4 Explaining the Slump in Japanese Business Investment

Tamim Bayoumi, and Charles Collyns
Published Date:
March 2000
  • ShareShare
Show Summary Details
Ramana Ramaswamy

Business investment in Japan declined for five consecutive quarters through end-1998, with the decline for 1998 as a whole being about 11¼ percent. Despite the pickup in the first quarter of 1999, surveys of business investment intentions and monthly orders data point toward renewed declines later in the year. The decline in business investment was more precipitous than during any other episode in Japan’s postwar history; even with the recent increase, business investment has declined by almost 20 percent between the first quarter of 1997 and the first quarter of 1999.

The sharp decline in the rate of business investment since the bursting of the asset price bubble in 1990–91 has included both cyclical as well as structural elements, though the latter appears to have been the dominant driving force. The share of business investment in GDP fell from about 20 percent in 1990 to about 16 percent in 1998—with a transient recovery in 1996 spurred in part by new opportunities created by deregulation (Figure 4.1). The contrast between trends in business investment in Japan and in the United States since 1990 is striking. The share of business investment in GDP has risen in the United States by approximately the same magnitude as it has fallen in Japan in the 1990s. Consequently, while the share of business investment in GDP in Japan was almost 11 percentage points higher than in the United States in 1991, by 1999 it was only about 4 percentage points higher.

Figure 4.1.Trends in Japanese Investment

Sources: Nikkei Telecom; WEFA; and IMF, World Economic Outlook.

While the share of aggregate business investment in GDP has declined significantly in the 1990s, the disaggregated trends across sectors provide some interesting contrasts (Figure 4.2). The steepest declines in the rate of business investment in the 1990s occurred in manufacturing and construction—sectors that had expanded most closely in tandem with the buildup in asset prices in the latter half of the 1980s. Investment declined sharply in the immediate aftermath of the asset price collapse in the banking, insurance, and services sectors, but has picked up robustly since 1996. In both transportation and communications, and electricity—sectors in which deregulation initiatives have been prominent—the share of investment in output continued to rise even after asset prices collapsed in the early 1990s.

Figure 4.2.Sectoral Trends in Business Investment

(Rates of investment, 1990 prices)

Sources: Nikkei Telecom; WEFA, Nomura database; and IMF staff estimates.

This chapter aims to explain the reasons for the steep decline in the rate of business investment in Japan in the 1990s. The analysis is carried out in two stages. The first discusses the various alternatives proposed to explain the decline in business investment in Japan. This is followed by econometric estimates of the determinants of aggregate business investment in Japan. The main conclusions emerging from the regression analysis are that structural factors have been a dominant force driving the slump in business investment in the 1990s. In particular, the unwinding of the capital stock overhang from the 1980s and the impact of the debt burden have played prominent roles. The decreasing effectiveness of the system of corporate governance provided the milieu in which firms incurred large debts in pursuit of rapid market expansion at the expense of profitability, a strategy that backfired when growth expectations declined. The collapse in equity prices in the 1990s also contributed to the slump in business investment. While the regression analysis does not assign a prominent role to cyclical factors in accounting for the slump in business investment, other evidence indicates that business investment was constrained by a credit crunch during 1997–98.

Alternative Explanations of Business Investment

Some of the more prominent hypotheses put forward to explain the slump in business investment in Japan are that it constitutes the unwinding in the 1990s of the overinvestment in the latter half of the 1980s; that business investment was held back by the effects of an excessive buildup of debt; that the decline in business investment is related to a credit crunch arising from the interactions between the asset price collapse and an overextended banking system; and that the changing structure of the Japanese economy, including the trend decline in the growth of the labor force and deindustrialization, had a negative impact on business investment in the 1990s. This section discusses each of these hypotheses from the perspective of the relevant theoretical literature. This exercise provides guidance in selecting the specification used for estimating the aggregate business investment function in the following section.

The Overinvestment Hypothesis

In order to make a convincing case that the decline in business investment in the 1990s was a response to overinvestment during the 1980s, it is necessary to nail down the answers to the following set of questions. How is overinvestment to be measured? And what were the conditions that allowed firms to overinvest?

