## _wp03215

## Source details

**Canonical URL:** [_wp03215](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2003/_wp03215.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2003/_wp03215.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2003/_wp03215.pdf.json)

---

### I. INTRODUCTION
- Deflation has continued unabated in Japan for over half a decade.
- Price measures:
  - GDP deflator has fallen by over 9 percent since 1995.
  - CPI has fallen by 3 percent since 1998.
- Central questions examined:
  - Have the persistent declines in the price level been very costly for Japan?
  - Or is deflation broadly tolerable?
- Observations and debate:
  - Some observers downplay the impact of the relatively modest decline in the price level vis-à-vis the sharp fall in asset prices through the 1990s.
  - It has been suggested that mild deflation might be a sign of price stability and entail very little cost.
  - On the policy front, necessity of vigorous response to deflation has been questioned because potential costs of aggressive policies may be higher than costs of allowing deflation to run under existing policy frameworks.
- Counterarguments:
  - Prolonged, unanticipated deflation has impeded monetary policy efficacy, hampered financial market activities, squeezed corporate profitability, and raised the real burden of private and public debt.
- Organization of the paper:
  - Section II: theoretical arguments on the cost of deflation.
  - Section III: historical price-related developments in Japan.
  - Section IV: evidence on the impact of deflation on monetary policy-making, labor market, financial market, households, and the public sector.
  - Section V: concluding comments.

### II. CONCEPTUAL ISSUES
- Core asymmetry:
  - The clearest asymmetry between inflation and deflation is the problem of a zero-bound on interest rates that tends to be associated with deflationary episodes.
- Costs depend on source, extent, and duration of deflation:
  - Temporary price declines due to strong aggregate supply expansion or productivity spurts may not be too problematic.
  - Positive external shocks (falling import prices, aggressive trade liberalization) could push down domestic prices without entrenching deflationary expectations.
  - Recent productivity increases from information technology, globalization, and deregulation have reduced inflationary pressures; similar productivity spurts mattered in 19th-century deflationary episodes.
- Aggregate demand-driven deflation and nominal rigidities:
  - Nominal rigidities in financial markets and the labor market imply aggregate demand-driven deflation is likely costly.
  - Estimates due to the zero-interest rate floor are significant but hard to quantify precisely.
  - Studies of downward rigidity of nominal wages suggest deflation could impose costs amounting to several percentage points of GDP (Akerlof et. al., 1996).
- Collateral and balance sheet effects:
  - Fisher’s (1933) debt-deflation theory: unanticipated deflation redistributes wealth from debtors to creditors and reduces collateral values, especially when deflation accompanies reductions in asset prices.
  - As collateral loses value and bankruptcies rise, banks may raise financing charges (external finance premium rises) or cut back lending, reducing aggregate demand.
- Sticky wages:
  - Aggregate demand-induced deflation raises unemployment when nominal wages are rigid downward.
  - Akerlof et. al. (1996) estimate that with a sustained 1 percent deflation and downward rigidities in nominal wages, unemployment in the United States could rise from a long-run equilibrium rate of 5.8 percent to 10.0 percent.
  - Phillips curve estimates suggest output losses could amount to a multiple of the roughly 4 percentage point loss in employment.
- Deflation versus inflation or disinflation:
  - Inflexibilities in financial and labor markets create asymmetry: sustained deflation can have substantial adverse effects while moving from zero to very low inflation rates (say below 2 percent) likely has negligible efficiency losses.
  - Zero interest rate floor is more likely to bind under deflation than disinflation.
  - Collateral and balance sheet effects and credit cutbacks are likely stronger under unexpected deflation.
- Institutional factors:
  - Modern economies are substantially more credit dependent (especially long-term credit) than in the classical gold-standard era (Bernanke, 2000).
  - Deflation’s impact through the balance sheet channel by raising real debt burdens is therefore potentially more pronounced.
  - Long periods of rising prices have historically anchored expectations; agents often do not foresee deflation until it materializes.

