This Selected Issues paper analyzes the key features of the Japanese business cycle, and investigates whether the current recovery differs from past recoveries. In particular, this paper poses the following questions: what are the main characteristics of Japanese business cycles since 1980, and what happens to output, expenditure components, and prices over the cycle? The paper reviews the recent performance and policy issues with respect to the banking sector, insurance sector, and public financial sector in Japan. The stability of the financial sector is also assessed.

Abstract

This Selected Issues paper analyzes the key features of the Japanese business cycle, and investigates whether the current recovery differs from past recoveries. In particular, this paper poses the following questions: what are the main characteristics of Japanese business cycles since 1980, and what happens to output, expenditure components, and prices over the cycle? The paper reviews the recent performance and policy issues with respect to the banking sector, insurance sector, and public financial sector in Japan. The stability of the financial sector is also assessed.

I. Key Features of the Japanese Business Cycle: Is the Current Cycle Different?1

A. Introduction

1. Economic activity in Japan has been gradually expanding since the end of 2001. Prior to this expansion, Japan suffered from two transient recoveries during the period that followed the massive collapse of the asset-price bubble in 1991, which eventually led to deflation. At present, many commentators are arguing that the structural adjustments implemented thus far may be paving the way for Japan to return to a more sustainable economic growth path. The current recovery has gained pace since late 2003 and broadened, and the consensus near-term outlook is for continued expansion. Nevertheless, in light of the experience of the two previous recoveries, the durability of the current expansion is an open question.

2. The research presented in this paper documents the key features of the Japanese business cycle and investigates whether the current recovery differs from past recoveries. In particular, this paper poses the following questions: First, what are the main characteristics of Japanese business cycles since 1980? And, what happens to output, expenditure components, and prices over the cycle? Second, have the characteristics of these cycles changed over the years? Third, how does the current cycle differ from previous cycles? Fourth, is this recovery likely to be more durable than past recoveries?

B. Key Features of Japanese Business Cycles

What do Japanese business cycles look like?

3. Since 1980 Japan has experienced four complete business cycles and is currently in the recovery phase of a fifth cycle.2 A business cycle is defined as recurrent sequences of contractions and expansions in the level of economic activity. A complete business cycle is defined as running from the peak of economic activity to the next peak. Each cycle is split into two phases: recession and expansion. The recession phase starts at the peak and ends at the trough while the expansion phase begins at the end of the trough and ends at the next peak. The table provides dates for the five Japanese business cycles that have been identified since 1980, three of which have occurred since the asset-price bubble burst in 1991.

Description of the Japanese Business Cycle

article image
Source: Cabinet Office and Staff estimates.

4. A typical business cycle in Japan lasts approximately 5½ years, beginning with a recession of about 2½ years and followed by an expansion of 3 years (see table above). However, the overall duration of the Japanese business cycle seems to be shortening.3 In the last two cycles the recession phase became shorter, lasting 2 and 1½ years, respectively. Furthermore, in the last complete cycle the expansion was also shorter. Typically, clusters of short business cycles occur in the context of relatively high labor and product market rigidity (IMF 2002). Against this background, it is not surprising that Japan, with its problems of structural rigidities, has experienced three cycles since 1993.

5. In contrast to the Japanese experience, the typical or average business cycle for industrialized countries lasts about six years.4 These cycles begin with a recession of about one year, during which output falls, and is followed by an expansion of 5 years, during which output grows.5 To illustrate, in the United States—based on the National Bureau of Economic Research business cycle dating committee—over the period 1980–2003 the average recession phase is only 3 quarters and the average expansion phase is around 24 quarters. Furthermore, business cycles in most industrialized countries have become longer, reflecting longer expansions. The average length of business cycles increased from about four years during the 1970s to about six years during the 1980s and 1990s (IMF 2002).

What happens to output over the cycle?

6. An important characteristic of a business cycle is what happens to economic activity (in particular, output and prices). For output behavior a number of different methods can be used to describe the key features including: the overall trend in real GDP, the differences in growth profiles especially over the two phases of the cycle, and the extent of output loss (the amplitude of the cycle).

7. Figure I.1 shows that the trend in real GDP differs across cycles. The general upward trend in real GDP slowed after the asset-price bubble burst in the 1990s. During the first two complete business cycles, the trend in real GDP continued upward over both phases of the cycle. The fact that growth was positive, even in recessions, indicates that these episodes were “growth recessions.”6 For the last three cycles, the general trend in real GDP slowed and, in particular, was mostly negative during the recessions.

