Japan: Selected Issues

This Selected Issues paper analyzes Japan’s medium- and long-term fiscal challenges. It discusses the initiatives that will be necessary to cope with Japan’s medium- and long-term fiscal challenges. It describes medium- and long-term projections of the fiscal position, and assesses the size of the consolidation measures needed to restore long-term fiscal sustainability. The paper explores possible options for consolidation and offers a representative package of such measures. The paper also describes the projected baseline path for the long-term fiscal balance, incorporating the long-term pension reform plan formulated in 1994.

Abstract

This Selected Issues paper analyzes Japan’s medium- and long-term fiscal challenges. It discusses the initiatives that will be necessary to cope with Japan’s medium- and long-term fiscal challenges. It describes medium- and long-term projections of the fiscal position, and assesses the size of the consolidation measures needed to restore long-term fiscal sustainability. The paper explores possible options for consolidation and offers a representative package of such measures. The paper also describes the projected baseline path for the long-term fiscal balance, incorporating the long-term pension reform plan formulated in 1994.

III. Assessing The Risks of A Credit Squeeze Among Small- and Medium-Sized Enterprises in Japan 1/

1. Introduction

Whether continuing banking-sector problems in Japan will lead to a credit squeeze and impede economic growth is of new interest as the recovery gathers momentum. While no consensus view has emerged, this paper provides a framework for assessing the risks of such a credit squeeze. The analytical focus is on small- and medium-sized firms (SMEs), because this is the sector of the economy that is most reliant on bank-intermediated lending. The slow pace of the Japanese recovery to date has made the issue mainly academic so far, but this paper argues that several preconditions are in place that could lead to a broad credit squeeze should the pace of the recovery be somewhat faster than currently envisioned. Such a credit squeeze could distort conventional indicators of monetary conditions, serve as a governor to economic growth, and present a challenging macroeconomic problem to policy makers. By describing the conditions under which credit flows could be constrained, and suggesting certain early warning signals that might be associated with a credit squeeze, this paper may help authorities to better recognize and manage such risks.

By international standards, businesses in Japan have been well supplied with credit, at least historically. According to a study by Borio (1995), for example, total credit available to the non-government sector in Japan in 1993 was about two times the level of GDP, or roughly twice the ratio for the other industrial countries. Several factors, however, appear to increase the risks of a credit squeeze in Japan. First, the ratio of bad loans to total loans for the Japanese banking sector, estimated to be on the order of 9 or 10 percent for the entire industry and 15 percent or higher for certain categories of banks, is extremely high by historical and international standards. 2/ Experiences in other countries that have had banking problems suggest that in order to meet Bank for International Settlements (BIS) capital ratios, Japanese banks will have an incentive to rebalance their assets away from loans (which have a high capital charge) and toward government securities (which have no capital charge). As banks undertake this asset substitution to repair their capital ratios, they may well cause a credit squeeze that will hit SMEs the hardest. Second, because the capital markets, and the corporate bond market in particular, remain outside the purview of the vast majority of SMEs in Japan, alternate sources of financing appear unlikely to fill a possible bank credit void. Even the ability of the country’s largest banks to step in and provide substitute credit for weaker banks seems questionable, given that even these historically strong institutions are experiencing record bad loan problems. And third, SMEs, which are so important to the economic growth process in Japan, have been found to behave differently from large firms in tight credit conditions. According to Gertler and Gilchrist (1994), small firms tend to ratchet down their investments and their operations much faster and further than large firms when credit supply is restricted. With SME firms responsible for nearly 40 percent of all private investment spending in Japan, the risks that a credit squeeze could slow the overall economic recovery are clearly worth studying.

To provide a context for assessing the risks of a credit squeeze in Japan, this paper provides a discussion of the key issues dealing with SMEs and the bank credit channel. Section 2 provides background material. It begins with a brief theoretical discussion about the general relationship among banking sector health, credit supply, and the real economy. It then looks at questions such as: “What is the role of investment spending by SMEs in typical Japanese expansions? Where does the supply of credit for SMEs usually come from? What is the relative health today of the parts of the financial sector upon which SMEs are most dependent? And what evidence is there that banking sector problems have been associated with reductions in the supply of credit or with reduced investment spending?” Section 3 then provides arguments for why heavy bad loan burdens may cause Japanese banks to cut back on their lending. It also describes some “early warning indicators” of credit tightness and reviews empirical research that deals specifically with the credit supply issue in the Japanese context. Section 4 provides an overall assessment of the risks of a credit squeeze and provides a comparison to others’ assessments of these risks. It ends with some thoughts about possible implications for macroeconomic policy and banking policy. Section 5 then concludes.

2. Background issues

a. Theoretical considerations about the credit channel

There is a theoretical perspective that argues that movements in money and credit influence the macroeconomy through the so-called “credit channel.” 1/ According to this view, imperfect information and other frictions in credit markets and bank lending behavior are important to the operation of this credit channel, and help to explain the potency of monetary policy. That is, these credit forces can greatly amplify traditional interest rate effects on the macroeconomy. A key concept in this credit transmission is a mechanism sometimes referred to as the “external finance premium” or “disintermediation premium,” which can be defined as the difference between the cost of funds raised by firms externally (by issuing equity or debt) and funds generated internally (by retained earnings)—that is, the wedge between the expected return received by lenders and the costs faced by potential borrowers. This external finance premium is determined by movements in two channels: the balance sheet (or net worth) channel and the bank lending channel.

