## _wp1601 - Section  3  shows  some  stylized  facts  and  discusses  some  potential  theoretical  channels  for

## Source details

**Canonical URL:** [_wp1601 - Section  3  shows  some  stylized  facts  and  discusses  some  potential  theoretical  channels  for](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp1601.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp1601.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp1601.pdf.json)

---

### Literature Review
- Central themes:
  - Relationship between the exchange rate regime and the evolution of domestic prices in EMs.
  - Role of liability dollarization, "fear of floating", and "original sin" in amplifying ERPT (Calvo and Reinhart (2002); Hausmann et al. (2006)).
  - Taylor hypothesis: adoption of inflation targeting (IT) is associated with lower ERPT because lower inflation reduces firms' pricing power (Taylor (2000); Choudhri and Hakura (2006); Gagnon and Ihrig (2004); Edwards (2006); Mishkin and Schmidt-Hebbel (2007); Coulibaly and Kempf (2010)).
- Evidence on non-linearities and asymmetries:
  - Bussière (2013): finds non-linearities and asymmetries in trade prices for G7 with heterogeneity across countries.
  - Frankel et al. (2012): threshold effect for large devaluations; depreciations above 25 percent have proportionately larger pass-through; find asymmetries with possible downward price rigidity.
  - Carranza et al. (2009): in dollarized economies large real depreciation may reduce aggregate investment via balance sheet effects.
  - Burstein et al. (2005): after nine large post-1990 contractionary devaluations inflation rates were relatively low compared to devaluation magnitudes, due to distribution costs and import substitution.
  - Pollard and Coughlin (2004): industry-level asymmetries in the United States; pass-through higher when exchange rate movements are large.

### Stylized facts and theoretical underpinnings for asymmetries and non-linearities
- Data and basic patterns:
  - Sample period for country-level averages: 1980-2014.
  - Four stylized facts distinguishing EMs from AEs:
    - Inflation is higher and more volatile in EMs than in AEs.
    - Higher inflation in EMs is associated with higher depreciation rates compared to AEs.
    - Depreciation rates are positively associated with depreciation volatility in EMs; no apparent relationship in AEs.
    - Depreciation volatility and inflation volatility are positively associated in EMs; not evident for AEs.
- Potential economic channels generating asymmetries and non-linearities:
  - Pricing-to-market and downward rigidity of export prices:
    - Exporters can increase markups more easily than decrease them, so depreciations tend to have larger effects on import prices than appreciations.
    - Large appreciations on exporters' side exacerbate rigidity and increase pass-through on the importer side.
  - Firm heterogeneity and product quality (Berman et al. (2012); Corsetti and Dedola (2005) extension):
    - Higher productivity / higher-quality firms can absorb more of exchange rate movements in markups, implying lower pass-through for imports of high-quality goods.
    - Large depreciations may drive low-quality exporters out, leaving high-quality firms that absorb more of the exchange rate movement — generating non-linearities (possibly smaller pass-through for big depreciations).
    - Conversely, export sectors populated by small firms with small mark-ups may pass on a larger share of appreciation.
  - Upward rigidity of export quantities and capacity constraints:
    - Firms at full capacity facing a depreciation may raise markups rather than expand production, reducing pass-through on import prices.
  - Commodity exposure and CPI composition in EMs:
    - EMs often have large shares of food and fuel (imported) in CPI baskets; commodity price volatility can cause large exchange rate movements and sudden effects on import prices and domestic inflation.
- Unconditional data checks and suggestive patterns:
  - Conditional on thresholds, slopes differ:
    - Relationship between inflation and depreciation is steeper when the exchange rate depreciates by more than 10 or 20 percent (evidence of non-linearity).
    - Comparison across regimes: pass-through coefficient is higher for non-targeters (consistent with the Taylor hypothesis).
    - Asymmetry between depreciation and appreciation episodes: inflation reacts differently during depreciations versus appreciations.