Figure I.1.
Figure I.1.

Economic Activity

Citation: IMF Staff Country Reports 2004, 247; 10.5089/9781451820577.002.A001

Source: Cabinet Office.

8. The comparison of the profile of real GDP growth, divided by the two phases, indicates that differences in real GDP growth across cycles are more pronounced in recessions (Figure I.2). During the first two recessions growth continued, during the third recession real GDP was broadly flat, whereas in the current and previous recessions real GDP declined. In fact, the recession of the current cycle appears to have been the shortest with the most severe decline in output. In contrast, the lower panel of Figure I.2 shows that GDP growth profile for all expansions tends to be similar, with the main notable difference being the length (duration) of the expansion.

Figure I.2.
Figure I.2.

GDP Path During Recessions and Expansions

Citation: IMF Staff Country Reports 2004, 247; 10.5089/9781451820577.002.A001

Source: Cabinet Office and staff calculations.

9. A further means to characterize the phase (cycle) is a measure of its deepness (amplitude), which is captured by the change in GDP. This is equivalent to measuring the extent of output loss (gain) in recession (expansion). The table shows that recessions have become more severe over time. In particular, during the last recession real GDP declined 2.7 percent, while in the first recession output rose 7.6 percent. Differences in output movement have occurred during expansions, owing in particular to the differences in the length of those expansions.7 Output gains have fluctuated widely during expansions, ranging from 5 percent in the fourth expansion to nearly 25 percent in the second expansion. The increase in output in the current expansion is 7.6 percent; after eight quarters it is somewhat below the 5-cycle average of nearly 11 percent.

Output Loss in Terms of GDP1

article image
Source: Staff estimates.

Measured as movement in output in percent change.

Current duration is eight quarters.

10. Another way to calculate output loss is from the perspective of the percentage shortfall of industrial production (IP), measured from the starting peak. While the behavior of IP is similar to GDP, it does not mirror its movement completely (lower panel Figure I.1). The average output loss for the recession phase of the five cycles was 33 percentage points.8 There are, however, considerable differences among the cycles. For instance, for the first two business cycles industrial production hardly fell over the recession, declining only 3 percentage points below the initial level of IP by the end of the recession. For the last three recessions, output fell by more than 50 percentage points. Not surprisingly, output loss also varied widely for the complete cycle, with production increasing in the first two cycles and declining in the last three.

Output Loss in Terms of Industrial Production1

article image
Source: Staff estimates.Note: Output loss (measured in percentage points) is the sum of the percentage shortfall of industrial production from its peak level in each quarter between the peak and the return to peak.

11. The analysis of output behavior suggests the following key features of Japanese business cycles:

  • Recessions have become more severe over time. The two recessions in the 1980s were growth recessions, where output did not decline. The last recession was the shortest with the largest decline in output;

  • Expansions tend to display similar profiles, with the main difference in overall growth stemming from the duration of the expansion.

How do expenditure components move over the cycle?

12. Another key feature of the business cycle is the behavior of expenditure components over the cycle. As in the case of output, the behavior of the components of aggregate demand can be examined by looking at differences in their profile over the two phases and by considering each components contribution to growth.

13. The profile of investment differs among recessions as well as between recessions and expansions. For instance, investment declined sharply in the last two recessions, as it does in virtually all recessions in industrialized countries (see IMF 2002), while it was either stagnant or continued to grow at a more moderate pace in the first three recessions (Figure I.3). This robustness in investment even during recessions possibly foreshadows some of the subsequent problems of over-investment that were experienced in Japan. During expansions, investment behaved similarly across the cycles; it tended to be relatively flat at first and then increase steadily. Figure I.4 (left panels) shows that the behavior of consumption has been similar across both phases of the cycle and more or less mirrored the behavior of output. The only exceptions are stronger than “normal” consumption in the expansion phase of the second cycle and the large drop in consumption in the last quarter of the third cycle. Public demand has tended to move countercyclically, by increasing more rapidly during recessions and rising modestly in expansion (right panels). A notable exception has been the drag of public demand in the last expansion (see further details below).

Figure I.3
Figure I.3

Investment Path During Recessions and Expansions

Citation: IMF Staff Country Reports 2004, 247; 10.5089/9781451820577.002.A001

Source: Cabinet Office and staff calculations.
Figure I.4.
Figure I.4.

Consumption and Public Demand During Recessions and Expansions

Citation: IMF Staff Country Reports 2004, 247; 10.5089/9781451820577.002.A001

Source: Cabinet Office and staff calculations.

