## wpiea2020073-print-pdf

## Source details

**Canonical URL:** [wpiea2020073-print-pdf](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020073-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020073-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020073-print-pdf.pdf.json)

---

### I. INTRODUCTION
- COVID-19 pandemic context and immediate facts:
  - "As of end-May 2020, worldwide cases of coronavirus exceed 5.5 million and over 350 thousand deaths."
- Key macroeconomic concerns:
  - Containment measures (social distancing, lockdowns of non-essential businesses) have caused a global halt in economic activity; many countries likely to experience "the deepest recession since the Great Depression."
  - Central questions: shape and length of the recession, steepness of recovery, and calls for aggressive policy (Gourinchas 2020, Gali 2020, Krugman, 2020).
- Core conceptual point:
  - Cyclical deviations can leave persistent or permanent scars on trend output (hysteresis) — the trend itself may react to the cycle rather than cycles being mere movements along a fixed trend.

### II. CYCLE AND TREND
- Historical and methodological perspectives:
  - Burns and Mitchell (1946): business cycles as recurrent, non-periodic turning points; data-driven approach.
  - Koopmans (1947) criticized measuring cycles without theory.
  - Keynes (1936), Hicks (1933): cycles as deviations from long-run equilibrium driven by demand/price elements.
- Modeling approaches and key contributions:
  - Kydland and Prescott (1982): cycles as deviations from theoretically-grounded long-run equilibrium; trend may be stochastic (RBC).
  - Beveridge and Nelson (1981): decomposition approach.
  - Blanchard and Quah (1989): technology shocks persistent/permanent, demand shocks transitory.
  - Friedman (1964, 1993): "plucking model" — asymmetric cycles with recessions as downward deviations from a maximum potential.
- Endogenous-growth perspective and implications:
  - Temporary shocks can have permanent effects if they affect growth engines (King, Plosser, and Rebelo (1988); Stadler (1986, 1990); Stiglitz (1993); Martin and Rogers (1997); Fatás (2000a)).
  - Recessions can also produce "cleansing" or positive long-term effects in some models (Gali and Hammour (1992), Caballero and Hammour (1994), Aghion and Saint-Paul (1991), Hall (1991)).
- Definition emphasized:
  - Path dependence in steady states (hysteresis) — the level of GDP depends on the history of shocks.

### III. EMPIRICAL EVIDENCE
- Persistence as central empirical feature:
  - Nelson and Plosser (1982): many macro time series appear non-stationary; could not reject non-stationarity for US series.
  - Campbell and Mankiw (1987): "a one percent innovation in real GDP changes GDP forecasts over a long horizon by over one percent."
  - Bluedorn and Leigh (2018): "output forecasts are super-persistent—an unexpected 1 percent upward revision in current period output typically translates into a revision of ten year-ahead forecasted output by about 2 percent."
- Debate: persistent vs. permanent (unit root):
  - Cochrane (1988, 1991): evidence of trend reversion over many years; low variance ratio for US GDP (one-third).
  - Fractionally integrated models show long memory without unit root (Diebold and Rudebusch (1989)).
  - Christiano and Eichenbaum (1990): "we don't know" (unit root) and "we don't care"; tests and sample periods matter.
- Cross-country and crisis evidence:
  - Cogley (1990): variance ratio for nine OECD countries (1871–1985) average 1.16.
  - Cerra and Saxena series (2005a; 2005b; 2005c; 2008) and Cerra, Panizza, and Saxena (2013): robust evidence that shocks have permanent impacts across 192 countries.
  - Blanchard, Cerutti and Summers (2015): "roughly two-thirds" of recessions followed by lower output (hysteresis); incidence "around 63 percent" for intentional disinflations.
  - Jordà, Schularick, and Taylor (2011): financial-crisis recessions are costlier, slower recoveries; sample 1870–2008.
  - Reinhart and Rogoff (2014): average ~eight years to reach pre-crisis income level; median ~6.5 years; >40 percent experienced double dips across 100 systemic banking crises.
  - Ball (2014): average loss in potential output of 8.4 percent (23 countries, weighted by size) comparing current potential to 2007 pre-recession path.
  - Ma, Rogers, and Zhou (2020): in six modern health crises, "output remains below the pre-shock level five years later" with effects "as large as those resulting from systemic banking crises."
- Persistence after financial crises (quantified losses):
  - Average persistent loss ~5 percent for balance of payments crises.
  - Average persistent loss ~10 percent for banking crises.
  - Average persistent loss ~15 percent for twin crises.
- Mechanisms for persistence:
  - Labor-market scars: persistent effects of job losses, timing of entry, participation, and hours (Topel (1990); Ruhm (1991); Von Wachter and Bender (2006); Davis and Von Wachter (2011); Kahn (2010); Yagan (2018); Rinz (2019); Saez et al. (2019)).
  - Capital, R&D, innovation: procyclical investment in R&D and adoption leads recessions to reduce long-run productivity (Fatás (2000a,b); Barlevy (2007); Anzoategui et al. (2019)); Queralto (2019) finds banking crises → persistent 35 percent drop in R&D, 15 percent permanent decline in output, 7 percent decline in hours, 8½ percent decline in productivity in panel of 36 countries.
  - Policy shocks: Jordà, Singh, and Taylor (2020): "A one percentage point increase in domestic interest rates leads to a 6 percentage point decline in GDP over 12 years in the full sample from 1890, and a more muted decline of about 2.5 percent in the sample since WWII."
  - Fiscal consolidations: Fatás and Summers (2018) document persistent GDP effects visible 7 years later; larger consolidations → larger persistent effects.
- Policy effectiveness in recovery:
  - Countercyclical fiscal and monetary stimulus, real depreciation, foreign aid, flexible exchange rates can spur rebounds (Cerra, Panizza, and Saxena (2013); IMF (2009)); fiscal policy especially effective in advanced countries after banking crises.
  - Ma, Rogers, and Zhou (2020): aggressive fiscal response to disease outbreaks → larger medium-run bounce-back.
  - Reifschneider, Wascher, and Wilcox (2015): models with hysteresis imply more aggressive policy would have reduced long-term damage.
- Asymmetry and partial reversibility:
  - Expansions may not fully reverse recession-induced scars; policy may not easily engineer above-normal innovation to fully offset losses.

