## wpiea2025256-source-pdf

## Source details

**Canonical URL:** [wpiea2025256-source-pdf](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025256-source-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025256-source-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025256-source-pdf.pdf.json)

---

### 4.1 Monetary policy implementation — Taylor-rule estimates and reaction to inflation and output
- Estimation setup:
  - Estimated reaction function (Equation 2) uses policy rate r_{i,t}, one-year ahead inflation expectation E_t \bar{π}_{i,t+h} from consensus forecasts, the real-time GDP output gap \bar{y}_{i,t-3}, country fixed effects α_i, and time fixed effects τ_t.
  - Regressions exclude EMs with fixed exchange rate regimes and Argentina, Türkiye, and Ukraine. Standard errors double clustered at country and month level.
- Main empirical findings:
  - Taylor rule coefficients on inflation deviations declined in the post-GFC period, interpreted as improved central bank credibility and more strongly anchored long-term inflation expectations.
  - EM central banks shifted attention toward curbing output fluctuations after the GFC: the Taylor rule coefficient on the output gap is negative and not statistically significant pre-GFC and positive and statistically significant post-GFC.
  - AE estimates in the post-crisis period are strikingly similar to EMs in the post-crisis period.
- Key coefficients from Table 1 (Taylor rule / Augmented Taylor rule):
  - Inflation gap:
    - EMs, Pre-GFC: 2.047**
    - EMs, Post-GFC: 1.314***
    - AEs, Post-GFC (benchmark): 1.572***
    - Augmented EMs, Pre-GFC: 2.251***
    - Augmented EMs, Post-GFC: 1.279***
  - Output gap:
    - EMs, Pre-GFC: -0.323
    - EMs, Post-GFC: 0.398**
    - AEs, Post-GFC: 0.355*
    - Augmented EMs, Pre-GFC: -0.104
    - Augmented EMs, Post-GFC: 0.487**
  - NEER depreciation (augmented specification):
    - Pre-GFC: 0.514***
    - Post-GFC: 0.247**
  - Observations by column: 3,455; 3,455; 2,709; 3,455; 3,455
  - R^2 by column: 0.993; 0.993; 0.996; 0.993; 0.993
  - Notes: Pre-GFC = 1997–2009; Post-GFC = 2010–24.

### 4.1 Monetary policy implementation — Exchange rate pass-through and anchoring of inflation expectations
- Exchange rate pass-through estimation:
  - Equation 3 uses cumulative CPI response to a one percentage point depreciation of the nominal exchange rate (∆be_{i,t} measured around US monetary announcements), with controls including 12 lags of inflation, exchange rate depreciation, and the policy rate.
- Findings on pass-through:
  - Pre-GFC: two years after a one percentage point depreciation in the exchange rate, inflation was about 1.5 percentage point higher.
  - Post-GFC: no statistically significant exchange rate pass-through to inflation.
- Anchoring of expectations (Equation 4: rolling six-year regressions):
  - Sensitivity of long-term (three-year ahead) inflation expectations to changes in short-term (one-year ahead) inflation expectations declined substantially after the GFC.
- Quantitative results (Table A4, cumulative CPI response to 1 percentage point depreciation vis-à-vis the USD):
  - Exchange rate depreciation coefficients:
    - h=6: 0.332*** (0.121)
    - h=12: 0.704*** (0.206)
    - h=24: 1.358*** (0.403)
    - h=36: 1.767*** (0.608)
  - Interaction Exchange rate depreciation × post-GFC:
    - h=6: -0.319** (0.144)
    - h=12: -0.529** (0.258)
    - h=24: -1.189** (0.510)
    - h=36: -1.956** (0.787)
  - Observations: 5,291; 5,171; 4,931; 4,691
  - Adjusted R2: 0.662; 0.672; 0.682; 0.720
  - Notes: Pre-GFC period 1997–2009; post-GFC period 2010–24. Exchange rate depreciation instrumented using change between the day before and the day after scheduled US Federal Reserve monetary policy announcements. Standard errors clustered by time.

### 4.1 Monetary policy implementation — Perceived reaction function and markets’ perceptions
- Perceived monetary policy reaction function (Equation 5) estimated using forecaster-country level data from Consensus economics; regressions run separately by calendar year with minimum 1,000 observations and standard errors clustered at forecaster level.
- Finding:
  - Progressive decline in the magnitude of the Taylor rule coefficient on expected inflation over time and a marginal increase in the size of the output gap coefficient — interpreted as gains in monetary policy credibility.
- Selected yearly estimates (Table A5 examples):
  - 2004: Expected inflation coefficient 2.360 (standard error 0.440); Expected real GDP growth coefficient -0.120 (standard error 0.160); R2 = 0.900; Observations = 1,009
  - 2008: Expected inflation 2.990 (0.150); Expected real GDP growth -0.350 (0.120); R2 = 0.900; Observations = 1,951
  - 2016: Expected inflation 1.100 (0.060); Expected real GDP growth -0.070 (0.070); R2 = 0.820; Observations = 2,736
  - 2024: Expected inflation 1.330 (0.120); Expected real GDP growth 0.580 (0.140); R2 = 0.830; Observations = 2,713
- Notes: Sample excludes countries with fixed exchange rate regimes, Argentina, Türkiye, and Ukraine. Forecaster fixed effects included. Standard errors clustered at forecaster level.

