## _wp15230

## Source details

**Canonical URL:** [_wp15230](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15230.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15230.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15230.pdf.json)

---

### I. Introduction
- Sample and scope:
  - Analysis covers roughly 50 years for 23 advanced economies.
  - Quarterly data starting in 1960 (or whenever available).
  - 122 recessions are analyzed (recessions of the 1960s and after 2010 excluded for lack of pre- and post-data).
- Research questions:
  - Do recessions have permanent effects on the level of output relative to trend (hysteresis / super-hysteresis)?
  - Has the relation between inflation and activity (the Phillips curve) deteriorated, explaining the “missing disinflation’’?
- Key summary findings:
  - About two-thirds of recessions are followed by lower output relative to the pre-recession trend.
  - About one-half of those sustained-gap cases (equivalently 33% of all recessions on average across specifications) are followed by lower output growth (an “increasing gap”).
  - Recessions associated with oil price increases or financial crises are more likely to be followed by lower output later, but many recessions plausibly triggered by demand shocks (including intentional disinflations) are also followed by sustained gaps.

### II. Aftereffects of Recessions: Hysteresis? (Methodology and Results)
- Methodology (non-parametric, recession-focused):
  - Recessions identified using Harding and Pagan (2002) peak/trough method.
  - Pre-recession trend extrapolated using two alternative anchors:
    - Start at two years before the recession.
    - Start at two years before the recession or earlier if unusual credit growth (Crowe et al. (2011) episodes: annual growth of credit-to-GDP > 10% and deviation > 1.5 standard deviations) binds.
  - Two alternative trend estimation windows:
    - Four-year (log-linear) trend (t0-9 to t0-24 in quarters).
    - Ten-year trend (t0-9 to t0-48 in quarters).
  - Adjustments made:
    - Trend slope adjusted downward by one standard deviation of the trend coefficient.
    - Adjustment for secular decline in growth: regress log GDP on linear and quadratic time trends over the whole sample; adjust pre-recession trend downward by coefficient on quadratic term times square of time difference.
  - Post-recession output gaps defined as average gaps from three to seven years after the recession: t1+12 to t1+27.
  - Classification rules:
    - Average gap during t1+12 to t1+27 non-positive → “no sustained gap”.
    - Average gap positive → “sustained gap”.
    - Regress gaps from t1+12 to t1+27 on constant and time trend; if estimated time trend is positive and significant at the 1% level → “increasing gap”.

- Aggregate quantitative results (across four combinations of trend-anchor and window):
  - 31% of recessions had NO sustained gap.
  - 69% of recessions had a sustained gap.
  - 47% of the sustained-gap cases (33% of all cases) had an increasing gap.
- Robustness using log real GDP per working-age population:
  - 31% no sustained gap, 69% sustained gap, 46% increasing output gap (roughly 30%+ of all recessions with increasing gap).
- Visual/diagnostic note:
  - In roughly 80% of “increasing output gap” classifications the increase was unambiguous.

- Explanations considered:
  1. Hysteresis / super-hysteresis mechanisms:
     - Labor market: prolonged unemployment causing permanent dropout or unemployability; institutional changes raising the natural rate.
     - Capital and R&D: reduced investment, debt overhang, reduced R&D leading to permanently lower productivity levels.
     - Reallocation effects: lower job creation/destruction reducing productivity growth.
     - Super-hysteresis (permanent lower growth) harder to justify broadly — would require persistent reductions in total factor productivity growth.
  2. Common supply shocks:
     - Oil price increases or financial crises could cause both the initial recession and subsequent lower output/growth.
     - Changes in financial intermediation or bank regulation (e.g., higher capital ratios) could lower long-run efficiency.
  3. Reverse causality / news about lower potential growth:
     - Anticipation or realization of lower underlying growth reduces consumption and investment, causing recessions.
     - Supporting facts: productivity declines at end of expansions; firms over-accumulate capital in expansions.