### IV. THEORY
- Models without hysteresis:
  - RBC framework: Y_t = A_t f(K_t, L_t); A_{t+1} = c + ρ A_t + e_{t+1}. If ρ < 1 temporary technology shocks die out; ρ = 1 makes technology shocks permanent.
  - New Keynesian models: nominal rigidities permit temporary demand effects but long-run output determined by exogenous trend; do not generate hysteresis alone.
- Hysteresis-generating frameworks:
  - Endogenous growth models (learning-by-doing, AK, R&D-expanding-variety) where accumulable factors (K, H, Z) interact so temporary shocks can permanently affect growth engines (King, Plosser, and Rebelo (1988); Romer (1990); Aghion and Howitt (1992)).
  - Representative formulations: Y_t = A_t f(K_t, H_t, L_t); H_{t+1} = g(H_t) or Z_{t+1} = Z_t g(J_t).
  - Specific model findings:
    - Comin and Gertler (2006), Comin (2009): procyclical TFP and diffusion; recessions reduce R&D returns → persistent productivity declines.
    - Anzoategui et al. (2019): liquidity demand shocks drove post-Great Recession productivity slowdown via adoption declines.
    - Engler and Tervala (2018): New Keynesian DSGE with learning-by-doing; hysteresis raises fiscal multipliers (NPV fiscal output multiplier 4.5 vs. 0.8 absent hysteresis) and welfare multipliers (each dollar public spending → welfare gain equivalent to 2.2 dollars private consumption in benchmark).
    - Financial-friction extensions link credit constraints to persistent declines in investment, R&D, and TFP (Bianchi, Kung and Morales (2019); Queralto (2019); Guerron-Quintana and Jinnai (2014); Cerra, Hakamada, and Lama (forthcoming)).
- Expectations, ZLB, and multiple equilibria:
  - Models with self-fulfilling expectations, ZLB, liquidity traps, or wage rigidity can produce multiple equilibria or stagnation traps (Farmer and Woodford (1997); Benigno and Fornaro (2018); Garga and Singh (2018); Schmitt-Grohé and Uribe (2017); Eggertsson et al. (2019); Summers (2014, 2015)).
  - Benigno and Fornaro (2018): endogenous growth + ZLB → multiple equilibria; pessimistic expectations can lead to permanent output loss if monetary policy cannot offset at ZLB.
  - Policy remedies debated: higher inflation target, fiscal interventions, macroprudential measures.
- Common theoretical implication:
  - Temporary shocks can become permanent via endogenous growth channels, financial frictions, expectations, and policy constraints/non-linearities.