### 4.1 Additional estimation notes
- Real-time output gap constructed using real GDP data available up to period t and World Economic Outlook forecasts for the following five years, reflecting data available to policymakers (Orphanides and Norden, 2002).
- Regressions follow Carvalho et al. (2021) with OLS; additional specifications control for NEER depreciation to capture fear-of-floating concerns.

### 4.5 The contribution of policy frameworks to macroeconomic stabilization — Empirical approach
- Objective: quantify the contribution of policy frameworks relative to benign external conditions in improving emerging market (EM) resilience to risk-off events.
- Main regression (Equation 10) predicts the 12‑month change in log real GDP or log CPI (y_{i,τ+12}−y_{i,τ−1}) from:
  - Pre-determined policy variables P_{k,i,τ−1} (degree of anchoring of inflation expectations, reserve adequacy, FX mismatches, stringency of macroprudential regulation, external debt burden, cyclically-adjusted fiscal balance, stringency of CFM measures);
  - Lags of changes in log real GDP and log CPI (∆X_{i,τ−j});
  - Episode fixed effects α_τ to compare countries during the same risk-off episodes.
- To isolate external conditions, the common episode component ̄y_τ is regressed (Equation 11) on three proxies for external conditions G_{m,τ}: real GDP growth in advanced economies relative to trend, changes in commodity terms-of-trade, and US financial conditions (all expressed as cumulative changes over 12 months).

### 4.5 Regression findings: role of individual policy metrics
- Countries entering risk-off episodes with stronger policy frameworks experienced higher real GDP growth and lower inflation (panel (a) of Figure 8; see Table A8).
- Key marginal effects (one standard deviation changes):
  - A country entering a risk-off episode at a one standard deviation lower FX mismatches is expected to experience 0.9 percentage point higher growth relative to a country at the sample mean of FX mismatches.
  - A country at a one standard deviation greater anchoring of long-term inflation expectations tends to experience 2.3 percentage points lower inflation.
- Selected coefficients from Table A8 (policy variables expressed in standard deviations):
  - Panel A (Real GDP growth):
    - Macropru stringency: 0.475* (0.268) in column (2)
    - External debt burden: -0.834*** (0.243) in column (3)
    - Reserve adequacy: 0.614** (0.253) in column (4)
    - FX mismatches: -0.948*** (0.332) in column (5)
    - Adjusted R2 across columns: 0.306; 0.311; 0.331; 0.350; 0.335; 0.303; 0.308
  - Panel B (CPI inflation):
    - Inflation expectation anchoring: -2.349** (0.932) in column (1)
    - Macropru stringency: -0.529** (0.261) in column (2)
    - External debt burden: 1.122*** (0.365) in column (3)
    - FX mismatches: 0.939** (0.464) in column (5)
    - Adjusted R2 across columns: 0.590; 0.525; 0.533; 0.609; 0.619; 0.568; 0.524
  - Notes: Policy variable definitions preserved exactly (e.g., external debt burden = stock of external debt as a percentage of exports; reserve adequacy = stock of FX reserves as a percentage of short-term external debt; FX mismatches = foreign currency liabilities as a percentage of exports). Robust standard errors reported in parentheses. ***, **,* denote statistical significance at 1, 5, and 10 percent levels.

### 4.5 Quantified overall contributions (pre-GFC vs post-GFC)
- Improved policy frameworks (accounting for observed changes in policy variables before and after the GFC) contributed to EM resilience during recent risk-off episodes by:
  - Raising growth by 0.5 percentage point in the period since the GFC compared with the period before the crisis.
  - Lowering inflation by 0.6 percentage point in the period since the GFC compared with the period before the crisis.
- Contribution of external conditions:
  - More benign external conditions contributed to faster growth in EMs after the GFC by 0.5 percentage point.
  - External conditions did not ease inflationary pressures.
- Robustness: estimates robust to exclusion of individual risk-off episodes or inclusion of the 2013 Taper tantrum episode (see Table A9 examples):
  - Excluding 1997m8: Policy frameworks — Real GDP growth 0.55; CPI inflation -0.69; External conditions — Real GDP growth 0.46; CPI inflation 0.03
  - Mean row: 0.55; -0.57; 0.45; 0.03
  - Including Taper Tantrum: 0.55; -0.56; 0.45; 0.03

### 4.5 Structural analysis: QIPF model and key frictions
- Framework: two-country New Keynesian QIPF model incorporating four key frictions:
  1. Limited risk-bearing capacity in FX markets (shallow FX markets) leading to a risk-augmented UIP premium: 1 = E_t[(I^*_t / I_t) (ε_{t+1}/ε_t)] + Γ B_{F,t} / P_t (Equation 12), where Γ≥0 is FX market shallowness.
  2. An occasionally binding external debt limit: −B_t / P_t ≤ m E_t Y_{t+1} (Equation 14), with m>0; when binding, a domestic spread Θ_t arises (I^b_t = I_t + Θ_t).
  3. Balance sheet FX mismatches via the law of motion for net foreign assets (Equation 15), with ownership share parameter 0≤ω≤1 governing effective currency composition of economy-wide net foreign assets and exposure to exchange rate risk.
  4. Imperfectly anchored inflation expectations achieved via a Kimball aggregator (non-linear Phillips curve) and dynamic indexation mechanisms that raise inflation persistence.
- Other features: imperfect substitution between domestic and foreign goods, home bias, labor and capital inputs, standard monetary policy rule responding to inflation deviations and output gap, fiscal authority with lump-sum taxes for debt stabilization.

