firms have low leverage. Firms with low leverage is due to financial slack or due to borrowing constraint.

Reference

Bah, R. and Dumontier, P., 2001. R&D intensity and corporate financial policy: Some international evidence. Journal of Business Finance and Accounting. 28. 672–692. Balakrishnan, S. and Fox, I., 1993. Asset specificity, firm heterogeneity and capital structure. Strategic Management Journal. 14. 3–16. Barney, J. B., 1991. Firm resources and sustained competitive advantage. Journal of Management. 17. 1:99–120. Baysinger, B. D., Kosnik, R. D. and Turk, T. A., 1991. Effects of board and ownership structure on corporate R&D strategy. Academy of Management Journal. 34. 205–214. Bhagat, S. and Welch, I., 1995. Corporate research & development investments: International comparison. Journal of Accounting and Economics. 19. 443–470. Bradley, M., Jarrell, G. A. and Kim, E. H., 1984. On the Existence of an Optimal Capital Structure: Theory and Evidence. Journal of Finance. 39. 857-878. Dittmar, A. and Thakor, A., 2007. Why do firms issue equity? Journal of Finance. 62. 1–54. Frank, M. Z. and Goyal, V. K., 2009. Capital structure decisions: Which factors are reliably important? Financial Management. 38. 1–37. Garvey, G. T. and Mawani, A., 2005. Risk-taking incentives of executive stock options and the asset substitution problem. Accounting and Finance. 45. 3–23. Graham, J. R., Lemmon, M. L. and Schallheim, J. S., 1998. Debt, leases, taxes, and the endogeneity of corporate tax status. Journal of Finance. 53. 1: 31–162.

Ho, Y. K., Xu, Z. and Yap, C. M., 2004. R&D investment and systematic risk. Accounting and Finance. 44. 393–418. Ho, Y. K., Tjahjapranata, M. and Yap, C. M., 2006. Size, leverage, concentration and R&D investment in generating growth opportunities. The Journal of Business. 79. 2: 851–876. Hovakimian, A., Opler, T. and Titman, S., 2001. The debt-equity choice. Journal of Financial and Quantitative Analysis. 36. 1: 1–24. Jensen, M. C. and Meckling, W. H., 1976. Theory of the firm: Managerial behavior, agency cost and ownership structure. Journal of Financial Economics. 3. 305–360. Jensen, M. C., 1986. Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers. American Economic Review. 76. 323-329. Kale, J. R. and Shahrur, H., 2007. Corporate capital structure and the characteristics of suppliers and customers. Journal of Financial Economics. 83. 321–365. Kochhar, R., 1996. Explaining firm capital structure: The role of agency theory vs. transaction cost economics. Strategic Management Journal. 17. 713–728. Kor, Y. Y., 2006. Direct and interaction effects of top management team and board compositions on R&D investment strategy. Strategic Management Journal. 27. 1081–1099. Liu, Q. and Wong, K. P., 2009. Intellectual Capital and Financing Decisions: Evidence from the U.S. Patent Data. SSRN Electronic Journal. ssrn.1358366. Long, M. and Malitz, I., 1985. The investment-financing nexus: Some empirical evidence. Midland Corporate Finance Journal. 3. 53–59. Modigliani, F. and Miller, M., 1958. The cost of capital, corporation finance and the theory of investment. American Economic Review. 48. 261–297.

Myers, S. C. and Majluf, N. S., 1984. Corporate financing and investment decisions when firms have information that investors do not have. Journal of financial Economics. 13. 187– 221. Myers, S. C., 1977. Determinants of corporate borrowing. Journal of Financial Economics. 5. 147-175. Myers, S. C., 1977. Determinants of corporate borrowing. Journal of Financial Economics. 13. 117–161. O’Brien, J. P., 2003. The capital structure implications of pursuing a strategy of innovation. Strategic Management Journal. 24. 5: 415–431. Ou, C. and Haynes, G. W., 2006. Acquisition of additional equity capital by small firms-findings from the National Survey of Small Business Finances. Small Business Economics. 27. 157–168. Parsons, C. and Titman, S., 2009. Empirical Capital Structure: A Review. Foundations and Trends in Finance. 3. 1: 1-93. Rajan, R. and Zingales, L., 1995. What do we know about capital structure? Some evidence from International data. Journal of Finance. 50. 5: 1421–1460. Schilling, M. A. and Hill, C. W. L., 1998. Managing the new product development process: Strategic imperatives. Academy of Management Executives. 12. 67–75. Singh, M. and Faircloth, S., 2005. The impact of corporate debt on long term investment and firm performance. Applied Economics. 37. 875–883. Titman, S. and Wessels, R. 1988. The Determinants of Capital Structure Choice. Journal of Finance. 43. 1-19.

Appendix

Variable List Variable Definition Dependent Variable Leverage Ratio Ratio of total debt to total assets Independent Variables R&D Ratio Ratio of R&D expenditure to total sales Industry 1, if the industry is computer and software industry; 0, if the industry is medical technology and surgical equipment industry Interaction Variable R&D Ratio × Industry Control Variables Firm Size Natural log of total assets Market-to-Book Ratio Ratio of market value of assets to book value of assets Profitability Ratio of earnings before income interest and taxes to total assets

Table 1 This table shows the regression results of leverage ratio on a R&D expenditure variable, a dummy industry variable, an interaction variable and three control variables. To determine the correlation between R&D expenditure and leverage level, data from recent two fiscal years, 2014 and 2015, are used. Regression 1 and 3 estimate the correlation with a dummy variable and an interaction variable for 2014 and 2015, respectively. Regression 2 and 4 estimate the correlation without the dummy variable and the interaction variable for 2014 and 2015, respectively. These two groups are used to investigate the impact of the dummy variable and the interaction variable on leverage level. Variables 2014 (1) 2014 (2) 2015 (3) 2015 (4) R&D ratio -0.000301 (0.000183) -0.0003789* (0.0001981) -0.00125* (0.000656) -0.00131** (0.000601) Industry 0.216 (0.198) 0.0811 (0.195) R&D ratio*Industry -0.000361 (0.000363) -0.000145 (0.000886) Firm Size -0.0524 (0.130) -0.051966 (0.130) -0.119 (0.145) -0.119 (0.145) M/B ratio 0.0000288 (0.0000461) 0.0000345 (0.0000484) 0.000544 (0.000346) 0.000523 (0.000333) Profitability -0.731*** (0.189) -0.732*** (0.189) -0.506** (0.231) -0.506** (0.231) Constant 0.248 (0.692) 0.3397762 (0.722195) 0.761 (0.783) 0.794 (0.832) Observations 880 880 930 930 R-squared 0.584 0.584 0.574 0.574 Robust standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1

Table 2 This table shows the effect of a change of R&D expenditure. Dependent variable is the change of leverage ratio from 2014 to 2015. The main independent variable R&D ratio and the interaction variable are also defined as the change from 2014 to 2015, as well as control variables. The difference between regression 5 and regression 6 is regression 5 use the dummy variable and the interaction variable, while regression 6 only use delta R&D ratio as independent variable. Variables (5) (6) del R&D ratio -0.0000378 (0.000452) -0.0000834 (0.000397) Industry -0.0788 (0.137) del R&D ratio*Industry -0.000184 (0.000948) del Firm Size 0.318** (0.148) 0.317** (0.148) del M/B ratio -0.0000607 (0.000104) -0.00005.84 (0.000103) del Profitability -1.012*** (0.0120) -1.012*** (0.0119) Constant -0.0143 (0.0905) -0.0480 (0.0688) Observations 827 827 R-squared 0.905 0.905 Robust standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1