14. The contribution to growth of each expenditure component—divided according to the two phases of the business cycle—also reveals the differences these components have played over the business cycle. During recessions, the major sources of the decline in output were from private investment—fixed and residential—and inventories. During the expansionary phase, private consumption and private investment are important contributors to growth in output. Over the last two cycles, however, the contribution of consumption has become slightly weaker perhaps owing to increased uncertainty and higher unemployment.

Contribution to Growth

article image
Source: Staff estimates.

How do prices behave over the cycle?

15. Another important feature of the business cycle is the movement of prices over the cycle. However, the relationship between prices (goods and asset prices) and business cycles is quite complex and has been extensively debated in the literature.9 Theory predicts that movements in prices over the business cycle depend on the relative importance of different shocks that drive the cycle.10 For instance, a recession induced by monetary tightening may be associated with an appreciation of the exchange rate and/or a decline in prices, while an exogenous contraction in aggregate demand may be associated with an exchange rate depreciation and/or decline in prices.

16. Despite these difficulties, several empirical regularities relating to movement of asset prices over the phases of the cycle emerge. These include:

  • The exchange rate tends to appreciate during the recession phase of the cycle—except for the current cycle—and there is no particular tendency during expansions.

  • Stock prices tend to decline during recessions and increase during expansions, except for the current expansion.

  • Interest rates (short- and long-term) tend to decrease during recessions and there is no particular tendency during expansions.

  • Land prices declined during both phases of the cycle, after the bubble burst.

  • Inflation (based on CPIs) rose modestly in both phases until deflation set in the late 1990s, and have fallen since then.

The movement in prices is mixed; they tend to behave similarly during recessions and less consistently during expansions.

Changes in Prices Over the Cycle1

article image
Source: Staff estimates.

Measured as cumulative change of variable in percent. Interest rates are in percentage points.

C. Are All Business Cycles Alike?11

17. Another way of assessing whether business cycles are similar is to look at the comovement among economic indicators and output. Business cycles may be similar across cycles if economic fluctuations stem from an accumulation of small economic shocks where each shock is unimportant when viewed in isolation, but together can generate fluctuations that look similar. Alternatively, cycles may differ considerably if large infrequent shocks trigger the fluctuations. Such a view is implicit in the description of specific periods such as the bursting of the asset bubble in 1991, the Asian financial crisis in 1997, or the IT bubble collapse in 2000.

18. Following in the Burns and Mitchell tradition, we study the co-movement of a reference variable, such as output (measured by real GDP), with various macroeconomic variables. To conduct these tests, we follow the common practice in the business cycle literature and decompose the time series into secular and cyclical components, using an econometric filtering technique.12 Once the data are filtered, the sample is divided into the five cycles, using the official reference dates. For each of these samples, and the sample as a whole, we compute the cross-correlations between the reference variable and the variable under consideration. A positive (negative) correlation between output and a macroeconomic variable indicates that the variable is procyclical (countercyclical). If the correlation is close to one, then the variable is highly correlated with the cycle, and if it is close to zero, the variable is largely uncorrelated with the cycle or acyclical.

19. Figure I.5 shows the co-movement of output relative to consumption, inflation, and the exchange rate. The upper panel shows that there is clearly a highly positive correlation between output and consumption over much of the sample. The bottom panels indicate that inflation and the exchange rate are also positively correlated, but this comovement is clearly not as strong as consumption. The table below further documents these correlations. For instance, the contemporaneous correlation of expenditure components (consumption and investment) with output is usually highly positive. In contrast, the correlation of stock prices and interest rates tend to be quite low, indicating little comovement over the cycle. The co-movements of the exchange rate, real wages, and inflation with output have been somewhere in-between these two other sets of variables.

Figure I.5.
Figure I.5.

Cyclical Movements

Citation: IMF Staff Country Reports 2004, 247; 10.5089/9781451820577.002.A001

Source: Staff estimates.

Co-Movement of Variable with Output1

article image
Source: Staff estimates.

All data in the table are in logarithms (except interest rates) and detrended using Corbae-Ouliaris filter.

20. These co-movements also differ widely across cycles. For instance, the comovement of consumption with output in the first four cycles is high (0.8) and positive, but for the current cycle the co-movement is negative (-0.5). This result may be reflecting the weak labor markets and sluggish household income growth that has led to lackluster consumption growth. Another striking difference is the unusually high correlation of real wages in the third and fourth business cycles. This increased association may be linked to the declines in real wages and output after the bubble burst. In general, these differences suggest that all business cycles are indeed not alike.