- Controlling by type of recession (selected quantitative breakdowns):
  - Financial crises (Laeven and Valencia definition):
    - 23 recessions coincided with financial crises.
    - Recessions with financial crises: on average 83% show sustained output gaps versus 66% without financial crises.
    - 35% of all financial-crisis-associated recessions are followed by an increasing output gap.
  - Oil price increases:
    - 18 recessions coincided with oil price increases (mostly during the 1970s).
    - Recessions with oil price increases: on average 90% show sustained gaps versus 65% without oil-price increases.
    - 76% of oil-price-increase recessions are followed by an increasing gap.
  - Inflation during recessions:
    - Recessions with increasing inflation (supply-type): 72% show sustained gaps versus 63% for recessions with decreasing inflation (demand-type).
  - Intentional disinflations:
    - 28 recessions identified as associated with intentional disinflations (large increase in nominal interest rates followed by disinflation).
    - Recessions with intentional disinflations: 63% show sustained gaps (versus 70% for others).
    - 20% of all intentional-disinflation cases are followed by lower output growth (increasing gap).
  - Relative output gaps (large recessions only):
    - For 68 recessions with output falls > 2 percent, the mean and median relative output gaps are between 1 and 2 (i.e., average gap larger than the initial fall), inconsistent with a simple hysteresis expectation of a ratio between 0 and 1.

- Main takeaways:
  - Two-thirds of recessions are followed by lower output relative to pre-recession trend.
  - Almost one-half of those sustained-gap cases are followed by lower output growth (increasing gap).
  - Larger proportions for supply-shock-associated recessions, but demand-shock and intentional-disinflation recessions also frequently show sustained gaps.
  - Likely all three explanations (hysteresis, supply shocks, reverse causality) are relevant; intentional-disinflation evidence is suggestive of hysteresis effects in some cases.

### III. Does Unemployment Affect Inflation? (Phillips Curve Evidence)
- Context and framing:
  - Post-crisis inflation decreased but less than anticipated (“missing disinflation”).
  - Expectations formation shifted: more weight on long-term expected inflation and central-bank perceived target, less on past inflation.
  - Shift from accelerationist Phillips curve to a level Phillips curve interpretation.

- Time-varying Kalman-filter estimation (country-by-country, quarterly CPI headline inflation, annualized):
  - Estimated specification includes time-varying weights on long-term expected inflation (λt) and a time-varying slope on the unemployment gap (θt), with controls for lagged inflation and import-price inflation.
  - Constraints: θt, μt ≥ 0; 0 ≤ λt ≤ 1; parameters follow constrained random walks.

- Median time-varying results (summary):
  - λt (weight on long-term expected inflation) has steadily increased since the mid-1970s.
  - θt (slope of the Phillips curve) has flattened over time, with most decline from mid-1970s to early 1990s; roughly constant since early 1990s and no further decrease during the crisis.
  - For most countries θt in recent years is small and statistically insignificant under the time-varying specification (one-standard-deviation bands reach zero for many countries by the mid-1990s).
    - By 2014, coefficient insignificant for all but four countries in the Kalman-filter time-varying estimates.

- Simpler constant-coefficient approach (motivated by stability since early 1990s):
  - Estimated: πt = λ eπt + θ (u* − u) + μ πm t with constant coefficients over 1990-2014 and 2007-2014.
  - Two choices for u*:
    - Time series u* from Kalman filter.
    - Constant u* = ū (sample average unemployment rate).
  - Key findings on θ (slope of Phillips curve):
    - θ typically significant at 95% confidence in simpler regressions:
      - Using u* from Kalman filter: significant in 15 out of 20 countries.
      - Using ū: significant in 14 out of 20 countries.
    - Countries where θ is not significant in either specification: Germany, the United Kingdom, Norway, Denmark.
    - Coefficients vary widely across countries (examples reported in source): Japan θ = -1.09, United States θ = -0.29, Spain θ = -0.08 (1990-2014, Kalman u* specification shown as examples).
    - Coefficients typically larger when u* is used (time-varying natural rate), likely because u* moves with u and reduces the measured gap.
  - Crisis-period stability:
    - Coefficients do not appear to have decreased during the crisis for most countries.
    - Exceptions with much smaller and insignificant coefficients over 2007-2014: Netherlands, Portugal.
  - Overall conclusion:
    - The slope of the Phillips curve decreased from mid-1970s to early 1990s, but has been roughly stable since; it remains significant in most countries under simpler constant-coefficient specifications.

### IV. Implications for Monetary Policy
- From hysteresis / aftereffects findings:
  - If recessions cause hysteresis/super-hysteresis:
    - Deviations of output from optimal are longer-lasting and costlier.
    - Monetary policy should react more strongly to output movements relative to inflation (i.e., be more aggressive early to limit unemployment increases and long-term unemployment).
    - Stabilizing inflation alone may not be optimal if increases in actual unemployment raise the natural rate, making unemployment gaps misleading.
  - If recessions reflect anticipation of lower potential growth (reverse causality):
    - Real-time estimates of potential output may be too optimistic, leading to overestimation of output gaps and an overly strong policy response to output movements during and after recessions.
    - Illustration in the source: if potential growth declines at time t and agents adjust with a lag, measured negative output gap in real time overstates true gap until expectations and behavior adjust.