### III. DEVELOPMENT OF DEFLATION IN JAPAN
- Historical summary:
  - In the fifteen years prior to the onset of deflation in the mid-1990s, Japan’s annual core inflation rate averaged just 2.1 percent.
  - In contrast, over the same period the United States’ core inflation averaged nearly 4 percent.
  - During the mid-1980s, core inflation (CPI excluding food) in Japan fell to near zero, followed by several quarters of decline in the GDP deflator.
  - Factors associated with low inflation in mid-1980s: fairly tight monetary policy after oil shocks, rapid capacity buildup, appreciating yen, and gradual removal of trade barriers.
- Late-1980s boom and asset prices:
  - Mid- to late-1980s: strong growth amid an asset price boom but only modest CPI inflation.
  - Broad indexes of land and equity prices peaked at the end of the decade, at four to five times their levels in 1980.
  - Economy overheated, operating at 2–3 percentage points over potential GDP in the late-1980s and early-1990s, but goods and services prices rose only moderately.
  - Core inflation peaked at slightly above 3 percent in early 1991, then trended down.
- Bust and aftermath:
  - Collapse of asset prices and private demand followed tightening of monetary policy and market fatigue.
  - Bernanke and Gertler (2001) argue monetary policy was behind the curve during the boom-bust cycle; this view has been questioned by Okina and Shiratsuka (2001).
  - Unraveling of the asset price boom led to:
    - Sharp fall in land and equity prices.
    - Real GDP growth slowed markedly.
    - Widening output gap and sizable excess capacity in manufacturing and construction, exerting downward pressure on prices.
    - Banks burdened by bad loans to real estate and construction firms, reducing new lending.
    - Broad money (M2+CDs) growth declined from over 11 percent in 1990 to 0.6 percent in 1992.
- Price sector specifics:
  - With demand in sharp decline, inflationary pressures virtually dissipated.
  - Tradable sector prices affected by further opening of the economy and competitive pressures.
  - Nontradable sector prices faced downward pressures owing to deregulation and innovations.
- Monetary policy response:
  - Bank of Japan eased monetary policy, lowering the uncollateralized overnight call rate (uncollateralized overnight call rate was maintained at around 4 percent between 1986 and 1989).

### Deflation in Japan: timeline and broad characteristics
- Core CPI deflation materialized fully in 1998 with the onset of a recession; the GDP deflator began its near-continuous decline earlier (in 1995).
- Short-lived recovery around the Y2K-related investment boom in the late-1990s did little to arrest deflation.
- Asset prices continued to decline, with both land and equity prices sitting at two-decade lows in mid-2003.
- Inflation expectations:
  - Remained positive until the beginning of actual price declines, then turned negative and became entrenched.
  - Surveys (e.g., the 2002 Nissan Business Conditions Survey with over 3,000 companies) reported that over 80 percent of respondents reported declining sales prices and felt deflation was harmful to their businesses; nearly 40 percent saw deflation continuing for at least three more years.
- Two key characteristics:
  - (i) Deflation has been hard to predict:
    - Official and private forecasts consistently failed to anticipate occurrence, magnitude, and duration of deflation (Ahearne et al (2002)).
    - Time-series forecasting exercise using monthly CPI (excluding food, and adjusted for the impact of the increase in consumption tax in 1989 and 1997) year-on-year inflation data from 1980 to present shows a distinct set of dynamics:
      - One step-ahead forecasts from a regression with a single lagged dependent variable: through the 1980s and first half of the 1990s, positive forecast errors were consistently equal or greater than negative errors; incidence of negative errors mounted in the late-1990s.
      - Contrasting three-year periods: during 1981–83 and 1993–95 forecast errors were roughly equally frequent; over 60 percent of observations had negative errors between 2000–2002.
      - In the 12-month period starting from October 1997 (when prices began to decline continuously), there were no positive forecast errors.
  - (ii) Deflation has been broad-based:
    - Very few CPI items experienced price increases or stability.
    - Items with declining trends include clothes and footwear, furniture, transportation and communication, private housing rent, reading and recreation.
    - The GDP deflator has declined even more than the CPI.
    - Likely causes include a combination of international competition, banking sector difficulties, insufficiently loose monetary policy, and stagnant demand rather than sector-specific factors alone.