### 4.5 Calibration
- Two EM calibrations:
  - Pre-GFC calibration: average EM prior to the GFC (weaker policy frameworks).
  - Post-GFC calibration: average EM after the GFC (stronger policy frameworks: more strongly anchored inflation expectations, smaller balance sheet mismatches).
- FX market depth Γ held fixed across calibrations.
- Balance sheet targets set to match observed EM net foreign assets:
  - Pre-GFC net foreign assets: −38.4% of annual GDP.
  - Post-GFC net foreign assets: −25.9% of annual GDP.
- External borrowing limit set so the economy is in the constrained regime approximately 3 percent of the time under the pre-GFC calibration.
- All other parameters and shock properties calibrated based on Adrian et al. (2021).

### 4.5 Model simulations: policy trade-offs and sudden stop probability
- Risk-off shock scenario: portfolio capital outflow normalized so that exchange rate depreciates by 10 percent under the pre-GFC calibration.
- Policy-tradeoff outcomes:
  - EM with strong policy frameworks:
    - Moderate increase in inflation because inflation expectations are strongly anchored.
    - Central bank does not need to tighten policy aggressively.
    - Domestic demand does not strongly decline; output may expand supported by higher net exports.
  - EM with weak policy frameworks:
    - Greater exchange rate pass-through to domestic prices produces a substantial increase in inflation and second-round wage pressures.
    - Central bank tightens much more aggressively, depressing domestic demand and causing a strong contraction in output.
    - Monetary authority may consider deploying FX interventions or CFMs to improve the trade-off.
- Sudden stop probability and severity:
  - Post-GFC improvements (including an average increase in net foreign assets of about 12.5 percent of GDP relative to pre-GFC) reduce sudden stop likelihood by two channels: being farther from the external debt limit and lower macroeconomic volatility from better-anchored inflation and lower FX mismatches.
  - Stochastic simulations keeping average distance from external debt unchanged show:
    - Probability of a sudden stop drops from about 3 percent to 1.5 percent.
    - Average credit spread during sudden stops falls by 1 percentage point (from 6.2 percent to 5.2 percent).

### 5.4 Delayed monetary tightening with weak frameworks — Context and simulation setup
- Exercise considers an EM facing one-off shocks to capital flows and domestic costs that affect the exchange rate and inflation.
- Two monetary policy strategies compared:
  - Baseline: central bank follows a simple policy rule in all periods.
  - Alternative: central bank keeps nominal interest rates lower than in the baseline for the year following the shocks, then switches to policy rates higher than in the baseline thereafter; the switch is not foreseen by agents ex ante but becomes perfectly credible thereafter.
- Alternative interpreted as a period of wait-and-see followed by a pivot toward tighter policy consistent with a revision in the central bank’s inflation forecasts.

### 5.4 Key model outcomes and mechanisms
- Baseline rule-following strategy:
  - Trade-off between inflation and output stabilization managed via a modest rise in real interest rates.
  - Output converges to potential from below and inflation converges to target from above.
- Alternative (delayed tightening) strategy:
  - Real rates initially low, preventing immediate decline in output but reducing downward pressure that slack exerts on inflation.
  - Larger price and wage increases occur than in the baseline, allowing adverse expectational dynamics to take hold.
  - Inflation becomes stickier and requires greater countervailing policy force to return to target.
  - Although nominal policy rates under the alternative strategy rise 150bps above their level in the baseline, the peak output decline is around twice as large relative to the baseline, and inflation outcomes remain inferior.

### 5.4 Quantitative highlights (from Figure 10 narrative)
- Delayed tightening leads to:
  - Sustained higher inflation relative to the baseline.
  - A deeper contraction in domestic output (peak decline ~ twice that under baseline).
  - Nominal policy rates rising 150bps above baseline in the tightening phase, yet still delivering worse inflation outcomes.

### 5.4 Policy implications
- Simulations validate concerns about “looking through” supply shocks in weak-framework EMs: delay in tightening can de-anchor inflation expectations and raise output costs.
- Pre-commitment to robust inflation stabilization is valuable when policy fundamentals are relatively weak.
- Central banks in weak-framework EMs should resist the temptation to delay tightening in response to supply-driven inflation shocks.

