This Selected Issues paper analyzes macroeconomic developments and prospects for Japan during the 1990s. Following a surge in activity during 1996 and early 1997, the economy fell into recession in the second quarter of 1997. Real GDP fell by 3¾ percent during the four quarters ended March 1998; the unemployment rate reached historical highs; and deflationary pressures reemerged. The downturn was largely unexpected, and most forecasters had projected growth of about 2 percent in 1997. This paper also examines fiscal policy issues for Japan.


This Selected Issues paper analyzes macroeconomic developments and prospects for Japan during the 1990s. Following a surge in activity during 1996 and early 1997, the economy fell into recession in the second quarter of 1997. Real GDP fell by 3¾ percent during the four quarters ended March 1998; the unemployment rate reached historical highs; and deflationary pressures reemerged. The downturn was largely unexpected, and most forecasters had projected growth of about 2 percent in 1997. This paper also examines fiscal policy issues for Japan.

Japan: Selected Economic Indicators, 1992–97

Nominal GDP: US$4190 billion (1997)

Population: 126.0 million (1997)

GDP per capita: US$33,250 (1997)

Quota: SDR 8,241.5 million

article image
Sources: Nikkei Telecom; and staff estimates.

Contribution to GDP growth.

Includes provision for a FY 1998 Supplementary Budget.

Staff assumptions (annual average).

Based on normalized unit labor costs; 1990=100.

I. Macroeconomic Developments and Prospects1

1. Following the collapse of the asset price bubble in 1990, Japan has experienced a protracted period of weak growth performance (Figure I.1, and Chart I.1). GDP growth has averaged only 1½ percent during the last eight years, only a third of the rate registered during the previous decade. The initial slowdown resulted from a stalling of domestic demand, as the excesses of the bubble period began to be unwound. Business investment declined as firms sought to reverse the over-accumulation of capital during the 1980s, and consumption was dampened by falling stock and real estate prices. Weak domestic demand was compounded by the yen’s sharp appreciation during 1994–95, which resulted in a substantial withdrawal of external demand.



Citation: IMF Staff Country Reports 1998, 113; 10.5089/9781451820447.002.A001



Citation: IMF Staff Country Reports 1998, 113; 10.5089/9781451820447.002.A001

Sources: Nikkei Telecom, WEFA, and staff estimates.

2. Following a surge in activity during 1996 and early 1997, the economy fell into recession in the second quarter of 1997 (Table I.1). Real GDP fell by ¾ percent during the four quarters ended March 1998, the unemployment rate has reached historical highs, and deflationary pressures have re-emerged. The downturn was largely unexpected, and most forecasters had projected growth of around 2 percent in 1997.2 The recent downturn suggests that the strength in late 1996 and early 1997 mainly reflected the temporary effects of fiscal stimulus, the earlier interest rate cuts and exchange rate correction, and anticipatory demand ahead of the April 1997 consumption tax hike. Thus, in the absence of a resolution of the underlying balance sheet problems that had constrained recovery since the collapse of the asset price bubble, the economy was unable to withstand the effects of the withdrawal of fiscal stimulus, banking sector fragilities, and the Asia crisis.

Table I.1.

Japan: Growth of Real GDP and Demand Components, 1993–98 1/

(Percent change from the previous period)

article image
Sources: Nikkei Telecom; and WEFA.

At 1990 prices. Quarterly data are seasonally adjusted.

Final private domestic demand is the sum of private consumption, residential investment, and business fixed investment. Final domestic demand is final private domestic demand plus government consumption and investment.

Contribution to real GDP growth.

Government consumption and public investment.

3. Fiscal policy helped tip the economy back into recession. The general government balance had deteriorated markedly during the first half of the 1990s, owing to efforts to stimulate recovery.3 However, with signs in 1996 that the economy was recovering, the policy emphasis shifted toward consolidation and achieving a sustainable fiscal position ahead of impending demographic pressures associated with population aging. Accordingly, the FY1997 (fiscal year beginning April 1997) budget was sharply contractionary, reducing the structural deficit by 1¼ percentage points of GDP to 2½ percent of GDP. Tax measures included (i) the withdrawal of temporary income tax cuts that had been introduced in 1995; (ii) a 2 percentage point hike in the consumption tax rate on April 1, 1997; and (iii) an increase in medical insurance copayments effective in September 1997. Public investment was also brought down quickly following the spending surge related to the September 1995 stimulus package.

4. Banking sector instability was a critical factor behind the slowdown. Confidence in the Japanese banking sector has declined markedly since early 1997, as deregulation and a tightening of supervisory requirements brought more attention to the fragility of balance sheets in the banking sector. Concerns were reinforced by the sharp fall in share prices from mid-1997, which eroded bank capital, and the crisis in Asia, since Japanese banks are heavily exposed to the rest of the region.

5. Banking sector problems affected both corporate and household demand, particularly in late 1997. In November 1997, a major bank and two top-tier securities firms had to be closed. As a result, funding costs for Japanese banks both domestically and abroad rose sharply, and domestic credit conditions tightened markedly—particularly for small- and medium-sized enterprises—which further dampened capital expenditures. Concern regarding the effect of financial sector consolidation on employment and wages further undermined consumer confidence and led a steep decline in consumption and residential investment.