The balance sheet channel is based upon the theory that the external finance premium facing a borrower should be related to his financial health. For example, the weaker the borrower’s balance sheet position, the higher the expected premium because presumably the borrower has less collateral to offer or is constrained to make a smaller down payment. Such a higher external finance premium would represent a higher cost for a firm, suggesting that balance sheet weakness would tend to reduce a firm’s investment outlays—an effect sometimes referred to as the “financial accelerator.” The bank lending channel works the following way. If the supply of bank loans is disrupted, bank-dependent borrowers, such as SMEs, will incur costs associated with finding a new lender or establishing a new credit relationship, and as a result, the external finance premium will increase. According to this theoretical framework, an increase in this premium will increase the cost of doing business for these firms, and hence reduce their investment spending and pace of real activity.

Reports of tighter credit by SMEs and wider interest rate spreads between bank loan rates and a bank’s cost of funds are evidence that the bank lending channel is tightening up, that the external finance premium is increasing, and that it is probably restricting macroeconomic growth. While it is very difficult to isolate empirically the balance sheet channel effects (including changes in balance sheets, income statements, cash flow, or liquid assets) from the bank lending channel effects, there are persuasive reasons to think both effects can be significant. This is particularly likely to be the case for a country like Japan, where bank lending remains the dominant external source of corporate financing.

b. The role of investment and especially SME investment in Japanese recoveries

Private investment spending on plant and equipment typically accounts for about one fifth of GDP in Japan, a noticeably higher share of output than in other industrial countries. Small- and medium-sized enterprises—those with share capital of ¥1 billion or less and referred to as SMEs in this paper—have been responsible for a relatively large share of this investment spending. 1/ Over the 1980s, for example, the SME share averaged between 43 and 46 percent of total investment spending (Chart III.1), and, not surprisingly, increases in SME investment spending have been critical to Japanese economic recoveries in the postwar period. Over the five business cycle recoveries in the 1970s and 1980s, for example, business investment spending increased by an average of 32 percent over the first eight quarters after the recession trough. Investment spending by SMEs has traditionally driven this increase in overall spending. Over the three business cycle recoveries prior to the current one for which a breakdown by firm size is available (since 1976), SME investment spending jumped by an average of 41 percent over the eight quarters after recession troughs, or about 2/3 faster than the 26 percent growth pace by large firms.

Chart III.1.
Chart III.1.

Small- and Medium-Sized Enterprises Share of Total Investment Spending

(Percent)

Citation: IMF Staff Country Reports 1996, 114; 10.5089/9781451820522.002.A003

Source: Japan Development Bank; Small- and medium-sized enterprises are firms with share capital of 1 billion yen or less.

A very different investment pattern has emerged in the current business cycle (Charts III.2 and III.3). Overall investment spending continued to decline after the recession trough in the fourth quarter of 1993 and had barely re-attained its recession trough level by the fourth quarter of 1995—the eighth quarter of recovery. Significantly, SME investment spending, which traditionally exceeds overall investment spending in this phase by a wide margin, has continued to decline over the past two years. This weakness in SME investment spending in the current recovery phase is highlighted in Chart III.1, which shows that the SME share of total investment spending plummeted from roughly 45 percent in the 1988-90 period to just 38 percent or so in 1995. In short, instead of fueling growth, SME investment has remained very depressed and has shrunk as a share of total investment spending and as a share of national output.

Chart III.2.
Chart III.2.

Investment Spending Over Recovery Cycles, Large Enterprises

(4 quarter moving average, indexed to quarter indicated)

Citation: IMF Staff Country Reports 1996, 114; 10.5089/9781451820522.002.A003

Source: Japan Development Bank; Large enterprises are firms with share capital of more than 1 billion yen.
Chart III.3.
Chart III.3.

Investment Spending Over Recovery Cycles, Small- and Medium-Sized Enterprises

(4 quarter moving average, indexed to quarter indicated)

Citation: IMF Staff Country Reports 1996, 114; 10.5089/9781451820522.002.A003

Source: Japan Development Bank; Small- and medium-sized enterprises are firms with share capital of 1 billion yen or less.

c. The supply of credit to small- and medium-sized enterprises

There are many plausible arguments for why SME investment spending has been so depressed over the current recovery phase: structural change caused by yen appreciation, the slow overall pace of the recovery that has prevented the buildup of any noticeable pressures on capacity, the economy’s increased openness to imports, and possible credit market disruptions caused by the bursting of the asset bubble in the early 1990s. This section focuses on the last issue. What are the sources of funds for investment spending in Japan? Where does credit for SME firms come from? What is the health of the banks and financial institutions that tend to supply credit to SMEs? Is the health of these institutions poorer than that of the financial institutions that supply funds for large firms?

According to national income accounts data, investment spending in Japan tends to be funded mainly through two sources: internally out of retained earnings, and externally through bank lending. Capital markets—net equity and net bond market issues—played a fairly important role in the late 1980s, but have tended to be relatively minor sources of funds for investment projects over the past half decade. Over the period from 1990 to 1993, for example, roughly two thirds of investment spending in Japan was funded out of retained earnings and roughly one third was funded by banks. On average less than 7 percent of funding over these years came from net bonds issues and less than 3 percent came from net equity issues. 1/ Although this breakdown on the sources of funds is not available by firm size, these data suggest that the most important factors affecting the investment spending of SME firms also are their profitability and the availability of bank credit. 1/

Table III.1 provides constructed estimates of the sources of non-bond, market credit for SMEs in Japan. 2/ City banks are the largest single source of such credit, supplying about 32 percent, regional I banks provide about 21 percent, regional II banks about 10 percent, shinkin banks and credit cooperatives about 24 percent altogether, and government loans to SMEs about 4 percent. This mix of credit supplying the SME sector is very different from the mix supplying larger firms. According to Gibson (1995a), for example, in FY 1994 the largest 385 listed firms on the Tokyo Stock Market (by share capital) relied on city banks for 31 percent of their bank borrowing, long-term credit banks for 34 percent, trust banks for 31 percent, and regional banks for just 4 percent. Virtually none came from shinkin banks, credit cooperatives, and other small associations. That is, while both large firms and SMEs tend to get about a third of their non-bond market credit from city banks, SMEs are much more dependent on regional banks, shinkin banks, and credit cooperatives and associations than are large firms.