*Source: wpiea2025256-source-pdf*

### 4.1    Monetary policy implementation

### 4.1 Monetary policy implementation

### Taylor-rule estimates and reaction to inflation and output
- Estimated reaction function (Equation 2) uses policy rate r_{i,t}, one-year ahead inflation expectation E_t \bar{π}_{i,t+h} from consensus forecasts, the real-time GDP output gap \bar{y}_{i,t-3}, country fixed effects α_i, and time fixed effects τ_t. Regressions exclude EMs with fixed exchange rate regimes and Argentina, Türkiye, and Ukraine. Standard errors double clustered at country and month level.
- Main empirical findings:
  - Taylor rule coefficients on inflation deviations declined in the post-GFC period, interpreted as improved central bank credibility and more strongly anchored long-term inflation expectations.
  - EM central banks shifted attention toward curbing output fluctuations after the GFC: the Taylor rule coefficient on the output gap is negative and not statistically significant pre-GFC and positive and statistically significant post-GFC.
  - AE estimates in the post-crisis period are strikingly similar to EMs in the post-crisis period.

- Key coefficients from Table 1 (Taylor rule / Augmented Taylor rule)
  - Inflation gap:
    - EMs, Pre-GFC: 2.047**
    - EMs, Post-GFC: 1.314***
    - AEs, Post-GFC (benchmark): 1.572***
    - Augmented EMs, Pre-GFC: 2.251***
    - Augmented EMs, Post-GFC: 1.279***
  - Output gap:
    - EMs, Pre-GFC: -0.323
    - EMs, Post-GFC: 0.398**
    - AEs, Post-GFC: 0.355*
    - Augmented EMs, Pre-GFC: -0.104
    - Augmented EMs, Post-GFC: 0.487**
  - NEER depreciation (augmented specification):
    - Pre-GFC: 0.514***
    - Post-GFC: 0.247**
  - Observations:
    - Columns report 3,455; 3,455; 2,709; 3,455; 3,455 (by column)
  - R^2 reported as 0.993, 0.993, 0.996, 0.993, 0.993 (by column)
  - Notes: Pre-GFC = 1997–2009; Post-GFC = 2010–24.

### Exchange rate pass-through and anchoring of inflation expectations
- Exchange rate pass-through estimates (Equation 3) use cumulative CPI response to a one percentage point depreciation of the nominal exchange rate (∆be_{i,t} measured around US monetary announcements), with controls including 12 lags of inflation, exchange rate depreciation, and the policy rate.
- Findings:
  - Pre-GFC: two years after a one percentage point depreciation in the exchange rate, inflation was about 1.5 percentage point higher.
  - Post-GFC: no statistically significant exchange rate pass-through to inflation.
- Anchoring of expectations (Equation 4: rolling six-year regressions):
  - Sensitivity of long-term (three-year ahead) inflation expectations to changes in short-term (one-year ahead) inflation expectations declined substantially after the GFC.

### Perceived reaction function and markets’ perceptions
- Perceived monetary policy reaction function (Equation 5) estimated using forecaster-country level data from Consensus economics; regressions are run separately by calendar year with minimum 1,000 observations and standard errors clustered at forecaster level.
- Finding:
  - Progressive decline in the magnitude of the Taylor rule coefficient on expected inflation over time and a marginal increase in the size of the output gap coefficient — interpreted as gains in monetary policy credibility.

### Additional notes on estimation and samples
- Real-time output gap constructed using real GDP data available up to period t and World Economic Outlook forecasts for the following five years, reflecting data available to policymakers (Orphanides and Norden, 2002).
- Regressions follow Carvalho et al. (2021) with OLS; additional specifications control for NEER depreciation to capture fear-of-floating concerns.

*Italic line: Source: wpiea2025256-source-pdf - 4.1    Monetary policy implementation*

### 4.5    The contribution of policy frameworks to macroeconomic stabilization

### 4.5    The contribution of policy frameworks to macroeconomic stabilization

### Empirical approach
- Objective: quantify the contribution of policy frameworks relative to benign external conditions in improving emerging market (EM) resilience to risk-off events.
- Main regression (Equation 10): predicts the 12‑month change in log real GDP or log CPI (y_{i,τ+12}−y_{i,τ−1}) from:
  - A set of pre-determined policy variables P_{k,i,τ−1} (degree of anchoring of inflation expectations, reserve adequacy, FX mismatches, stringency of macroprudential regulation, external debt burden, cyclically-adjusted fiscal balance, stringency of CFM measures);
  - Lags of changes in log real GDP and log CPI (∆X_{i,τ−j});
  - Episode fixed effects α_τ to compare countries during the same risk-off episodes.
- To isolate external conditions, the common episode component (episode fixed effects minus policy contributions) ̄y_τ is regressed (Equation 11) on three proxies for external conditions G_{m,τ}: real GDP growth in advanced economies relative to trend, changes in commodity terms-of-trade, and US financial conditions (all expressed as cumulative changes over 12 months).

### Regression findings: role of individual policy metrics
- Countries entering risk-off episodes with stronger policy frameworks experienced higher real GDP growth and lower inflation (panel (a) of Figure 8; see Table A8 for regression results).
- Key marginal effects (one standard deviation changes):
  - A country entering a risk-off episode at a one standard deviation lower FX mismatches is expected to experience 0.9 percentage point higher growth relative to a country at the sample mean of FX mismatches.
  - A country at a one standard deviation greater anchoring of long-term inflation expectations tends to experience 2.3 percentage points lower inflation.