6. Besides its effect on domestic financial markets, the Asia crisis has begun to contribute to the Japanese recession through its effect on external demand. Initially, the drop in export volumes to the region that began in mid-1997 was partially offset by stronger exports to other partner countries, as well as a decline in import volumes. However, by the first quarter of 1998, total export volumes began to fall and net exports exerted a substantial negative contribution to growth.

A. Macroeconomic Developments

Output developments

7. Following a temporary surge in activity, the Japanese economy fell into recession in the second quarter of 1997.4 GDP expanded by 4 percent in 1996, and rose a further 2 percent in first quarter of 1997, reflecting the lagged effects of interest rates cuts and the yen’s depreciation, as well as household expenditures ahead of the consumption tax hike (Table I.1). However, aggregate demand fell by 2¾ percent in the second quarter, marking the end of the recovery. Real GDP was little changed in the latter half of 1997, but fell by further 1¼ percent in the first quarter of 1998. The cumulative fall in real GDP from the first quarter of 1997 to the first quarter of 1998 was 3¾ percent.

8. A substantial proportion of the slowdown reflected sharp cuts in public investment (Figure I.2). Public investment had risen sharply during the 1990–95 period, owing to a series of fiscal stimulus packages and to reconstruction efforts following the Hanshin earthquake of January 1995, and reached a peak of almost 10 percent of GDP in the second quarter of 1996. However, reflecting the increasing emphasis on fiscal consolidation, the December 1996 supplementary budget contained relatively modest spending commitments and the FY1997 budget set strict spending limits. As a result, public investment fell by nearly 20 percent (around 2 percent of GDP) between the second quarter of 1996 and the first quarter of 1997, and by a further 2 percent during the subsequent four quarters.



In percent of GDP

Citation: IMF Staff Country Reports 1998, 113; 10.5089/9781451820447.002.A001

9. Household consumption was dampened by the effect of tax increases and weaker labor market conditions on disposable incomes, as well as the effect of financial market strains on consumer sentiment (Chart I.2). Consumption surged in the second half of 1996 and the first quarter of 1997, rising by 5¼ percent over three quarters. With hindsight, the pickup in demand was largely due to anticipatory purchases of durables ahead of the consumption tax hike in April 1997, and during the subsequent four quarters consumption fell by 4½ percent, as the earlier surge was unwound. Consumer demand was also constrained by the effects on real disposable incomes of the income tax hikes (which began to affect pay checks in late June), as well as by the increase in health-care copayments and charges in September 1997. Labor incomes have been further dampened by the rise in unemployment and a decline in real monthly earnings. However, the decline in consumption has been considerably larger than can be explained by income and tax developments alone, and the household saving rate rose to historical highs in early 1998 (Section B contains a further discussion of recent household behavior).



Citation: IMF Staff Country Reports 1998, 113; 10.5089/9781451820447.002.A001

Source: Nikkei Telecom and WEFA.1/ Seasonally adjusted data.

10. Residential investment plunged in 1997. Housing demand had begun to recover in late 1995 in response to interest rate cuts and pent up demand, and expanded further in 1996 ahead of the consumption tax hike (Chart I.2).5 However, residential investment fell by 27½ percent during 1997, and its ratio to GDP fell to a level not seen since the 1960s. Although residential investment rebounded in the first quarter, rising by 1¾ percent, the decline in housing starts since January 1998 suggests that the pick up will not be sustained in the second quarter.

11. Business investment has slowed markedly since early 1997, reflecting the effects of tighter credit conditions and slumping profits (Chart I.3). Investment growth accelerated sharply in 1996, largely owing to improved corporate profitability and the effects of deregulation in the cellular telecommunications and retail sectors. Nonetheless, the recovery of capital expenditures, particularly by SMEs, was significantly weaker than typical in previous cycles, in large part reflecting balance sheet fragilities in the business sector. Although investment continued to rise strongly in the first quarter of 1997, it fell sharply in the second quarter.6 As domestic sales declined and credit conditions began to tighten, corporate profitability was eroded in the second half of 1997 and early 1998. At the same time, the number of bankruptcies, which had stabilized in 1992–96, rose sharply and the liabilities of bankrupt firms reached new historical peaks by mid-1998. Inventories also rose sharply, most recently reflecting a build-up of production goods related to the slowdown of exports of capital goods to Asia. These factors caused investment growth to slow and turn sharply negative at the end of 1997 and early 1998.



Citation: IMF Staff Country Reports 1998, 113; 10.5089/9781451820447.002.A001

Source: Nikkei Telecom and WEFA1/ Troughs defined as: FY 1993 for current cycle; FY 1987 for 1986–87 cycle; and FY 1983 for 1982–83 cycle.

12. Net exports rose strongly in 1997, but the Asia crisis severely eroded the support from external demand in early 1998 (Figure I.3 and Chart I.4). Following three years of decline, net exports contributed 1¼ percentage point to growth in 1997. The turnaround reflected the lagged response of export and import volumes to the massive correction in the real effective exchange rate of the yen that occurred in 1995—in real effective terms the yen depreciated by 30 percent from the second quarter of 1995 to the fourth quarter of 1996. However, much of the strength of net exports during the latter half of 1997 reflected the effects of weakening domestic demand, as import volumes fell sharply. Export volume growth slowed in the latter half of 1997, as the impact of the Asia crisis began to be felt, and turned negative in the first quarter of 1998. As a result, net exports withdrew ½ percentage point from GDP growth in the first quarter.



(January 1997=100)

Citation: IMF Staff Country Reports 1998, 113; 10.5089/9781451820447.002.A001

Sources: Bank of Japan and staff estimates.