Table III.1.

Japan: Where Do Small- and Medium-Sized Enterprises (SMEs) Get Credit?

article image
Source: Data from Bank of Japan, Economic Statistics Annual. 1995, Research and Statistics Department, April 1995, Table 64 (Loans and Discounts by Industry). Notes: data for some smaller cooperatives and associations was for FY 1994, not FY 1995. Assumes 100 percent of Shinkin and credit cooperative loans are to SMEs and therefore may overstate these percentages somewhat.Definition: SMEs: Firms with capital of ¥100 million or less or 300 persons or less, or if in wholesale trade, firm capital of ¥30 million or less or 100 employees or less.

Japan Finance Corporation for Small Businesses, People’s Finance Corporation, and Small Business Credit Insurance Corporation.

Not only do capital markets in Japan supply a significantly smaller share of external financing to businesses than the share supplied by banks, but the non-bank capital markets have been providing a decreasing share of total economy-wide loans and credit. According to Borio (1995), for example, between 1983 and 1993 the non-bank capital market share of total loans supplied in Japan decreased by nearly 10 percentage points. 3/ In addition, as Smith (1995) points out, there are a number of reasons why the bond market in particular has remained beyond the reach of SMEs in Japan. Serving to limit SME access to the capital markets in Japan, for example, have been high bond placement fees, regulatory restrictions, such as the requirement that domestic bonds be rated at investment grade, and the fact that many bond issues are effectively similar to bank loans because they are bought entirely by banks.

Indeed, Gibson (1995a) shows that access to the corporate bond market in Japan is strongly related to firm size. Among the largest 385 Japanese publicly traded firms—those with share capital of more than ¥14.3 billion—he estimates that 52 percent of all long-term credit came from the bond market. But this percentage of bond financing fell rapidly as firm size decreased. Among firms with capital of between ¥2.3 and 5.7 billion, for example, Gibson calculated that just 33 percent of credit was from the bond market, and this dropped to just 22 percent for the publicly traded firms with share capital of less than ¥2.3 billion. 1/ It is impossible to calculate the share of bond financing in the total borrowing of the small- and medium-sized firms defined in this paper—those SMEs with share capital of ¥1 billion or less—because of data limitations for non-listed firms. But based upon Gibson’s data it seems likely that such SME firms as a group get less than 10 percent of their total long-term credit from the bond market, and perhaps less than 5 percent. Similarly, while Gibson’s group of smaller firms (share capital of less than ¥2.3 billion) depended upon banks for 60 percent of their long-term credit, it seems likely that the SME group with share capital of ¥1 billion or less is probably at least 85-90 percent dependent on banks for credit.

d. Condition of the banking sectors that supply credit to SMEs

What is the relative health of the banking sectors upon which SMEs rely most heavily? The International Banking Credit Agency (IBCA) has analyzed the nonperforming loans, loans to jusen, and restructured loans of the various segments of the Japanese banking system, and has summarized the relative health of each banking sector via a ratio of adjusted bad loans to overall loans. These data suggest that for FY 1995, the total bad loan ratio ranged from under 7 percent for regional I banks as a group, to 8 to 8 1/2 percent for city banks, shinkin banks, and regional II banks, to 14 to 17 percent for trust banks, credit cooperatives, and the agricultural sector (Table III.2). Clearly the biggest problem area for SMEs is their reliance on credit cooperatives. Also a number of shinkin and regional II banks are weak, reflecting the wide variance in the health of the institutions in those sectors. Altogether, it seems likely that somewhere between 20 and 35 percent of the borrowing of SMEs originates in those banking sectors that are experiencing the worst bad loan problems.

Table III.2.

Japan: International Banking Credit Agency (IBCA) Estimates of Total Bad Loans by Type of Deposit-Taking Institution, September 1995

article image
Source: International Banking Credit Agency, staff estimates.

The biggest problem for large firms is their heavy dependence on trust banks and long-term credit banks, which have high estimated ratios of bad loans, and which typically account for 40 to 60 percent of the bank credit for these firms. When taking into account the fact that large firms typically get 30 to 50 percent of their overall long-term credit from the bond market, however, large firms probably are somewhat less dependent on banks that have high bad loan ratios than are SMEs. Of course, for these large firms, even the option of tapping the bond market would allow for the reconfiguration of credit supply should a firm have difficulty obtaining credit through its normal banks.