While there is no standard way for measuring overinvestment, it can nevertheless be deduced in a number of ways. A rising capital-output ratio, for instance, could indicate a declining response of output to investment, and a potential case of overinvestment. Drawing inferences about overinvestment from trends in the capital-output ratio, however, has to be done with care—for instance, the increase in the capital-output ratio could reflect the rising capital intensity of production accompanying certain phases of economic development, and not the declining efficiency of investment. One way of overcoming the interpretation problems associated with measures of the capital-output ratio is to compare how the ratio of capital to output has changed over time in relation to its own trend.1Figure 4.3 shows that the ratio of net business capital to potential output in Japan increased sharply in relation to trend in the latter half of the 1980s. Despite the prolonged decline in the rate of investment in the 1990s, this ratio is yet to fully revert back to trend, suggesting that overinvestment did take place in the 1980s and is yet to be unwound fully.

Figure 4.3.Indicators of Overinvestment

Sources: OECD; and Nikkei Telecom.

1Potential GDP is generated using a production function approach.

An alternative criterion for measuring overinvestment is the dynamic efficiency test proposed by Abel and others (1989), which compares trends in gross investment with those in gross capital income. In the steady state, if investment exceeds capital income, then the economy is clearly overinvesting, as capital accumulation is absorbing more resources than all past accumulation is making available for consumption. In the case of Japan, the share of gross operating surplus in GDP was significantly higher than the rate of business investment in the mid-1980s, suggesting that the dynamic efficiency criterion was being satisfied easily. However, the gap between the share of gross operating surplus in GDP and the rate of business investment narrowed sharply in the latter half of the 1980s, and continued at the compressed levels into the 1990s despite the sharp declines in the rate of business investment, suggesting a decline in the efficiency with which investment was being deployed (see Figure 4.3). This experience is also consistent with the trend decline in corporate profitability (Figure 4.4). Thus, the trends in the relationship between capital income and business investment in Japan also provide indirect evidence supportive of the overinvestment hypothesis.

Figure 4.4.Corporate Profitability, 1980–98

Source: Bank of Japan.

What were the driving forces behind the overinvestment in the latter half of the 1980s? During this period, high investment reflected a blend of exuberant expectations of growth and an easy access to finance, that in turn allowed growth expectations to be largely validated. Large firms, particularly those in manufacturing, invested heavily in equipment and buildings in this period by tapping into the buoyant equity market, and also by issuing convertible bonds. The booming stock market made both these sources of capital relatively cheap. Deprived to some extent of their traditional client base as a consequence of financial liberalization, banks turned to funding the investment plans of small and medium-sized firms, particularly those in the real estate sector, using overvalued collateral as the basis for such loans. The macroeconomic environment also abetted the high levels of borrowing. The official discount rate was halved in several steps to 2½ percent between end-1985 and early-1987, and then remained unchanged until mid-1989.

There were a number of reasons why the system of corporate governance in Japan failed to provide the necessary checks and balances against overinvestment. Despite some warning signals, both shareholders and banks may have been lulled by past successes into not constraining the corporate sector’s strategy of attempting to expand market share. Moreover, at times shareholders were not fully aware of the true health of firms, masked as they were by nontransparent accounting practices. The “keiretsu” system, involving close relationships—including those of cross-share ownership—between banks and industrial groups, offered firms a relatively cheap source of funds and favored strategies of market expansion, given the minimal threat of being taken over in case profitability was adversely affected.2 While banks could intervene when contractual obligations associated with routine payments of interest and principal were not met, they often tended in practice to bail out the firms in trouble rather than forcing them to restructure.3 Japanese firms were therefore able to implement their market expansion strategies without being too constrained by considerations of profitability. The investment boom finally petered out when monetary policy was tightened sharply in 1989–90, and the asset price bubble collapsed thereafter, resulting in a drastic reduction in access to cheap sources of funds.