Citation: IMF Staff Country Reports 1998, 113; 10.5089/9781451820447.002.A001

Sources: Nikkei Telecom, WEFA, and staff estimates.1/ Real net exports of goods and services on a national accounts basis.

13. Industrial production has been severely affected by the decline in domestic and external demand. Industrial production fell by ½ percent in 1997 (fourth-quarter to fourth-quarter basis), following growth of ¾ percent during 1996. Initially, declines were concentrated in the consumer durables sector (especially automobiles), which was most affected by the consumption tax hike, but the downturn subsequently spread owing to the effects of declining domestic and regional demand for intermediate and capital goods. Particularly hard hit were the chemicals and electric machinery sectors. The downturn intensified in 1998 after inventories reached historical highs in early 1998. Substantial production cutbacks were announced, particularly in the auto sector, and industrial production fell by a cumulative 9½ percent during February-May 1998.

External developments

14. The current account strengthened significantly in 1997 (Chart I.4 and Table I.2). The surplus rose from ¥7.2 trillion (1½ percent of GDP) in 1996 to ¥11.4 trillion (2¼ percent of GDP) in 1997, reflecting both an increase in the trade surplus and a decline in the invisibles deficit. The trade balance was bolstered by the effect on export volumes of strong partner-country demand, principally from the United States and Europe, and the lagged effects of the yen’s depreciation from its peak in mid-1995 (Figure I.4). These factors more than offset the sharp decline in export volumes to the ASEAN-4 and Korea in the latter half of 1997. The trade balance also was bolstered by a slowing of import volume growth in response to the yen’s depreciation and the decline in domestic demand. Imports from the Asian region were particularly hard hit, partly reflecting the effects of the slowdown of housing activity and cuts in public investment on the demand for construction materials. Imports from regional partners may also have been affected by a decline in “reverse imports,” as Japanese firms reduced their purchases of components from Asian factories in the face of declining domestic demand.7



(Four-quarter percent charges)

Citation: IMF Staff Country Reports 1998, 113; 10.5089/9781451820447.002.A001

Table I.2.

Japan: Current Account Summary, 1993–98

(In billions of yen)

article image
Sources: Nikkei Telecom; WEFA, and staff estimates.

Sum of the seasonally adjusted trade balance and the seasonally unadjusted invisibles balances.

Seasonally adjusted.

Seasonally unadjusted.

15. The external surplus rose further in early 1998, reaching ¥16 trillion on an annualized basis (3 percent of GDP) in the first quarter, notwithstanding a contraction of net exports on a volume basis. The further improvement in the surplus reflected lower import values, largely due to the decline in oil and commodity prices at the beginning of the year, as well as the weakness in domestic demand. Export values fell sharply in the first quarter, owing to a substantial decline in export volumes in February-March, particularly to Asia, as well as lower export unit values. The invisibles balance continued its upward trend in the first quarter, and reached a surplus for the first time in many years. This reflected the effect of the growing net foreign asset position and yield differentials on investment income, the impact of the yen’s depreciation and weak domestic demand on tourism, and spillover of the slowdown in Asia on shipping and other service receipts.

16. Capital outflows have increased in line with the widening of the current account surplus (Table I.3). Bank lending abroad rose strongly, particularly in the fourth quarter, as banks responded to the substantial increase in the Japan premium that followed the failure of Yamaichi Securities and Hokkaido Takushoku Bank in November by supporting their overseas subsidiaries. These outflows more than offset substantial net portfolio inflows in 1997, which reflected the repatriation of proceeds from securities sales by Japanese banks to bolster their capital positions, as well as large purchases by nonresidents of Japanese bonds on the expectation of continued bond yield declines. In the first quarter of 1998, the bank-related outflows were largely reversed, as funding pressures eased, balanced by portfolio outflows as nonresidents’ positions in Japanese bonds were unwound.

Table I.3.

Japan: Capital and Financial Account Summary, 1993–98

(In billions of yen, not seasonally adjusted)

article image
Source: Nikkei Telecom and WEFA.

17. Deregulation has eased restrictions on capital outflows during the past year. Restrictions on pension fund investment in foreign assets were lifted beginning in 1998,8 and the Foreign Exchange Law—which required pre-notification of all major foreign exchange transactions—was relaxed as of April 1998. These measures, and the relatively low yield on Japanese investments, have facilitated a shift in investor preference toward foreign assets. For example, although portfolio outflows fell in March 1998, owing to the need to bolster domestic liquidity at the fiscal-year end, portfolio outflows rose sharply in May, mainly reflecting equity investments abroad. Moreover, there has been a rapid growth of Japanese residents’ holdings of investment trusts (mutual funds) denominated in foreign currencies. The value of foreign assets held by domestic funds rose by 32 percent in April 1998 from the previous year, and holdings by domestic residents of overseas funds rose by nearly 80 percent.9

Labor market developments

18. Employment conditions have deteriorated considerably during the past year (Chart I.5). After accelerating to 2 percent in February 1997, 12-month employment growth slowed during the rest of 1997 and turned negative in 1998, falling to minus ½ percent in May 1998 (Figure I.5). The construction sector was the first affected, reflecting the impact of the decline in public works spending that began in mid-1996, as well as the decline in residential construction. In early 1997, the drop in household consumption began to adversely affect manufacturing employment. During the latter half of 1997 and the first half of 1998, employment declines spread more generally to include the services sectors.