Given the differing levels of bad loan problems among these different segments of the banking industry, is there evidence that the banking groups facing the biggest problems (such as the credit cooperatives, the trust banks, and the long-term credit banks) have behaved differently in their lending patterns than the healthier banking groups (such as the regional I banks)? Table III.3 suggests that the evidence is confusing, with the supposedly weak trust banks actually leading all other banking sectors in loan growth over the past three years, and the supposedly stronger regional I banks providing loans at only about the average overall pace. Chart III.4 shows graphically that the expected inverse relationship between estimated bad loan burden and loan supply does not hold very strongly. The exact underlying reasons for this are not clear, however. Certainly it is possible that some of the loan growth among weaker institutions may simply reflect additional loans extended to borrowers that have trouble meeting repayment schedules, and may provide a misleading picture of true commercially viable developments. On the other hand, the prima facia case that the weaker banking sectors would tend to cut back on their lending is not directly confirmed by this aggregate data. A more detailed examination of this relationship is required.

Table III.3.

Japan: Bank Loan Growth, Various Bank Categories, FY 1990-95

(Percent change year ago)

article image
Source: Bank of Japan, Economic Statistics Annual. 1995, Research and Statistics Department, Tokyo, 1996.

Calendar year basis.

Chart III.4.
Chart III.4.

International Banking Credit Agency (IBCA) Estimate of 1995 Bad Loan Percentage vs. Average 1993-95 Loan Growth, Various Bank Groups

Citation: IMF Staff Country Reports 1996, 114; 10.5089/9781451820522.002.A003

Source: International Banking Credit Agency (IBCA) staff estimates, May 1996, and Bank of Japan Economic Statistics Annual, 1995.

3. The relationship between banking sector problems and the supply of credit

Does it matter if the portion of the financial system that supplies credit to SMEs has bad loan problems? How would bank lending behavior tend to respond to a heavy bad loan burden? Is there evidence that these problems might restrict the supply of credit, and if so, how severe might this restriction be? This section provides arguments about how the BIS capital ratio requirements may encourage banks to cut back on their lending. It also reports on a couple of early warning signals that might indicate tightness in the credit channel and reviews several empirical studies that attempt to quantify the link between bank health and credit supply.

a. Capital requirements and why Japanese banks might cut back on their lending

Japanese banks have been making large loan loss provisions over the past couple of years and, although profits have provided the bulk of the funds for these provisions, large amounts of bank capital have also been used. In FY 1995, for example, the largest 21 banks used over ¥2 trillion in BIS tier 1 capital—about 10 percent of their total tier 1 capital—for write-off purposes. This use of capital dropped the aggregate BIS tier 1 capital ratio for these banks from about 5.2 percent of total loans to 4.6 percent, with the floor requirement for tier 1 capital being 4 percent. In order to satisfy BIS capital requirements, banks have an incentive to rebalance their assets away from loans, which have a full capital charge, and toward government securities;, which have no capital charge for BIS purposes. It is precisely this asset substitution to repair capital ratios that has been offered by Thakor (1996) and others as an explanation for the credit squeeze in the United States during 1989-93. In addition, large Japanese banks must allocate additional capital to cover market risk in their trading books by the end of 1997, as required by the Amendment to the Basle Capital Accord to Cover Market Risks. These two capital ratio pressures suggest that Japanese banks will have an incentive to reduce their loan portfolios over the next couple of years, and the business sector that appears most likely to be affected is SME firms.

b. Early warning signals of credit tightness for SMEs

Performance measures suggested by the credit channel approach already show some evidence of credit tightness for SMEs in Japan. First, survey data, such as the Bank of Japan Tankan report on banks’ willingness to lend, suggests that SMEs are feeling some credit tightness. For example, even though the average lending rate charged by banks plunged from over 8 percent in early 1991 to under 2 percent in early 1996, the Tankan survey judgement by small- and medium-sized enterprises about bank willingness to lend over this period increased by only a modest 10 points (Chart III.5). In contrast, the same judgement by major enterprises on bank willingness to lend jumped by a dramatic 80 points or so over the same period. In fact, there appears to have been a breakdown in the correlation of these two surveys. Statistical tests, for example, find evidence of a structural break in the relationship between the bank-willingness-to-lend-attitude indexes of large firms and SMEs in 1990. 1/ Over the previous period of large-scale recession and sharp interest rate declines—1980 to 1983—the lending-attitude index for SMEs increased sharply and roughly in line with the lending index for large enterprises. But so far in this cycle SMEs are not reporting the easing of lending conditions that would have been expected given the sharp increase in the lending-attitude index of large firms.

Chart III.5.
Chart III.5.

Judgement on Lending Attitude

(Percentage points)

Citation: IMF Staff Country Reports 1996, 114; 10.5089/9781451820522.002.A003

Source: Bank of Japan Tankan Survey; Small- and medium-sized enterprises are firms with share capital of 1 billion yen or less.

Another early warning indicator of credit tightness suggested by the “credit channel” approach is to look at movements in the “external finance premium”—that is, the interest rate spread between bank loan rates and a proxy for banks’ cost of funds. 2/ A study by Utsunomiya (1995) compared the movements in these interest rate spreads over the recoveries from the 1985-86 recession and the 1991-93 recession, and in particular the movements in the interest rate spreads experienced by SMEs and large firms in each episode. According to his regression results, the interest spreads experienced by SMEs were only slightly greater than those experienced by large firms over the period from 1985 to 1988, and the spreads for both classes of firms drifted downward almost in lockstep over the economic rebound phase. His results for the current recovery phase show a very different story. Overall, interest rate spreads on bank loans increased noticeably over the period from 1991 to 1994 (the last year of his data), increasing on average from about 0-1 percent per annum in 1991, to 1 1/2-2 percent in 1994. Even more dramatic was the upward shift in the interest spread on loans to SMEs relative to the spread on loans to large firms. In 1994, for example, Utsunomiya found roughly a 50-basis point premium on loans to SMEs over loans to large firms, whereas in the mid-1980s period there was almost no SME premium.