The Debt Hypothesis

The high levels of indebtedness of Japanese firms has often been adduced as an important factor behind the slump in business investment as it amplifies the vulnerability of firms to negative shocks. Figure 4.5 shows that the buildup in debt was substantial during the latter half of the 1980s for a significant part of the corporate sector, with the notable exception of large manufacturing firms that financed capital formation largely through retained earnings and the issue of equity. While the link between high indebtedness and vulnerability to negative shocks is intuitive, however, the relationship between debt and investment is a complex one from an analytical point of view. The theoretical literature in this area indicates that there are both advantages as well as disadvantages to financing investment through debt, though, as discussed below, the Japanese experience appears to have been more one of being saddled with the disadvantages.

Figure 4.5.Japanese Corporate Indebtedness, 1980–98

Source: Bank of Japan.

Early models of the relationship between debt and investment—for instance, the Modigliani-Miller theorem—concluded that the source of financing should not have an impact on the investment decisions of firms. The Modigliani-Miller theorem is, however, predicated on a series of restrictive assumptions regarding the nature of capital markets, and recent developments in the theory of asymmetric information and imperfect markets have shown that financial policies of the firm do have an impact on investment.4

Developments in the theory of asymmetric information led to a line of research where it was argued that financing investment through debt rather than equity could in principle have advantages for firms. The existence of debt contracts allows firms to tap into funds that may not otherwise be available. For instance, since lenders cannot in general observe and monitor fully the actions of firms, risk-averse creditors would prefer noncontingent contracts that assure them of a regular stream of earnings and legal rights to principal, rather than enter into contingent contracts entailed by equity ownership. Thus, other things being equal, the existence of debt contracts allows investment to be higher than it otherwise would be, by making it possible to tap into the funds of risk-averse agents. Firms can also in principle fund their investments more cheaply with debt than with equity under certain circumstances. In cases where the “equity premium” arises more on account of asymmetric information rather than the “true” riskiness of the investment projects, it is cheaper for the firm to obtain finance through debt rather than equity.5 Also, as stressed by Jensen (1988) and Shleifer and Vishny (1988), another potential advantage of debt over equity is that it constrains firms with large cash flows, but low growth prospects, from diversifying into unconnected and wasteful projects because of the contractual obligation to set aside part of the cash flows for interest payments, and the potential threat of bankruptcy when contracts are violated.

While the asymmetric information literature argues that some debt is better than no debt, incurring high levels of debt can nullify its potential advantages. High levels of debt increase the sensitivity of the firm to adverse shocks, implying a higher probability of bankruptcy, and lowers the range of circumstances over which the firm can offer noncontingent contracts. Thus, debt financing will cease to be relatively cheaper than equity financing when a firm’s debt levels are already high. High levels of debt also magnify the sensitivity of firms’ net worth and equity values to negative aggregate shocks, thereby amplifying normal business cycle fluctuations. High debt has a particularly deleterious impact on activity during deflationary episodes, or more generally when inflation turns out to be lower than prior expectations, as the contract is fixed in nominal rather than real terms.

In the case of Japan, some of the potential advantages of debt financing for investment were negated, both on account of the high levels of debt that had been built up by the time a series of negative shocks hit the economy in the early 1990s, and by both inflation and growth thereafter being lower than were expected when many of the debt contracts were initially incurred. A sense of complacency based on past successes had undermined corporate governance, with firms continuing to expand production capacity despite declining profitability, and slippages in contractual obligations associated with debt tended at times to be tolerated. Moreover, given the weakened state of the banking system in the 1990s, banks that had provided much of the debt to the corporate sector were not in a position to exercise the necessary corporate oversight function—particularly by acting preemptively to deal with emergency problems—while insolvency law was not well suited to provide for corporate reorganization through the legal system.6

The Credit Crunch and Asset Price Collapse Hypotheses

The bank lending channel usually tends to get treated separately in the literature from the asset price channel as a determinant of investment. The impact of equity price fluctuations on investment is generally incorporated into investment functions through some variant of Tobin’s Q—defined as the ratio of the market value of equity to the replacement value of physical capital. The bank lending channel—that is, the one linking bank loans and activity—is assumed to have an impact on investment that is independent of the effects of asset price fluctuations, because asymmetric information or institutional factors render bank loans as imperfect substitutes for either bonds or equities.7

In the case of Japan, however, the effects of the bank lending and asset price channels on investment have in practice been significantly intertwined. The fact that banks in Japan have counted part of their hidden capital gains on holdings of equity as bank capital implies that fluctuations in equity prices have a direct impact on bank capital, and therefore on their capacity to lend. Disruptions to bank lending, in turn, have the potential to have strong adverse effects on investment in Japan because of the important role played by banks in the process of financial intermediation.