- From Phillips curve findings:
  - The Phillips curve slope is smaller than pre-1990s levels but generally significant; no evidence of disappearance since the crisis.
  - Policy-rule implication:
    - Interest rate rules within an inflation-targeting framework should put relatively more weight on the unemployment/output gap versus inflation because stabilizing inflation may require very large unemployment movements and entail large welfare losses.
  - Broader policy implication:
    - Results reinforce the case for a dual mandate—stabilizing inflation and employment—since a small unemployment effect on inflation implies large employment movements would be required to stabilize inflation alone.

*Source: _wp15230 - 1. Analysis of the differences between output level and trend across recessions; Phillips curve evidence; monetary policy implications (PDF).*

### 1.  Analysis of the differences between output level and trend across recessions _________11

### 1.  Analysis of the differences between output level and trend across recessions _________11

### Major sections covered
- 1.  Analysis of the differences between output level and trend across recessions _________11
- 2.  Recessions with/without financial crises ______________________________________14
- 3.  Recessions with/without oil price increases  ___________________________________14
- 4.  Recessions with/without increasing inflation  __________________________________15
- 5.  Recessions with/without intentional disinflations _______________________________16
- 6.  Slope of the Phillips Curve  ________________________________________________23

### Figures included
- 1.  Advanced Economies Real GDP   ____________________________________________5
- 2.  Actual GDP and Extrapolated Trends _________________________________________7
- 3.  United States –Evolution of log real GDP and Extrapolated Trends __________________8
- 4.  Portugal –Evolution of log real GDP and Extrapolated Trends  _____________________9
- 5.  Histogram of Average Adjusted Output Gaps __________________________________10
- 6.  Histogram of Average Relative Output Gaps  __________________________________17
- 7.  Advanced Economies CPI Headline Inflation  _________________________________18
- 8.  Median estimates  _______________________________________________________20
- 9.  Estimates for Germany and the UK    ________________________________________21
- 10.  Decreases in growth, recessions, and output gaps  _____________________________26

*Source: _wp15230 - 1.  Analysis of the differences between output level and trend across recessions _________11 (PDF).*

### References ________________________________________________________________27

### _wp15230 - References ________________________________________________________________27

### I. Introduction
- Sample and scope:
  - Analysis covers roughly 50 years for 23 advanced economies.
  - Quarterly data starting in 1960 (or whenever available).
  - 122 recessions are analyzed (recessions of the 1960s and after 2010 excluded for lack of pre- and post-data).
- Research questions:
  - Do recessions have permanent effects on the level of output relative to trend (hysteresis / super-hysteresis)?
  - Has the relation between inflation and activity (the Phillips curve) deteriorated, explaining the “missing disinflation’’?
- Key summary findings:
  - About two-thirds of recessions are followed by lower output relative to the pre-recession trend.
  - About one-half of those sustained-gap cases (equivalently 33% of all recessions on average across specifications) are followed by lower output growth (an “increasing gap”).
  - Recessions associated with oil price increases or financial crises are more likely to be followed by lower output later, but many recessions plausibly triggered by demand shocks (including intentional disinflations) are also followed by sustained gaps.

---

### II. Aftereffects of Recessions: Hysteresis?
- Methodology (non-parametric, recession-focused):
  - Recessions identified using Harding and Pagan (2002) peak/trough method.
  - Pre-recession trend extrapolated using two alternative anchors:
    - Start at two years before the recession.
    - Start at two years before the recession or earlier if unusual credit growth (Crowe et al. (2011) episodes: annual growth of credit-to-GDP > 10% and deviation > 1.5 standard deviations) binds.
  - Two alternative trend estimation windows:
    - Four-year (log-linear) trend (t0-9 to t0-24 in quarters).
    - Ten-year trend (t0-9 to t0-48 in quarters).
  - Adjustments made:
    - Trend slope adjusted downward by one standard deviation of the trend coefficient.
    - Adjustment for secular decline in growth: regress log GDP on linear and quadratic time trends over the whole sample; adjust pre-recession trend downward by coefficient on quadratic term times square of time difference.
  - Post-recession output gaps defined as average gaps from three to seven years after the recession: t1+12 to t1+27.
  - Classification rules:
    - Average gap during t1+12 to t1+27 non-positive → “no sustained gap”.
    - Average gap positive → “sustained gap”.
    - Regress gaps from t1+12 to t1+27 on constant and time trend; if estimated time trend is positive and significant at the 1% level → “increasing gap” (i.e., gap widening over time).