It can be argued, of course, that an upward shift in the interest spread between the bank lending rate and the bank cost of funds rate might represent an appropriate change in the market assessment of lending risks. That is, perhaps it really has become much riskier to lend to SMEs in the current period. While it is impossible to know the true answer to this question, several pieces of information suggest that some additional tightening factor is at work. Profit growth at SMEs, for example, though lagging behind that at large firms in 1995, is projected to be much higher in 1996, suggesting that the overall credit-worthiness of these smaller firms should not have been noticeably lower. Also, there remains the question of why indexes of perceived bank willingness to lend to SMEs and large firms diverge so much now, after having moved more closely together over prior business cycles.

c. Empirical studies of the link between bank health and credit supply in Japan

This information on surveys of lending attitudes and interest rate spreads provides a general characterization of the lending environment in Japan over the past few years. In addition, a number of empirical studies have attempted to directly measure the relationship between the poor health of banks on one hand, and their lending behavior or the level of investment spending by their customers on the other. This section reviews recent studies by Gibson (1995b, 1996), Maeda (1996), and Yoshikawa (1994) et al. Generally these studies are constrained by data availability and tend to concentrate only on the largest banks or sometimes just the publicly-traded banks, and only on the impact on investment spending of their publicly-traded clients. Overall, the studies do find evidence that poor bank health, as captured by credit ratings or a ratio of nonperforming loans to total loans, does have some negative impact on bank lending and investment activity. The range of estimates is fairly wide, with poor bank health associated with anywhere from a negligible to a 50 percent cutback in lending and/or customer investment spending.

The Gibson (1995b, 1996) papers look at investment outlays for roughly 1600 publicly-traded companies and try to determine whether the credit rating of the firm’s main bank is significant in explaining these outlays. In the 1995 paper that looked at data from 1991-92, Gibson found that firms whose main bank had the lowest category credit rating tended to experience a 30 percent decline in investment spending, compared with firms with “neutral” main banks. The coefficients on the bank credit variables were statistically significant. Gibson (1996) repeated his previous analysis with a 1994-95 data set and found that the credit rating variables were no longer statistically significant and that the estimated coefficients were smaller. Investment spending at firms whose main bank had the lowest category credit rating tended to be reduced by just 10 percent. Gibson concluded in the 1996 paper that having a weak main bank in 1994-95 did not tend to depress investment much overall, and that the differences in the results were explained by changes in economic conditions between 1991-92 and 1994-95.

But Gibson (1996) did find that whether or not a firm had access to the corporate bond market mattered a lot. Of the firms in his 1994-95 data set, he classified 26 percent as “bank dependent,” meaning that they had never tapped the corporate bond market, and the other 74 percent as “non-bank dependent,” because they had issued corporate bonds at least once. Among the bank-dependent firms, Gibson found that having a weak main bank could reduce investment spending by over 50 percent, while for non-bank dependent firms there was no significant effect of having a main bank with a weak credit rating. He concluded that “…investment at small firms may depend more strongly on bank health, as results from bank-dependent firms… suggest,” and “… perhaps the poor health of the banking sector is restraining investment by small firms.” 1/

Yoshikawa et al (1994) looked at the lending patterns of different groups of banks and tried to quantify the impact of the ratio of bad assets to total assets on these credit flows. Loans to bankrupt borrowers and loans on which interest payments were at least 6 months late were considered in the calculation of the bad asset ratio. Using data through September 1994, Yoshikawa found little effect of bad assets on the lending patterns of the 11 city banks or the 3 long-term credit banks. However, there were significant negative effects on the lending patterns of the 7 trust banks and the regional banks. As an additional step, Yoshikawa looked for industry-specific lending effects for the 21 major banks, and found that loans to the real estate sector in particular had been retarded because of higher bad asset ratios.

Maeda (1996) examined bank loan patterns for the 11 city banks and the 3 long-term credit banks for both loans to all firms and for loans to SMEs. Loan supply was specified as a function of a bank’s nonperforming loans as a ratio of total loans, the bank’s BIS capital ratio, the bank’s special loan loss provisions, the bank’s hidden reserves, the lending rate, and the deposit rate. Using data through April 1995, Maeda found that the ratio of nonperforming loans had a significant negative effect on bank lending. Overall, a 1 percentage point increase in this nonperforming loan ratio was shown to reduce overall bank lending by about 1 percent. For bank lending to SMEs Maeda showed a more negative effect of nonperforming loans—a 1 percentage point increase in the bad loan ratio was found to cause a 1.33 percent decline in loans to these smaller firms. Other explanatory factors had about the same effects on loans to all firms and on loans to SMEs.

4. Risks for the Japanese economy and policy perspectives

The first issue in determining whether the potential for a credit squeeze exists is to judge whether the overall scope of banking-sector problems in Japan is of a sufficient scale to raise concerns. Although the estimates of bad loan ratios in Japan cited earlier are certainly high by historical and international standards (on the order of 9 or 10 percent of total loans), Berg (1995) has argued that the lack of international definitions for bad loans makes it impossible to compare bad loan stock estimates across countries. Instead, he argues that the best way to estimate and compare the financial strain on a country’s banking system is to look at the flow of yearly loans losses and loan loss provisions. According to Berg’s experiences with the Scandinavian banking crises in the early 1990s, if these annual banking costs exceed 4 percent of total loans, a banking system is at high risk, and will likely face numerous bank failures, large-scale restructuring, forced mergers and consolidations, and heavy government bailouts. Berg notes that this 4 percent rule of thumb ratio can be related to the 8 percent Bank for International Settlements (BIS) capital adequacy ratio, half of which can be made up of subordinated capital, which he points out can be notoriously difficult to use in restructuring situations. That is, he argues that the 4 percent rule of thumb ratio provides a logical threshold of strain for financial institutions trying to remain solvent.

