## _wp1492

## Source details

**Canonical URL:** [_wp1492](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1492.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1492.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1492.pdf.json)

---

### Abstract and Central Claim
- Many central banks target an inflation rate near two percent; this essay argues policymakers would do better to target four percent inflation.
- Primary benefit: ease constraints from the zero bound on nominal interest rates, making economic downturns less severe.
- Cost assessment: four percent inflation does not harm an economy significantly, according to history and research cited.
- Context: discusses choice of a long-run target for the inflation rate; does not take a side on proposals for immediate but temporary increases in inflation.

### Rationale for Raising the Long-Run Inflation Target
- Central banks control long-run (steady-state) inflation and can choose a target π* to keep inflation close to this level on average.
- A higher π* raises long-run nominal interest rates (r* + π*), allowing larger cuts in nominal rates before hitting the zero bound.
- Example comparison:
  - If r* = 2% and π* = 2%, nominal rate can be reduced by up to 4 percentage points.
  - If r* = 2% and π* = 4%, nominal rate can be reduced by up to 6 percentage points.
- Real rate lower bound: because i ≥ 0, real rate r = i − π cannot fall below −π; a larger π reduces this lower bound and permits greater monetary stimulus.

### The Zero-Bound Problem (Historical Illustration)
- Liquidity-trap concept gained practical relevance in:
  - Japan in the 1990s.
  - The 2007-2009 financial crisis.
- Historical episodes and policy responses:
  - Japan: policy rate fell from 6% in 1992 to 0.1% in 1999 and stayed near zero until 2006.
  - United States (2007-2009): federal funds target fell from 5.25% in August 2007 to a range of 0 to 0.25% in December 2008; target remained in that range in mid-2014.
  - U.S. employment-population ratio fell from 63% in 2006 to 59% in 2009 and remained 59% in mid-2014.
- Quantitative easing provided limited stimulus; with near-zero nominal rates, further adverse shocks could push economies into prolonged high unemployment episodes because of the zero bound.

### Quantified Economic Impact (Back-of-the-Envelope)
- Ball (1999) dynamic IS calibration:
  - y_t = -(1.0) r_{t-1} + (0.8) y_{t-1}, where y is log output, r is real interest rate, time period = year.
- Counterfactual: If Fed had 4% target instead of 2% in 2000s (neutral away from zero bound), nominal and real rates would have been 2 percentage points higher prior to the crisis; in late 2008 Fed could have reduced rates by 2 additional points.
- Using the IS calibration:
  - If interest rates had been two points lower during 2009, output in 2010 would have been 2% higher.
  - Output gain for 2013 would be 5.9%.
  - Cumulative gain over 2010-2013 would be 16.4% of annual output.
  - Assuming Okun’s Law coefficient of one half, cumulative reduction in unemployment would be 8.2 percentage points.

### Future Risks from the Zero Bound — Historical Evidence
- Frequency and severity of zero-bound episodes matter for evaluating a higher inflation target.
- Key historical observations:
  - Two largest economies that adopted ~2% targets both hit the zero bound within twenty years (Japan and U.S.).
  - Rudebusch (2009): Taylor rule fitting pre-2008 U.S. policy implies federal funds rate of −5% in 2009, suggesting rates were 500 basis points above the level needed to restore full employment.
  - Eight U.S. recessions since 1960 provide evidence on initial core inflation, unemployment peaks, and lowest nominal and real federal funds rates.
- Recessions starting with low initial inflation (between 2% and 3%):
  - Three such recessions: 1960-61, 2001, and 2008-09.
  - 1960-61: nominal funds rate fell to 1.2% following the recession.
  - 2001: nominal funds rate fell to 1.0% in 2003 after the recession.
  - 2008-09: funds rate hit the zero bound.
  - Observation: mild recessions reduced rates to about +1%; severe recessions could push optimal rates below zero, implying the zero bound would bind in a typical recession starting at 2% inflation.
- Recessions starting with high initial inflation (above 4%):
  - Five of eight recessions since 1960 began with inflation above 4%; nominal rates did not approach zero in these cases.
  - Interpreting real rate lower bound as −π and accounting for typical declines in inflation during recessions (2–3% initial inflation episodes fell to ~1%), a 2% initial inflation target implies a realistic lower bound around −1% on the real rate.
  - Empirical minima for real rates in five high-inflation-start recessions:
    - 1973-75 and 1980: real rate fell below −4% (a −1% bound would have severely constrained policy).
    - 1969-70: minimum real rate was −2.3% (problematic for a −1% bound).
    - 1990-91: minimum real rate was −0.6% (near-miss with a −1% bound).
    - 1981-82: minimum real rate was above zero (only exception).
- Conclusion from history: with a 2% inflation target, the lower bound is likely to constrain policy in a large fraction of recessions; given eight recessions in 50 years, a 2% target appears to have large costs.

