Home > Announcements > Speeches and Statements > Speeches 2003 > Japan's Deflation and Policy Response

Japan's Deflation and Policy Response

Based on a speech given by Kazuo Ueda, Member of the Policy Board, at the Meeting on Economic and Financial Matters in Nara City, Nara Prefecture, on April 24, 2003.

April 24, 2003
Bank of Japan

Click on ko0304d.pdf (52KB) to download figures.



There is much confusion in popular discussion of Japan's deflation and associated economic problems. This confusion tends to arise from a failure to distinguish between three related, but different phenomena: the stagnation of the real side of the economy, the deflation of general prices, and the deflation of asset prices. The deflation of general prices has certainly persisted since the mid- or late 1990s, depending on the price index one looks at. However, the extent of the price decline has been mild. The cumulative decline in the consumer price index (CPI) since its peak in 1998 has been no more than about 3%. It is hard to believe that such mild declines in general prices have been the root cause of the stagnation of the economy.

Declines in asset prices since the 1990s in Japan have been as large as they were during the Great Depression. Both TOPIX and the price index of urban commercial land have plummeted by 70-80% from their peaks. The collapse in asset prices has had serious effects on Japan's financial system, and in turn, on the economy, including general prices. In this short paper I attempt to discuss the relationship between such deflationary forces.

In section one, I first discuss some salient features of the recent deflation in Japan. This section contains a quick survey of analyses of the causes of the recent deflation of general prices. In section two, I briefly summarize the behavior of the price level, and of nominal and real interest rates during the Great Depression. I also refer to the literature on the causes of the Depression, paying particular attention to the so-called debt deflation theory and the role of the negative financial accelerator. The discussion in this section provides a benchmark against which to evaluate Japan's deflation experience since the 1990s. This is done in section three, where I basically show that Japan in recent years has not seen as serious a debt deflation as that experienced in either Japan or the U.S. during the Great Depression. There is no evidence of a sharp rise in real interest rates and thus in the real debt burden as a result of the deflation in general prices. It is also clear, however, that any significant rise hereafter in the rate of general price deflation substantially raises the risk of throwing the economy into a deflationary spiral.

As stated above, it has been the deflation of asset prices, not that of general prices, that has generated serious negative effects on the balance sheets of borrowers and, over time, on those of lenders. Through this route a negative financial accelerator has set in, adding to deflationary forces in the economy. This process is described in section four. Finally, in section five, I turn to the discussion of the appropriate policy response to deflation. Debt deflation caused by deflation in the general price level can be cured by macroeconomic policy to stop deflation, and measures to address problems in financial intermediation. Japan's recent case is, however, a difficult one as the deflation of general prices has not been the root cause of these problems. Moreover, asset price deflation was a natural market response to the bubble of the late 1980s. In addition, there is some evidence that the return on capital has been on a secular declining trend since sometime in the 1980s, giving support to the view that "reforms" are a key to sustained recovery. Worse still, the failure to address financial system instabilities at an early stage has substantially reduced the effectiveness of macroeconomic policy in easing even general price deflation. Such a complicated chain of events has not been well understood and has led to confusion in popular discussion on "deflation." Anyway, one thing seems clear, namely, the need to address the financial system problems as soon as possible.

I. Japan's Deflation since the 1990s

The behavior of Japanese final goods prices has been fairly stable since the early 1990s. Figure 1 shows movements of the CPI and the GDP deflator.1 The average annual rate of change in the indexes is 0.3% and -0.75% respectively, over the period of 1992-2002. The larger decline in the GDP deflator reflects the large secular decline in the deflator for investment resulting from technological improvements, and also its nature as a Paasche-type price index based on a commodity basket that changes over time. The two indexes have been falling since the mid- and late 1990s respectively. At the same time, however, there is at present no clear tendency for the deflation to accelerate.