Given this background, it is noteworthy that in FY 1995, the largest 21 banks in Japan faced loan loss/loan loss provision costs of about 5 percent of total loans, slightly above Berg’s 4 percent “risk” threshold. 1/ Although actual loan loss provisions for the other banking sectors—the regional banks, shinkin banks, cooperatives, etc.—were somewhat lower that year, these banks in aggregate actually face somewhat higher bad loan to total loan ratios than the major 21 banks, suggesting that their true underlying loan loss/loan loss provision costs would be somewhat higher than for the major banks if they were moving aggressively to try to resolve their problems. In addition, there is the pressure that the BIS capital requirements impose upon banks. The experience of the United States in the early 1990s, where bad loan difficulties of a smaller overall scale did lead to the type of asset re-balancing away from loans and toward government securities described above, and did result in credit pressures, also must be taken into account. Altogether these factors suggest that there are good reasons to believe that the banking-sector problems in Japan are of a sufficient scale to raise concerns about a credit squeeze.

The second step is to try to make an overall assessment about the risks of a credit squeeze based upon the other evidence presented above. On balance, this evidence suggests that there is at least a moderate risk that some form of credit squeeze could materialize in Japan over the next couple of years. The risk is probably highest in the SME business sector. Already there are scattered signs that credit tightness has developed, although it is not clear that this tightness has had a noticeable retarding effect on Japanese investment spending to date. One sign is that SME firms do not yet indicate the degree of willingness of banks to lend that would have been expected at this stage of the business cycle. This is especially true given the sharp decline in interest rates that has occurred and the dramatic recovery in the perceptions by large firms about bank willingness to lend. Another sign is that the “external finance premium” that SME firms face has increased over the past couple of years above its expected historical differential with large-scale firms. Finally, most empirical studies that have examined the relationship between the lending behavior of Japanese banks and the quality of their loan portfolios find some evidence of a negative correlation between bad asset problems and credit supply. This seems to be particularly true for smaller borrowers. Studies that do not find such a linkage have tended to concentrate on data for larger Japanese firms that typically have access to the corporate bond market.

Gibson (1996), for example, concludes that the risks of a credit crunch in Japan appear to be quite low, based on the fact that with 1994-95 data he could not find a statistical connection between the health of large corporations’ main banks and their lending to those clients. But this conclusion was based upon a sample of the 1600 or so largest publicly-traded corporations in Japan, roughly two thirds of which were active in the corporate bond market. These firms collectively were responsible for just 46 percent of all fixed assets in Japan, 35 percent of all sales, and just 17 percent of all employment, suggesting that while they are important to the Japanese economy, they are not very representative of overall patterns and do not necessarily give a clear picture of the risks of a systemic credit squeeze. Although it would be difficult to assemble a consistent data set to comprehensively analyze the supply of credit to firms that are smaller than those in the Gibson data set, it seems likely that corporate bond financing is effectively unavailable to smaller firms, that they are much more dependent on the banking system for credit, and that they probably do face some credit constriction if their banks have a heavy bad loan burden. Importantly, SME firms as defined in this paper are responsible for roughly 40 percent of all investment spending in Japan and roughly three quarters of all employment. This suggests that rather than having only minor potential effects on the Japanese economy, a credit squeeze that did bind on the spending plans of these smaller firms could have noticeable macroeconomic effects.

a. Assessing the balance of risks

Whether or not a more full-blown credit squeeze could develop in Japan over the coming years depends upon the interplay of a number of factors, so it is appropriate to think about the “balances of risks” of a credit squeeze under various scenarios. It is useful to think about this “balance of risks” as being influenced by a competition between the balance sheet channel of credit transmission on one hand, which influences the demand for credit, and the bank lending channel on the other, which influences the supply of credit. Consider, for example, what would happen to these two channels if land prices continue to languish and the balance sheet positions of firms remain weak. In this case, investment demand and the overall pace of economic growth would tend to remain moderate, and the balance sheet channel of credit transmission would tend to keep downward pressure on credit demand and therefore keep the risks of a credit squeeze further fairly low. On the other hand, if land prices began to increase equities and investment demand picked up more smartly, the balance sheet channel would tend to put upward pressure on credit, and tend to increase the risks of a credit squeeze. The question then would be what would happen to the bank lending channel in these circumstances. Presumably higher land and equity prices would be good for bank health and encourage banks to increase their supply of loans. This could roughly compensate for the higher loan demand and leave the balance of risks of a credit squeeze unchanged, or it could over or under compensate. The biggest concern and biggest unknown in providing an overall risk assessment, therefore, involves knowing the whether bank lending can be maintained or increased in the face of heavy bad loan writeoffs.

Other factors could also influence whether or not there might be a credit squeeze. First, monetary policy could play a role. Continued low short-term interest rates would tend to boost economic activity and investment demand—factors that would tend to raise the risks of a credit squeeze—while at the same time tending to assist in the recovery of bank profits and providing a working margin for the repair of bank balance sheets. On balance, it is probably the case that the risks of a credit squeeze would tend to be lower the longer short-term interest rates remained low. Profit and cash flow trends could also work in both directions. Stronger cash flow would tend to boost business confidence and encourage investment spending, which would increase the risks of a credit squeeze, but it would also provide a good portion of the funding—in the form of retained earnings—for that additional spending. Given the high reliance of Japanese firms on retained earnings for investment spending, it seems likely that on balance, higher SME profitability would probably tend to reduce the risks of a credit squeeze.

