2002 Japan Conference: A Summary of the Papers

The literature on relationship banking postulates that close bank ties can mitigate the asymmetric information and moral hazard problems that afflict public capital markets. Most notably, Hoshi, Kashyap, and Scharfstein (1991) (hereafter HKS91) investigate the investment-cash flow sensitivity of bank affiliated and independent firms in Japan, and provide evidence that strong ties have helped to alleviate liquidity constraints. However, a body of work has emerged that focuses instead on the associated costs of "main banking" in Japan. ⁽¹⁾ This paper builds on that literature by examining the investment-cash flow sensitivity of Japanese manufacturing firms, and uses the explicit bond issuing criteria in a simple test of whether firms enjoyed positive net benefits from close bank ties in the 1980s and 1990s.

The literature on investment-cash flow sensitivity rests on the idea that Tobin's Q is a "sufficient statistic" that governs firm investment, while other factors, liquidity in particular, should not be important. A positive and significant coefficient on liquidity is interpreted as evidence of financing constraints, because this correlation presumably reflects the degree to which firms rely on internal funds for investment. Typically, researchers speculate on the nature of financing constraints, and then compare the investment-cash flow sensitivity across firm groups. However, isolating truly constrained firms is difficult. Ideally, exogenous criteria should be used to identify financing constraints, but researchers often are forced to rely on behavioral variables, which can lead to endogeneity problems. Fortunately (for this analysis), Japanese capital markets were regulated for much of the post-war period, and firms had to meet explicit criteria to issue debt in the domestic bond market in the 1980s. These criteria are the lynchpin upon which this analysis rests, and can be used to address the empirical difficulties.

There has been considerable research on Japanese firms in this area, typically in the context of assessing the benefits of affiliation with large domestic banks. In their seminal paper in both the Japan and Q literatures, HKS91 examine a panel of 145 manufacturing firms, and find that firms with strong bank ties exhibited significantly lower investment-cash flow sensitivity than did independent firms. Since main banks acquire inside knowledge of client firm's investment opportunities, the asymmetric information problems that force firms to rely on internal funds for investment are reduced. More recently, however, Hayashi (2000) re-examines the HKS91 firm sample (with different data); after excluding outliers, he finds no significant difference in the cash flow sensitivity of bank-affiliated and independent firms. ⁽²⁾ While this discrepancy ultimately may be driven by econometric technicalities, it does cast a shadow on the robustness of the original HKS91 results, and calls for further investigation.

There is mounting evidence that bank affiliation did not come without costs to client firms. Deregulation during the 1980s provided firms with alternative funding sources for the first time in the post-war era. This expansion of non-bank financing options led to increased heterogeneity in capital structure across firms; researchers have been able to exploit this in identifying the costs of bank affiliation. Hoshi et al. (1990a) examine the shift toward non-bank financing for a sample of 109 firms, and find higher sensitivity for firms that decreased their reliance on bank debt. This suggests that the net benefit to firms of bank affiliation may have been negative (at least in this period), since these firms presumably could have maintained close bank ties after deregulation.⁽³⁾ In a more direct test, Weinstein and Yafeh (1998) estimate a model in which banks can influence firm investment through shareholding and force firms to borrow as though their cost of capital is lower than it actually is. They argue that bank pressure induced artificially high loan flows and inefficient investment strategies, which in turn led to the over-capitalization of client firms.

Building on this literature, my paper investigates the investment-cash flow sensitivity of Japanese manufacturing firms during 1980-96. While similar in spirit to Hoshi et al. (1990a), it uses a much larger panel, includes the more detailed data described in Hayashi (2000), and uses the bond eligibility criteria as an exogenous firm-sorting mechanism. In addition, it relies on recently developed empirical techniques, and carries the analysis of cash flow sensitivity and profitability into the 1990s, when bond market deregulation was complete.

After separating firms into "Restricted" and "Unrestricted" groups, I show in the first section of the paper that sensitivity was lowest for those firms that were restricted from the bond market during the 1980s. This result is very much at odds with the standard prediction in the Q literature. That is, the investment of firms known to have faced capital market restrictions was less sensitive to measures of internal net worth than the investment of firms with a wider range of financing options. This result survives several robustness tests, including controls for negative observations as outlined in Allayannis and Mozumdar (2001), and may indicate that sensitivity analysis is not an appropriate technique for identifying financing constraints.⁽⁴⁾ However, restricted firms were, by definition, bank-dependent; this means that the results presented here are consistent with established theory, and lend support to HKS91 (as opposed to Hayashi (2000)).