### Quantified overall contributions (pre-GFC vs post-GFC)
- Improved policy frameworks (accounting for observed changes in policy variables before and after the GFC) contributed to EM resilience during recent risk-off episodes by:
  - Raising growth by 0.5 percentage point in the period since the GFC compared with the period before the crisis.
  - Lowering inflation by 0.6 percentage point in the period since the GFC compared with the period before the crisis.
- Contribution of external conditions:
  - More benign external conditions contributed to faster growth in EMs after the GFC by 0.5 percentage point.
  - External conditions did not ease inflationary pressures.
- Robustness: these estimates are robust to exclusion of individual risk-off episodes or inclusion of the 2013 Taper tantrum episode (see Table A9).

### Structural analysis: QIPF model and key frictions
- Framework: a two-country New Keynesian QIPF model incorporating four key frictions:
  1. Limited risk-bearing capacity in FX markets (shallow FX markets) leading to a risk-augmented UIP premium: 1 = E_t[(I^*_t / I_t) (ε_{t+1}/ε_t)] + Γ B_{F,t} / P_t (Equation 12), where Γ≥0 is FX market shallowness.
  2. An occasionally binding external debt limit: −B_t / P_t ≤ m E_t Y_{t+1} (Equation 14), with m>0; when binding, a domestic spread Θ_t arises (I^b_t = I_t + Θ_t).
  3. Balance sheet FX mismatches captured via the law of motion for net foreign assets (Equation 15), with ownership share parameter 0≤ω≤1 governing effective currency composition of economy-wide net foreign assets and exposure to exchange rate risk.
  4. Imperfectly anchored inflation expectations achieved via a Kimball aggregator (non-linear Phillips curve) and dynamic indexation mechanisms that raise inflation persistence (as in Erceg et al. (2024)).
- Other model features: imperfect substitution between domestic and foreign goods, home bias, labor and capital inputs, standard monetary policy rule responding to inflation deviations and output gap, fiscal authority with lump-sum taxes for debt stabilization.

### Calibration
- Two EM calibrations:
  - Pre-GFC calibration: represents the average EM prior to the GFC (weaker policy frameworks).
  - Post-GFC calibration: represents the average EM after the GFC (stronger policy frameworks: more strongly anchored inflation expectations, smaller balance sheet mismatches).
- FX market depth Γ held fixed across calibrations to highlight effects of other frictions.
- Balance sheet targets set to match observed EM net foreign assets:
  - Pre-GFC net foreign assets: −38.4% of annual GDP.
  - Post-GFC net foreign assets: −25.9% of annual GDP.
- External borrowing limit set so the economy is in the constrained regime approximately 3 percent of the time under the pre-GFC calibration.
- All other parameters and shock properties calibrated based on Adrian et al. (2021).

### Model simulations: policy trade-offs and sudden stop probability
- Risk-off shock scenario: portfolio capital outflow normalized so that exchange rate depreciates by 10 percent under the pre-GFC calibration.
- Policy-tradeoff outcomes:
  - EM with strong policy frameworks:
    - Moderate increase in inflation because inflation expectations are strongly anchored.
    - Central bank does not need to tighten policy aggressively.
    - Domestic demand does not strongly decline; output may expand supported by higher net exports.
  - EM with weak policy frameworks:
    - Greater exchange rate pass-through to domestic prices produces a substantial increase in inflation and second-round wage pressures.
    - Central bank tightens much more aggressively, depressing domestic demand and causing a strong contraction in output.
    - Monetary authority may consider deploying FX interventions or CFMs to improve the trade-off.
- Sudden stop probability and severity:
  - Post-GFC improvements (including an average increase in net foreign assets of about 12.5 percent of GDP relative to pre-GFC) reduce sudden stop likelihood by two channels: being farther from the external debt limit and lower macroeconomic volatility from better-anchored inflation and lower FX mismatches.
  - Stochastic simulations keeping average distance from external debt unchanged show:
    - Probability of a sudden stop drops from about 3 percent to 1.5 percent.
    - Average credit spread during sudden stops falls by 1 percentage point (from 6.2 percent to 5.2 percent).

*Source: IMF working paper section 4.5, "The contribution of policy frameworks to macroeconomic stabilization."*

### 5.4    Delayed monetary tightening with weak frameworks

### 5.4    Delayed monetary tightening with weak frameworks

### Context and simulation setup
- The exercise considers an EM facing one-off shocks to capital flows and domestic costs that affect the exchange rate and inflation.
- Two monetary policy strategies are compared:
  - Baseline: central bank follows a simple policy rule in all periods.
  - Alternative: central bank deviates by keeping nominal interest rates lower than in the baseline for the year following the shocks, then switches to policy rates higher than in the baseline thereafter; the switch is not foreseen by agents ex ante but becomes perfectly credible thereafter.
- The alternative strategy is interpreted as a period of wait-and-see followed by a pivot toward tighter policy consistent with a revision in the central bank’s inflation forecasts.