### Section IV: ℤ_t, 핪_t, and the evolution of effective productivity
- Role and interpretations of ℤ_t:
  - ℤ_t represents technological innovations developed endogenously (e.g., new varieties of intermediate goods).
  - In AK models, ℤ_t is the aggregate stock of physical capital.
  - In learning-by-doing models, ℤ_t is the stock of human capital.
  - For endogenous growth, the production function for effective productivity accumulation needs to be linear homogeneous in ℤ_t.
  - Diminishing returns specification: ψ logℤ_t with ψ < 1 (ψ less than but close to unity) ⇒ shocks have highly persistent, albeit not permanent, impacts on output; policy implications similar to unit-root case.
- Role and modeling of 핪_t (cyclical component):
  - 핪_t can reflect the state of economic activity or one of its components.
  - Earlier models: 핪_t associated with a “scale” variable such as employment, producing permanent effects via learning-by-doing or innovation incentives.
  - More recent models: 핪_t is the gap between output and its flexible price level equilibrium; policy objective becomes minimizing deviations from the flexible price equilibrium with larger costs of inaction due to supply effects of low demand.
  - 핪_t modeled as function of policy and non-policy factors: 핪_t = 핪_t(핡_t, 합_t).
  - Policy interventions feed into supply potential indirectly via aggregate demand.
- Exact law of motion reproduced for effective productivity:
  - logℤ_{t+1} = g_t + logℤ_t + η f(핪_t(핡_t,합_t))
- Hysteresis parameter and ZLB implications:
  - Larger hysteresis parameter (η) ⇒ larger costs of inaction during recessions.
  - Diminishing-returns (ψ < 1) ⇒ high persistence but not permanence.
  - Costs of hitting the ZLB are larger because inability to restore the efficient equilibrium has permanent supply-side consequences.
  - Quantitative benefits of monetary policy under hysteresis given in Garga and Singh (2018).

### V. POLICY IMPLICATIONS
- General stabilization and cost considerations:
  - Traditional view: cycles temporary and symmetric; costs of cycles small under some utility assumptions (Lucas (1987)); "divine coincidence" — stabilizing inflation can stabilize output.
  - Asymmetric/plucking view (Friedman): reducing downward deviations can increase average output.
- Optimal stabilization with hysteresis:
  - If cyclical states feed into long-run supply, earlier and more aggressive policy action reduces permanent scars; fiscal and monetary policy can have permanent effects on GDP level (Martin and Rogers (1997); Engler and Tervala (2018); Garga and Singh (2016); Jordà, Singh, and Taylor (2020)).
  - Policy design challenge: potential output is endogenous and reacts to policy; closing a static output-gap is insufficient since policy affects the trend itself.
  - Reduced-form hysteresis mapping emphasized: logℤ_{t+1} = g_t + logℤ_t + η f(ψ_t).
- Specific policy findings:
  - Early, aggressive countercyclical fiscal and monetary stimulus tends to reduce medium-term permanent losses after crises (Cerra, Panizza, and Saxena (2013); IMF (2009); Ma, Rogers, and Zhou (2020)).
  - Fiscal consolidations post-Great Recession associated with persistent output losses visible 7 years later (Fatás and Summers (2018)); fiscal multipliers larger when hysteresis present (Engler and Tervala (2018)).
  - Monetary policy shocks can have long-lasting negative effects (Jordà, Singh, and Taylor (2020)); ZLB may constrain monetary response and exacerbate hysteresis (Benigno and Fornaro (2018); Garga and Singh (2018); Schmitt-Grohé and Uribe (2017)).
  - Financial stability considerations complicate choices: aggressive stimulus reduces long-term scars but may affect financial stability (Reifschneider, Wascher, and Wilcox (2015)).
  - High-pressure economy rationale:
    - Running the economy close to potential may upgrade skills, boost investment and R&D, and generate positive hysteresis.
    - Reversibility and feasibility uncertain; some hysteresis effects hard to reverse in expansions.
- Measurement and policymaker behavior:
  - Potential output estimates are procyclical; a negative surprise of 1 percent of annual GDP leads to immediate revisions to potential GDP of between 0.6-0.8 percent among advanced economies (Fatás (2019); Dovern and Zuber (2019); Claeys, Darvas and Leandro (2016); Coibion, Gorodnichenko and Ulate (2018); Martin, Munyan, and Wilson (2015)).
  - If policymakers treat measured "structural" changes as exogenous (equivalent to assuming η = 0), they will underreact to downturns, making recessions deeper and creating large permanent GDP losses via hysteresis.
  - Avoiding policy errors requires models that allow for hysteresis and endogenous supply reactions — not merely different statistical filters.
- Practical policy prescriptions:
  - Aggressive and fast policy action during recessions to avoid permanent supply-side damage.
  - Careful use of macroprudential policy during expansions to prevent future large recessions while avoiding premature tightening that could hinder potential growth.
  - In major crises (e.g., health crisis), continue macroeconomic stimulus where policy space exists; costs of policy mistakes are very large, so avoid excessive caution.