### Empirical model: local projections
- Method and motivation:
  - Use Jordà (2005) local projection (LPs) method to estimate dynamic responses of inflation to exchange rate movements while allowing for non-linearities and asymmetries.
  - Define asymmetries as differences between appreciation and depreciation episodes.
  - Define non-linearities as:
    - i) depreciation above or below certain thresholds (e.g., >10 percent, >20 percent, >25 percent noted in literature),
    - ii) countries operating under IT vs. non-targeters.
- Advantages of LPs:
  - Flexible semi-parametric technique that directly estimates sequences of linear projections for future values of dependent variables.
  - Avoids non-linear transformations of slope coefficients required for VAR impulse responses.
  - Easier to extend to panel data with interaction terms to capture state dependence, asymmetries, and non-linearities.
  - OLS-based estimation provides straightforward inference without asymptotic transformations.
- Limitations and mitigation:
  - LPs lose observations at longer horizons due to forecasting horizon; IRFs may oscillate at long horizons.
  - LPs can be erratic at long horizons; use monthly data to maximize sample length.
  - Comparisons with VARs: if VAR is a good approximation, it is optimal, but LPs are more robust to misspecification and less subject to bias from non-linear transformations of estimated parameters.

### Baseline model (linear LP specification)
- Baseline local projection specification (linear model):
  - Equation (1):
    - ∆CPI_{i,t+h} = α_h + Σ_{s=1}^{p} ρ_{s,h} ∆CPI_{i,t−s} + β_h ∆NEER_{it} + μ_h crisis_t + Σ_{s=1}^{p} γ_{s,h} ∆CPI^*_{i,t−s} + ε_{i,t+h}
  - Definitions and choices:
    - ∆CPI: year-on-year percent change in the CPI of country i at time t.
    - ∆NEER: year-on-year percent change in the nominal effective exchange rate.
    - crisis: dummy equal to one from 2009 to 2012 to proxy for the financial crisis.
    - ∆CPI^*: percent change in foreign prices, extracted from the Relative Price Index (RPI) reported in the Information Notice System (INS) database.
    - Number of lags p chosen: 3 (tested country by country; on average the correct number of lags is 3).
  - Interpretation:
    - β_h traces the response of inflation at t+h to a depreciation/appreciation occurring at t.
- Econometric challenges and solutions:
  - Overlapping data and moving average error structure due to yoy percent changes with monthly data:
    - Serial correlation and cross-sectional dependence addressed using Driscoll-Kraay standard errors (Driscoll and Kraay (1998)).
  - Stationarity and data exclusions:
    - Estimate in differences to target stationarity.
    - Exclude periods when yoy percent changes are larger than 100.
    - Panel stationarity/unit root tests implemented: Maddala and Wu (1999) (Phillips and Perron and ADF versions); Im et al. (2003); Pesaran (2006). Tests confirm stationarity (Table 1).
  - Endogeneity:
    - Potential endogeneity between depreciation and inflation acknowledged.
    - Structural VAR identification and instrument issues discussed; authors indicate more comprehensive treatment in final section.

### 6.1 Linear model — Baseline estimates (full EM sample, 28 countries)
- ERPT on consumer prices:
  - around 22 percent after 12 months after the initial shock
  - reaches 25 percent after two years
- Example interpretation:
  - "for a depreciation of 100 percent yoy inflation would increase by 22 percent."
- Comparison with selected literature:
  - Choudhri and Hakura (2006): ERPT coefficient of 14 percent after 1 quarter, 24 percent after 4 quarters and 27 percent after 20 quarters; higher-inflation countries around 50 percent after 4 quarters.
  - Bussière and Peltonen (2008): pass-through on import prices is 35 percent after 1 quarter.
  - Kohlscheen (2010): pass through equal to 5 percent, 17 percent, 20 percent, 24 percent (after 3, 6, 9, 12 months).
- Selected baseline coefficient estimates (Table 2):
  - Horizon 1 month: ERPT 0.07; Std. Error 0.00
  - Horizon 12 months: ERPT 0.22; Std. Error 0.05
  - Horizon 14 months: ERPT 0.26; Std. Error 0.10