### Key model outcomes and mechanisms
- Under the baseline rule-following strategy:
  - Trade-off between inflation and output stabilization is managed via a modest rise in real interest rates.
  - Output converges to potential from below and inflation converges to target from above in a manner consistent with good policy.
- Under the alternative (delayed tightening) strategy:
  - Real rates are initially low, which prevents an immediate decline in output but removes some downward pressure that slack exerts on inflation.
  - Larger price and wage increases occur than in the baseline, allowing adverse expectational dynamics to take hold.
  - Inflation becomes stickier and requires greater countervailing policy force to return to target.
  - Although nominal policy rates under the alternative strategy rise 150bps above their level in the baseline, the peak output decline is around twice as large relative to the baseline, and inflation outcomes remain inferior.

### Quantitative highlights (from Figure 10 narrative)
- Delayed tightening leads to:
  - Sustained higher inflation relative to the baseline.
  - A deeper contraction in domestic output (peak decline ~ twice that under baseline).
  - Nominal policy rates rising 150bps above baseline in the tightening phase, yet still delivering worse inflation outcomes.

### Policy implications
- The simulations validate concerns about “looking through” supply shocks in weak-framework EMs: delay in tightening can de-anchor inflation expectations and raise output costs.
- Pre-commitment to robust inflation stabilization is valuable when policy fundamentals are relatively weak.
- Central banks in weak-framework EMs should resist the temptation to delay tightening in response to supply-driven inflation shocks.

---

*Source: IMF staff analysis — "5.4    Delayed monetary tightening with weak frameworks" (chapter text).*

### References

### References

### Bibliographic references
- Comprehensive list of cited works on emerging markets, monetary policy spillovers, exchange rate regimes, macroprudential policy, capital flows, and related topics. Notable repeated themes include: US monetary policy and global spillovers; exchange rate pass-through; macroprudential and capital flow measures; credibility and inflation expectations; and the global financial cycle. (Full citations appear in the source reference list.)

### Online Appendix for: “Emerging Market Resilience: Good Luck or Good Policies?”
- Title and authors: “Emerging Market Resilience: Good Luck or Good Policies?” Marijn A. Bolhuis, Francesco Grigoli, Marcin Kolasa, Roland Meeks, Andrea F. Presbitero, and Zhao Zhang. Date: December 2, 2025.

- A-I Data
  - Table A1: Data sources (variables and exact sources listed). Key variable–source pairs (preserved exactly as in source):
    - CFM measures — Integrated Macroprudential Policy database iMaPP
    - Commodity terms of trade — Gruss and Kebhaj (2019)
    - Consumer prices — IMF, World Economic Outlook; Haver Analytics
    - Cyclically-adjusted balance — World Bank Cross-Country Database of Fiscal Space
    - De jure central bank independence — Romelli (2024)
    - Domestic monetary policy shock — Checo et al.(2024)
    - EMBI spread — Bloomberg; J.P. Morgan
    - External debt — IMF, World Economic Outlook; World Bank, Cross-Country Database of Fiscal Space
    - FX interventions — Adler et al.(2025)
    - FX mismatches — Allen and Juvenal (2025)
    - Government bond yields — Bloomberg; Haver Analytics
    - Inflation expectations — Consensus Economics
    - Inflation expectations anchoring — Bemset al.(2021)
    - Inflation targets — Haver Analytics
    - Macroprudential policy — Integrated Macroprudential Policy database iMaPP
    - Reserve adequacy — IMF, World Economic Outlook
    - RORO index — Bloomberg; Chariet al.(2023); Haver Analytics
    - Sovereign spreads — Haver Analytics; J.P. Morgan
    - US financial conditions — Haver Analytics
    - US monetary policy shock — Bauer and Swanson 2023

- A-II Additional Tables and Figures (table/figure inventory and select quantitative results preserved exactly)
  - Table A2: Dates, magnitude and duration, and events during risk-off episodes
    - Lists 16 risk-off episodes with Start, End, Magnitude (SD), Duration (Months), and Main events. Example entries preserved exactly:
      - 1997m8 — 1997m10 — 0.53 — Asian financial crisis
      - 1998m7 — 1998m9 — 1.73 — Russian financial crisis; Long-Term Capital Management collapse
      - 2000m9 — 2001m3 — 0.87 — Unraveling of dotcom bubble
      - 2001m6 — 2001m9 — 1.34 — Worsening of dotcom slump; September 11 attacks
      - 2002m4 — 2002m9 — 1.36 — Enron and WorldCom corporate scandals
      - 2002m12 — 2003m2 — 0.53 — Iraq war; Severe Acute Respiratory Syndrome epidemic
      - 2007m6 — 2008m4 — 0.81 — Collapse of Bear Stearns and BNP Paribas subprime funds
      - 2008m6 — 2008m11 — 2.66 — Failures of Indymac, Fannie Mae, Freddie Mac and Lehman Brothers
      - 2010m4 — 2010m6 — 1.03 — Start of Greek sovereign debt crisis; 2010 flash crash
      - 2011m5 — 2011m9 — 1.15 — US rating downgrade; worsening of European sovereign debt crisis
      - 2012m4 — 2012m5 — 1.12 — Peak of European sovereign debt crisis
      - 2015m5 — 2016m2 — 0.40 — China growth slowdown; commodity price collapse
      - 2018m8 — 2018m12 — 0.65 — Escalation of US-China trade tensions; concerns over Fed tightening
      - 2019m3 — 2019m5 — 0.33 — Lingering US-China trade tensions; global manufacturing slowdown
      - 2020m1 — 2020m3 — 2.93 — COVID-19 pandemic
      - 2022m4 — 2022m6 — 0.73 — Rising global inflation and monetary policy tightening; Russia-Ukraine war