- Aggregate quantitative results:
  - Across four combinations of trend-anchor and window: averages show:
    - 31% of recessions had NO sustained gap.
    - 69% of recessions had a sustained gap.
    - 47% of the sustained-gap cases (33% of all cases) had an increasing gap.
  - Robustness using log real GDP per working-age population:
    - Same proportions: 31% no sustained gap, 69% sustained gap, 46% increasing output gap (roughly 30%+ of all recessions with increasing gap).
  - Visual robustness checks: in roughly 80% of “increasing output gap” classifications the increase was unambiguous.

- Explanations considered:
  1. Hysteresis / super-hysteresis mechanisms:
     - Labor market: prolonged unemployment causing permanent dropout or unemployability; institutional changes raising the natural rate.
     - Capital and R&D: reduced investment, debt overhang, reduced R&D leading to permanently lower productivity levels.
     - Reallocation effects: lower job creation/destruction reducing productivity growth.
     - Super-hysteresis (permanent lower growth) harder to justify broadly — would require persistent reductions in total factor productivity growth.
  2. Common supply shocks:
     - Oil price increases or financial crises could cause both the initial recession and subsequent lower output/growth.
     - Changes in financial intermediation or bank regulation (e.g., higher capital ratios) could lower long-run efficiency.
  3. Reverse causality / news about lower potential growth:
     - Anticipation or realization of lower underlying growth reduces consumption and investment, causing recessions.
     - Supporting facts:
       - Productivity declines at end of expansions (Gordon 1980, 2003).
       - Firms over-accumulate capital in expansions (Beaudry et al. 2014).

- Controlling by type of recession (selected quantitative breakdowns):
  - Financial crises (Laeven and Valencia definition):
    - 23 recessions coincided with financial crises.
    - Recessions with financial crises: on average 83% show sustained output gaps versus 66% without financial crises.
    - 35% of all financial-crisis-associated recessions are followed by an increasing output gap.
  - Oil price increases:
    - 18 recessions coincided with oil price increases (mostly during the 1970s).
    - Recessions with oil price increases: on average 90% show sustained gaps versus 65% without oil-price increases.
    - 76% of oil-price-increase recessions are followed by an increasing gap.
  - Inflation during recessions:
    - Recessions with increasing inflation (supply-type): 72% show sustained gaps versus 63% for recessions with decreasing inflation (demand-type) — but recessions with decreasing inflation still show sustained gaps in 63% of cases.
  - Intentional disinflations:
    - 28 recessions identified as associated with intentional disinflations (large increase in nominal interest rates followed by disinflation).
    - Recessions with intentional disinflations: 63% show sustained gaps (versus 70% for others).
    - 20% of all intentional-disinflation cases are followed by lower output growth (increasing gap).
  - Relative output gaps (large recessions only):
    - For 68 recessions with output falls > 2 percent, the mean and median relative output gaps are between 1 and 2 (i.e., average gap larger than the initial fall), inconsistent with a simple hysteresis expectation of a ratio between 0 and 1.

- Main takeaways from this section:
  - Two-thirds of recessions are followed by lower output relative to pre-recession trend.
  - Almost one-half of those sustained-gap cases are followed by lower output growth (increasing gap).
  - Larger proportions for supply-shock-associated recessions, but demand-shock and intentional-disinflation recessions also frequently show sustained gaps.
  - Conclusion: likely all three explanations (hysteresis, supply shocks, reverse causality) are relevant; intentional-disinflation evidence is suggestive of hysteresis effects in some cases.

---

### III. Does Unemployment Affect Inflation? (Phillips Curve Evidence)
- Context:
  - Post-crisis inflation decreased but less than anticipated (“missing disinflation”).
  - Expectations formation shifted: more weight on long-term expected inflation and central-bank perceived target, less on past inflation.
  - Shift from accelerationist Phillips curve to a level Phillips curve interpretation.