*Content extracted from wpiea2020073-print-pdf - References................................................................................................39*

### References................................................................................................39

### References................................................................................................39

### I. INTRODUCTION
- COVID-19 pandemic: "As of end-May 2020, worldwide cases of coronavirus exceed 5.5 million and over 350 thousand deaths."
- Containment measures (social distancing, lockdowns of non-essential businesses) have caused a global halt in economic activity; many countries likely to experience "the deepest recession since the Great Depression."
- Key issues examined: shape and length of the recession, steepness of recovery, calls for aggressive policy (Gourinchas 2020, Gali 2020, Krugman, 2020).
- Core conceptual point: cyclical deviations can leave persistent or permanent scars on trend output (hysteresis) — the trend itself may react to the cycle rather than cycles being mere movements along a fixed trend.

### II. CYCLE AND TREND
- Historical perspectives:
  - Burns and Mitchell (1946): business cycles as recurrent, non-periodic turning points (peaks/troughs); data-driven approach avoiding imposing a theoretical trend.
  - Koopmans (1947) criticized measuring cycles without theory.
  - Keynes (1936), Hicks (1933): cycles as deviations from long-run equilibrium driven by demand/price elements.
- Modeling trend and cycle:
  - Kydland and Prescott (1982): formalized cycles as deviations from theoretically-grounded long-run equilibrium; trend can be stochastic (Real Business Cycle, RBC).
  - Beveridge and Nelson (1981): decomposition approach to trend and cycle.
  - Blanchard and Quah (1989): identify technology shocks as persistent/permanent and demand shocks as transitory.
  - Friedman (1964, 1993): "plucking model" — asymmetric cycles; recessions are downward deviations from a maximum potential.
- Endogenous growth perspective:
  - Temporary shocks may have permanent effects if they affect the long-term growth engine (King, Plosser, and Rebelo (1988); Stadler (1986, 1990); Stiglitz (1993); Martin and Rogers (1997); Fatás (2000a)).
  - Recessions can also produce "cleansing" or positive long-term effects in some models (Gali and Hammour (1992), Caballero and Hammour (1994), Aghion and Saint-Paul (1991), Hall (1991)).
- Definition emphasized: path dependence in steady states (hysteresis) — the level of GDP depends on the history of shocks.

### III. EMPIRICAL EVIDENCE
- Persistence as a central empirical feature:
  - Nelson and Plosser (1982): many macro time series appear non-stationary; could not reject non-stationarity for US series.
  - Campbell and Mankiw (1987): "a one percent innovation in real GDP changes GDP forecasts over a long horizon by over one percent."
  - Bluedorn and Leigh (2018): "output forecasts are super-persistent—an unexpected 1 percent upward revision in current period output typically translates into a revision of ten year-ahead forecasted output by about 2 percent."
- Debate: persistent vs. permanent (unit root)
  - Cochrane (1988, 1991): evidence of trend reversion over many years; low variance ratio for US GDP (one-third).
  - Fractionally integrated models (Diebold and Rudebusch (1989)) find long memory (persistence) without a unit root.
  - Christiano and Eichenbaum (1990): "we don't know" (unit root) and "we don't care"; Diebold and Senhadji (1996) and Cushman (2016) argue tests matter and unit roots may not be rejected for some periods.
- Cross-country evidence:
  - Cogley (1990): variance ratio for nine OECD countries (1871–1985) average 1.16 (greater low-frequency variability than US).
  - Cerra and Saxena (2005a); Cerra, Panizza, and Saxena (2013); Cerra and Saxena (2017): robust evidence that shocks have permanent impacts on levels of GDP across 192 countries.
  - Blanchard, Cerutti and Summers (2015): "roughly two-thirds" of recessions followed by lower output (hysteresis); incidence of hysteresis "around 63 percent" even for intentional disinflations.
  - Haltmaier (2013): depth of recession affects potential loss in advanced countries; recession length matters for emerging markets.
- Persistence after financial crises and deep recessions:
  - Cerra and Saxena (2005b, 2005c, 2008): evidence of permanent output losses after crises; average persistent loss ~5 percent for balance of payments crises, 10 percent for banking crises, 15 percent for twin crises.
  - Jordà, Schularick, and Taylor (2011): financial-crisis recessions are costlier, slower recoveries; long historical sample (1870–2008).
  - Reinhart and Rogoff (2014): average ~eight years to reach pre-crisis income level; median ~6.5 years; >40 percent experienced double dips across 100 systemic banking crises.
  - Ball (2014): estimates average loss in potential output (23 countries) of 8.4 percent (weighted by size) comparing current potential to 2007 pre-recession path.
  - Ma, Rogers, and Zhou (2020): in six modern health crises, "output remains below the pre-shock level five years later" with effects "as large as those resulting from systemic banking crises."
- Sources/mechanisms of persistence:
  - Blanchard and Quah (1989) identification (supply shocks permanent, demand shocks transitory) may be invalid in presence of hysteresis.
  - Labor-market scars:
    - Blanchard and Summers (1986): European unemployment persistence termed hysteresis; participation and hours affected.
    - Micro evidence: Topel (1990), Ruhm (1991), Von Wachter and Bender (2006), Davis and Von Wachter (2011), Kahn (2010), Yagan (2018), Rinz (2019), Saez et al. (2019) — persistent effects of job losses, timing of entry, and tax policy on labor outcomes; Hotchkiss and Moore (2018) show positive effects of tight labor markets.
  - Capital accumulation, R&D, innovation:
    - Endogenous-growth channels (capital, R&D, human capital) imply procyclicality can create persistence (Fatás (2000a,b); Barlevy (2007); Anzoategui et al. (2019)).
    - Financial constraints amplify procyclicality of R&D and innovation (Aghion et al. (2012); Ouyang (2011); Duval, Hong, and Timmer (2019)).
    - Queralto (2019): banking crises → persistent 35 percent drop in R&D, 15 percent permanent decline in output, 7 percent in hours, 8½ percent in productivity in panel of 36 countries.
  - Policy shocks as persistency sources:
    - Jordà, Singh, and Taylor (2020): "A one percentage point increase in domestic interest rates leads to a 6 percentage point decline in GDP over 12 years in the full sample from 1890, and a more muted decline of about 2.5 percent in the sample since WWII."
    - Fatás and Summers (2018): fiscal consolidations after the Great Recession produced persistent GDP effects visible 7 years later; larger consolidations → larger persistent effects.
    - Blanchard et al. (2015): 70–80 percent of 122 recessions (23 advanced economies) show an "unambiguous" persistent gap; demand-driven recessions are almost as likely to be persistent as other shocks.
  - Policy effectiveness in recovery:
    - Cerra, Panizza, and Saxena (2013); IMF (2009): countercyclical fiscal and monetary stimulus, real depreciation, foreign aid, flexible exchange rates can spur rebounds; fiscal policy especially effective in advanced countries after banking crises.
    - Ma, Rogers, and Zhou (2020): aggressive fiscal response to disease outbreaks → larger medium-run bounce-back.
    - Reifschneider, Wascher, and Wilcox (2015): modified FRB/US calibration with hysteresis shows more aggressive policy would have reduced long-term damage, though concerns about financial stability remain.
  - Asymmetry and reversibility:
    - Some evidence that expansions do not fully reverse recession-induced scars (Reifschneider et al. (2015)); policy may not easily engineer above-normal innovation to fully offset losses.