As effective firm monitors, banks may have facilitated the efficient use of capital, and helped firms to achieve and maintain their first-best investment path. On the other hand, high monitoring costs may have meant that firms paid a premium for bank financing. In a more sinister scenario, banks may have enjoyed information monopolies, or market power from the remaining capital controls, that allowed them to extract rents through higher interest rates, severe collateral demands, or compensating balance requirements. This is of particular concern because of the gradual nature of bond market deregulation; banks may have been able to capitalize on their market power over firms that did not meet the bond criteria. Thus, the results introduced above do not provide clear evidence on whether bank ties were a net benefit to firms. Sensitivity may have been lower for restricted firms because they enjoyed better access to funding (from close banks), although at higher effective costs. Alternatively, sensitivity may have been driven by artificially high loan flows if firms were pressured to borrow (Weinstein and Yafeh (1998)).

Splitting the sample using both a standard proxy for main bank affiliation and the bond criteria sheds light on this issue. As in Hoshi et al. (1990a), this section of my paper implicitly hypothesizes that if firms enjoyed a positive net benefit from close bank ties, then sensitivity should be lower for bank-affiliated firms, regardless of bond market access. In addition, there should be some ex post observable difference in performance between affiliated and independent firms that reflects the benefits of bank ties. If the net benefit accrued to banks (because of high monitoring costs or rent extraction) then bond-eligible, main bank-client firms may have had a stronger incentive to move to bond financing, even if this increased the likelihood of asymmetric information problems. In other words, the implications of higher sensitivity may have been more palatable to firms than the bank relationship.

The evidence I present supports the latter case. For the four firm groups, sensitivity was highest for main bank-client firms with access to the bond market in the 1980s, and lowest for main bank client-firms without such access. That is, the spread in sensitivity was larger for bank-affiliated firms than for independent firms, once external financing options were introduced. In addition, bank-affiliated firms accessed the bond market more often, and with (slightly) larger issues than did independent firms, despite their lower profitability.

These results are consistent with banks capturing the rents from main bank ties. With deregulation, mature and healthy firms chose to reduce bank dependence; this implies that internal financing (and its associated costs) was less than the cost of a main bank relationship. Importantly, these results do not imply that firms never benefited from having a main bank. Indeed, it is often argued that the bank-centered financial system was an important factor behind Japan's rapid growth prior to the mid-1970s, and only became obsolete once Japan reached the technological frontier. The lack of outside financing options in this earlier period precludes the firm division proposed in this paper.

4. There is considerable debate about the use of sensitivity analysis. Most recently, Allayannis and Mozumdar (2001) overturned the Kaplan and Zingales (1997) critique of Fazzari et al. (1988), as well as the results in Cleary (1999). They argue that firms with negative cash flow have driven investment down to its lowest possible level, making it unable to respond to further reductions (or small fluctuations) in cash flow.

Allayannis, G., and A. Mozumdar (2001): "The Investment-Cash Flow Sensitivity Puzzle: Can Negative Cash Flow Observations Explain it?", Virginia Tech Working Paper 98-5

Cleary, S. (1999): "The Relationship Between Firm Investment and Financial Status," The Journal of Finance, LIV(2), 673-692

Fazzari, S., G. Hubbard, and B. Petersen (1988): "Financing Constraints and Corporate Investment," Brookings Papers on Economic Activity, pp. 141-195.

Hayashi, F. (2000): "The Main Bank System and Corporate Investment: An Empirical Reassessment," in Finance, Governance, and Competitiveness in Japan, ed. by M. Aoki, and G. Saxonhouse, pp. 81-98. Oxford University Press, Oxford.

Hoshi, T., (2000): "The Main Bank System and Corporate Investment: Further Robustness Tests," in Finance, Governance, and Competitiveness in Japan, ed. by M. Aoki, and G. Saxonhouse, pp. 99-105. Oxford University Press, Oxford.

Hoshi, T., and A. Kashyap (2001): Corporate Financing and Governance in Japan: The Road to the Future, MIT Press

Hoshi, T., A. Kashyap, and D. Scharfstein (1990a): "Bank Monitoring and Investment: Evidence from the Changing Structure of Japanese Corporate Banking Relationships," in Asymmetric Information, Corporate Finance, and Investment, ed. by G. Hubbard, pp. 105-26, University of Chicago Press

Hoshi, T., A. Kashyap, and D. Scharfstein (1991): "Corporate Structure, Liquidity, and Investment: Evidence from Japanese Industrial Groups," Quarterly Journal of Economics, 106.

Kaplan, S. N., and L. Zingales (1997): "Do Investment-Cash Flow Sensitivities Provide Useful Measures of Financing Constraints?," Quarterly Journal of Economics, CXII, 169-215.

Morck, R., M. Nakamura, and A. Shivdasani (2000): "Banks, Ownership Structure, and Firm Value in Japan," Journal of Business, 73(4), 539-67.

Weinstein, D., and Y. Yafeh (1998): "On the Costs of a Bank-Centered Financial System: Evidence from the Changing Main Bank Relations in Japan," The Journal of Finance, LIII(2)