A few factors would have clearer one-directional effects on the balance of risks of a credit squeeze. On the credit demand side, consider an unexpected increase in investment demand, say because of technological change and the need for the business sector to autonomously update its capital stock—say to add more computers and office equipment—to remain internationally competitive. In this case, the risks of a credit squeeze would tend to increase rather clearly. On the credit supply side, consider revelations that banking-sector problems were actually worse than currently recognized, that is, that a larger share of total loans were bad. This also would tend to work in the direction of increasing the risks of a credit squeeze.

b. If a credit squeeze were to materialize, how would it appear?

If a credit squeeze were to materialize, it likely would show up through the early warning signals described above. Larger proportions of businesses (especially SMEs) would report that bank credit was “tighter,” lending spreads by banks would widen, especially for SMEs, and investment spending would remain weaker than expected for the stage of the business cycle, especially spending by SMEs. The result would be lower aggregate demand coming from business investment and slower economic growth. It is sometimes argued that if a credit squeeze were to materialize, it might not be observable through interest rates. That is, a gradual credit constriction might lead to a gradual reduction in SME investment, but no sharp spike in interest rates. In fact, this could well be the way events would unfold—with no clearly visible financial-sector effects, but just a slower pace of growth than otherwise would have been the case.

c. Policy perspectives

If the bank lending channel of credit transmission were to be constricted and a credit squeeze were to develop, investment spending be reduced, especially at SMEs, GDP growth would tend to be slower than otherwise, and the output gap would close less quickly. These conditions would suggest that monetary policy be biased in the direction of ease. An easier monetary policy would also mean a lower cost of funds for banks than would otherwise be the case, and that would tend to boost bank profitability, help to repair bank balance sheets, and stimulate the bank lending channel. Easier monetary policy would also help to improve corporate balance sheets, and therefore have a positive balance sheet effect. Both movements would thereby help to restore economic growth.

Another policy consideration in dealing with a potential credit squeeze is to make sure that the financial system is sufficiently open and deregulated to minimize such risks in the first place. The more open and deregulated a country’s capital markets, the less likely it is that problems at a particular class of institution will disrupt the flow of credit to worthy borrowers. Even though the Japanese financial system has made much progress in recent years in deregulation, including the easing of restrictions on corporate bond market activities in early 1996, the bond market remains less active than say, the U.S. bond market. Further steps toward deregulation could be helpful in making corporate bonds available to smaller companies. Institutional reforms to promote a more active venture capital business in Japan would also help to reduce the risks of a credit squeeze, especially among SMEs.

There are also good arguments for financial authorities to better clarify responsibilities for bad debt so that healthier banks and institutions can get on with their lending operations, and not be saddled with undeserved handicaps. The emergence of the “Japan premium” over the past year, which added to the cost structure of relatively troubled, but also relatively strong and healthy banks to some extent, illustrates the problem that failing to resolve systemic banking sector imbalances can cause. Some observers would also argue that Japan is currently oversupplied with banking services, that a shakeout is inevitable, and that it is better to pare the weak performers from the banking system, reward the remaining healthier firms, and hasten the day of true banking sector recovery in Japan.

Finally, it is possible that the bank lending system in Japan is overly dependent on the use of land as loan collateral. It is probably desirable for the banking system to develop new ways to evaluate loans and make loans on other bases besides land. New ways of valuing business plans, patents, trademarks, and other business assets may need to be devised.

5. Conclusions

There is evidence that small- and medium-sized enterprises in Japan have been experiencing at least a mild credit squeeze over the past few years. Surveys of SMEs indicate that they are not experiencing the rebound of bank lending that typically would be the case for this stage of recovery, and interest rate spreads on loans to SMEs remain high by historical standards and high relative to the normal spread over loans to larger enterprises. Moreover, most statistical studies of the relationship between bank health and credit supply in Japan in the mid 1990s report some contraction in credit flows because of banks’ bad loan burdens. The evidence that firms are facing credit restrictions is stronger for SMEs than it is for larger companies. In fact, the main reason a few studies do not find evidence of some bank lending constraints is probably because they concentrate only on relatively large Japanese firms—firms that also have access to the corporate bond market.

Given the heavy bad loan burden on the Japanese banking sector, further credit restrictions may well become more apparent over the next couple of years, especially if the economic recovery begins to pick up faster than the 2 to 3 percent growth pace that most forecasters have been expecting for 1996 and 1997. The pressure on banks to meet the even tougher BIS capital requirements being imposed may reinforce what would probably have been a downward movement in desired bank loan supply. While larger firms will probably be able to shift their credit mix more toward the capital markets, and the bond market in particular, these markets are likely to remain beyond the reach of most SMEs. The result may well be that investment spending, especially spending by the important SME sector, may continue to under-perform relative to historical experience, and overall economic growth may therefore be more lackluster than normal. Policy makers need to monitor developments in the credit markets that SMEs operate in, including interest rate spreads and other measures of credit tightness. Among the policy steps that might be helpful are: a bias toward easier monetary policy to try to boost bank profits and overall banking-sector health, the promotion of measures to deregulate the capital markets, and efforts to clarify the responsibilities for bad loans problems, so that bad assets can be removed from the balance sheets of healthy institutions and they can resume more normal lending operations.