### The zero-bound debate (theoretical vs. historical)
- Some papers (Schmidt-Grohe and Uribe (2011)) find that interest rates will rarely hit zero if the inflation target is two percent.
- Other studies (Reifschneider and Williams (2000); Coibion et al (2012)) find zero-bound episodes occur with some frequency, but conclude the welfare costs are small.
- Counterarguments in the section:
  - Historical evidence on the zero bound suggests greater concern than some model-based results imply.
  - Results depend on forward-looking IS and Phillips curves of New Keynesian models; these may not capture inertia in real-world inflation and output (e.g. Mankiw, 2001; Rudd and Whelan, 2006; Barnes et al., 2011).
  - In New Keynesian models, expectations that the zero bound will cease to bind raise output and inflation immediately; the paper questions whether expected future policies have such strong real-world effects.
- Footnotes and qualifications:
  - Williams (2009) finds substantial costs of zero-bound episodes with a two percent inflation target, but that result depends on assumptions of high macroeconomic volatility and a low neutral real interest rate (see Woodford, 2009).
  - Some analyses use inflation specifications with greater inertia than the New Keynesian Phillips curve, but these specifications still do not fit the data; Reifschneider and Williams describe their inflation equation as a mix of the New Keynesian model and the Fuhrer-Moore (1995) model, which, as discussed by Mankiw (2000), do not capture the degree of inertia in real-world inflation.

### Price-level targeting as an alternative
- Proposal overview:
  - Some economists (Eggertsson and Woodford, 2003; Coibion et al., 2012) advocate a target for the price level rather than an inflation-rate target.
  - This policy produces low inflation on average but raises inflation temporarily if a zero-bound episode has pushed the price level below its long run path.
- Mechanism:
  - If the nominal interest rate hits zero under price-level targeting, expected inflation rises, reducing the real interest rate and boosting the economy out of a slump.
- Credibility concerns:
  - Success depends on whether the policy “can actually be made credible to the public, so that inflation expectations are affected in the desired way.” (Woodford (2009) quoted)
  - Historical experience offers limited guidance: policymakers have successfully promised lower inflation (disinflations of the 1980s and 1990s) but typically only achieved reductions after actual inflation fell due to monetary contractions causing deep recessions.
  - Empirical observations cited:
    - Expected inflation generally followed actual inflation with a lag; disinflation announcements seldom shifted expectations before actual disinflation.
    - Countries with explicit inflation targets did not achieve lower sacrifice ratios than other countries (Bernanke et al., 1999).
    - Countries with highly independent central banks had higher sacrifice ratios (Debelle and Fischer, 1994).
- Skepticism about promises of higher inflation:
  - Promises of higher inflation may be even less effective than promises of lower inflation, particularly if policymakers lack tools to raise inflation at zero interest rates.