- Time-varying Kalman-filter estimation (based on IMF WEO / Matheson and Stavrev (2013) specification):
  - Equation estimated for each country separately (quarterly CPI headline inflation, annualized) with time-varying parameters:
    - πt = λt eπt + (1−λt)πt−1* + θt (u* t − ut) + μt πm t + εt  (presentation condensed to match source variables)
    - Variables: πt (headline CPI inflation), ut (unemployment rate), u*t (natural rate), eπt (long-term inflation expectations), πt−1* (average of last four quarterly inflation rates), πm t (import price inflation relative to headline).
    - Parameters λt, θt, μt and u*t follow constrained random walks (θt, μt ≥ 0; 0 ≤ λt ≤ 1).
  - Median (across countries) results and interquartile ranges show:
    - λt (weight on long-term expected inflation) has steadily increased since the mid-1970s — short-run inflation expectations more stable.
    - θt (slope of the Phillips curve) has flattened over time, with most decline from mid-1970s to early 1990s; roughly constant since early 1990s and no further decrease during the crisis.
  - However, for most countries θt in recent years is small and statistically insignificant under the time-varying specification (one-standard-deviation bands reach zero for many countries by the mid-1990s).
    - By 2014, coefficient insignificant for all but four countries in the Kalman-filter time-varying estimates.

- Simpler constant-coefficient approach (1990-2014 and 2007-2014 samples):
  - Motivated by stable θt since early 1990s; drop lagged inflation given well-anchored expectations.
  - Estimated: πt = λ eπt + θ (u* − u) + μ πm t  with constant coefficients over 1990-2014 and 2007-2014.
  - Two assumptions for natural rate u*:
    - Time series u* from Kalman filter.
    - Constant u* = ū (sample average unemployment rate).
  - Key table results for θ (slope):
    - θ typically significant at 95% confidence in simpler regressions:
      - Using u* from Kalman filter: significant in 15 out of 20 countries.
      - Using ū: significant in 14 out of 20 countries.
    - Countries where θ is not significant in either specification: Germany, the United Kingdom, Norway, Denmark.
    - Coefficients vary widely: e.g., Japan θ = -1.09 (1990-2014, Kalman u*), United States θ = -0.29, Spain θ = -0.08 (examples from table).
    - Coefficients typically larger when u* is used (time-varying natural rate), likely because u* moves with u and reduces the measured gap.
  - Crisis-period stability:
    - Coefficients do not appear to have decreased during the crisis for most countries.
    - Exceptions with much smaller and insignificant coefficients over 2007-2014: Netherlands, Portugal.
  - Overall conclusion:
    - The slope of the Phillips curve decreased from mid-1970s to early 1990s, but has been roughly stable since; it remains significant in most countries under simpler constant-coefficient specifications.

---

### IV. Implications for Monetary Policy
- From hysteresis / aftereffects findings:
  - If recessions cause hysteresis/super-hysteresis:
    - Deviations of output from optimal are longer-lasting and costlier.
    - Monetary policy should react more strongly to output movements relative to inflation (i.e., be more aggressive early to limit unemployment increases and long-term unemployment).
    - Stabilizing inflation alone may not be optimal if increases in actual unemployment raise the natural rate, making unemployment gaps misleading.
  - If recessions reflect anticipation of lower potential growth (reverse causality):
    - Real-time estimates of potential output may be too optimistic, leading to overestimation of output gaps and an overly strong policy response to output movements during and after recessions.
    - Illustration: if potential growth declines at time t and agents adjust with a lag, measured negative output gap in real time overstates true gap until expectations and behavior adjust (Figure 10 conceptual discussion).
- From Phillips curve findings:
  - The Phillips curve slope is smaller than pre-1990s levels but generally significant; no evidence of disappearance since the crisis.
  - Implication for policy rules:
    - Interest rate rules within an inflation-targeting framework should put relatively more weight on the unemployment/output gap versus inflation because stabilizing inflation may require very large unemployment movements and entail large welfare losses.
  - A more general conclusion: these results reinforce the case for a dual mandate—stabilizing inflation and employment—since a small unemployment effect on inflation implies large employment movements would be required to stabilize inflation alone.

---

*Source: _wp15230 - References ________________________________________________________________27*

### References

### _wp15230 - References

### Financial intermediation, banking crises, and macrofinancial stability
- Adrian, T. and Shin, H.S (2010), “Financial intermediaries and monetary economics”, in Friedman, B. and Woodford, M. (eds.), Handbook Monetary Economics, Vol. 3, pp. 601-650.
- Crowe, C., Dell’Ariccia, G., Igan, D. and Rabanal, P. (2011), “Policies for macrofinancial stability: Options to deal with real estate booms”, IMF Staff Discussion Note, No 11/02.
- Ennis, H. and Keister, T. (2003), “Economic growth, liquidity, and bank runs”, Journal of Economic Theory, Vol. 109(2), pp. 220-245.
- Laeven, L. and Valencia, F. (2013), “Systemic banking crises database", IMF Economic Review, Vol. 61(2), pp. 225-270.
- Ramirez, C. (2009), “Bank fragility, “Money under the mattress”, and long run growth: US evidence from the perfect Panic of 1893”, Journal of Banking and Finance, Vol. 33, pp. 2185-2198.
- Reinhart, C. and Rogoff, K. (2009), “The aftermath of financial crises”, American Economic Review, Vol. 99(2), pp. 466-72.