The collapse of asset prices in the early 1990s precipitated the problems in the banking sector. Equity prices fell by about 60 percent between 1990–92, and, despite the pickup in early 1999, are still more than 50 percent below peak levels. Land prices began declining from 1991 onward, with the average price of land in the six largest Japanese cities in 1998 being only about 40 percent of the peak values in 1990. As noted above, one effect of the collapse in asset prices was to erode bank capital and, therefore, their capacity to lend. The decline in asset prices, at the same time, also eroded the collateral of firms, making it riskier for banks to expand lending, even in situations where bank capital was not a binding constraint. Assessing the role of the credit crunch in holding back business investment in Japan involves examining the effects of the interactions among deteriorating loan collateral, declining bank capital, and the nature of the regulatory environment on business investment.

Recent research has come up with mixed evidence regarding the role of the credit crunch in constraining investment. Bayoumi (in Chapter 2 of this book) argues that while a strict credit crunch did not occur until 1997–98, part of the effects of the asset price collapse on investment in Japan was channeled through bank lending.8 One reason for the absence of stronger evidence of a credit crunch in a period when bank capital was being eroded by the collapse in asset prices could be related to the forbearance provided by the prevailing regulatory environment on the expectation that economic conditions would soon improve. The regulatory environment in Japan underwent a major strengthening from 1997 onward in response to the deepening of the banking crisis, and the realization that the underlying problems were not just cyclical. With new regulations guiding the operations of the banking sector, banks responded to the erosion of capital by cutting back lending sharply. A study by Woo (1999), for instance, found that there was a positive and statistically significant correlation between bank capital and lending growth in 1997, contrary to the experience in earlier years when the deteriorating capital positions of the banks failed to be a binding constraint on their lending activities.

The Changing Structure of the Japanese Economy Hypothesis

There are a number of strands to the relationship between the changing structure of the Japanese economy and business investment. One argument is that supply-side factors, such as the decline in the growth of the labor force in the 1990s, implied the need for a slower growth of capital stock in the steady state, and therefore warranted a downward transitional adjustment of investment in relation to GDP from previous levels (Figure 4.6).9 A closely related argument is concerned more with the future—that projected trends of a declining labor force will imply a relatively slower growth of capital stock and potential output in the years ahead. In as much as these projected trends have already been factored into growth expectations, they could offer a potential supply-side explanation for the trend decline in the rate of business investment. Another supply-side argument is that the sharp decline in the share of manufacturing employment associated with the rapid pace of deindustrialization in the 1990s, and the growing importance of the service sector, imply a lower steady state capital-labor ratio; the slump in business investment in the 1990s can be interpreted in part as the transitional adjustment to a lower equilibrium capital-labor ratio.

Figure 4.6.Labor Market Trends in Japan

Sources: Nikkei Telecom; WEFA; and IMF staff estimates.

While the decline in the trend growth of the labor force may have been one of the factors contributing to the slump in business investment, it appears unlikely to have been a major determinant. While the growth of the labor force did decline in the 1990s, this decline was not very pronounced—it grew on average at about 0.8 percent between 1990–98, compared with 1.2 percent between 1980–90, because slower population growth was partly offset by a rising participation rate. The decline in hours worked in the 1990s was sharper when compared with the previous decade; nevertheless, the difference, on average, was under 1 percentage point a year. In order for the labor supply explanation to hold, it is necessary to explain why a decline in the growth of the labor force, or of hours worked, of this magnitude ought to have led to a substantial slump in business investment in the 1990s.