### Opposition to higher inflation: assessing costs
- Commonly cited costs of inflation (Mishkin (2011)):
  - Distortions in cash holdings
  - Overinvestment in the financial sector
  - Greater uncertainty about relative prices and the aggregate price level
  - Distortions of the tax system
  - Redistributions of wealth
  - Difficulties in financial planning
- Empirical assessment:
  - Research motivated by the 1970s double-digit inflation has not produced a compelling case that inflation is highly harmful; Krugman (1997): efforts to measure costs yield “embarrassingly small numbers.”
  - Few empirical studies target costs of single-digit inflation because evidence is scarce for even double-digit inflation costs.
- Cross-country growth studies:
  - Common finding: inflation rates above some threshold reduce growth, but lower levels are neutral.
  - Threshold estimates vary considerably: 8% (Sarel, 1995) to 40% (Bruno and Easterly, 1996).
  - 4% is clearly below these estimated thresholds.

### Is 4% inflation destabilizing?
- Central bankers’ objections:
  - Concern that accepting 4% inflation may lead to higher or more volatile inflation and less stable inflation expectations (Bernanke, 2010a).
  - Bernanke (2010b) quote: long-run Fed credibility in keeping inflation low, around 2%, could be jeopardized by moving to 4%—risking escalation to 6% or higher and difficulty “to tie down expectations at 4%.”
  - Mishkin (2011) and Woodford (2010) echo similar concerns: history suggests stabilizing inflation at 4% may be harder than at 2%.
- Characterization of the objection:
  - Labeled the “addictive theory of inflation”: 4% may be harmless in itself but viewed as the first step toward uncontrollable higher inflation.
- Rebuttals and questions raised:
  - If policymakers can determine, explain, and carry out optimal policy, why could they not explain and maintain a target of 4% (contrasting Bernanke/Mishkin’s concerns with the practical argument for raising the target to mitigate zero-bound costs)?
  - Historical record does not clearly show that modest rises in inflation lead to runaway inflation; survey-based expectations have generally followed actual inflation with a lag, suggesting expectations are unlikely to overshoot modest target increases.

### Understanding policymakers’ attitudes
- Historical perspective:
  - Under Chairman Paul Volcker, after ending double-digit inflation, the Federal Reserve allowed inflation to settle at about 4% from 1985 through 1988 and did not try to reduce it further until inflation started rising at the end of 1988 (Romer and Romer, 1994).
  - Support for 2% targets grew in the 1990s (Canada, New Zealand, and others); in some countries, declines to near 2% were partly accidental (U.S. recessions of 1990-91 and 2001).
- Two reasons for current aversion to 4%:
  1. Fighting the last war: the scarring experience of 1970s high inflation led central bankers to prioritize preventing any repeat, exaggerating perceived dangers of inflation.
     - DeLong (1997) contrasted the earlier Fed’s fear of high unemployment (shadow of the Great Depression) with later fear of inflation.
  2. Academic theory: Kydland and Prescott (1977) and Rogoff (1985) provided theoretical justification for hawkish policies by highlighting dynamic inconsistency and advocating high inflation aversion; Bernanke (2004) describes these papers as intellectual grounding for hawkish monetary policy.

### Conclusion and policy proposition
- Current consensus: the optimal long-run inflation rate is about 2%; most advanced-economy central banks target near that level.
- Argument advanced in the paper:
  - Central banks would do better to target 4% inflation.
  - Raising the inflation target from 2% would ease the constraints of the zero bound on interest rates, reducing severity of economic downturns.
  - The important benefit would come at minimal cost because 4% inflation does not harm an economy significantly.
- Other recent proposals mentioned (not endorsed or fully analyzed here):
  - Immediate but temporary bursts of inflation to stimulate slumping economies (e.g., Japan).
  - Higher inflation in Germany to restore competitiveness in the Euro area periphery.
  - Rogoff (2008) advocated a temporary rise in U.S. inflation to 6% to reduce the real value of government debt.
- Research recommendation:
  - Future research should consider proposals for temporary or targeted inflation increases; the analysis in this paper does not address their desirability since their costs and benefits differ from raising the long-run inflation target.