A more disaggregated look at price developments turns out to be useful. Figure 2 presents cumulative inflation and output growth by industry for the period 1990-2000. The data points seem to lie on a downward-sloping schedule, suggesting the importance of supply side forces as determinants of cross sectional differences in the rate of price changes. Among the components of the CPI, although the data are not presented, the goods component has been falling faster than the others. Among services, those sectors that have experienced significant deregulation such as transportation and communications have seen larger declines in the rate of inflation. Deregulation in the non-manufacturing sector has certainly been an important part of the background for the deflation of general prices. A closer look at the goods component of the CPI reveals that import-competing goods have suffered larger price declines than the rest, as shown in Figure 3. This again lends support to the view that supply side forces have been important.

Turning to demand side factors, one can immediately point to the possible existence of a large GDP gap as the dominating force behind the deflation of general prices. Most estimates of the GDP gap are very large. For example, assuming 2% growth in trend output and the absence of a gap in the early 1990s, the GDP gap for 2002 must be larger than 10%. This is, however, hard to reconcile with the mild deflation of about 1% and the absence of a tendency for this deflation to accelerate. One needs to argue that the gap is much smaller (the growth rate of trend output is much lower), or that the effect of the gap on prices has been very small. In addition, one needs to rely heavily on supply side forces as a determinant of inflation.

Hirose and Kamada (2002) presents one estimate of trend growth and the GDP gap. It shows that the trend growth rate is now around one percent, much lower than in other literature on the subject. It also presents a statistical analysis pointing to the importance of low-price imported goods as a contributing factor to the deflation since the mid-1990s. Using time series analysis, Kamada and Hirakata (2002) attempts to decompose the causes of changes in the rate of inflation into demand side and supply side factors. They find that during the recent period of deflation supply side factors such as international comparative advantage shocks have played dominant roles.

It seems fair to say that more analyses and, perhaps, more data are necessary to determine the relative contributions of supply side versus demand side factors. The seemingly small effect of cyclical factors on inflation remains a puzzle.

In contrast to the behavior of general prices, the volatility of land and stock prices during the last two decades is worth pointing out. TOPIX went up by almost 400% between 1980 and 1989 and has fallen by about 70% from its peak. Similarly, the price index of urban commercial land in six large cities rose by almost 500% between 1980 and 1992 and has declined by 85%. Moreover, as of March 2003, i.e., more than ten years since its peak, it is still not clear whether they have reached bottom.

Asset price volatility has been as high as it was during the Great Depression, but has not been accompanied by volatility in general prices. In fact, this represents a shared feature in the experiences of many industrialized economies with respect to their stock market booms-and-busts since the mid-1990s. The asymmetry between asset and general prices is surely an important topic for future study.

  • 1  In this paper, the CPI and GDP deflator have been adjusted for the effects of the 1997 hike in the consumption tax rate. Specifically, the rate of change in the indexes has been adjusted downward by 1.5% for 1997 and by 0.5% for 1998 to undo the effects of the tax change.

II. Why Is Deflation a Problem?

Deflation of general prices, if unanticipated, creates a transfer from debtors to creditors by raising the real interest rate (ex post). Even an anticipated deflation raises the real interest rate, if nominal rates are at the zero bound and cannot be reduced further.

To the extent that debtors have higher propensities to spend out of income than creditors, such transfers reduce aggregate demand, adding to deflationary forces in the economy.2 In addition, under asymmetric information, banks may reduce lending in response to a decline in the net worth of debtors, setting in motion a negative financial accelerator process. Accelerator effects become more serious if financial institutions also suffer from balance sheet problems.3

Examples of serious debt deflation can be found in the experiences of industrialized countries in the 1920s and 1930s. Figure 4 shows Japan's call market rate, the rate of change in the GDP deflator, and the real interest rate (defined by the difference between the two). Deflation exceeded 10% and the real interest rate 15% in the early 1930s. As a result, the debt burden of borrowers rose sharply. For example, net interest payment relative to cash flow rose from about 80% in 1929 to more than 200% in 1930. As is well known, a similar pattern of movement in the variables is found for the U.S. in the 1930s. In addition, as Bernanke (1983) documents well, the debt deflation was exacerbated by the decline in the economy's ability to carry out financial intermediation.

  • 2  See I. Fisher (1933) or M. King (1994).
  • 3  See, for example, Bernanke (1983) and Bernanke and Gertler (1990).