The link between deindustrialization and the slump in business investment may have been a more important one for Japan in recent years than that between trends in the labor force and investment. All advanced economies have experienced a gradual trend decline in the share of manufacturing employment since about the early 1970s, driven in major part by the relatively faster growth of productivity in manufacturing.10 While Japan experienced a slower pace of deindustrialization until the early 1990s, its share of manufacturing employment declined sharply by about 3 percentage points to 21 percent between 1994–98—that is, it declined by about the same magnitude in the last 4 years as it did in the prior 20 years. The recent decline in the share of manufacturing employment in Japan was also relatively steep when compared with the pace of deindustrialization over any 4–5 year span in other industrial countries. Since manufacturing is relatively more capital intensive, the recent changes in the structure of Japanese employment may have accounted in part for the slump in business investment as part of the transition toward a lower capital-labor ratio.

Econometric Estimation

There has been, in general, less of a consensus over the appropriate econometric strategy for estimating aggregate investment functions than has been the case with some other macro relationships, in part due to the analytical controversies over the determinants of business investment. The empirical strategy adopted in this chapter is an eclectic one, combining both the neoclassical and Keynesian approaches to estimating investment functions, and is influenced by the issues that are particularly important to Japan, as discussed above.

The main purpose of the econometric estimation is to identify the extent to which business investment in Japan has been influenced by “structural” factors such as debt and capital stock overhangs, and the asset price collapse, as opposed to “cyclical” factors such as changes in output, interest rates, and credit. The general form of the proposed specification is:

I/Y = f(ΔY, K/Y, D/S, ΔC, Π, Q, Pκ/P, ρ/ω, Δr, T)

where I is real business investment; Y is real output; K is real business capital stock; D/S is the ratio of debt to sales; C is credit; Π is profitability; Q is Tobin’s Q (the ratio of the value of equity to the value of physical capital); Pk/P is the relative price of capital; ρ/w is the ratio of the cost of capital to the cost of labor; r is the real interest rate; and T is a time trend.

While economic theory broadly influences the choice of variables for the general specification of the investment function, the specific choice of the functional form adopted was based on pragmatic considerations as well. For instance, when it comes to choosing between incorporating the cost of capital or relative factor rentals in the specification, the choice of variable was based on the one that “fits” better, despite the fact that using the cost of capital instead of relative factor rentals in the econometric specification implies that only one of the first order conditions for profit maximization is utilized.

The debt ratio and credit are also used as explanatory variables, under the hypothesis that financial policies do matter for business investment, and that firms are liquidity constrained. However, given the complex nature of the theoretical relationship between debt and business investment that was discussed earlier, the debt ratio is divided by expected growth to get a measure of overindebtedness (D*). The rationale for this choice is that it is debt in excess of growth expectations rather than debt itself that should have a negative impact on business investment. The capital output ratio is used as an explanatory variable to capture the effects of the capital stock overhang on the rate of business investment. The specification incorporates Tobin’s Q to capture the impact of equity price movements on business investment. The changes in output, credit, and the interest rate are incorporated, as noted earlier, to capture the impact of cyclical effects on the rate of investment. For the empirical specification, profitability is defined as the ratio of recurring profits to sales. A trend term is included to proxy other structural influences on business investment.

There is a statistical conundrum in estimating aggregate investment functions for Japan in the post–oil shock period. Aggregate time series data for a number of variables exhibit significant changes in trend, reflecting the fact that there was an asset price bubble and the overheating of the economy in the latter half of the 1980s, followed by a collapse of asset prices and a prolonged period of stagnation of activity in the 1990s and, more recently, a severe recession. Consequently, time series data on variables such as the first differences of output and credit, as well as the rates of investment and profits, which can generally be expected on a priori grounds to be stationary, appear to be nonstationary over the sample period when formal tests are conducted.

One way to approach the econometric estimations is to adopt a purely statistical approach—that is, any variable that tests indicate to be nonstationary is differenced one stage further. However, while adopting such an approach would satisfy statistical criteria, it has the problem of ignoring potentially important levels of relationships. Moreover, differencing variables such as output and credit twice over is not useful for drawing meaningful economic inferences about the determinants of business investment. Consequently, the approach adopted in this chapter is to estimate the investment function in its economically meaningful form. This implies the inclusion of apparently nonstationary variables in the estimated equation. The regressions provide stationary residuals, however, indicating that the estimates are consistent.