Citation: IMF Staff Country Reports 1998, 113; 10.5089/9781451820447.002.A001



Citation: IMF Staff Country Reports 1998, 113; 10.5089/9781451820447.002.A001

Source: Nikkei Telecom and WEFA.1/ Seasonally adjusted data.

19. As a result, the unemployment rate, which had hovered around 3½ percent since late 1995, rose sharply in 1998, reaching a historical high of 4.1 percent in April and May. Unemployment among the youth and middle aged is especially severe—the unemployment rate for males aged 15–24 reached 9.1 percent in May, and the rate for males aged 55–64 percent was 7.0 percent (Figure I.6).



In percent

Citation: IMF Staff Country Reports 1998, 113; 10.5089/9781451820447.002.A001

20. Earnings also have been adversely affected by the slowdown. Average monthly compensation growth fell from nearly 3 percent in the first quarter of 1997 (four-quarter rate) to just less than zero in the second quarter of 1998. The slowdown in compensation growth and the rise in prices due to the consumption tax hike has left real earnings roughly 1¾ percent lower in the first five months of 1998 than in the same period in 1997. Last year’s shunto yielded an increase of close to 3 percent, but compensation growth was dampened by declining overtime hours and bonus payments in the face of cyclical pressures. This year’s spring shunto is expected to result in wage increases of about 2½ percent, but bonuses, which typically are the equivalent of five months’ salary, are expected to be roughly unchanged in 1998.10

21. A number of important structural changes have occurred in the labor market. Most notably, employment costs have risen as a result of the increased prevalence of the five-day, 40-hour work week (the 40-hour requirement was introduced with a 1987 amendment to Labor Standards Law, but was only applied to small- and medium-sized enterprises in April 1997).11 At the same time, social security contribution rates, which only apply to full-time employees, have risen sharply from 10.6 percent in the early 1980s to 17.35 percent in October 1997. The increased cost of full-time labor, and the increased supply of part timers (particularly among females, whose participation rate has been on a secular increase and also tends to rise during cyclical downturns) has encouraged a substantial increase in the proportion of part-time employees, particularly at the lower end of the wage scale (Figure I.7). The increase in part-time employment has also been facilitated by deregulation that has increased firms’ access to temporary placement firms (see Chapter VI for further discussion).



In percent of total employment

Citation: IMF Staff Country Reports 1998, 113; 10.5089/9781451820447.002.A001

22. These labor market changes have complicated assessment of the “natural” rate of unemployment and of deflationary pressures. Recent trends in the unemployment rate suggest that the natural rate has increased from around 2½ percent in the late 1980s to around 3 percent in early 1998, and the rapid rise in unemployment among the relatively aged supports the view that structural unemployment has increased significantly.12 Moreover, the fact that the vacancy rate has risen with the unemployment rate provides evidence of a structural increase in frictional unemployment (Figure I.8). At the same time, however, the greater use of part-time employees, and evidence that the lifetime employment system is eroding in favor of merit-based wage system, may mean that firms have begun to be less averse to layoffs in response to changes in demand conditions. Thus, both the natural rate and the responsiveness of the unemployment rate to the cycle may have increased in recent years.



Unemployment rate

Citation: IMF Staff Country Reports 1998, 113; 10.5089/9781451820447.002.A001

Price developments

23. Deflationary pressures mounted during the first half of 1998 (Chart I.5). The twelve-month rate of CPI inflation rose sharply in 1997, reaching a high of 2½ percent in October, largely reflecting the 2 percentage point increase in the consumption tax in April 1997 and the September rise in health care fees.13 Since then, the inflation rate has fallen steadily to around ½ percent in May 1998. The twelve-month core inflation rate (excluding fresh food and energy) has fallen more rapidly, dropping from a recent high of 2¼ percent in October 1997 to zero in May 1998; indeed, during the first five months of 1998, the seasonally-adjusted core CPI fell by just over ½ percent.

24. Besides the effect of economic slack, deflationary pressures have been exacerbated by declining commodity prices. Import unit values fell by 2¼ percent in the first quarter of 1998, compared with the previous year, reflecting the decline in commodity prices and effects of the depreciation of Asian partner-country currencies. Wholesale prices fell by 1¾ percent in May 1998 from the previous year, while the domestic component of wholesale prices fell by 2¼ percent during the same twelve-month period, as the early 1997 decline in commodity prices fed through to domestic producer prices.

B. What Explains the Recent Weakness of Household Spending?

25. The downturn in consumption and residential investment after April 1997 was much larger and more persistent than expected. Although a degree of “demand shifting” was anticipated around the time of the increase in the consumption tax, most forecasters projected a much milder dip in spending than actually occurred. The sharp increase in inventories immediately after April indicates that most businesses were similarly surprised. The downturn in spending was also considerably larger than observed at the time the consumption tax was introduced in 1989 at a rate of 3 percent, when the tax hike was one-and-a-half times as large.

26. The effect of the consumption tax increase was larger for durable goods than nondurable goods, as would be expected given their larger cost and the fact that their benefits are more evenly spread over time (Chart I.6).14 The spike and subsequent fall in sales before April 1 was larger for durable goods (measured using automobile registrations) than for overall consumption.15 By contrast, spending on non-durable goods and services showed very little impact from the tax hike. Most consumption indicators also pointed to a recovery within a few months of the tax hike (retail sales being an exception).