III. Japan's General Price Deflation since the 1990s and the Debt Burden of Borrowers

Japan since the 1990s is nowhere close to the U.S. or Japan of the 1930s in terms of the impact of general price deflation on the debt burden. Figure 5 plots the real interest rates faced by major borrowers, non-financial firms and the central government. The real interest rates are calculated as gross interest payments divided by total debt minus the rate of increase in the deflator for domestic demand. As may be seen, the real interest rates have declined slowly since the mid-1990s. Of course, one could say that we would have liked to see much lower real interest rates to stimulate aggregate demand.4 There is, however, at least no evidence that deflation has increased real interest rates. A gauge like interest payments relative to cash flows tells the same story. For non-financial corporations this ratio has been falling steadily since the early 1990s. It is now around 12%, compared with a level of more than 40% in 1991 and 1992.5

Several caveats have to be kept in mind here. First, the zero bound constraint will become increasingly serious if the rate of deflation rises from here. Nominal interest rates are virtually zero on debt instruments with less than a year to maturity. Even on those with longer horizons, rates are not far from zero. For example, as of April 2003, the rate on five-year government bonds (JGBs) is around 0.25%; that on the ten-year JGBs is around 0.6%. Obviously, the room for downward adjustment is limited. Consequently, it is very important to avoid an acceleration in the deflation of general prices.

Second, some debtors are in trouble. A typical example is the banks. Given that most of the banks' liabilities are short term, it would come as no surprise if banks were suffering heavily under deflation and the zero bound on nominal interest rates. Figure 7 shows the margin between lending and borrowing rates for Japanese and U.S. banks along with their credit cost ratios, i.e., bad loan write-offs and provisions relative to total loans outstanding. Although margins are much lower in Japan than the U.S., they have not been shrinking. In this sense, the deflation of general prices and the zero bound have not bitten into banks' profitability. However, the figure also shows that these margins have not covered credit costs in some recent years. This is the non-performing loan (NPL) problem and we will return to it shortly.

  • 4  One reason for the small declines in real interest rates is shown in Figure 6, where the spread between the interest rate paid by non-financial firms and the 10-year government bond rate is plotted. The spread has clearly risen since 1997. This is, however, not directly due to the deflation of general prices, but, as we discuss in the next section, to declining borrower creditworthiness and the increased cost of financial intermediation.
  • 5  Of course, deflation decreases this variable by lowering nominal interest payments faster than cash flows. This effect is offset by a rise in the real value of the principle. In order to see the net impact one needs to look at the real interest rate, which is what we have in Figure 5.

IV. The Negative Financial Accelerator in Japan since the 1990s

Without doubt the sharp fall in asset prices has been the major reason for the recent instability in the Japanese financial system. Less clear is the causality between the financial system problems and the deflation of general prices. To shed some light on this issue, let us first look at Figure 8, where the relationship, by industry, between inflation and the degree of seriousness of the NPL problem is shown. The figure clearly reveals that the lower the rate of inflation in an industry, the less serious is the NPL problem for that industry. Although a correct interpretation of the relationship in the figure requires further research, the relationship is evidently at odds with the view that the deflation of general prices has been the cause of the NPL problem.

In Figure 9 we show the relationship, by industry, between NPLs and the extent of land holding (land as a share of total assets) at the peak of the bubble. Assuming that the real estate observation does indeed contain significant information in this regard, there would seem to be a positive relationship between the two variables. That is, the larger the land holding, the more serious the NPL problem, providing evidence of causation running from asset price deflation to NPLs.6 One might wonder why NPLs have not disappeared after a decade of bad loan write-offs and provisions. The total amount of money banks and the government have spent on NPLs amounts to about 20% of GDP. Figure 10 provides some clue as to why. The figure presents the evolution of bank loans by industry since the late 1980s. Loans to "bubble industries" such as real estate and construction did not begin to decrease until the late 1990s. Only loans to non-banks began to decline in the mid-1990s as public discussion and the government's handling of NPLs focused on this industry. Thus, there was clearly reluctance on the part of banks to dispose of NPLs swiftly, leading to forbearance lending and to much larger NPLs later.