### IV. THEORY
- Traditional models without hysteresis:
  - RBC models: Y_t = A_t f(K_t, L_t); A_t+1 = c + ρ A_t + e_{t+1}. If ρ < 1 temporary technology shocks die out; setting ρ = 1 makes technology shocks permanent.
  - New Keynesian models: nominal rigidities permit temporary demand effects but long-run output determined by exogenous trend; do not generate hysteresis alone.
- Hysteresis-generating theoretical frameworks:
  - Endogenous growth models:
    - Learning-by-doing (LBD), AK ("Romer (1986) AK"), and R&D-expanding-variety models (Romer (1990), Aghion and Howitt (1992), Grossman and Helpman (1991)).
    - Production with accumulable factors (K, H, Z) and linear homogeneity can make temporary shocks permanent via impacts on growth engines.
    - Representative formulations: Y_t = A_t f(K_t, H_t, L_t); H_{t+1} = g(H_t) or Z_{t+1} = Z_t g(J_t) (R&D-driven varieties).
  - Key model mechanisms and examples:
    - King, Plosser, and Rebelo (1988): endogenous growth integrated in RBC — temporary shocks can permanently affect levels.
    - Stadler (1990): endogenous learning by doing with monetary shocks can create permanent output effects (monetary hysteresis).
    - Comin and Gertler (2006), Comin (2009): procyclical TFP and procyclical technology diffusion; recessions reduce R&D returns → less R&D/adoption → persistent productivity declines.
    - Anzoategui et al. (2019): liquidity demand shocks drove post-Great Recession productivity slowdown via endogenous adoption declines.
    - Vinci and Licandro (2019): AK endogenous growth + monetary policy; deep recessions can lower potential and reduce policy stimulus under reaction functions.
    - Engler and Tervala (2018): New Keynesian DSGE with learning-by-doing; hysteresis raises fiscal multipliers (NPV fiscal output multiplier 4.5 vs. 0.8 absent hysteresis) and welfare multipliers (each dollar public spending → welfare gain equivalent to 2.2 dollars private consumption in benchmark).
    - Financial-friction extensions (Bianchi, Kung and Morales (2019); Queralto (2019); Guerron-Quintana and Jinnai (2014); Cerra, Hakamada, and Lama (forthcoming)) link credit constraints to persistent declines in investment, R&D, and TFP.
  - Expectations-driven multiple equilibria and non-linearities:
    - Models with self-fulfilling expectations, zero lower bound (ZLB), liquidity traps, or wage rigidity can produce multiple equilibria or stagnation traps (Farmer and Woodford (1997); Gunn and Johri (2011); Benigno and Fornaro (2018); Garga and Singh (2018); Schmitt-Grohé and Uribe (2017); Eggertsson et al. (2019); Summers (2014, 2015)).
    - Benigno and Fornaro (2018): endogenous growth + ZLB → multiple equilibria; pessimistic expectations can lead to permanent output loss if monetary policy cannot offset at ZLB.
    - Liquidity/growth traps and coordination failures can make economies stuck in low-output equilibria; policies (higher inflation target, fiscal interventions) debated as remedies.
    - Agent-based and heterogeneous-agent models (Dosi et al. (2018)) show hysteresis with endogenous innovation/adoption and imperfect information.
- Common theoretical implication: temporary shocks can become permanent via channels (endogenous growth, financial frictions, expectations, non-linear policy constraints).