### 6.2 Non-linearities (threshold episodes)
- Episode definitions (two alternative criteria):
  - 1) monthly yoy percent change > 10 percent
  - 2) monthly yoy percent change > 20 percent
- Dummy: depisode = 1 if ∆NEER > 0 and ∆NEER > Ψ; 0 otherwise (Ψ per above thresholds)
- Main findings:
  - High depreciation episodes have been quite frequent since 1990; frequency of >20 percent episodes has been reduced during the last 10 years.
  - Responses of inflation during episodes of large depreciation are statistically different from responses during normal times.
  - For 10 percent depreciation episodes:
    - effect of prices is significant for almost a year
    - ERPT = 40 percent after 6 months
    - ERPT = 57 percent after 12 months
    - in normal times ERPT = 10 percent
    - implication: yoy inflation increases by 4 percentage points in the 6 months after a 10 percent depreciation episode
  - For 20 percent depreciation episodes:
    - ERPT = 44 percent after 6 months
    - ERPT = 45 percent after 12 months
  - Dynamics:
    - During big depreciation episodes ERPT is faster: after 1 month ERPT is almost 20 percent and reaches 40 percent after 6 months.
    - Depreciations of 20 percent and more have an even faster effect.
  - Interpretation:
    - Large movements may reduce the wedge between importers and consumers, forcing importers to pass through more to consumers.
  - Related literature:
    - Frankel et al. (2012) find depreciation of 25 percent have proportionately larger pass-through.

### 6.3 Asymmetries (depreciation vs appreciation)
- Specification: adds interaction with ddepr, where ddepr = 1 if ∆NEER > 0; 0 otherwise
- Main findings:
  - Significant evidence of asymmetries in the first 8 months after the initial shock.
  - Appreciation episodes generate a positive but not significant reaction in inflation.
  - Depreciation episodes:
    - about 38 percent pass-through after 12 months
  - Appreciation episodes:
    - less than 10 percent after 12 months

### 6.4 Targeters vs non-targeters (Taylor hypothesis test)
- Specification includes IT dummy (IT = 1 if country is an inflation targeter).
- Main finding:
  - Pass-through for inflation targeters is considerably lower than for non-inflation targeters.
  - Non-inflation targeters display more than 20 percent pass-through after 12 months.

### 7 Robustness
- 7.1 Addressing endogeneity:
  - Analytical approach: derive direction of bias using simple bivariate model:
    - ∆CPI = α + β ∆NEER + ε
    - ∆NEER = δ + γ_i ∆CPI + ν
  - Key implication:
    - Expect γ_IT > γ_NON−IT, implying targeters should suffer a higher upward bias than non-targeters (B_IT > B_NON−IT).
    - If endogeneity were driving results, targeters would display higher coefficients than non-targeters.
    - Empirical result: targeters do not display higher coefficients than non-targeters (Figure 9), which attenuates concerns about reverse causality bias.
- 7.2 Permanent vs temporary episodes:
  - Permanent episode: exchange rate depreciates by 20 percent for more than 3 consecutive months.
  - Results:
    - In baseline permanent episode case ERPT > 40 percent after 6 months and then becomes insignificant.
    - Additional combinations tested: 20 percent for more than 6 months; 10 percent for more than 3 months; 10 percent for more than 6 months.
    - Turkey has a higher number of episodes than average; re-estimation without Turkey yields slightly lower coefficients but results remain in line with baseline.
- 7.3 Dealing with outliers:
  - Procedures:
    - Dropped hyper-inflation episodes as in Fischer et al. (2002).
    - Implemented median regression (minimizes absolute residuals) with bootstrapped standard errors.
  - Findings:
    - Results align with baseline, ERPT highly significant but slightly slower.
    - Non-linear interaction results hold and show higher significance than baseline.
    - For 20 percent episodes median regression yields ERPT almost 100 percent after 20 months.
    - Advanced economies comparison (sample of 27): ERPT = 6 percent after 1 year.
- 7.4 Dealing with omitted variables:
  - Controls added:
    - Banking and sovereign crises variables (Laeven and Valencia (2012)).
    - Country fixed effects to control for country-specific time trends.
  - Findings:
    - ERPT coefficients robust to inclusion of crisis controls and country fixed effects (Figure 18).
    - Comparison of targeters vs non-targeters robust to these inclusions (Figure 19).