I hasten to add that the recent stagnation in the economy, especially its weakness since 2001, is adding to the NPL problem. Although official bad loans are still dominated by loans to bubble industries, loans requiring caution are of a wider variety and seem to reflect the weakening of the overall economy.

Researchers have had difficulty obtaining solid results on the effects of financial instability on the economy. With the accumulation of data and use of fine techniques, however, the literature is finally obtaining some interesting findings in this area. Nagahata and Sekine (2002) carries out a time series cross section analysis of the determinants of business fixed investment. Along with other determinants, their analysis examines the importance of the balance sheet conditions of both borrowers and their main banks. They find that the deterioration of borrower balance sheets has had significant negative effects on investment. They also find that the deterioration of lenders' balance sheets has exerted significant negative effects on the investment of firms without access to the bonds market. The deterioration of balance sheets can be mostly explained by declines in asset prices and by NPLs (in the case of banks). Moreover, most of the declines in bank lending since the mid-1990s can be attributed to these two factors, together with the liquidity problems of banks during 1997-98. Thus, a negative financial accelerator has clearly been working.7

The negative effects of financial instability spread throughout the economy during the credit crunch in 1997-1998. The Asian economic crisis, a premature tightening of fiscal policy in 1997, and the Russian crisis in 1998 were the triggers for the crunch. Several financial institutions went under. Risk premiums and the demand for liquidity rose sharply across the financial system. Japanese banks, already suffering from NPLs, were now facing difficulty in raising funds. Naturally, they started calling in their loans to non-financial companies. Even large companies were feeling the pressure of the credit crunch. Many said in our interviews that all but the main bank are saying that they will not be able to roll over loans at the year's end. It followed that businesses had to cut back on their investment. In retrospect, the failure to have resolved the NPL problem at an early stage resulted in the credit crunch and has become one of the reasons for the stagnation of the economy since.

  • 6  See Ueda (2000) for a more careful analysis of this point along with the discussion of other causes of the NPL problem.
  • 7  One could say that a similar mechanism was working during the Great Depression, which was only aggravated by the deflation of general prices.

V. Policies to Deal with the Problems

The prescription for 1930s' type debt deflation has been proposed by, among others, Bernanke and Gertler (1990). The financial accelerator problem may be dealt with by transferring income to debtors with promising projects, at least to the extent that such projects and borrowers are identifiable. The same thing may be said about banks. Macroeconomic policy can take care of general price deflation.

The case of Japan since the 1990s is more complicated. As I have hopefully made clear, the deflation of general prices has not been the major problem. Instead, asset price deflation and its interaction with the financial system and the economy have been at the core of the problem. There are two views on the asset price deflation of the period. One view holds that the bubble of the late 1980s had to burst. The other recognizes a secular decline in the return on capital starting sometime in the late 1980s and associated declines in asset prices and investment.8 The two are not inconsistent with each other, but mutually reinforcing. The second view seems to need the first to explain the temporary rise in asset prices. The two differ in their estimates of where prices will end up, but are in accordance in claiming that adjustments in asset prices and investment were inevitable. To be sure, the negative accelerator story suggests that the economy and asset prices may fall further than is necessary. Hence, beyond a certain point, policies to support asset prices may be justifiable. It would, however, be difficult to determine whether the economy had reached such a stage. Also, a more fundamental policy response would be to address the problems of the financial system and other inefficiencies in the economy.

The role of macroeconomic policy is also not straightforward. Stopping the deflation in final goods prices will surely ease the pain of the type of adjustment I have just described, but it does not mean that the economy can dispense with the need for adjustment. Raising the rate of inflation of general prices by a couple of percentage points would not mean much for asset prices.9 In fact, it is the ratio between asset and general prices that has had to adjust.10 Moreover, as I explain below for monetary policy, the problems in the financial system have lowered the effectiveness of macroeconomic policies in stimulating the economy.