### Recessions, potential output, and long-term effects
- Ball, L. (2014), “Long-term damage from the Great Recession in OECD countries”, NBER Working Paper, No 20185, May.
- Beaudry P., Galizia, D. and Portier, F (2014), “Reconciling Hayek's and Keynes' views of recessions”, CEPR Discussion Paper, No 9966; NBER Working Paper, No 20101.
- Cerra, V. and Saxena, S. (2008), “Growth dynamics: The myth of economic recovery,” American Economic Review, Vol. 98(1), pp. 439-457.
- Fernald, F. (2014), “Productivity and potential output before, during, and after the Great Recession”, in NBER Macroeconomics Annual 2014, Vol. 29.
- Haltmaier, J. (2012), ““Do recessions affect potential output?”, International Finance Discussion Paper, No1066, Federal Reserve Board, December.
- Martin and Wilson (2013), “Potential output and recessions: Are we fooling ourselves?”, manuscript.
- Matheson, T. and Stavrev, E. (2013), “The Great Recession and the inflation puzzle”, Economics Letters, Vol. 120(3), pp. 468-472.

### Unemployment, hysteresis, and labor market dynamics
- Blanchard, O. and Summers, L. (1986), "Hysteresis and European Unemployment", in Fischer, S. (ed.), NBER Macroeconomics Annual, MIT Press, September, pp. 15-77.
- Gali, J. (2015), “Hysteresis and the European unemployment problem revisited”, slides presented at the NBER summer institute, June.
- Broadbent, B. (2014), “Unemployment and the conduct of monetary policy in the UK”, Jackson Hole, August.

### Inflation, Phillips curve, and expectations
- Andrle, M., Bruha, J. and Solmaz, S. (2013), “Inflation and output comovement in the euro area: Love at second sight”, IMF Working Paper, No 13/192.
- Coibion, O. and Gorodnishenko, Y. (2015), “Is the Phillips curve alive and well after all? Inflation expectations and the missing disinflation’’, American Economic Journal, Vol. 7(1), pp.197-232.
- IMF World Economic Outlook (2013), Chapter 3, “The dog that didn’t bark: Has inflation been muzzled or was it just sleeping?”, April.
- Matheson, T. and Stavrev, E. (2013), “The Great Recession and the inflation puzzle”, Economics Letters, Vol. 120(3), pp. 468-472.

### Business cycles, news, and productivity
- Comin, D. and Gertler, M. (2006), “Medium-term business cycles”, American Economic Review, Vol. 96(3), pp. 523-551.
- Blanchard, O., L’Huillier, J.P. and Lorenzoni, G. (2013), “News, noise, and fluctuations: An empirical exploration”, American Economic Review, Vol. 103(7), December, pp. 3045-3070.
- Harding, D. and Pagan, A. (2002), "Dissecting the cycle: a methodological investigation," Journal of Monetary Economics, Vol. 49(2), pp. 365-381.
- Gordon, R. (2003), “Exploding productivity growth: Context, causes, and implications”, Brookings Papers on Economic Activity, Vol. 2, pp. 207-298.
- Fernald, F. (2014), “Productivity and potential output before, during, and after the Great Recession”, in NBER Macroeconomics Annual 2014, Vol. 29.

### Monetary policy theory
- Woodford, M. (2003), Interest and prices: Foundations of a theory of monetary policy, Princeton University Press.
- Adrian, T. and Shin, H.S (2010), “Financial intermediaries and monetary economics”, in Friedman, B. and Woodford, M. (eds.), Handbook Monetary Economics, Vol. 3, pp. 601-650.

### IMF analyses and reports
- IMF World Economic Outlook (2009), Chapter 3, “From recession to recovery: How soon and how strong?”, April.
- IMF World Economic Outlook (2013), Chapter 3, “The dog that didn’t bark: Has inflation been muzzled or was it just sleeping?”, April.

*References list from: _wp15230 - References*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15230.pdf_