### 8 Conclusion and policy implications
- Main empirical findings:
  - During depreciations greater than 10 percent and 20 percent ERPT is equal to 40 percent compared to 20 percent during normal times.
  - Depreciations elicit faster and more pronounced responses in domestic prices.
  - Evidence of asymmetries: depreciations have larger pass-through than appreciations.
  - Adoption of inflation targeting reduces the degree of ERPT.
  - Results robust to tests for reverse causality, permanent vs temporary episodes, outliers, and omitted variables.
- Policy implications:
  - Policymakers should acknowledge non-linear and state-dependent dynamics when designing policy responses to exchange rate shocks.
  - Monetary policy responses should not assume time-invariant or state-independent parameters for the transmission of exchange rate shocks to inflation.
  - Caution is warranted in policy-rate adjustments after/during depreciation episodes and/or changes in the economy’s nominal anchor to avoid harming price stability and anchoring of inflation expectations.

*Source: Excerpt from the supplied content unit.*

### Section  3  shows  some  stylized  facts  and  discusses  some  potential  theoretical  channels  for

### _wp1601 - Section  3  shows  some  stylized  facts  and  discusses  some  potential  theoretical  channels  for

### Literature Review
- Central themes:
  - Relationship between the exchange rate regime and the evolution of domestic prices in EMs.
  - Role of liability dollarization, "fear of floating", and "original sin" in amplifying ERPT (Calvo and Reinhart (2002); Hausmann et al. (2006)).
  - Taylor hypothesis: adoption of inflation targeting (IT) is associated with lower ERPT because lower inflation reduces firms' pricing power (Taylor (2000); Choudhri and Hakura (2006); Gagnon and Ihrig (2004); Edwards (2006); Mishkin and Schmidt-Hebbel (2007); Coulibaly and Kempf (2010)).
- Evidence on non-linearities and asymmetries:
  - Bussière (2013): finds non-linearities and asymmetries in trade prices for G7 with heterogeneity across countries.
  - Frankel et al. (2012): threshold effect for large devaluations; depreciations above 25 percent have proportionately larger pass-through; find asymmetries with possible downward price rigidity.
  - Carranza et al. (2009): in dollarized economies large real depreciation may reduce aggregate investment via balance sheet effects.
  - Burstein et al. (2005): after nine large post-1990 contractionary devaluations inflation rates were relatively low compared to devaluation magnitudes, due to distribution costs and import substitution.
  - Pollard and Coughlin (2004): industry-level asymmetries in the United States; pass-through higher when exchange rate movements are large.

### Stylized facts and theoretical underpinnings for asymmetries and non-linearities
- Data and basic patterns:
  - Sample period for country-level averages: 1980-2014.
  - Four stylized facts distinguishing EMs from AEs:
    - Inflation is higher and more volatile in EMs than in AEs.
    - Higher inflation in EMs is associated with higher depreciation rates compared to AEs.
    - Depreciation rates are positively associated with depreciation volatility in EMs; no apparent relationship in AEs.
    - Depreciation volatility and inflation volatility are positively associated in EMs; not evident for AEs.
- Potential economic channels generating asymmetries and non-linearities:
  - Pricing-to-market and downward rigidity of export prices:
    - Exporters can increase markups more easily than decrease them, so depreciations tend to have larger effects on import prices than appreciations.
    - Large appreciations on exporters' side exacerbate rigidity and increase pass-through on the importer side.
  - Firm heterogeneity and product quality (Berman et al. (2012); Corsetti and Dedola (2005) extension):
    - Higher productivity / higher-quality firms can absorb more of exchange rate movements in markups, implying lower pass-through for imports of high-quality goods.
    - Large depreciations may drive low-quality exporters out, leaving high-quality firms that absorb more of the exchange rate movement — generating non-linearities (possibly smaller pass-through for big depreciations).
    - Conversely, export sectors populated by small firms with small mark-ups may pass on a larger share of appreciation.
  - Upward rigidity of export quantities and capacity constraints:
    - Firms at full capacity facing a depreciation may raise markups rather than expand production, reducing pass-through on import prices.
  - Commodity exposure and CPI composition in EMs:
    - EMs often have large shares of food and fuel (imported) in CPI baskets; commodity price volatility can cause large exchange rate movements and sudden effects on import prices and domestic inflation.
- Unconditional data checks and suggestive patterns:
  - Conditional on thresholds, slopes differ:
    - Relationship between inflation and depreciation is steeper when the exchange rate depreciates by more than 10 or 20 percent (evidence of non-linearity).
    - Comparison across regimes: pass-through coefficient is higher for non-targeters (consistent with the Taylor hypothesis).
    - Asymmetry between depreciation and appreciation episodes: inflation reacts differently during depreciations versus appreciations.