### Monetary policy constraints and costs (zero bound)
- Bank of Japan actions:
  - Lowered short-term interest rates to their floor.
  - Since March 2001 pursued a quantitative easing framework targeting bank and non-bank current account balances.
  - Monetary base grew by 58 percent between March 2002 and June 2003.
  - Liquidity injection has so far fallen short of reviving inflationary expectations.
- Zero bound implications:
  - At the zero bound it is more challenging for the central bank to successfully guide inflation expectations.
  - Yates (2002): zero bound need not be costly if no large output gap, but costly when output gap widens and deflationary pressures mount (as in Japan).
  - Taylor rule analysis (following McCallum (2003)):
    - Taylor rule prescription used: R = (long-run real rate assumed 3 percent; inflation target 2 percent), functional form presented as: )(5.0)2(5.03 tt a t a tt yyppR − + − ∆ + ∆ +=
    - Using quarterly data from 1990 to 2002, Taylor rule estimates suggest the need for nominal rates to be negative since the second quarter of 1998 (not feasible).
    - Indicates importance of short-term real rates being negative to stimulate the economy.
- Output cost estimation from zero bound:
  - Interest rate elasticity of output growth estimated around -0.5 (1991 to 2002): a 100 basis points cut in the short-term interest rate raises real GDP growth by 0.5 percent with a one quarter lag.
  - Taylor-rule estimates suggested rate cuts between 100 and 200 basis points from mid-1998.
  - Simulated alternative GDP path incorporating these cuts implies a cumulative loss of about 6 percent of GDP through 2002.
- Implication: an end to deflation is required to bring about negative real interest rates; with nominal rates at floor, only a return to inflation and inflation expectations can accomplish this.

### Labor market and wage rigidity
- Real wage developments:
  - During the 1980s, real wages grew by about 15 percent in an environment of economic growth and price stability.
  - Post-bubble years: growth slowed and price pressures dissipated; real wages did not adjust commensurately.
  - Wages began to adjust in real terms from 1998 onward, but only gradually.
  - Real wages are still at about (or above) the levels prevailing at the peak of the bubble around 1990.
- Empirical evidence:
  - Kuroda and Yamamoto (2003a and 2003b) find nominal wage change distributions statistically skewed to the right, indicative of downward wage rigidity.
- Labor market outcomes:
  - Relative inflexibility of wages has squeezed corporate profits and may have partly been reflected in rising unemployment.
  - Unemployment rose from 2½ percent in 1993 to around 5½ percent in early 2003.
  - Social and economic costs include rising real expenditures on safety net measures.
- Firms’ wage adjustment methods:
  - Switching from hiring full-time to part-time workers (part-time compensation tends to be lower).
  - Cutting significantly the bonus component, reducing overall compensation.

### Financial sector effects
- Low nominal interest rates necessary to stabilize financial system and prevent further deflation acceleration; higher interest rates could have severe negative impacts.
- Adverse side effects of low rates:
  - Most debt instruments do not adjust for deflation; nominal returns on deposits and bonds are not designed to fall below zero, placing a floor on real returns at the zero bound.
  - Interbank money market activity dampened as zero short-term rates cause transaction costs to outweigh returns.
  - Flattening yield curve compressed credit spreads, pressuring bank profitability.
  - Deposit rates close to floor reduced franchise value of retail banking.
  - Difficulty in price discovery and pricing of risk; increased bilateral trades or private placements instead of brokered trades.
  - Reduced liquidity in interest-rate-hedging markets: JGB futures market average open interest positions (weekly basis) fell by 66 percent between 1997 and May-2003.
- Search costs for information and hedging have been magnified substantially.