D. Is the Current Cycle Different?

21. While business cycles are persistent features of market-oriented economies and share many common elements, they also have distinguishing features that make each cycle unique. There are two notable differences in the current business cycle. First, the recession phase of the current business cycle differs from the previous cycles as it was shorter than previous cycles and the decline in output was particularly pronounced. Second, the behavior of expenditure components has differed:

uA01fig01

Contribution to Growth: Recessions

Citation: IMF Staff Country Reports 2004, 247; 10.5089/9781451820577.002.A001

Source: Staff calculations.
uA01fig02

Contribution to Growth: Expansions

Citation: IMF Staff Country Reports 2004, 247; 10.5089/9781451820577.002.A001

Source: Staff calculations.
  • Business fixed investment has played a larger role, relative to other expenditure components, in both the recession and expansion phase than in the past;

  • Private consumption has contributed only modestly to growth in the early stages of this expansion rather than being the driving force of the expansion;

  • Public demand has contributed less to growth in this recession than in the past. Furthermore, it has been a drag on the economy in the current expansion;

  • Net exports contributed to the slowdown in output in the recession, but have provided early and continuous stimulus to the economy during the expansion phase.

E. Is the Current Recovery Sustainable?

22. The current recovery has continued to exceed growth expectations. A key question remains as to whether this recovery can be sustained for a long period or whether as with earlier recoveries after the bursting of the bubble it too will falter. All indicators suggest that the current recovery is likely to be more durable than the previous recoveries in the 1990s for the following reasons:

  • First, the current expansion is more broadly based than previous expansions, with a firmer basis in private domestic demand. Furthermore, investment is apt to be sustained going forward by strong corporate profits and improving business sentiment. In addition, consumption has begun to strengthen against the backdrop of rising consumer confidence and firming labor market conditions;

  • Second, corporate restructuring has fostered structural gains in profitability, which should also support investment going forward. In this expansion firms have progressed in cutting excess leverage and labor cost, improving their position to reap profits as demand grows. (See Chapter III for further discussion);

  • Third, the corporate and financial sectors have become more resilient to economic shocks. Firms have deleveraged and banks have shed nonperforming loans and equity holdings. As a result, these sectors are less vulnerable to shocks such as a drop in stock market prices.

uA01fig03

Tankan Survey for Large Manufacturers1/

Citation: IMF Staff Country Reports 2004, 247; 10.5089/9781451820577.002.A001

Source: Bank of Japan.1/ Percentage of respondents reporting good business conditions less those reporting poor conditions.

23. While the economy is well positioned for a continuing recovery, additional progress in policy reforms would further help to promote a self-sustaining recovery that will enable the economy to withstand being buffeted by most economic shocks. As noted earlier, one striking difference between business cycles in the United States and Japan is that U.S. recessions are short-lived partly because of a highly flexible and mobile economy that enables a rapid response to negative economic shocks as exhibited in 2001. In contrast, in the post-bubble environment imbalances in Japan were allowed to linger rather than being resolved quickly.

F. Conclusions

24. This paper documents the key features of the Japanese business cycle during the period 1980–2003. Since the empirical methodology does not impose any strong theoretical priors on the data, it allows us to focus on the simple regularities observed in the data.

25. The key characteristic of the Japanese business cycle is that it lasts about 5½ years, slightly shorter than other industrialized countries. However, the duration of recessions and expansions is quite different from other industrialized countries. In Japan, recessions and expansions are roughly of equal length, whereas in most industrialized countries there is greater asymmetry, with recessions lasting one year and expansions extending five years.

26. The results also suggest that there are considerable differences in the behavior of output and prices across the five business cycles. This difference appears to be even more pronounced in the last cycle. For instance, the largest decline in output occurred in the recession of the current cycle. Also, the behavior of the expenditure components over the current cycle has been somewhat different from previous expansions. For instance, public demand has been a drag on the economy in the current expansion.

27. Finally, the pace of the current recovery has quickened and it appears to be broadening and becoming more durable. Furthermore, most economic indicators point to continuing signs of improvement. It seems that the recent progress in addressing Japan’s underlying structural problems has led to a more favorable growth picture for the period ahead.

References

  • Artis, Michael J., Zenon G. Kontolemis, and Denise, R. Osborn, 1997, “Business cycles for G7 and European Countries,The Journal of Business, vol. 70, pp. 249279.