*Source: WP/14/92, "The Case for a Long-Run Inflation Target of Four Percent", Laurence Ball, June 2014.*

### Section 1

### _wp1492 - Section 1

### Abstract and Central Claim
- Many central banks target an inflation rate near two percent; this essay argues policymakers would do better to target four percent inflation.
- Primary benefit: ease constraints from the zero bound on nominal interest rates, making economic downturns less severe.
- Cost assessment: four percent inflation does not harm an economy significantly, according to history and research cited.
- Context: discusses choice of a long-run target for the inflation rate; does not take a side on proposals for immediate but temporary increases in inflation.

### Rationale for Raising the Long-Run Inflation Target
- Central banks control long-run (steady-state) inflation and can choose a target π* to keep inflation close to this level on average.
- A higher π* raises long-run nominal interest rates (r* + π*), allowing larger cuts in nominal rates before hitting the zero bound.
- Example comparison:
  - If r* = 2% and π* = 2%, nominal rate can be reduced by up to 4 percentage points.
  - If r* = 2% and π* = 4%, nominal rate can be reduced by up to 6 percentage points.
- Real rate lower bound: because i ≥ 0, real rate r = i − π cannot fall below −π; a larger π reduces this lower bound and permits greater monetary stimulus.

### The Zero-Bound Problem (Historical Illustration)
- The liquidity-trap concept (Keynes, 1936) became practically relevant starting in Japan in the 1990s and during the 2007-2009 financial crisis.
- Historical episodes and policy responses:
  - Japan: policy rate fell from 6% in 1992 to 0.1% in 1999 and stayed near zero until 2006.
  - United States (2007-2009): federal funds target fell from 5.25% in August 2007 to a range of 0 to 0.25% in December 2008; target remained in that range in mid-2014.
  - U.S. employment-population ratio fell from 63% in 2006 to 59% in 2009 and remained 59% in mid-2014.
- Quantitative easing provided limited stimulus; with near-zero nominal rates, further adverse shocks could push economies into prolonged high unemployment episodes because of the zero bound.

### Quantified Economic Impact (Back-of-the-Envelope)
- Ball (1999) dynamic IS calibration:
  - y_t = -(1.0) r_{t-1} + (0.8) y_{t-1}, where y is log output, r is real interest rate, time period = year.
- Counterfactual: If Fed had 4% target instead of 2% in 2000s (neutral away from zero bound), nominal and real rates would have been 2 percentage points higher prior to the crisis; in late 2008 Fed could have reduced rates by 2 additional points.
- Using the IS calibration:
  - If interest rates had been two points lower during 2009, output in 2010 would have been 2% higher.
  - Output gain for 2013 would be 5.9%.
  - Cumulative gain over 2010-2013 would be 16.4% of annual output.
  - Assuming Okun’s Law coefficient of one half, cumulative reduction in unemployment would be 8.2 percentage points.

### Future Risks from the Zero Bound — Historical Evidence
- Frequency and severity of zero-bound episodes matter for evaluating a higher inflation target.
- Key historical observations:
  - Two largest economies that adopted ~2% targets both hit the zero bound within twenty years (Japan and U.S.).
  - Rudebusch (2009): Taylor rule fitting pre-2008 U.S. policy implies federal funds rate of −5% in 2009, suggesting rates were 500 basis points above the level needed to restore full employment.
  - Eight U.S. recessions since 1960 provide evidence on initial core inflation, unemployment peaks, and lowest nominal and real federal funds rates.
- Recessions starting with low initial inflation (between 2% and 3%):
  - Three such recessions: 1960-61, 2001, and 2008-09.
  - 1960-61: nominal funds rate fell to 1.2% following the recession.
  - 2001: nominal funds rate fell to 1.0% in 2003 after the recession.
  - 2008-09: funds rate hit the zero bound.
  - Observation: mild recessions reduced rates to about +1%; severe recessions could push optimal rates below zero, implying the zero bound would bind in a typical recession starting at 2% inflation.
- Recessions starting with high initial inflation (above 4%):
  - Five of eight recessions since 1960 began with inflation above 4%; nominal rates did not approach zero in these cases.
  - Interpreting real rate lower bound as −π and accounting for typical declines in inflation during recessions (2–3% initial inflation episodes fell to ~1%), a 2% initial inflation target implies a realistic lower bound around −1% on the real rate.
  - Empirical minima for real rates in five high-inflation-start recessions:
    - 1973-75 and 1980: real rate fell below −4% (a −1% bound would have severely constrained policy).
    - 1969-70: minimum real rate was −2.3% (problematic for a −1% bound).
    - 1990-91: minimum real rate was −0.6% (near-miss with a −1% bound).
    - 1981-82: minimum real rate was above zero (only exception).
- Conclusion from history: with a 2% inflation target, the lower bound is likely to constrain policy in a large fraction of recessions; given eight recessions in 50 years, a 2% target appears to have large costs.