### Redistribution of wealth and borrower effects
- Unanticipated deflation implies transfer of resources from debtors to creditors.
- Illustrative exercise (Bernanke (2000) framework):
  - Assume a borrower took a ten-year loan in 1997 at interest rate 2.1 percent (yield on long-term government bonds then).
  - Borrower’s expected inflation ~1.1 percent per year (average of previous decade).
  - Through 2003, borrower’s real obligations would have been 12 percent higher than anticipated.
  - Allowing for trend decline in deflation in coming years, by loan maturity in 2007 the real debt burden would be about 20 percent higher.
  - If the loan were refinanced in 2000 as interest rates came down and deflation set in, the borrower’s debt burden would still be about 7 percent higher by 2003 than initially anticipated.
- Collateral value decline:
  - If using land as collateral and expecting real value maintenance, real value would be 34 percent lower than anticipated through 2003.
- Combined effects:
  - Increase in real debt burden and decrease in collateral value harmed financial intermediaries.
  - Indebted households and corporates curtailed spending and investment; bankruptcies increased; banks saddled with bad loans with substantially lower recovery value.

### Fiscal costs and government finances
- Public gross debt stock rose from around 70 percent of GDP in the early 1990s to 160 percent of GDP at end-2002.
- Government attempts to revive the economy via tax cuts and spending measures contributed to the debt increase, but deflation was also a key factor:
  - Government’s real debt burden increased owing to unanticipated deflation.
  - Revenues declined alongside contracting nominal GDP while expenditures continued to rise.
  - Declining prices and weak activity put severe downward pressure on revenue collection.
- Government finances and vulnerability:
  - Government revenue items decline when prices fall; expenditure items may not be indexed downwardly to reflect deflation.
  - Social security payments rise to the extent that unemployment increases accompany deflation; even without rising unemployment, the real burden of such payments would increase with continued deflation.
  - An indexation scheme for social security payments existed in Japan but was suspended between 1998 and 2002; otherwise the scheme would have necessitated a reduction in such payments to reflect the decline in the price level.
- Fiscal cost of deflation and government borrowing:
  - Between 1997 and 1999, the government of Japan issued bonds worth about 31 percent of GDP.
  - The unanticipated increase in the real debt burden owing to deflation from just these three years' borrowing is estimated to amount to over 3 percent of GDP through 2003.
  - Calculations are a better approximation for Japanese government debt than for household debt, since government debt has been serviced without refinancing (barring some smoothing operation-related buyback of JGBs).
  - Some of the unanticipated increase in the real debt burden may have been mitigated by the Bank of Japan’s purchasing of government bonds during this period, as some of the higher real payments owed on the bonds would have been offset through profit transfers from the BoJ’s bond portfolio.
- Additional fiscal markers cited in the source:
  - 31.8 percent in 1992 to 29.3 percent in 2002, a very large decline given the size of the [text context].

### Macroeconomic and financial sector effects (summary)
- Unanticipated deflation has led to substantial transfers of resources from debtors to creditors; however, creditors have not fully benefited because the increased debt burden, compounded by falling collateral value, has contributed to defaults and diminished loan recovery value.
- Wage rigidities have caused unemployment to mount in the presence of deflation.
- Interest rates reaching their floor have hampered the normal intermediation process in the financial sector.
- Deflation has raised the public debt burden substantially and constrained monetary policy.

### Policy implications and recommendations
- Japan’s ongoing experience is a warning to policy-makers elsewhere about the costs of even mild deflation and the need to prevent it from manifesting rather than face the challenge of curing it.
- For Japan, the paper’s clear lesson is: deflation, however mild, continues to impose significant costs on the economy.
- Policies to revive inflation expectations are therefore critically needed.

### Empirical evidence (CPI components)
- The document presents year-on-year percentage change series for Japan’s Consumer Price Index and components (Housing; Fuel, Light and Water Charge; Furniture and Household Utensils; Clothes and Footwear; Medical Care; Transportation and Communication; Education; Reading and Recreation; Food; Private House Rent) covering the period Jan-90 through Jan-03, sourced from the CEIC database and not corrected for the 1997 increase in consumption tax.

*Source: Excerpt from IMF Working Paper content unit _wp03215.*

### References..............................................................................................................

### _wp03215 - References..............................................................................................................