(1987Q4/1995Q4 INDEX=100)

Citation: IMF Staff Country Reports 1998, 113; 10.5089/9781451820447.002.A001

Sources: Nikkei Telecom, WEFA, and staff estimates.

27. Residential investment showed an even larger demand shift in response to the consumption tax hike (Chart I.7). The introduction of the consumption tax in 1989 had a relatively limited impact on residential investment, perhaps because the increases in the price of land associated with the asset price bubble dwarfed the impact of the tax. By contrast, in 1997 residential investment rose sharply in the year preceding the tax hike, but declined very sharply afterwards.




Citation: IMF Staff Country Reports 1998, 113; 10.5089/9781451820447.002.A001

Sources: Nikkei Telecom, WEFA, and staff estimates.

28. Theory suggests that household consumption responds to movements in real interest rates, disposable income, and consumer confidence.16 Real interest rates define the value of consumption over time; in particular, the “demand shifting” observed in early 1997 reflects the anticipated impact of the consumption tax hike on the real interest rate. Changes in disposable income matter because some individuals are liquidity constrained, and hence vary their consumption in line with their income. Uncertainty about economic prospects—reflected in consumer confidence indicators—signals households’ expectations of permanent income, which determines the spending of forward-looking individuals, and can also encourage precautionary saving.

29. Disposable income in FY1997 was eroded by higher taxes and social security contributions.17 In addition to the consumption tax hike, the FY1997 budget increased income taxes by ¥2 trillion as existing temporary tax rebates were ended,18 while social security charges were raised in both September 1996 (pension contribution rates) and September 1997 (mainly health care copayments). The overall impact was to lower real disposable by over ¥7¾ trillion (2¼ percent of disposable income) in FY1997.

Effects of Fiscal Measures on Real Disposable Income, FY1997

article image
Notes: Only those parts of the pension and health increases relevant for FY1997 are included in the calculations.

30. An estimated consumption function was used to assess how far the recent movements in consumption were predictable. The model assumes that in the long run consumption depends on the level of the real interest rate, wealth, demographic variables, and disposable income, and that in the short run the change in consumption also responds to the change in disposable income (reflecting the influence of liquidity constraints) and the change in the real interest rate. In order to capture the return to households from varying the timing of their purchases in response to changes in indirect taxes, the real interest rate was calculated in a forward-looking manner. The model implies that a 5 percent fall in real disposable income lowers consumption by around 1 percent in the short run, while an anticipated 7 percentage point rise in the annualized real interest rate has a similar impact.

Model of Household Consumption

The following consumption equation was estimated, using quarterly data for 1980–96:



ecmt = ct-0.11wt - 0.22 landt - 0.34 yt - 1.96 dept - 0.10rt,

where ct is the log of consumption, yt is the log of disposable income, rt is the forward-looking real short-term interest rate, wt is the log of household financial net assets, landt is the logarithm of the real price of land, dept is the dependency ratio, ecm, is the difference between actual consumption and its long-run equilibrium and t-statistics are reported in parentheses.

As both disposable income and the real interest rate are potentially endogenous (disposable income because it may be correlated with unexpected shocks to permanent income, and the real interest rate because it contains future outcomes for inflation), the dynamic equation was estimated using two-stage least squares. The instruments were a constant term, lagged values of consumption, lagged values of income (movements in income should help predict consumption in a permanent income model), the lagged real interest rate, lagged values of inflation and nominal interest rates, and a dummy for the introduction of the consumption tax.

31. Applying this equation to 1997–98 data, the projections significantly underestimated the observed shift in consumption demand, with a 2 percent under-prediction in the first quarter of 1997 being roughly offset by the over-prediction in the next quarter (Chart I.8).19 Actual consumption moved closer to the predicted value in the third quarter of 1997, but subsequently diverged again. The renewed weakness of consumption in late 1997 and early 1998 probably reflected the impact of other factors, including financial closures and the sharp increase in unemployment, both of which might be expected to raise precautionary savings.



Citation: IMF Staff Country Reports 1998, 113; 10.5089/9781451820447.002.A001

Source: Nikkei Telecom, WEFA.

32. Similar results were found for residential investment. An econometric equation was estimated in which the short-run behavior of residential investment depends upon changes in activity and tax hikes, while the long-run equilibrium depends upon real wealth, real land prices and real interest rates.20 The predictions from this equation, shown in the lower panel of Chart I.8, indicate that the rise in residential investment spending before the consumption tax hike and, in particular, the downturn after the tax hike were much larger than might have been anticipated based on earlier behavior.

33. These results suggest that the recent weakness in household spending was larger than would have been expected on the basis of tax and income developments. Even taking into account the impact of tax hikes on real interest rates and disposable income, consumption and residential investment are weaker than would be expected from past behavior. The most likely explanation is that failures of financial institutions, higher unemployment, and reduced job security lowered expectations of permanent income, reduced consumer confidence, and caused a rise in precautionary savings. While some of these factors may be transitory, others may be associated with more fundamental structural shifts in the economy, suggesting that at least some of the current weakness in spending may be protracted.

C. To What Extent Does the Recent Weakness Reflect Lower Potential Growth?

34. The normal uncertainties surrounding estimates of potential growth are exacerbated in the present situation by the volatility of output and the fact that the economy appears to have deviated significantly from historical trends. The staff bases its estimates of potential output on a Cobb-Douglas function of the net capital stock, the trend value of labor inputs, and the trend value for total factor productivity (Chart I.9).21



Citation: IMF Staff Country Reports 1998, 113; 10.5089/9781451820447.002.A001

Source: Staff estimates.