### Theoretical Research (Brief Note)
- A growing literature uses New Keynesian models to quantify zero-bound risks; such theoretical models generally suggest smaller risks than the historical evidence implies.
- The paper views theoretical estimates as less credible than the historical record for gauging zero-bound frequency and severity.

*Source: WP/14/92, "The Case for a Long-Run Inflation Target of Four Percent", Laurence Ball, June 2014.*

### Section 2

### _wp1492 - Section 2

### The zero-bound debate
- Some papers (Schmidt-Grohe and Uribe (2011)) find that interest rates will rarely hit zero if the inflation target is two percent.
- Other studies (Reifschneider and Williams (2000); Coibion et al (2012)) find zero-bound episodes occur with some frequency, but conclude the welfare costs are small.
- Counterarguments in the section:
  - Historical evidence on the zero bound suggests greater concern than some model-based results imply.
  - Results depend on forward-looking IS and Phillips curves of New Keynesian models; these may not capture inertia in real-world inflation and output (e.g. Mankiw, 2001; Rudd and Whelan, 2006; Barnes et al., 2011).
  - In New Keynesian models, expectations that the zero bound will cease to bind raise output and inflation immediately; the paper questions whether expected future policies have such strong real-world effects.
- Footnotes and qualifications preserved from the source:
  - Williams (2009) finds substantial costs of zero-bound episodes with a two percent inflation target, but that result depends on assumptions of high macroeconomic volatility and a low neutral real interest rate (see Woodford, 2009).
  - Some analyses use inflation specifications with greater inertia than the New Keynesian Phillips curve, but these specifications still do not fit the data; Reifschneider and Williams describe their inflation equation as a mix of the New Keynesian model and the Fuhrer-Moore (1995) model, which, as discussed by Mankiw (2000), do not capture the degree of inertia in real-world inflation.

### Price-level targeting as an alternative
- Proposal overview:
  - Some economists (Eggertsson and Woodford, 2003; Coibion et al., 2012) advocate a target for the price level rather than an inflation-rate target.
  - This policy produces low inflation on average but raises inflation temporarily if a zero-bound episode has pushed the price level below its long run path.
- Mechanism:
  - If the nominal interest rate hits zero under price-level targeting, expected inflation rises, reducing the real interest rate and boosting the economy out of a slump.
- Credibility concerns:
  - Success depends on whether the policy “can actually be made credible to the public, so that inflation expectations are affected in the desired way.” (Woodford (2009) quoted)
  - Historical experience offers limited guidance: policymakers have successfully promised lower inflation (disinflations of the 1980s and 1990s) but typically only achieved reductions after actual inflation fell due to monetary contractions causing deep recessions.
  - Empirical observations cited:
    - Expected inflation generally followed actual inflation with a lag; disinflation announcements seldom shifted expectations before actual disinflation.
    - Countries with explicit inflation targets did not achieve lower sacrifice ratios than other countries (Bernanke et al., 1999).
    - Countries with highly independent central banks had higher sacrifice ratios (Debelle and Fischer, 1994).
- Skepticism about promises of higher inflation:
  - Promises of higher inflation may be even less effective than promises of lower inflation, particularly if policymakers lack tools to raise inflation at zero interest rates.