### I. INTRODUCTION
- Deflation has continued unabated in Japan for over half a decade.
- Price measures:
  - GDP deflator has fallen by over 9 percent since 1995.
  - CPI has fallen by 3 percent since 1998.
- Central questions examined:
  - Have the persistent declines in the price level been very costly for Japan?
  - Or is deflation broadly tolerable?
- Observations and debate:
  - Some observers downplay the impact of the relatively modest decline in the price level vis-à-vis the sharp fall in asset prices through the 1990s.
  - It has been suggested that mild deflation might be a sign of price stability and entail very little cost.
  - On the policy front, necessity of vigorous response to deflation has been questioned because potential costs of aggressive policies may be higher than costs of allowing deflation to run under existing policy frameworks.
- Counterarguments:
  - Prolonged, unanticipated deflation has impeded monetary policy efficacy, hampered financial market activities, squeezed corporate profitability, and raised the real burden of private and public debt.
- Organization of the paper:
  - Section II: theoretical arguments on the cost of deflation.
  - Section III: historical price-related developments in Japan.
  - Section IV: evidence on the impact of deflation on monetary policy-making, labor market, financial market, households, and the public sector.
  - Section V: concluding comments.

### II. CONCEPTUAL ISSUES
- Core asymmetry:
  - The clearest asymmetry between inflation and deflation is the problem of a zero-bound on interest rates that tends to be associated with deflationary episodes.
- Costs depend on source, extent, and duration of deflation.
  - Temporary price declines due to strong aggregate supply expansion or productivity spurts may not be too problematic.
  - Positive external shocks (falling import prices, aggressive trade liberalization) could push down domestic prices without entrenching deflationary expectations.
  - Recent productivity increases from information technology, globalization, and deregulation have reduced inflationary pressures; similar productivity spurts mattered in 19th-century deflationary episodes.
- Aggregate demand-driven deflation and nominal rigidities:
  - Nominal rigidities in financial markets and the labor market imply aggregate demand-driven deflation is likely costly.
  - Estimates due to the zero-interest rate floor are significant but hard to quantify precisely.
  - Studies of downward rigidity of nominal wages suggest deflation could impose costs amounting to several percentage points of GDP (Akerlof et. al., 1996).

- Collateral and balance sheet effects:
  - Fisher’s (1933) debt-deflation theory: unanticipated deflation redistributes wealth from debtors to creditors and reduces collateral values, especially when deflation accompanies reductions in asset prices.
  - As collateral loses value and bankruptcies rise, banks may raise financing charges (external finance premium rises) or cut back lending, reducing aggregate demand.

- Sticky wages:
  - Aggregate demand-induced deflation raises unemployment when nominal wages are rigid downward.
  - With sticky wages, price declines cause real wages to rise, profit margins to fall, and employment to be cut back.
  - Akerlof et. al. (1996) estimate that with a sustained 1 percent deflation and downward rigidities in nominal wages, unemployment in the United States could rise from a long-run equilibrium rate of 5.8 percent to 10.0 percent.
  - Phillips curve estimates suggest output losses could amount to a multiple of the roughly 4 percentage point loss in employment.
  - Other studies find smaller but still significant costs.

- Deflation versus inflation or disinflation:
  - Inflexibilities in financial and labor markets create asymmetry: sustained deflation can have substantial adverse effects while moving from zero to very low inflation rates (say below 2 percent) likely has negligible efficiency losses.
  - Zero interest rate floor is more likely to bind under deflation than disinflation.
  - Collateral and balance sheet effects and credit cutbacks are likely stronger under unexpected deflation.
  - Deflation more likely reveals macroeconomic imbalances manifesting in a generalized decline in the price level.

- Institutional factors:
  - Modern economies are substantially more credit dependent (especially long-term credit) than in the classical gold-standard era (Bernanke, 2000).
  - Deflation’s impact through the balance sheet channel by raising real debt burdens is therefore potentially more pronounced.
  - Long periods of rising prices have historically anchored expectations; agents often do not foresee deflation until it materializes.

