مکانیزم انتقال سیاست پولی در ژاپن : شواهدی از ترازنامه بانک ها

We examine how banks' responses to monetary policy vary according to their balance sheet using Japanese bank data from 1975 to 1999. We find that the effect of monetary policy on lending is stronger for banks that are smaller, less liquid, and more abundant with capital. The effects of bank balance sheet on monetary transmission are different by bank types, policy stances and borrowers' industries. Our results imply that a lending channel of monetary transmission exists, that the effect of expansionary monetary policy is attenuated if banks' capital is scarce, and that the effect of monetary policy on the allocation of funds depends on banks' balance sheets. J. Japanese Int. Economies20 (3) (2006) 380–405.

Do the effects of monetary policy depend on banks' balance sheets? The recent Japanese experience has drawn renewed attention to this issue. Some economists argue that Japan's expansionary monetary policy during the 1990s was less effective than in earlier periods because banks' non-performing loans and scarce capitals put constraints on their ability to extend new loans (e.g., (Nagahata and Sekine, 2002)), while others argue that such a credit crunch is not warranted (e.g., (Krugman, 1998)). In addition to the volume of loans, the allocation of loans has also attracted economists' attention. Hoshi (2000), for example, suggests that non-performing loans distorted banks' allocation of loans and restrained the flow of funds towards growing firms. But empirical evidence on these issues is still scarce. To analyze the link between banks' balance sheets and the effects of monetary policy, the so-called “bank-lending view” of monetary transmission is a useful benchmark. According to this view, monetary policy shifts banks' loan supply schedules and thereby affects the expenditures by bank-dependent borrowers (e.g., (Bernanke and Blinder, 1988 and Kashyap and Stein, 1994)). Shifts in loan supply are not identical among banks but dependent on their balance sheets. For example, when the central bank drains a bank's reserve, the bank must decrease loans unless they make up for a shortfall in deposits by selling their securities or financing by nonreservable instruments. Banks with less liquid assets tend to decrease loans by a greater extent if nonreservable liabilities are not available or available only at higher costs than deposits. Stein (1998) develops a formal model of credit rationing in bank financing markets based on the observation that most of the bank liabilities that escape reserve requirements are not covered by deposit insurance and hence are likely to be subject to adverse selection. Our purpose is to test this micro implication of the bank-lending view using a set of panel data for Japanese banks from 1975 through 1999. With this microdata set, we can overcome the problem of distinguishing loan supply from loan demand. For example, a tight monetary stance restrains corporate investment through a rise in interest rates and/or currency appreciation and thereby decreases the demand for bank loans. We can regard such policy-induced changes in loan demand as a macro shock. On the other hand, if there are differences among banks in changes in loans, we can safely regard these microlevel differences as differences of shifts in the loan supply. It is an advantage of microdata analysis that we can overcome an identification problem. In contrast, time series analysis using macrodata potentially suffers from identification problems (e.g., Romer and Romer, 1990, Bernanke and Blinder, 1992, Kashyap et al., 1993, Hoshi et al., 1993a, Ueda, 1993, Hosono, 1995 and Ramey, 1993; Hatakeda, 1997 and Hatakeda, 2000Miyagawa and Ishihara, 1997). The first to analyze monetary transmission mechanisms in the US using a microdata set for banks were Kashyap and Stein (2000). They found that the impact of monetary policy on lending is stronger for banks with less liquid balance sheets. Using bank data for four European countries, France, Germany, Italy, and Spain, Favero et al. (1999) found that the effect of monetary contraction in 1992 depended on banks' size, liquidity, and the countries in which they operated. The present study contributes to this literature in four ways. First, we analyze the monetary transmission in Japan using banks' balance sheet data. Japanese banks, until recently, were subject to stricter regulation of nonreservable, uninsured financing when compared with US banks. It is interesting to examine the nature of bank-lending channels under different regulatory environments, because regulations are likely to affect monetary transmission mechanisms by restricting financial instruments that are available to banks. Because the Japanese banking sector has been burdened with huge non-performing loans since the early 1990s, our analysis also helps to reveal whether and to what extent a weak banking sector strengthens or weakens the effects of monetary policy. Second, we develop a new approach that focuses on capital ratios as well as sizes and liquidity ratios. Banks with abundant capital are less likely to be subject to asymmetric information problems such as adverse selection or moral hazard (Holmstrom and Tirole, 1997) and hence tend to be less sensitive to monetary policy. On the other hand, if capital adequacy requirements are imposed, banks with scarce capital cannot increase loans as a response to expansionary monetary policy. Even without capital regulation, managers of banks with scarce capital may hesitate to increase loans as a response to expansionary monetary policy if managers are risk averse and fear being fired when banks fail. We will examine empirically which effect, i.e. the moral hazard (or adverse selection) effect or the capital constraint (or risk aversion) effect, is prevailing. Considering the difficulty in measuring capital ratios accurately, we use several proxies for capital ratios: accounting measures, market measures, BIS ratios, and non-performing loan ratios. Third, we divide the sample years into expansionary and contractionary periods and see if the effects of monetary policy are asymmetric. Banks' liquidity, which serves as a buffer stock, may make a significant difference under tight monetary policy. Banks may become capital constrained under a loose monetary policy stance to meet the capital adequacy requirements. We also compare the effects of monetary policy during crisis periods and normal periods. Demand for liquidity increased sharply during the banking crisis in the late 1990s and may have affected the effectiveness of monetary policy. Fourth, we analyze loans to each borrowing firms' industry as well as total loans and see if banks' balance sheets affect the impacts of monetary policy on the allocation of funds. It also contributes to a robustness check. The purpose of this study thus is quite different from much of the literature in this field. For example, a number of studies have examined the microlevel relationship between loans and deposits (Jayaratne and Morgan, 2000 and Ogawa and Kitasaka, 2000) or the relationship between loans and capital (Bernanke and Lown, 1991, Peek and Rosengren, 1997, Woo, 1999, Sasaki, 2000 and Ito and Sasaki, 2002). Though they are related to the “bank-lending view” in the sense that they paid attention to bank balance sheet conditions as a determinant of loan supply, none of these papers studies the effects of monetary policy. Other papers study the effects of borrowers' net worth on their investment (Kwon, 1998 and Ogawa and Suzuki, 1998; Bayoumi, 1998; Ogawa, 2000 and Hosono and Watanabe, 2002). However, in the present study, we focus on banks and not on borrowers. It should be noted that if monetary policy is to affect the real sector through banks' loan supply, there must be some borrowers that cannot frictionlessly tap alternative source of funds such as commercial papers or bonds. Though it is beyond the scope of this paper to examine this issue, it is well known that some types of firms, in particular, small firms, are heavily dependent on bank loans in Japan as well as in the US (Hoshi et al., 1993b, Anderson and Makhija, 1999 and Hosono, 2003). The remainder of the paper is organized as follows. Section 2 reviews the regulatory environment in Japan and compares it with the situation in the US. Section 3 presents some hypotheses based on the theoretical background of the “bank-lending view.” Section 4 describes the data and the specification of our econometric model. Section 5 presents baseline results and various robustness checks. The effects of BIS ratios and non-performing loans are also reported in this section. Section 6 presents the estimation results using sub-samples, classified by bank type, monetary policy stance, and borrowing firms' industry. Section 7 concludes the paper.