  - Table A3: Macroeconomic dynamics during risk-off episodes (coefficients and sample/period definitions preserved exactly)
    - Panel A. Portfolio outflows (percent of GDP)
      - Risk-off episode, pre-GFC: -0.102*** (h=1); -0.155*** (h=2); -0.106** (h=3); -0.073** (h=4)
      - Risk-off episode, post-GFC: 0.048; -0.033; -0.056; -0.098
      - Observations: 1,151 (for h=1–4)
      - R2: 0.183, 0.242, 0.236, 0.232
    - Panel B. Nominal exchange rate depreciation (percent)
      - Risk-off episode, pre-GFC: 2.108*** (h=1); 8.789*** (h=2); 12.260*** (h=3); 10.035*** (h=4)
      - Risk-off episode, post-GFC: 3.884***; 6.106***; 4.384**; 3.058
      - Observations: 1,302
      - R2: 0.277, 0.358, 0.382, 0.398
    - Panel C. EMBI spread (percent)
      - Risk-off episode, pre-GFC: 0.794***; 1.633***; 1.496***; 1.146***
      - Risk-off episode, post-GFC: 0.678***; 0.959**; 0.876; 1.185*
      - Observations: 1,293; 1,291; 1,289; 1,286
      - R2: 0.250, 0.278, 0.267, 0.288
    - Panel D. Real GDP growth (percent)
      - Risk-off episode, pre-GFC: -0.647***; -1.867***; -2.803***; -3.021***
      - Risk-off episode, post-GFC: -0.020; -1.000*; -1.741***; -4.716***
      - Observations: 1,524
      - R2: 0.150, 0.229, 0.299, 0.332
    - Panel E. CPI inflation (percent)
      - Risk-off episode, pre-GFC: 0.682***; 0.878**; 0.943*; 0.747
      - Risk-off episode, post-GFC: 0.437*; -0.309**; -1.521; -0.891
      - Observations: 1,523; 1,522; 1,521; 1,520
      - R2: 0.632, 0.656, 0.691, 0.705
    - Notes: Pre-GFC period is 1997–2009 and post-GFC period is 2010–24. Robust standard errors reported in parentheses. ***, **,* denote significance at 1, 5, and 10 percent levels.

  - Table A4: Exchange rate pass-through to inflation (cumulative CPI response to 1 percentage point depreciation vis-à-vis the USD)
    - Exchange rate depreciation coefficients:
      - h=6: 0.332*** (0.121)
      - h=12: 0.704*** (0.206)
      - h=24: 1.358*** (0.403)
      - h=36: 1.767*** (0.608)
    - Interaction Exchange rate depreciation × post-GFC:
      - h=6: -0.319** (0.144)
      - h=12: -0.529** (0.258)
      - h=24: -1.189** (0.510)
      - h=36: -1.956** (0.787)
    - Observations: 5,291; 5,171; 4,931; 4,691
    - Adjusted R2: 0.662; 0.672; 0.682; 0.720
    - Notes: Pre-GFC period 1997–2009; post-GFC period 2010–24. Exchange rate depreciation instrumented using change between the day before and the day after scheduled US Federal Reserve monetary policy announcements. Standard errors clustered by time.

  - Table A5: Estimates of perceived monetary policy reaction function (Taylor rule coefficients by year)
    - Years with reported coefficients (exact values preserved). Examples:
      - 2004: Expected inflation coefficient 2.360 (standard error 0.440); Expected real GDP growth coefficient -0.120 (standard error 0.160); R2 = 0.900; Observations = 1,009
      - 2008: Expected inflation 2.990 (0.150); Expected real GDP growth -0.350 (0.120); R2 = 0.900; Observations = 1,951
      - 2016: Expected inflation 1.100 (0.060); Expected real GDP growth -0.070 (0.070); R2 = 0.820; Observations = 2,736
      - 2024: Expected inflation 1.330 (0.120); Expected real GDP growth 0.580 (0.140); R2 = 0.830; Observations = 2,713
    - Notes: Sample includes years with at least 1,000 observations and excludes countries with fixed exchange rate regimes, Argentina, Türkiye, and Ukraine. Forecaster fixed effects included. Standard errors clustered at forecaster level. ***, **,* denote statistical significance at 1, 5, and 10 percent levels.