### III. DEVELOPMENT OF DEFLATION IN JAPAN
- Historical summary:
  - In the fifteen years prior to the onset of deflation in the mid-1990s, Japan’s annual core inflation rate averaged just 2.1 percent.
  - In contrast, over the same period the United States’ core inflation averaged nearly 4 percent.
  - During the mid-1980s, core inflation (CPI excluding food) in Japan fell to near zero, followed by several quarters of decline in the GDP deflator.
  - Factors associated with low inflation in mid-1980s: fairly tight monetary policy after oil shocks, rapid capacity buildup, appreciating yen, and gradual removal of trade barriers.
- Late-1980s boom and asset prices:
  - Mid- to late-1980s: strong growth amid an asset price boom but only modest CPI inflation.
  - Broad indexes of land and equity prices peaked at the end of the decade, at four to five times their levels in 1980.
  - Economy overheated, operating at 2–3 percentage points over potential GDP in the late-1980s and early-1990s, but goods and services prices rose only moderately.
  - Core inflation peaked at slightly above 3 percent in early 1991, then trended down.
- Bust and aftermath:
  - Collapse of asset prices and private demand followed tightening of monetary policy and market fatigue.
  - Bernanke and Gertler (2001) argue monetary policy was behind the curve during the boom-bust cycle (central bank tightened too late during the bubble and delayed easing thereafter); this has been questioned by Okina and Shiratsuka (2001) on ex ante data grounds.
  - Unraveling of the asset price boom led to:
    - Sharp fall in land and equity prices.
    - Real GDP growth slowed markedly.
    - Widening output gap and sizable excess capacity in manufacturing and construction, exerting downward pressure on prices.
    - Banks burdened by bad loans to real estate and construction firms, reducing new lending.
    - Broad money (M2+CDs) growth declined from over 11 percent in 1990 to 0.6 percent in 1992.
- Price sector specifics:
  - With demand in sharp decline, inflationary pressures virtually dissipated.
  - Prices in the tradable sector were affected by further opening of the economy and competitive pressures.
  - Prices in the nontradable sector faced downward pressures owing to deregulation and innovations.
- Monetary policy response:
  - Bank of Japan eased monetary policy, lowering the uncollateralized overnight call rate (un-collateralized overnight call rate was maintained at around 4 percent between 1986 and 1989).

*Source: _wp03215 - References..............................................................................................................*

### 8.5 percent in early 1991 to 0.5 percent by late 1995, but that proved to be insufficient in the

### _wp03215 - 8.5 percent in early 1991 to 0.5 percent by late 1995, but that proved to be insufficient in the

### Deflation in Japan: timeline and broad characteristics
- Core CPI deflation materialized fully in 1998 with the onset of a recession; the GDP deflator began its near-continuous decline earlier (in 1995).
- Short-lived recovery around the Y2K-related investment boom in the late-1990s did little to arrest deflation.
- Asset prices continued to decline, with both land and equity prices sitting at two-decade lows in mid-2003.
- Inflation expectations:
  - Remained positive until the beginning of actual price declines, then turned negative and became entrenched.
  - Surveys (e.g., the 2002 Nissan Business Conditions Survey with over 3,000 companies) reported that over 80 percent of respondents reported declining sales prices and felt deflation was harmful to their businesses; nearly 40 percent saw deflation continuing for at least three more years.
- Two key characteristics:
  - (i) Deflation has been hard to predict:
    - Official and private forecasts consistently failed to anticipate occurrence, magnitude, and duration of deflation (Ahearne et al (2002)).
    - Time-series forecasting exercise using monthly CPI (excluding food, and adjusted for the impact of the increase in consumption tax in 1989 and 1997) year-on-year inflation data from 1980 to present shows a distinct set of dynamics:
      - One step-ahead forecasts from a regression with a single lagged dependent variable: through the 1980s and first half of the 1990s, positive forecast errors were consistently equal or greater than negative errors; incidence of negative errors mounted in the late-1990s.
      - Contrasting three-year periods: during 1981–83 and 1993–95 forecast errors were roughly equally frequent; over 60 percent of observations had negative errors between 2000–2002.
      - In the 12-month period starting from October 1997 (when prices began to decline continuously), there were no positive forecast errors.
  - (ii) Deflation has been broad-based:
    - Very few CPI items experienced price increases or stability.
    - Items with declining trends include clothes and footwear, furniture, transportation and communication, private housing rent, reading and recreation.
    - The GDP deflator has declined even more than the CPI.
    - Likely causes include a combination of international competition, banking sector difficulties, insufficiently loose monetary policy, and stagnant demand rather than sector-specific factors alone.