### V. POLICY IMPLICATIONS
- Stabilization policy and costs:
  - Traditional view: cyclical fluctuations are temporary and symmetric; costs of cycles small under some utility assumptions (Lucas (1987)); stabilizing inflation can also stabilize output (divine coincidence).
  - Asymmetric/plucking view (Friedman): reducing downward deviations can increase average level of output.
- Optimal stabilization with hysteresis:
  - If cyclical states feed into long-run supply, earlier and more aggressive policy action reduces permanent scars; both fiscal and monetary policy can have permanent effects on level of GDP (Martin and Rogers (1997); Engler and Tervala (2018); Garga and Singh (2016); Jordà, Singh, and Taylor (2020)).
  - Policy design challenge: potential output is endogenous and reacts to policy; closing a static output-gap measure is insufficient since policy affects the trend itself.
  - Conceptual reduced-form hysteresis mapping: log ℤ_{t+1} = g_t + log ℤ_t + η f(ψ_t) where ψ_t is aggregate demand/cyclical GDP component, η is hysteresis parameter, and f' (ψ_t) > 0 — highlighting how demand amplifies effective productivity growth.
- Specific policy findings from literature:
  - Early, aggressive countercyclical fiscal and monetary stimulus tends to reduce medium-term permanent losses after crises (Cerra, Panizza, and Saxena (2013); IMF (2009); Ma, Rogers, and Zhou (2020)).
  - Fiscal consolidations post-Great Recession are associated with persistent output losses evident 7 years later (Fatás and Summers (2018)); fiscal multipliers larger when hysteresis is present (Engler and Tervala (2018)).
  - Monetary policy shocks can have long-lasting negative effects on output and TFP (Jordà, Singh, and Taylor (2020)); ZLB may constrain monetary response and exacerbate hysteresis (Benigno and Fornaro (2018); Garga and Singh (2018); Schmitt-Grohé and Uribe (2017)).
  - Financial stability considerations complicate policy choices: aggressive stimulus can reduce long-term scars but may have adverse effects on financial stability (Reifschneider, Wascher, and Wilcox (2015)).
  - Creating "high-pressure" economies can, in principle, generate positive hysteresis (upgrading skills, boosting investment and R&D) but reversibility and practicality are uncertain; some hysteresis effects may be hard to reverse in expansions.

*Italic source attribution: Content extracted from wpiea2020073-print-pdf - References................................................................................................39*

### Section IV. In some of the models, ℤ

### Section IV. In some of the models, ℤ_t represents technological innovations that are developed endogenously in a R&D sector

### Theoretical framework and role of ℤ_t and 핪_t
- ℤ_t:
  - Represents technological innovations developed endogenously (e.g., new varieties of intermediate goods).
  - In AK models, ℤ_t is the aggregate stock of physical capital.
  - In learning-by-doing models, ℤ_t is the stock of human capital.
  - For endogenous growth, the production function for effective productivity accumulation needs to be linear homogeneous in ℤ_t.
  - If there are diminishing returns in the evolution of effective productivity, ψ logℤ_t with ψ < 1 (ψ less than but close to unity), shocks have highly persistent, albeit not permanent, impacts on output; policy implications are fairly similar to the unit-root case.

- 핪_t (cyclical component):
  - Can reflect the state of economic activity or one of its components.
  - Earlier models: 핪_t associated with a “scale” variable such as employment, producing permanent effects via learning-by-doing or innovation whose incentives depend on economic activity.
  - More recent models: 핪_t is the gap between output and its flexible price level equilibrium; optimal policy becomes minimizing deviations from the flexible price equilibrium, but with larger costs of inaction due to supply effects of low demand.
  - 핪_t is modeled as a function of policy and non-policy factors:
    - 핪_t = 핪_t(핡_t, 합_t)
    - Policy interventions feed into supply potential indirectly via aggregate demand.