References

  • Berg, Jesper (1995), “The Nordic Banking Crisis: Lessons to be Learned,presented at an IMF Research Department Seminar, Washington, DC, July 21, 1995.

    • Search Google Scholar
    • Export Citation
  • Bernanke, Ben and Mark Gertler (1995), “Inside the Black Box: The Credit Channel of Monetary Policy Transmission,in Journal of Economic Perspectives. Vol. 9, No. 4, Fall, pp. 2248.

    • Search Google Scholar
    • Export Citation
  • Borio, Claudio (1995), “The Structure of Credit to the Non-Government Sector and the Transmission Mechanism of Monetary Policy: A Cross-Country Comparison,Basle: Bank for International Settlements Working Paper No. 24, April.

    • Search Google Scholar
    • Export Citation
  • Gertler, Mark and Simon Gilchrist (1994), “Monetary Policy, Business Cycles, and the Behavior of Small Manufacturing Firms,Quarterly Journal of Economics. Vol. 109, May, pp. 30940.

    • Search Google Scholar
    • Export Citation
  • Gibson, Michael (1995a), “Borrowing Patterns of Large and Small Listed Japanese Firms,” memorandum, December 26.

  • Gibson, Michael (1995b), “Can Bank Health Affect Investment? Evidence from Japan,” in Journal of Business. Vol. 68, No. 3, pp. 281308.

    • Search Google Scholar
    • Export Citation
  • Gibson, Michael (1996), “More Evidence on the Link Between Bank Health and Investment in Japan,Federal Reserve System International Discussion Papers Number 549, May.

    • Search Google Scholar
    • Export Citation
  • Maeda, Tsutoma (1996), “Slow Growth in Bank Lending in Japan: An Empirical Analysis of Banks’ Unwillingness to Lend,” (translated from Japanese),” in Japan Ministry of Finance Financial Review. Institute of Fiscal and Monetary Policy (March), pp. 13153.

    • Search Google Scholar
    • Export Citation
  • Smith, Todd (1995), “Markets for Corporate Debt Securities,IMF Working Paper 95/67, July.

  • Thakor, Anjan (1996), “Capital Requirements, Monetary Policy, and Aggregate Bank Lending: Theory and Empirical Evidence,” in Journal of Finance. Vol. 51, No. 1, March, pp. 279324.

    • Search Google Scholar
    • Export Citation
  • Utsunomiya (1995), “An Analysis of Banks Lending Interest Rate Spreads,” in 1995 Annual White Paper (Tokyo: Economic Planning Agency, December), pp. 142144 (in Japanese).

    • Search Google Scholar
    • Export Citation
  • Yoshikawa, Hiroshi, Masaru Eto, and Toshihiro Ike (1994), “Unwillingness of Banks to Lend to Small- and Medium-Sized Enterprises,” (translated from Japanese), Japan Economic Planning Agency, Economic Research Institute (March), pp. 143.2

    • Search Google Scholar
    • Export Citation
1/

This chapter was prepared by Robert Wescott.

2/

See, for example, the estimates by the International Banking Credit Agency in Table III.2.

1/

This view about the credit channel draws upon analysis by Bernanke and Gertler (1995) and others.

1/

Data for this analysis are from the Japan Development Bank.

1/

These shares add up to slightly over 100 percent of total net sources of funds because of statistical discrepancies in the national accounts. In the late 1980s, the sources of funds were significantly different. In 1989, for example, net equity issues were used to fund nearly 10 percent of all investment spending, and net bond issues were used to fund over 20 percent of such spending. Bank lending in 1989 supplied less than 30 percent of the funds for investment.

1/

Although this paper argues that constraints on bank lending may tend to limit overall investment spending, it must be recognized that the outlook for business profitability and cash flow is probably the single most important factor influencing such spending, and that movements in this outlook may tend to dominant movements in bank credit.

2/

Because financial reports for banks, credit unions, and other lending associations do not all use the same breakdown of firm sizes in their data, some of these data are not strictly comparable. These data are intended to present the approximate weights of credit sources to SMEs.

3/

This decline contrasts with the trend toward a rising capital market share of total credit in most industrial countries over this period.

1/

These firms with less than ¥2.3 billion in share capital were the 385 corporations ranked between the 1,155th largest and 1,540th largest on the Tokyo Stock market.

1/

Chow test results based upon a regression of the SME index of bank willingness to lend on the large firm index of bank willingness to lend provide evidence of a structural break in the first quarter of 1990 that was significant at the 1 percent confidence level.

2/

Again, this indicator is suggested by Bernanke and Gertler (1995).

1/

Gibson (1996), pp. 7-8.

1/

This total ratio may understate the seriousness of the problems at specific banks. In fact, some of the weakest banks may face lower figures, just because they are delaying in making loan loss provisions.

Japan: Selected Issues
Author: International Monetary Fund
  • View in gallery

    Small- and Medium-Sized Enterprises Share of Total Investment Spending

    (Percent)

  • View in gallery

    Investment Spending Over Recovery Cycles, Large Enterprises

    (4 quarter moving average, indexed to quarter indicated)

  • View in gallery

    Investment Spending Over Recovery Cycles, Small- and Medium-Sized Enterprises

    (4 quarter moving average, indexed to quarter indicated)

  • View in gallery

    International Banking Credit Agency (IBCA) Estimate of 1995 Bad Loan Percentage vs. Average 1993-95 Loan Growth, Various Bank Groups

  • View in gallery

    Judgement on Lending Attitude

    (Percentage points)