### Monetary policy constraints and costs (zero bound)
- Bank of Japan actions:
  - Lowered short-term interest rates to their floor.
  - Since March 2001 pursued a quantitative easing framework targeting bank and non-bank current account balances.
  - Monetary base grew by 58 percent between March 2002 and June 2003.
  - Liquidity injection has so far fallen short of reviving inflationary expectations.
- Zero bound implications:
  - At the zero bound it is more challenging for the central bank to successfully guide inflation expectations.
  - Yates (2002): zero bound need not be costly if no large output gap, but costly when output gap widens and deflationary pressures mount (as in Japan).
  - Taylor rule analysis (following McCallum (2003)):
    - Taylor rule prescription used: R = (long-run real rate assumed 3 percent; inflation target 2 percent), functional form presented as: )(5.0)2(5.03 tt a t a tt yyppR − + − ∆ + ∆ +=
    - Using quarterly data from 1990 to 2002, Taylor rule estimates suggest the need for nominal rates to be negative since the second quarter of 1998 (not feasible).
    - Indicates importance of short-term real rates being negative to stimulate the economy.
- Output cost estimation from zero bound:
  - Interest rate elasticity of output growth estimated around -0.5 (1991 to 2002): a 100 basis points cut in the short-term interest rate raises real GDP growth by 0.5 percent with a one quarter lag.
  - Taylor-rule estimates suggested rate cuts between 100 and 200 basis points from mid-1998.
  - Simulated alternative GDP path incorporating these cuts implies a cumulative loss of about 6 percent of GDP through 2002.
  - Figure reference: real GDP data normalized to 1998=100; simulated alternative path vs actual shows divergence for 1998–2002.
- Implication: an end to deflation is required to bring about negative real interest rates; with nominal rates at floor, only a return to inflation and inflation expectations can accomplish this.

### Labor market and wage rigidity
- Real wage developments:
  - During the 1980s, real wages grew by about 15 percent in an environment of economic growth and price stability.
  - Post-bubble years: growth slowed and price pressures dissipated; real wages did not adjust commensurately.
  - Wages began to adjust in real terms from 1998 onward, but only gradually.
  - Real wages are still at about (or above) the levels prevailing at the peak of the bubble around 1990.
- Empirical evidence:
  - Kuroda and Yamamoto (2003a and 2003b) find nominal wage change distributions statistically skewed to the right, indicative of downward wage rigidity.
- Labor market outcomes:
  - Relative inflexibility of wages has squeezed corporate profits and may have partly been reflected in rising unemployment.
  - Unemployment rose from 2½ percent in 1993 to around 5½ percent in early 2003.
  - Social and economic costs include rising real expenditures on safety net measures.
- Firms’ wage adjustment methods:
  - Switching from hiring full-time to part-time workers (part-time compensation tends to be lower).
  - Cutting significantly the bonus component, reducing overall compensation.

### Financial sector effects
- Low nominal interest rates necessary to stabilize financial system and prevent further deflation acceleration; higher interest rates could have severe negative impacts.
- Adverse side effects of low rates:
  - Most debt instruments do not adjust for deflation; nominal returns on deposits and bonds are not designed to fall below zero, placing a floor on real returns at the zero bound.
  - Interbank money market activity dampened as zero short-term rates cause transaction costs to outweigh returns.
  - Flattening yield curve compressed credit spreads, pressuring bank profitability.
  - Deposit rates close to floor reduced franchise value of retail banking.
  - Difficulty in price discovery and pricing of risk; increased bilateral trades or private placements instead of brokered trades.
  - Reduced liquidity in interest-rate-hedging markets: JGB futures market average open interest positions (weekly basis) fell by 66 percent between 1997 and May-2003.
- Search costs for information and hedging have been magnified substantially.