- Evolution of effective productivity (exact equation reproduced):
  - logℤ_{t+1} = g_t + logℤ_t + η f(핪_t(핡_t,합_t))

### Hysteresis, persistence, and implications for the zero lower bound (ZLB)
- Hysteresis parameter and persistence:
  - The larger the hysteresis parameter (η), the larger the costs of inaction during recessions.
  - Diminishing-returns specification (ψ logℤ_t with ψ < 1) implies high persistence though not permanence.

- ZLB and policy constraints:
  - Costs of hitting the ZLB are larger than previously assumed because inability to restore the efficient equilibrium has permanent consequences on the supply side.
  - Quantitative estimates of benefits of monetary policy under hysteresis are provided in Garga and Singh (2018).

### Optimal policy during recessions and expansions
- Recessions — optimal stabilization policy:
  - Stabilization policy should offset the damaging impact of an adverse shock as fast as possible during a recession.
  - Aggressive and fast action during recessions becomes optimal to avoid permanent supply-side scars.

- Expansions — macroprudential considerations and trade-offs:
  - Strong role for macroprudential policy and pre-emptive countercyclical measures to avoid excessive credit growth and future large recessions.
  - Prudential standards should be tightened and buffers increased when economic and credit growth are strong.
  - Risk of acting too early: overly conservative policy during expansions can, via hysteresis, negatively affect the supply side and shorten the expansion.
  - Running the economy as close to potential as possible can bring large benefits (Okun (1973) intuition; recent policy references to “high-pressure” economy, e.g., Yellen (2016)).
  - Experience of the US labor market in the current expansion (natural rate of unemployment being revised downwards) highlights drawbacks of reacting too early to fears of excesses.

### Measuring the output gap and implications for policy design
- Traditional approach and its failures under hysteresis:
  - Output gap measures rely on estimates of potential output or natural rate of unemployment derived from models that separate cyclical dynamics from long-term trends.
  - In traditional exogenous-trend frameworks, policy reacts only to temporary cyclical components; errors create volatility but not permanent supply scars.
  - In presence of hysteresis, estimation errors propagate to the supply side and can generate permanent scars on GDP.

- Procyclicality of potential output estimates:
  - Potential output estimates are procyclical due to reliance on traditional business cycle models and filters.
  - A negative surprise of 1 percent of annual GDP leads to immediate revisions to potential GDP of between 0.6-0.8 percent among advanced economies (cited studies: Fatás (2019), Dovern and Zuber (2019), Claeys, Darvas and Leandro (2016), Coibion, Gorodnichenko and Ulate (2018), Martin, Munyan, and Wilson (2015)).

- Policy-maker behavior and learning:
  - If policy makers treat measured “structural” changes as exogenous productivity (equivalent to assuming η = 0), they will be hesitant to use stabilization tools and will underreact.
  - Underreaction makes recessions deeper and longer and causes large permanent GDP losses via hysteresis.
  - Ignoring hysteresis can lead policymakers to misinterpret ex-post negative long-term effects as confirmation of pessimistic potential-output estimates, impeding learning and causing repeated mistakes (Fatás (2019) evidence for fiscal policy in 2010-14 among European countries).

- Required change in modeling and policy design:
  - Avoiding these policy errors requires designing stabilization policy using economic models that allow for hysteresis and endogenous reactions of the supply side—not merely choosing different statistical filters.

### Conclusions and policy message
- Conceptual shift:
  - Business-cycle models emphasizing history dependence show that shocks can have permanent effects on output (hysteresis), differing from traditional cycle-trend decompositions.

- Empirical and theoretical convergence:
  - Theoretical literature: hysteresis is a natural outcome in the presence of endogenous growth.
  - Empirical literature: strong evidence of persistent fluctuations and demand-related shocks contributing to long-term effects.

- Policy prescriptions:
  - Aggressive and fast policy action during recessions to avoid permanent supply-side damage.
  - Careful use of macroprudential policy during expansions to prevent future large recessions while avoiding premature tightening that could hinder potential growth.
  - Understand and target the costs of deviating from potential output; running a “high-pressure” economy may have large supply-side benefits.
  - In major crises (e.g., health crisis), continued macroeconomic stimulus where policy space exists is needed; costs of policy mistakes are very large, so avoid excessive caution.