Focusing now on monetary policy, I show in Figure 11 the difficulty the Bank has been facing in combating deflation. The ratio of nominal GDP to base money shown in the chart displays a sharp downward deviation from its trend prior to around 1995, with the ratio roughly halving since then. Yet deflation has persisted, albeit at moderate rates. This is a clear case of monetarism failing to explain the relationship between money and inflation. Possible reasons behind this are suggested in the same figure. The rate of growth of bank loans has been either around zero or negative during the same period, in line with financial sector problems. As I mentioned above, this has reduced the ability of low interest rates to stimulate the economy. Moreover, as the chart makes clear, short-term interest rates hit the zero bound during this period, depriving the Bank of further room to use standard monetary policy.11

Given the nature of the problem, it would be nice if measures could be found that address financial sector problems as well as macroeconomic deflationary forces. The Bank, in fact, has been adopting such steps. In an attempt to ease corporate financial problems, the Bank has been accepting commercial paper, corporate bonds and some asset-backed securities as collateral when carrying out its operations to supply liquidity. In the last monetary policy meeting the Bank decided to study the possibility of buying some asset-backed securities outright in order to strengthen such efforts. Since late last year, the Bank has been buying equities from banks, helping them to reduce equity holdings to below TIER-I capital, as dictated by law. Its intention has been to ease the pain of asset price adjustment felt by the banking system without necessarily supporting asset prices. We would like to try harder to find other measures to achieve the same end.

  • 8  See, for example, Hayashi and Prescott (2002) for an expression of such a view.
  • 9  This could be a debatable point. A rise in the rate of general price inflation will leave asset prices unaffected if the real interest rate remains unchanged. If, however, the zero bound on nominal rates were to have prevented the real interest rate from falling, the real rate would decline as inflation rises. Discussions in section 3 suggest that evidence is mixed on this point. There is no clear sign that the real interest rate has risen because of deflation. But it is also difficult to refute the possibility that the real rate is higher than it should be.
  • 10 The ratio of asset prices to the GDP deflator has declined sharply since the peak around 1990 and is now back at where it was around 1983 for both stock and land prices. Thus, a fair amount of adjustment has already taken place. As pointed out above, however, it is difficult to determine whether the process is over, incomplete or has gone too far.
  • 11 See Ueda (2001),(2002) for more detailed accounts of monetary policy during the period.


  • Bernanke, B. (1983) "Nonmonetary Effects of the Financial Crisis in the Propagation of the Great Depression," American Economic Review, Vol. 73, 257-276.
  • ______ and M. Gertler (1990) "Financial Fragility and Economic Performance," Quarterly Journal of Economics, 87-114.
  • Fisher, I. (1933) "The Debt-Deflation Theory of Great Depressions," Econometrica, Vol.1, October, 337-357.
  • Hayashi, F. and E. Prescott (2002) "The 1990s in Japan: A Lost Decade," mimeo.
  • Hirose, Y. and K. Kamada (2002) "Estimation of Japan's Potential Growth Rate Assuming Variable NAIRU (in Japanese)," Working Paper 02-8, Research and Statistics Department, The Bank of Japan.
  • Kamada, K. and N. Hirakata (2002) "Import Penetration and Consumer Prices," Working Paper 02-1, Research and Statistics Department, The Bank of Japan.
  • King, M. (1994) "Debt Deflation: Theory and Evidence," European Economic Review, Vol.38:419-445.
  • Nagahata , T. and T. Sekine (2002) "The Effects of Monetary Policy on Firm Investment After the Collapse of the Asset Price Bubble: An Investigation Using Japanese Micro Data," Working Paper 02-3, Research and Statistics Department, The Bank of Japan.
  • Ueda, K. (2000) "Causes of Japan's Banking Problems in the 1990s," in Crisis and Change in the Japanese Financial System, Hoshi, T. and H. Patrick (eds.), Kluwer Academic Publishers.
  • ______ (2001) "Japan's Liquidity Trap and Monetary Policy," speech given at the meetings of the Japan Society of Monetary Economics
  • ______ (2002) "The Transmission Mechanism of Monetary Policy Near zero Interest Rates: The Japanese Experience, 1998-2000," in Monetary Transmission in Diverse Economies, Mahadeva, L. and P. Sinclair (eds.), Cambridge University Press. The original draft is available.