    • Search Google Scholar
    • Export Citation
  • Baxter, Marianne, and Robert King, 1999, “Measuring Business Cycles: Approximate Band-Pass Filters for Economic Time Series,” Review of Economics and Statistics, vol. 82, pp. 864888.

    • Search Google Scholar
    • Export Citation
  • Blanchard, Olivier J. and Mark W. Watson, 1986, “Are Business Cycles All Alike?” in “The American Business Cycle, edited by Robert J. Gordon, (Chicago, Illinois: University of Chicago Press).

    • Search Google Scholar
    • Export Citation
  • Cashin, Paul and Sam Ouliaris, 2004, “Key Features of Australian Business Cycles, Australian Economics Papers, March, pp. 3958.

  • Chadha, Bankim and Eswar Prasad, 1994, “Are Prices Countercyclical? Evidence from the G-7,Journal of Monetary Economics, vol. 34, pp. 239257.

    • Search Google Scholar
    • Export Citation
  • Chadha, Bankim and Eswar Prasad, 1997, “Real Exchange Rate Fluctuations and the Business Cycle: Evidence from Japan,IMF Staff Papers, vol. 44, pp 328354.

    • Search Google Scholar
    • Export Citation
  • Corbae, Dean, and Sam Ouliaris, 2003, “Extracting Cycles from Non-Stationary Data”, forthcoming Econometric Theory and Practice: Frontiers of Analysis and Applied Research: Essays in Honor of Peter C. B. Phillips, edited by Dean Corbae, Bruce Hansen, and Steven Durlauf.

    • Search Google Scholar
    • Export Citation
  • Cooley, Thomas F. and Lee E. Ohanian, 1991, “The Cyclical Behavior of Prices,Journal of Monetary Economics, vol. 28, pp. 2560.

  • International Monetary Fund, “Recessions and Recoveries,World Economic Outlook, April 2002.

  • Kim, Sunghyun Henry, M. Ayhan Kose, and Michael G. Plummer, 2003, “Dynamics of Business Cycles in Asia: Differences and Similarities,Review of Development Economics, vol. 7, pp. 462477.

    • Search Google Scholar
    • Export Citation
  • Kydland, Finn and Edward Prescott, 1990, “Business Cycles: Real Facts and a Monetary Myth,” Quarterly Review, Federal Reserve Bank of Minneapolis, pp. 318.

    • Search Google Scholar
    • Export Citation
  • Romer, Christina D., 1999, “Changes in Business Cycles: Evidence and Explanations,Journal of Economic Perspectives, pp 2344.

  • Saito, Jun, 1997, “The Japanese Business Cycle After 1991: How is it Different, and Why?Journal of Asian Economics, Volume 8, pp. 263293.

    • Search Google Scholar
    • Export Citation
1

Prepared by Hali Edison (ext. 36946).

2

The source for the dating of Japanese business cycle comes from the Working Group of Indexes of Business Conditions within the Cabinet Office (CAO) and is reported at the following website: http://www.esri.cao.go.jp/en/stat/di/011221rdates.html.

3

Historically (from 1950 to 1980), the average length of a cycle was only four years as opposed to 5½ years and these cycles were asymmetric, with short recessions and longer recoveries—a characteristic that is currently observed in most industrialized economies, but not Japan.

4

See IMF (2002) for a more detailed discussion of business cycles characteristics in industrialized economies.

5

Artis, Kontolemis, and Osborn (1997) first noted the result that the duration of recession and expansions are asymmetric in industrialized countries.

6

Saito (1997) describes these episodes as growth recessions. Such recessions have also occurred in the United Kingdom and in the United States.

7

Such differences are reduced if one examines the average quarterly amplitude, as it scales the change in output by the duration of the phase.

8

This method follows from Romer (1999). Output loss is defined as the sum of the quarterly percentage shortfall of industrial production (IP) from its peak level over the prescribed period (trough or next peak). For example, a recession in which IP is 5 percent below its peak for six quarters would lead to an output loss of 30 percentage points.

10

For Japan, see for example, Chadha and Prasad (1994, 1997).

12

Specifically, the problem is to decompose a time series yt into a trend component (gt) and a cyclical component (ct) such that yt = gt + ct. There are many ways of performing this decomposition: (i) a Hodrick-Prescott (HP) filter; (ii) a band-pass filter (see Baxter and King (1999)); (iii) an optimal band-pass filter using information about the business cycle (Corbae and Ouliaris (2003)). In the exercise above, each time series was filtered by all methods, but only the results for the Corbae/Ouliaris filter method are reported.

Japan: Selected Issues
Author: International Monetary Fund