*Source: wpiea2020073-print-pdf - Section IV. In some of the models, ℤ_t represents technological innovations that are developed endogenously in a R&D sector.*

### 21726. National Bureau of Economic Research, Cambridge, MA. Available at:

### 21726. National Bureau of Economic Research, Cambridge, MA. Available at: http://www.nber.org/papers/w21726.

### Major themes and topics covered (as reflected in the cited works)
- Real wage rigidities and New Keynesian modeling (Blanchard and Galí, 2007, Journal of Money, Credit and Banking 39(s1):35–65).
- Growth forecast errors and fiscal multipliers (Blanchard and Leigh, 2013, American Economic Review 103(3):117–20).
- Dynamic effects of aggregate demand and supply disturbances (Blanchard and Quah, 1989, American Economic Review 79(4):655–673).
- Hysteresis in unemployment and labor markets (Blanchard and Summers, 1986; O’Shaughnessy, 2011; Yagan, 2019; others).
- Measurement and properties of business cycles, trends, and unit roots (Burns and Mitchell, 1946; Nelson and Plosser, 1982; Perron, 1988; Campbell and Perron, 1991; Cochrane, 1991).
- Financial crisis effects, recoveries, and lost recoveries (Cerra and Saxena series, 2005a/b/c; Cerra and Saxena, 2008; Cerra, Saxena, and Panizza, 2013; Reinhart and Rogoff, 2014; Jordà, Schularick, and Taylor, 2011).
- Liquidity traps, secular stagnation, and policy responses (Eggertsson and Woodford, 2004; Krugman, 1998 and 2020; Summers, 2014 and 2015; Woodford, 2012).
- Long-run growth, endogenous technical change, and interactions between cycles and growth (Solow, 1956; Romer, 1986 and 1990; King, Plosser, and Rebelo, 1988; Rebelo, 1991).
- Fiscal stimulus, austerity, and long-term effects (Gechert, Horn, and Paetz, 2019; Fatás, 2019; Fatás and Summers, 2018).
- Job displacement, scarring, and life-cycle effects from recessions (Davis and Von Wachter, 2011; Kahn, 2010; Rinz, 2019; Ruhm, 1991; Topel, 1990).

### Methodological emphases and empirical approaches (as indicated by titles and outlets)
- Use of macro time-series methods to study trends, unit roots, and stochastic trends (Campbell and Perron, 1991; Perron, 1988; Stock and Watson, 1988).
- Structural and theoretical modeling: New Keynesian frameworks, real-business-cycle and endogenous growth models (Calvo, 1983; Kydland and Prescott, 1982 and 1990; Lucas, 1972, 1973, 1977; Romer, 1986, 1990).
- Evidence from cross-country and historical episode studies of crises and recoveries (Reinhart and Rogoff, 2014; Cerra and Saxena series; Jordà, Schularick, and Taylor, 2011).
- Micro evidence on labor-market scarring and firm/worker dynamics (Haltiwanger, Jarmin, and Miranda, 2013; Von Wachter and Bender, 2006; Saez, Schoefer, and Seim, 2019).
- Policy simulation and quantitative evaluation of models addressing liquidity traps, hysteresis, and fiscal multipliers (Eggertsson, Mehrotra, and Robbins, 2019; Mertens and Ravn, 2014; Bluedorn and Leigh, 2019).

### Recurring policy-relevant implications (as signaled by titles)
- Fiscal policy effectiveness and risks: studies address fiscal multipliers, the permanent effects of fiscal consolidations, and proposals to revive fiscal frameworks (Blanchard and Leigh, 2013; Fatás and Summers, 2018; Claeys, Darvas, and Leandro, 2016).
- Monetary policy at the lower bound: multiple works discuss policy options and accommodation at the interest-rate lower bound and in liquidity traps (Eggertsson and Woodford, 2004; Krugman, 1998; Woodford, 2012).
- The importance of addressing hysteresis and long-term labor-market damage in policy design (Blanchard and Summers, 1986; Engler and Tervala, 2018; Yagan, 2019).
- Financial frictions, leverage, and slow recoveries: role of credit and financial crises in producing protracted output losses (Jordà, Schularick, and Taylor, 2011; Queralto, 2019; Cerra, Hakamada, and Lama, forthcoming).

### Representative citations indicating scope and chronology
- Fundamental and early contributions to business-cycle measurement and theory: Burns and Mitchell, 1946; Hicks, 1933; Keynes, 1936; Solow, 1956; Swan, 1956.
- Seminal methodological and empirical contributions on trends and unit roots: Nelson and Plosser, 1982; Perron, 1988; Campbell and Perron, 1991; Cochrane, 1991.
- Recent discussions on secular stagnation, liquidity traps, and pandemic-era policy: Summers, 2014 and 2015; Krugman, 2020; Gourinchas, 2020; Ma, Rogers, and Xhou, 2020.

*Compiled from the bibliography entries listed in the source PDF.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020073-print-pdf.pdf_