Corporate governance and research and development: Evidence from Japan

Abstract

This paper investigates the effects of the ownership structure on the R&D intensity. Using the Japanese machine-manufacturing firm data from 1987 till 1998, we first found that the effects of R&D on stock market valuation and TFP growth were significantly positive in the latter half of the 1990s. Next, analyzing the determinants of the R&D intensity in 1998, we found that the shareholding ratios of large shareholders and the leverage ratios were positively correlated with R&D intensity, while the proportion of bank loans to total debt was negatively correlated with it. These results are consistent with the hypotheses that stress the disciplinary roles of large shareholders and debt. It is also consistent with a bank's holdup hypothesis. Finally, comparing the results of 1998 with those of 1989, we found that the positive roles of keiretsu affiliation and cross-shareholdings disappeared during the last decade.

Keywords:

• R&D
• Corporate governance
• Ownership structure
• Japan

Joji Tokui kindly provided us with the R&D expenditures data on diskettes. Tsutomu Watanabe also generously provided us with the capital stock and Tobin's q data that he had constructed with Kaoru Hosono. Comments by two anonymous referees, Shin'ichi Hirota, Hiroyuki Odagiri and other seminar participants at Japan Economic Association, Hitotsubashi University, Gakushuin University, Nagoya City University are greatly acknowledged. Financial support by Economics and Management Institute at Gakushuin University is also acknowledged.

Notes

¹This definition is the same as Shleifer and Vishny (1997)'s one: “the ways in which the suppliers of finance to corporations assure themselves of getting a return on their investment.” We do not consider a broad definition of the “stakeholder society” advocated by Tirole (2001).

²In addition, Prowse (1990) explains the leverage ratio from the viewpoint that R&D shifts assets from creditors to shareholders because it depends on managerial discretion and is difficult for creditors to monitor.

³Samuel (2000) finds that the institutional ownership has a negative effect on R&D using U.S. firm data and interprets this as the managerial myopia.

⁴Kaplan and Minton (1994) show that the managers of the firms with large shareholders are more likely to be dismissed than those without large shareholders in Japan. If the monitoring by large shareholders is effective, keiretsu firms must be subject to strict monitoring than independent firms because in many cases large shareholders in keiretsu firms are the group banks who are also the large creditors of the keiretsu firm and have a strong incentive to monitor (See Hoshi et al., 1991).

⁵Burkart et al. (1997) argue that if large shareholders extract rents accrued from managers' and employees' accumulation of firm-specific human capital, managers and employees have little incentive to accumulate firm-specific human capital. We can apply a similar argument to R&D if managerial efforts to raise profits by devoting themselves to R&D are noncontractible firm-specific investments.

⁶Rajan (1992) considers a firm that is run by an owner-manager. But if managers' compensation is correlated with corporate profits, his conclusion is still valid for the firms where manager are different from owners. Using our sample, we have confirmed that directors' bonus per capita is positively correlated with the corporate profit rate.

⁷Stein (1989) refers to Abegglen and Stalk (1985) who insist that the difference between the managerial time horizons in Japan and those in the U.S. come from the difference of corporate governance such as inside directors versus outside directors.

⁸Hall et al. (1999) provided evidence that R&D was more highly sensitive to cash flow in the U.S. than in France and Japan. Huang and Xu (1999), however, demonstrate that the relationship between soft budget and R&D is not monotonic. If the risk of R&D is relatively small, a single source of funds that induces soft budget promotes R&D, while if the risk of R&D is relatively large, many sources of funds that induce hard budget promote R&D.

⁹In addition to the effects on managerial time horizons and effort levels, corporate governance may affect managerial behavior towards risk and thereby R&D. Under the limited liability, creditors are more risk averse than shareholders. Morck and Nakamura (1999), for example, show that Japanese banks send directors to financially distressed non-keiretsu firms. To take into consideration the effects of governance on corporate risks, we examined whether the standard deviation and the coefficient of variation in operating profits divided by total assets over 1987 through 1996 were affected by corporate governance. The estimation results showed no such correlation. For this reason, we do not discuss the effect of corporate governance on the behavior towards risks in details.

¹⁰Survey of Research and Development is carried out by Ministry of Public Management, Home Affairs, and Posts and Telecommunications and covers most Japanese firms that are active in R&D.

¹¹Our time convention is such that R&D expenditures during April 1988 to March 1989, for example, are referred to as R&D expenditures in 1989.

¹² Keiretsu no Kenkyu focuses on bank-centered keiretsu. Other definitions of keiretsu are possible, including six major business groups and corporate transaction links. Our definition of keiretsu is generally narrower than this broad definition.

¹³The indexes of main bank relationships are similar to Hirota (1999).

¹⁴Pakes (1985) and Griliches et al. (1991) explored the dynamic relationships among R&D expenditures, the number of patent applications, and the firm's market value using a recursive model to investigate the causes (demand factors vs. technological shocks) and effects of R&D activities. Here we focus on the market value of R&D activities.

¹⁵The estimation results for these small sample sets are available from the authors upon request.

¹⁶Because the correlation between the R&D stocks and R&D expenditures is high in the cross section, we believe that our results do not depend on our use of R&D expenditures as a proxy for R&D stocks. The estimation results are available from the authors upon request.

¹⁷The results are available from the authors upon request.

¹⁸Bond et al. (1998) and Mulkay et al. (2001) estimate R&D investment equations using instrumental variables. We did not use instrumental variables, however, because it was difficult to derive a structural equation from the optimizing behavior, given various, sometimes opposing, hypotheses concerning the effects of corporate governance on R&D.

¹⁹Other mechanism that affects managerial time horizons is a liquidity threat. If firms can obtain liquidity even when they run deficits, they do not have to interrupt long-term projects. When we define the financial distress as running the operating deficits in the current and preceding years, the numbers of the financially distressed firms are 11 in 1987 (4.78% of the sample) and 23 in 1996 (8.42% of the sample). We could not estimate the effect of the governance on R&D spendings using such small sample data sets.

²⁰To distinguish the two opposing effects of debt on investment as described in Section 2, McConnell and Servaes (1995) investigate the relationship between corporate value and leverage by distinguishing high-growth firms and low-growth firms using the U.S. firm data. and drew opposing interpretations.

²¹See Tomiyama (1998) for details.