35. These estimates of potential growth suggest that both cyclical and structural factors have contributed to the growth slowdown since the bursting of the asset-price bubble. At the cyclical peak in 1990, the excess demand gap reached roughly 4 percent of potential output, compared with an excess supply gap estimated at around 6 percent in the first quarter of 1998. However, during the same period the estimated underlying potential growth rate of the economy also slowed markedly, from around 3½ percent during 1980–90 to 2¾ percent during 1991–1997 (potential growth is estimated to have fallen to around 2 percent in early 1998). Thus, almost half of the decline in growth during the 1991–97 period compared to the previous decade can be ascribed to structural factors that have slowed potential growth.

Japan: GDP Growth Decomposition1

(In percent)

article image
Source: Staff estimates.

Period averages are the average of annual growth rates. Trend hours and trend employment are estimated on the basis of a Hodrick-Prescott filter, and trend TFP is assumed to be a log-linear time trend, segmented in 1972.

36. Labor’s contribution to potential growth has fallen sharply. This decline has reflected the effects of demographic changes—the growth of the working age population fell from around 1 percent during the 1980s to minus ¼ percent in 1996, as well as an increase in the natural unemployment rate from around 2½ during the late 1980s to 3 percent by 1998, which has been partly offset by continued increases in the participation rate. A significant decline in average hours worked—owing to legislation mandating a 5-day, 40-hour work week—also reduced the growth of the labor input.

37. A decline in the business investment rate also contributed to the slowdown in potential growth. The rate of capital accumulation fell sharply during the 1990s owing to the decline in investment as a ratio to GDP. This, in turn, reflected the effects of balance sheet difficulties, as well as efforts by firms to unwind excess capital stocks accumulated during the bubble period. During the late 1980s, the capital stock-to-potential GDP ratio rose well above the trend that had been established over the previous decade.

38. There is considerable uncertainty regarding the extent to which capital stock adjustments have been completed. Measures of the capital-to-potential output ratio suggest that the capital stock adjustment has not yet been completed, since the ratio still exceeds its historical trend level (Chart I.3). However, these estimates may overstate the effective capital stock, since large amounts of capital accumulated during the bubble period may be relatively unproductive, either because of technological change or because of excessive investment in the real estate or other sectors.22 Also, the estimate of the trend and actual capital-to-potential output ratios are highly dependent on the underlying estimate of potential output, which itself is subject to uncertainty. By contrast, to the extent that the trend increase in the “equilibrium” capital-output ratio has slowed in recent years—including in response to the fact that Japan’s capital-output ratio has broadly caught up with that of other industrial countries—then the required adjustment to the capital stock could be even greater.

39. Underlying total factor productivity (TFP) growth may also have slowed in the 1990s. In the estimates of potential GDP described above, TFP is assumed to have continued to grow at a 1¼ percent rate, as in the 1980s. However, the actual growth of total factor productivity—the measure of the economy’s overall rate of productivity growth—fell to only ½ percent during 1991–97, well below trend. A decline in TFP growth is typical in a cyclical downturn, particularly in the context of the Japanese labor market, since firms tend to avoid layoffs in favor of cutting real wages. However, the magnitude of the decline in actual TFP suggests that its underlying trend may have slowed, in which case the output gap and potential growth may be smaller than estimated.

D. Economic Prospects in 1998

40. Even with a pick-up in the second half of 1998, GDP is projected to decline by 1¾ percent in 1998 as a whole. After the sharp first-quarter drop, monthly data suggest that GDP growth was again negative in the second quarter. In the second half of 1998, domestic demand is expected to receive a significant boost from the additional spending contained in the April 1998 fiscal package and the Government’s commitment to frontload FY1998 spending. Nonetheless, fiscal stimulus is likely to be partially offset by underlying weakness in private sector and foreign demand.

41. Household consumption and residential investment are expected to remain subdued, despite income tax cuts. Factors expected to weigh on consumption include weak sentiment, and a sharp deceleration of disposable income growth related to a lack of employment growth, modest wage increases, and the continued shift toward low-wage, part-time employment. Although disposable income will be boosted by the income tax cuts announced in December 1997 and April 1998, the impact may be offset in part by continued high levels of precautionary savings, particularly given expectations that tax increases would need to be unwound to achieve medium-term fiscal consolidation.

42. Business investment is expected to decline sharply in 1998. The June Tankan survey suggested that principal enterprises expect to reduce investment outlays by 1¼ percent and small enterprises reducing capital spending by 19 percent in FY1998. Lower investment plans reflect a drop in profitability, continued weak confidence in growth prospects, and a substantial inventory overhang.23 Survey evidence suggests that the decline in spending will be especially severe in sectors that invested heavily in recent years to meet growing demands for information technology-related materials. The construction and real estate sectors are also expected to be hard hit, owing to continued high vacancy rates and the earlier drop in public investment. In addition, a reduction in business inventories is expected to have a negative impact on growth in 1998.

43. Net exports are expected to be roughly flat during the latter half of 1998. Export volumes to the rest of Asia would remain compressed owing to the regional crisis, although the impact on net exports is expected to be offset by weak domestic demand for imports. The current account surplus is projected to reach 3¼ percent of GDP, owing to the effect of weak commodity prices and regional export unit values on the terms of trade.