Table 4.1 summarizes the results from estimating the preferred specification. Quarterly data are used, spanning the period from 1981 to end-1998. The dependent variable is the rate of business investment—defined as the ratio of real business investment to real output lagged one period (to take account of endogeneity issues). The estimated equation closely follows the general specification outlined above, except for using the change in the real interest rate (defined as the 3-month CD rate minus the 12-month change in the CPI) instead of the change in the relative factor rentals as one of the dynamic factors. The relative price of capital is dropped in the preferred specification, as the coefficient is wrongly signed in regressions that included it.

Table 4.1.OLS Estimates of Business Investment1(Dependent variable is ratio of business-investment to lagged GDP)
RegressorCoefficientStandard ErrorT-Ratio [Prob]
Κ/Y (-1)-0.20910.0462-4.5230[.000]
D* (-1)-0.00290.6134E-3-4.7946[.000]
Δr (-1)-0.00160.8299E-3-1.8729[.066]
π (-1)-0.40540.3119-1.2997[.199]
Q (-1)53.638817.26663.1065[.003]
S.E. of regression0.0069F-stat. F(8, 61)69.6022[.000]
Mean of dependent variable0.1636S.D. of dependent variable0.0207
Residual sum of squares0.0029Equation log-likelihood253.6422
Akaike Info. criterion244.6422Shwarz Bayesian criterion234.5240
Augmented Dickey-Fuller Tests of the Residual
DFADF (1)ADF (2)ADF (3)ADF (4)
Test statistic 2-3.3319-3.0960-2.6056-2.5812-2.9943
Note: The sample period is 1981:Q2 to 1998:Q3 for the main regression and 1982:Q3 to 1998:Q3 for the augmented Dickey-Fuller tests.

OLS—Ordinary Least Squares.

The 95 percent critical value for the augmented Dickey-Fuller statistic is -2.9062.

Note: The sample period is 1981:Q2 to 1998:Q3 for the main regression and 1982:Q3 to 1998:Q3 for the augmented Dickey-Fuller tests.

OLS—Ordinary Least Squares.

The 95 percent critical value for the augmented Dickey-Fuller statistic is -2.9062.

The coefficients for both the capital output ratio and the adjusted debt ratio are negative, and the estimates highly statistically significant in both cases. Thus, the regression results provide empirical support to the intuition that the debt burden and the capital stock overhang have had an important influence in explaining the collapse of business investment in the 1990s. The coefficient of Tobin’s Q is positive and the estimate is statistically significant at the 5 percent level, suggesting that the decline in business investment was also driven by the collapse in equity prices. The finding of a strong positive relationship between business investment and Tobin’s Q is not surprising given that large firms had financed a substantial part of their capital expenditures by issuing equity, and much of the listed equities are held by the corporate sector rather than by households.11 The coefficient for the profit rate is wrongly signed but is not statistically significant, implying that there is no well-defined relationship between business investment and profitability, and is consistent with assessments that Japanese firms followed a market expansion strategy without paying due heed to profitability considerations. The time trend is positive and highly statistically significant, as might be expected given the rising capital-output ratio. The estimated relationships remain stable when changes are made either to the lag structure of the explanatory variables, to the estimation period, or when specific dynamic factors are excluded from the regressions.

The regression results indicate that none of the cyclical factors—the change in output, the change in credit, and the change in the real interest rate—has a statistically significant relationship with business investment. The cyclical factors are not statistically significant even when the regression is estimated with the level of business investment as the dependent variable rather than the ratio of business investment to lagged GDP as the dependent variable. The absence of a statistically significant relationship between credit and business investment appears at first sight puzzling, particularly given the severity of the recent banking crisis. However, the anecdotal evidence points toward a credit-crunch only between end-1997 and end-1998, as the regulatory system was tightened and the system of “prompt corrective action” went into operation. Consequently, a regression estimated on aggregate data spanning the period 1980–98 may find it difficult to capture this particular episode of the credit crunch.