### Redistribution of wealth and borrower effects
- Unanticipated deflation implies transfer of resources from debtors to creditors.
- Illustrative exercise (Bernanke (2000) framework):
  - Assume a borrower took a ten-year loan in 1997 at interest rate 2.1 percent (yield on long-term government bonds then).
  - Borrower’s expected inflation ~1.1 percent per year (average of previous decade).
  - Through 2003, borrower’s real obligations would have been 12 percent higher than anticipated.
  - Allowing for trend decline in deflation in coming years, by loan maturity in 2007 the real debt burden would be about 20 percent higher.
  - If the loan were refinanced in 2000 as interest rates came down and deflation set in, the borrower’s debt burden would still be about 7 percent higher by 2003 than initially anticipated.
- Collateral value decline:
  - If using land as collateral and expecting real value maintenance, real value would be 34 percent lower than anticipated through 2003.
- Combined effects:
  - Increase in real debt burden and decrease in collateral value harmed financial intermediaries.
  - Indebted households and corporates curtailed spending and investment; bankruptcies increased; banks saddled with bad loans with substantially lower recovery value.

### Fiscal costs
- Public gross debt stock rose from around 70 percent of GDP in the early 1990s to 160 percent of GDP at end-2002.
- Government attempts to revive the economy via tax cuts and spending measures contributed to the debt increase, but deflation was also a key factor:
  - Government’s real debt burden increased owing to unanticipated deflation.
  - Revenues declined alongside contracting nominal GDP while expenditures continued to rise.
  - Declining prices and weak activity put severe downward pressure on revenue collection; revenue as a percentage of GDP fell (text cuts off mid-sentence in source).

*Source: Excerpt from IMF Working Paper content unit _wp03215*

### 31.8 percent in 1992 to 29.3 percent in 2002, a very large decline given the size of the

### 31.8 percent in 1992 to 29.3 percent in 2002, a very large decline given the size of the

### Government finances and vulnerability to deflation
- Government revenue items decline when prices fall; expenditure items may not be indexed downwardly to reflect deflation.
- Social security payments rise to the extent that unemployment increases accompany deflation; even without rising unemployment, the real burden of such payments would increase with continued deflation.
- An indexation scheme for social security payments existed in Japan but was suspended between 1998 and 2002; otherwise the scheme would have necessitated a reduction in such payments to reflect the decline in the price level.

### Fiscal cost of deflation and government borrowing
- Between 1997 and 1999, the government of Japan issued bonds worth about 31 percent of GDP.
- Using the approach in the previous section, the unanticipated increase in the real debt burden owing to deflation from just these three years' borrowing is estimated to amount to over 3 percent of GDP through 2003.
- The calculations to obtain the increase in real debt burden are a better approximation of reality for Japanese government debt than for household debt, since government debt has been serviced without refinancing (barring some smoothing operation-related buyback of JGBs).
- Some of the unanticipated increase in the real debt burden may have been mitigated by the Bank of Japan’s purchasing of government bonds during this period, as some of the higher real payments owed on the bonds would have been offset through profit transfers from the BoJ’s bond portfolio.

### Macroeconomic and financial sector effects
- Unanticipated deflation has led to substantial transfers of resources from debtors to creditors; however, creditors have not fully benefited because the increased debt burden, compounded by falling collateral value, has contributed to defaults and diminished loan recovery value.
- Wage rigidities have caused unemployment to mount in the presence of deflation.
- Interest rates reaching their floor have hampered the normal intermediation process in the financial sector.
- Deflation has raised the public debt burden substantially and constrained monetary policy.

### Policy implications and recommendations
- Japan’s ongoing experience is a warning to policy-makers elsewhere about the costs of even mild deflation and the need to prevent it from manifesting rather than face the challenge of curing it.
- For Japan, the paper’s clear lesson is: deflation, however mild, continues to impose significant costs on the economy.
- Policies to revive inflation expectations are therefore critically needed.

### Empirical evidence (CPI components)
- The document presents year-on-year percentage change series for Japan’s Consumer Price Index and components (Housing; Fuel, Light and Water Charge; Furniture and Household Utensils; Clothes and Footwear; Medical Care; Transportation and Communication; Education; Reading and Recreation; Food; Private House Rent) covering the period Jan-90 through Jan-03, sourced from the CEIC database and not corrected for the 1997 increase in consumption tax.

*Source: IMF working paper excerpt (figures, notes, and text as provided).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2003/_wp03215.pdf_