  - Table A6: Use of FX interventions conditioning on policy frameworks (Percent of GDP)
    - Panel A (Conditioning on inflation expectation anchoring): UIP × inflation expectations anchoring coefficients:
      - h=3: -0.838*** (0.218)
      - h=6: -1.289*** (0.336)
      - h=9: -1.305*** (0.370)
      - h=12: -1.214*** (0.376)
      - Observations: 2,034; 2,032; 2,032; 2,032
      - Adjusted R2: -0.467; -0.546; -0.321; -0.143
    - Panel B (Conditioning on the stringency of macroprudential regulation): UIP × stringency of MPru regulation coefficients:
      - h=3: 0.035 (0.054)
      - h=6: -0.043 (0.077)
      - h=9: -0.195* (0.100)
      - h=12: -0.353*** (0.127)
      - Observations: 2,390; 2,385; 2,382; 2,379
      - Adjusted R2: -0.013; -0.002; -0.041; -0.128
    - Notes: Sample excludes EMs with fixed exchange rate regimes. Standard errors clustered at month level.

  - Table A7: Use of CFM measures conditioning on policy frameworks (Index)
    - Panel A (Conditioning on inflation expectation anchoring): UIP × inflation expectations anchoring coefficients:
      - h=3: -0.111*** (0.039)
      - h=6: -0.096 (0.059)
      - h=9: -0.162** (0.074)
      - h=12: -0.230** (0.090)
      - Observations: 2,944; 2,902; 2,860; 2,818
      - Adjusted R2: -0.130; -0.117; -0.122; -0.125
    - Panel B (Conditioning on the stringency of macroprudential regulation): UIP × stringency of MPru regulation coefficients:
      - h=3: -0.027 (0.032)
      - h=6: -0.052 (0.051)
      - h=9: -0.029 (0.065)
      - h=12: 0.028 (0.083)
      - Observations: 3,557 for all horizons
      - Adjusted R2: -0.137; -0.143; -0.093; -0.101
    - Notes: Sample excludes Argentina, Türkiye, and Ukraine. Standard errors clustered at month level.

  - Table A8: Policy determinants of resilience to risk-off events (one-year outcomes after start of episode)
    - Dependent variables: Panel A — Real GDP growth; Panel B — CPI inflation (one-year changes, expressed in percentage points (100×log points)). Policy variables expressed in standard deviations.
    - Selected coefficients (exact values preserved):
      - Panel A (Real GDP growth): Examples across specifications (columns (1)–(7)):
        - X (Inflation expectation anchoring): 0.138 (0.538) in column (1)
        - Macropru stringency: 0.475* (0.268) in column (2)
        - External debt burden: -0.834*** (0.243) in column (3)
        - Reserve adequacy: 0.614** (0.253) in column (4)
        - FX mismatches: -0.948*** (0.332) in column (5)
        - Observations vary: 273; 378; 378; 347; 270; 344; 348
        - Adjusted R2: 0.306; 0.311; 0.331; 0.350; 0.335; 0.303; 0.308
      - Panel B (CPI inflation):
        - X (Inflation expectation anchoring): -2.349** (0.932) in column (1)
        - Macropru stringency: -0.529** (0.261) in column (2)
        - External debt burden: 1.122*** (0.365) in column (3)
        - Reserve adequacy: -0.451 (0.408) in column (4)
        - FX mismatches: 0.939** (0.464) in column (5)
        - Observations same as Panel A
        - Adjusted R2: 0.590; 0.525; 0.533; 0.609; 0.619; 0.568; 0.524
    - Notes: Policy variable definitions preserved exactly (e.g., external debt burden = stock of external debt as a percentage of exports; reserve adequacy = stock of FX reserves as a percentage of short-term external debt; FX mismatches = foreign currency liabilities as a percentage of exports). Robust standard errors reported in parentheses. ***, **,* denote statistical significance at 1, 5, and 10 percent levels.

  - Table A9: Robustness of the contributions of policy frameworks and external conditions
    - Reports estimated contributions to Real GDP growth and CPI inflation during risk-off episodes excluding one episode at a time (start month listed) or including all episodes and the 2013 Taper Tantrum.
    - Example entries preserved exactly (selected rows):
      - Excluding 1997m8: Policy frameworks — Real GDP growth 0.55; CPI inflation -0.69; External conditions — Real GDP growth 0.46; CPI inflation 0.03
      - Mean row: 0.55; -0.57; 0.45; 0.03
      - Median row: 0.55; -0.58; 0.45; 0.03
      - Including Taper Tantrum: 0.55; -0.56; 0.45; 0.03
    - Notes: External condition proxies include real GDP growth in advanced economies, commodity terms-of-trade shocks, and the US FCI-G index.

  - Figure A1: Likelihood and severity of sudden stops (simulation results)
    - Panel (a): Probability of sudden stops (percent)
    - Panel (b): Credit spread during sudden stops (percent) — credit spread computed as borrowing rate minus policy rate.
    - Notes: Stochastic simulation results illustrating how stronger policy frameworks change the probability and severity of sudden stops.

- Overall appendix structure: Data provenance (Table A1), empirical identification of 16 risk-off episodes (Table A2), regression evidence on macroeconomic dynamics and policy responsiveness (Tables A3–A9), and a simulation figure on sudden stops (Figure A1).

*References and Online Appendix content as provided in the source PDF.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025256-source-pdf.pdf_