Business investment has declined steeply in the 1990s in Japan, and—as this Chapter has argued—the slump has been driven primarily by structural factors—the unwinding of the capital stock over-hang from the 1980s and the impact of the debt burden. Slippages in corporate governance provided the milieu in which firms incurred large debts and embarked on a strategy of market expansion. The collapse in equity prices in the 1990s also contributed to the slump in investment, both by making equity financing costly and by reducing the net worth of the corporate sector.


    AbelAndre B.N. GregoryMankiwLawrence H.Summers and Richard J.Zeckhauser1989“Assessing Dynamic Efficiency: Theory and Evidence,”Review of Economic StudiesVol. 56 pp. 120.

    BernankeBen S. and John Y.Cambell1988“Is there a Corporate Debt Crisis,”Brookings Papers on Economic Activity: 1Brookings Institution.

    BernankeBen S. and MarkGertler1995“Inside the Black Box: The Credit Channel of Monetary Policy Transmission,”Journal of Economic Perspectives (Fall) pp. 2748.

    FazzariSteven M.GlennHubbard and BrucePeterson1988“Financing Constraints and Corporate Investment,”Brookings Papers on Economic Activity: 1Brookings Institution.

    HoshiTakeoAnilKashayp and DavidScharfstein1991“Corporate Structure, Liquidity, and Investment: Evidence from Japanese Industrial Groups,”Quarterly Journal of EconomicsVol. 106 (February) pp. 3360.

    JensenMichael C.1988“Takeovers: Their Causes and Consequences,”Journal of Economic PerspectivesVol. 2 (Winter) pp. 2148.

    PrescottEdward C.1999“Some Observations on the Great Depression,”Federal Reserve Bank of Minneapolis Quarterly ReviewVol. 23:1 (Winter).

    RowthornRobert E. and RamanaRamaswamy1997“Deindustrialization: Causes and Implications” Staff Studies for the World Economic Outlook (Washington: International Monetary Fund).

    RowthornRobert E. and RamanaRamaswamy1999“Growth, Trade and Deindustrialization,”IMF Staff PapersVol. 46March (Washington: International Monetary Fund).

    SekineToshitaka1999“Firm Investment and Balance-Sheet Problems in Japan,”IMF Working Paper 99/111 (Washington: International Monetary Fund).

    ShleiferAndrei and VishnyRobert W.1988“Value Maximization and the Acquisition Process,”Journal of Economic Perspectives 2:1 (Winter) pp. 720.

    WooDavid1999“In Search of Capital Crunch: Supply Factors Behind the Credit Slowdown in Japan,”IMF Working Paper 99/3 (Washington: International Monetary Fund).

An upward trend in the real capital-output ratio (as has occurred in Japan) can also reflect the effects of a declining trend in the relative price of capital. For instance, with a Cobb-Douglas production function, capital income is a constant proportion of nominal output. Accordingly, if the relative price of capital falls, its real level will increase.

See Hoshi, Kashyap, and Scharfstein (1991) for a discussion of the Keiretsu system.

See Chapter 9 by Levy.

See, for instance, the discussion in Fazzari, Hubbard, and Peterson (1988).

For more detailed discussions of these issues, see Bernanke and Campbell (1988).

See Chapter 9 by Levy.

See, for instance, Bernanke and Gertler (1995) for a discussion of the analytical and empirical basis of the importance of the bank lending channel. Chapter 7 of this volume discusses financial intermediation and bank lending in the Japanese context in more detail.

Sekine (1999), using panel data, finds that small firms were credit constrained during the first half of the 1990s.

See Prescott (1999) for a discussion of the supply-side explanation of the investment decline in Japan.

For a more detailed analysis of deindustrialization, see Rowthorn and Ramaswamy (1997 and 1999).

The collapse of equity prices in Japan in 1990 was accompanied by a significant increase in volatility. Estimations with a bivariate VAR (not reported here for the sake of brevity) indicate that the increase in volatility had a negative impact on business investment. Thus, the dampening effect of the asset price collapse on business investment was likely amplified by the increased volatility.

    Other Resources Citing This Publication