44. The large output gap and the decline in wholesale prices is expected to put further downward pressure on prices. The CPI is expected to increase by about ½ percent in 1998, on a year-over-year basis, but to decline by nearly ½ percent on a fourth-quarter to fourth-quarter basis. However, downside risks to the inflation outlook are suggested by the fact that inflation in Japan has tended to be relatively responsive to changes in the employment gap (i.e., the difference between the actual unemployment rate and the natural rate),24 which is estimated to be in the range of 1 percentage point.

E. Medium-Term Prospects for the Saving-Investment Balance

45. The yen’s recent depreciation against the U.S. dollar raises questions about the consistency of the current exchange rate with medium-term fundamentals.25 The yen has depreciated significantly from mid-1995, and since early 1997 appears in real effective terms to be well below the trends established since the early 1970s (Chart I.10). Although exchange rate weakness would be expected in the face of a cyclical downturn, the current value of the yen, if maintained in real effective terms, would imply a substantial increase in the current account surplus over time. On the basis of the staffs trade model, an unchanged exchange rate, and current WEO projections for partner countries, the current account surplus would rise to over 4 percent of GDP in 2003, even assuming a substantial recovery of activity in Japan that largely eliminates the current output gap.26



Citation: IMF Staff Country Reports 1998, 113; 10.5089/9781451820447.002.A001

Sources: Nikkei Telecom, WEFA, and staff estimates.1/ CPI-based rate employs an average of G-7 partner countries; WPI-based rate employs an average of G-7 partner countries for which WPIs are available: US, Germany, UK; Relative export unit values are based on G-7 partner countries.

46. By contrast the underlying saving-investment (SI) balance is estimated to be in the range of ¼-2¼ percent of GDP (tabulation). Demographic factors are expected to weigh heavily on the SI balance over the medium term. In particular, the dependency ratio—the ratio of nonworking age population to the working age population—is projected to rise from around 45 percent to 50 percent between 1997 and 2003, which would be expected to reduce the private saving rate by around 1–2 percentage points of GDP.27 An increase in public saving—in line with the authorities’ commitment to fiscal consolidation and the effects of a cyclical recovery—would only partially offset the impact of demographics, since higher public saving would be partly offset by lower private saving. Moreover, the private investment rate in Japan has been below trend in recent years, owing to efforts by the business sector to work off the excesses of the bubble period. If previous investment trends were to be re-established, this also would tend to raise the investment ratio and lower the SI balance.

Saving-Investment Balances

(In percent of GDP)

article image

Current account projections are based on a trade model, assuming an unchanged real effective exchange rate and a substantial narrowing of the output gap by 2003.

The estimates of the underlying SI balance are based on the empirical work in Hamid Faruqee and Guy Debelle, “Saving-Investment Balances in Industrial Countries: An Empirical Investigation,” in Peter Isard and Hamid Faruqee (eds.), Exchange Rate Assessment: Recent Extensions and Applications of the Macroeconomic Balance Approach, IMF Occasional Paper No. 167 (1998 forthcoming). The three-equation model comprised panel estimates of the saving rate, the investment rate, and current account as a share of GDP, using output per capita, the dependency rate, the output gap, and general government balance as a share of GDP as exogenous variables. The single equation estimates related the SI balance to the same exogenous variables, as well as the capital-output ratio and net foreign assets as a share of GDP.

47. To close the substantial gap between the projected current account surplus and the underlying SI balance would require a significant real appreciation of the yen over the medium term. For example, to close a 2 percentage point gap between the projected current account balance in 2003 and the underlying SI balance would require around a 20–30 percent appreciation of the yen in real effective terms, based on conventional trade elasticities. Such an appreciation is somewhat larger than implied by real interest rate differentials, which imply a roughly 15 percent real appreciation over the next five years.28 However, the implications of these results for the exchange rate, must be viewed with caution given the considerable uncertainty attached to the estimates of the current account and underlying SI balance five years hence.


Prepared by Tamim Bayoumi and Christopher Towe.


The May 1997 WEO projected 1997 growth of 2¼ percent, and the private sector Consensus Forecast in August 1997 was for growth of 2 percent.


See Chapter II for a more detailed discussion of fiscal developments.


In June 1998, the Economic Planning Agency officially declared that the economy peaked in March 1997. Business cycle dating in Japan is performed by the Economic Planning Agency on the basis of its diffusion index of coincident indicators, and downturns are usually declared only with a significant lag.


The additional tax was not levied on homes for which contracts were signed before October 1996, regardless of completion date.


The second-quarter decline in investment was anomalous, as it exceeded the drop in most of the monthly indicators including capital goods shipments, something that also occurred in 1989, when the tax was first introduced. This suggests the possibility that the second-quarter decline partly reflected statistical problems in accounting for the consumption tax.


The relationship between developments in Japan and Asia are discussed in more detail in Chapter VII.


Pension funds’ holdings of foreign securities were limited to no more than 30 percent of total assets. However, while Japanese portfolio diversification has risen over time, this restriction was not generally a binding constraint.


Tomoko Fujii, “Private Capital Outflows Continue,” in Japan: Issues and Prospects, Salomon Smith Barney—Economic & Market Analysis (June 4, 1998).


A recent survey suggested that the average employee summer bonus will be only ½ percent higher than last year, while average bonuses fell in 22 out of 40 industries surveyed (Nihon Keizai Shimbun, June 18, 1998).