### Opposition to higher inflation: assessing costs
- Commonly cited costs of inflation (Mishkin (2011)):
  - Distortions in cash holdings
  - Overinvestment in the financial sector
  - Greater uncertainty about relative prices and the aggregate price level
  - Distortions of the tax system
  - Redistributions of wealth
  - Difficulties in financial planning
- Empirical assessment:
  - Research motivated by the 1970s double-digit inflation has not produced a compelling case that inflation is highly harmful; Krugman (1997): efforts to measure costs yield “embarrassingly small numbers.”
  - Few empirical studies target costs of single-digit inflation because evidence is scarce for even double-digit inflation costs.
- Cross-country growth studies:
  - Common finding: inflation rates above some threshold reduce growth, but lower levels are neutral.
  - Threshold estimates vary considerably: 8% (Sarel, 1995) to 40% (Bruno and Easterly, 1996).
  - 4% is clearly below these estimated thresholds.

### Is 4% inflation destabilizing?
- Central bankers’ objections:
  - Concern that accepting 4% inflation may lead to higher or more volatile inflation and less stable inflation expectations (Bernanke, 2010a).
  - Bernanke (2010b) quote: long-run Fed credibility in keeping inflation low, around 2%, could be jeopardized by moving to 4%—risking escalation to 6% or higher and difficulty “to tie down expectations at 4%.”
  - Mishkin (2011) and Woodford (2010) echo similar concerns: history suggests stabilizing inflation at 4% may be harder than at 2%.
- Characterization of the objection:
  - Labeled the “addictive theory of inflation”: 4% may be harmless in itself but viewed as the first step toward uncontrollable higher inflation.
- Rebuttals and questions raised:
  - If policymakers can determine, explain, and carry out optimal policy, why could they not explain and maintain a target of 4% (contrasting Bernanke/Mishkin’s concerns with the practical argument for raising the target to mitigate zero-bound costs)?
  - Historical record does not clearly show that modest rises in inflation lead to runaway inflation; survey-based expectations have generally followed actual inflation with a lag, suggesting expectations are unlikely to overshoot modest target increases.

### Understanding policymakers’ attitudes
- Historical perspective:
  - Under Chairman Paul Volcker, after ending double-digit inflation, the Federal Reserve allowed inflation to settle at about 4% from 1985 through 1988 and did not try to reduce it further until inflation started rising at the end of 1988 (Romer and Romer, 1994).
  - Support for 2% targets grew in the 1990s (Canada, New Zealand, and others); in some countries, declines to near 2% were partly accidental (U.S. recessions of 1990-91 and 2001).
- Two reasons for current aversion to 4%:
  1. Fighting the last war: the scarring experience of 1970s high inflation led central bankers to prioritize preventing any repeat, exaggerating perceived dangers of inflation.
     - DeLong (1997) contrasted the earlier Fed’s fear of high unemployment (shadow of the Great Depression) with later fear of inflation.
  2. Academic theory: Kydland and Prescott (1977) and Rogoff (1985) provided theoretical justification for hawkish policies by highlighting dynamic inconsistency and advocating high inflation aversion; Bernanke (2004) describes these papers as intellectual grounding for hawkish monetary policy.

### Conclusion and policy proposition
- Current consensus: the optimal long-run inflation rate is about 2%; most advanced-economy central banks target near that level.
- Argument advanced in the paper:
  - Central banks would do better to target 4% inflation.
  - Raising the inflation target from 2% would ease the constraints of the zero bound on interest rates, reducing severity of economic downturns.
  - The important benefit would come at minimal cost because 4% inflation does not harm an economy significantly.
- Other recent proposals mentioned (not endorsed or fully analyzed here):
  - Immediate but temporary bursts of inflation to stimulate slumping economies (e.g., Japan).
  - Higher inflation in Germany to restore competitiveness in the Euro area periphery.
  - Rogoff (2008) advocated a temporary rise in U.S. inflation to 6% to reduce the real value of government debt.
- Research recommendation:
  - Future research should consider proposals for temporary or targeted inflation increases; the analysis in this paper does not address their desirability since their costs and benefits differ from raising the long-run inflation target.

*Source: _wp1492 - Section 2*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1492.pdf_