If banks face incomplete capital markets and/or are subject to capital adequacy requirements, their responses to monetary policy depend upon their balance sheet conditions. Analyzing the balance sheet data of Japanese banks from 1975 to 1999, we found that the effect of monetary policy on lending is stronger for banks that are smaller, less liquid, and more abundant with capital, though the results for liquidity are somewhat less robust than those for size and equity. Banks’ responses are asymmetric between contractionary and expansionary policy stances: Liquidity is significant to monetary transmission in times of contractionary policies, while capital is significant in times of expansionary policies. During the banking crisis in the late 1990s, expansionary monetary policy relaxed the liquidity constraints of illiquid banks and promoted lending by these banks. Estimation results of loans by borrowing firms’ industry revealed that bank capital is significant to monetary transmission to the manufacturing sector. The implications of our results are as follows. First, a lending channel of monetary transmission exists. The difference in banks’ balance sheets makes an economically significant difference in responses to monetary policy. Second, the effect of expansionary monetary policy is attenuated if banks’ capital is scarce. The “bank-lending view” itself usually implies that well-capitalized banks are prone to asymmetric information problems and hence less sensitive to monetary policy shocks. But our finding that well-capitalized banks are more sensitive to monetary policy shocks, especially to expansionary shocks, suggest that capital ratios constrain banks’ loan extension under expansionary policies rather than attenuate information problems such as moral hazard or adverse selection problems. Finally, the effects of monetary policy on the allocation of funds depend upon banks’ balance sheets. Expansionary monetary policy decreases the relative share of loans to manufacturing industries if banks are less capitalized. Japanese banks’ financing patterns do not seem to have changed much after the end of our sample period, 1999. Whether or not and how bank financing and hence monetary transmission will change in the future depends on various factors, including further progress of deregulation, movements of interest rates, and reorganization of the banking industry.