The number of firms providing a five-day work week at least once a month rose from 78 percent in 1987 to 96 percent in 1995, and the number of firms providing a five-day work week each week rose from 7 percent to 26 percent over the same period.


The staff’s estimate of the natural rate is based on a Hodrick-Prescott filter of the actual unemployment rate. Historically, the unemployment gap has responded relatively modestly to the output gap, and estimates of the Okun coefficient (i.e., the percent change in the output gap that would be associated with a percent change in the unemployment rate) are around 5.


The consumption tax hike and the rise in medical costs are estimated to have increased the CPI by about 1½ percent and ½ percent, respectively.


Data on consumption come from various sources. The national income accounts (NIA) report quarterly data on consumption and (non-seasonally adjusted) compensation of employees with a lag of about 2 months, but the components of consumption—durable, semi-durable, and nondurable goods and services—and disposable income are not available. More timely monthly indicators of consumption include data on retail sales and car registrations, as well as the family income and expenditure survey (FIES). The FIES reports consumption by type of good, as well as labor and disposable income, thus covering most of the deficiencies in the NIA data. Unfortunately, the FIES and NIA data have very different underlying trends for variables such as the average propensity to consume, leading to concerns about the representativeness of the FIES survey methodology.


The rise in automobile registrations in 1989 reflects other tax changes implemented at the same time as the consumption tax hike.


Longer term influences on consumption include demographic trends and wealth.


In 1989, the introduction of the consumption tax was designed to be roughly revenue neutral in the short term. In 1997, by contrast, the increase was an offset to tax cuts enacted in 1995.


The temporary tax cuts were provided in June and December, so their elimination had particularly large effects on disposable income in those months.


Disposable income was projected using national income account data on wage incomes and FIES data on the gap between wage income and disposable income.


As the impact of the tax hike was less easy to define, it was modeled using a simple dummy variable rather than the real interest rate.


The labor input is assumed equal the product of employment and average monthly hours, compared with estimates presented in last year’s report (JapanEconomic and Policy Developments, IMF Staff Country Report No. 97/91, October 1997, Chapter II), which were based simply on total employment. The trends for the labor inputs are based on a Hodrick-Prescott filter of the actual data, and the trend for TFP is based on a log-linear time trend, broken in 1972.


A breakdown of investment into machinery and equipment and structures is not available, but loans to the real estate and construction sectors totaled just over 15 percent of total loans to the private sector during 1986–91. A further data complication is that official estimates of the net capital stock—which is a more relevant concept for gauging the economy’s productive capacity—are not available. Thus, the net capital stock series used in the estimates of potential GDP were constructed by subtracting from gross investment an ad hoc estimate of the depreciation rate, which may be significantly different from the true depreciation rate.


Among principal enterprises, manufacturers expected a 1 ½ percent decline in profits in FY1998 and nonmanufacturers expected a 4¼ percent decline, following drops of 6¼ percent and 8 percent in FY1997, respectively. A recent Ministry of Finance survey showed an even more pessimistic picture—nonfinancial firms expected recurring profits to decline by 11¾ percent in FY1998, following a 6½ percent decline in FY1997. Nonmanufacturers projected profit declines of 16 percent.


A simple inflation equation (estimated on quarterly data over 1983–97) illustrates the effect of the cycle on inflation:


where π is core inflation (quarter-over-quarter at an annual rate), μ is percentage change in the import deflator (national accounts basis), d89 and d97 are dummy variables for the consumption tax, and gap is the difference between the unemployment rate and the natural rate (estimated using a Hodrick-Prescott filter). In this case a 1 percentage point increase in the unemployment gap would lower the inflation rate roughly 2 percentage points (after taking into account the effect of the lagged inflation rate). This result is broadly consistent with the results reported by Fumihira Nishizaki “The NAIRU in Japan: Measurement and its Implications,” OECD Economics Department Working Paper No. 173 (1997).


For a discussion of the Fund’s approach to measuring exchange rate misalignments, see Peter Clark, Leonardo Bartolini, Tamim Bayoumi, and Steven Symansky, Exchange Rates and Economic Fundamentals: A Framework for Analysis, IMF Occasional Paper No. 115 (Washington DC: International Monetary Fund, 1994). The framework essentially involves comparing the saving investment balance predicted on the basis of demographics, fiscal policy and other fundamentals, with the current account balance that results from a standard trade model, assuming current levels of exchange rates and full employment.


For a description of the staffs approach to modeling the current account, see Bankim Chadha, “External Adjustment in Japan: Recent Developments and the Medium-Term Outlook,” in Japan—Selected Issues, IMF Staff Country Report No. 96/114 (October 1996), pp. 150–172. Medium-term SI prospects and the implications for the exchange rate are also discussed in more detail by Guy Meredith, “The Yen: Past Movements and Future Prospects,” in Bijan B. Aghevli, Tamim Bayoumi, and Guy Meredith (eds.), Structural Change in Japan: Macroeconomic Impact and Policy Challenges (Washington DC: International Monetary Fund, 1998), pp. 13–50.


Although there is considerable uncertainty about the empirical evidence regarding the effect of demographics on saving behavior, many recent estimates suggest that a 1 percentage point increase in the dependency ratio tends to lower the saving rate by around 0.2–0.4 percentage points of GDP.


This calculation is based on current differentials between 5-year government bond yields and on Consensus Forecast estimates of inflation rates for Japan and her major trading partners.

Japan: Selected Issues
Author: International Monetary Fund