### Empirical model: local projections
- Method and motivation:
  - Use Jordà (2005) local projection (LPs) method to estimate dynamic responses of inflation to exchange rate movements while allowing for non-linearities and asymmetries.
  - Define asymmetries as differences between appreciation and depreciation episodes.
  - Define non-linearities as:
    - i) depreciation above or below certain thresholds (e.g., >10 percent, >20 percent, >25 percent noted in literature),
    - ii) countries operating under IT vs. non-targeters.
- Advantages of LPs:
  - Flexible semi-parametric technique that directly estimates sequences of linear projections for future values of dependent variables.
  - Avoids non-linear transformations of slope coefficients required for VAR impulse responses.
  - Easier to extend to panel data with interaction terms to capture state dependence, asymmetries, and non-linearities.
  - OLS-based estimation provides straightforward inference without asymptotic transformations.
- Limitations and mitigation:
  - LPs lose observations at longer horizons due to forecasting horizon; IRFs may oscillate at long horizons.
  - LPs can be erratic at long horizons; use monthly data to maximize sample length.
  - Comparisons with VARs: if VAR is a good approximation, it is optimal, but LPs are more robust to misspecification and less subject to bias from non-linear transformations of estimated parameters.

### Baseline model
- Baseline local projection specification (linear model):
  - Equation (1):
    - ∆CPI_{i,t+h} = α_h + Σ_{s=1}^{p} ρ_{s,h} ∆CPI_{i,t−s} + β_h ∆NEER_{it} + μ_h crisis_t + Σ_{s=1}^{p} γ_{s,h} ∆CPI^*_{i,t−s} + ε_{i,t+h}
  - Definitions and choices:
    - ∆CPI: year-on-year percent change in the CPI of country i at time t.
    - ∆NEER: year-on-year percent change in the nominal effective exchange rate.
    - crisis: dummy equal to one from 2009 to 2012 to proxy for the financial crisis.
    - ∆CPI^*: percent change in foreign prices, extracted from the Relative Price Index (RPI) reported in the Information Notice System (INS) database.
    - Number of lags p chosen: 3 (tested country by country; on average the correct number of lags is 3).
  - Interpretation:
    - β_h traces the response of inflation at t+h to a depreciation/appreciation occurring at t.
- Econometric challenges and solutions:
  - Overlapping data and moving average error structure due to yoy percent changes with monthly data:
    - Serial correlation and cross-sectional dependence addressed using Driscoll-Kraay standard errors (Driscoll and Kraay (1998)) — nonparametric covariance estimator robust to heteroskedasticity, autocorrelation, and cross-sectional dependence.
  - Stationarity of the data generating process:
    - Estimate in differences to target stationarity.
    - Exclude periods of hyperinflation: exclude periods when yoy percent changes are larger than 100 (following Fischer et al. (2002)); additional exclusion similar to Bussière et al. (2014).
    - Battery of panel stationarity/unit root tests implemented: Maddala and Wu (1999) in both Phillips and Perron and ADF versions; Im et al. (2003) test; Pesaran (2006) t-test to account for cross-sectional dependence. Tests confirm stationarity (Table 1).
  - Potential endogeneity between depreciation and inflation:
    - Recognized possibility that depreciation may be correlated with the error term.
    - Structural VAR identification via short-run restrictions (Choleski) is common but can be restrictive on timing assumptions (Edwards (2006)).
    - Finding valid instruments for depreciation is difficult (Meese and Rogoff (1983); Edwards (2006)); many studies rely on OLS.
    - Authors indicate they will address endogeneity more comprehensively in the final section and discuss potential direction of bias.

*Source: Excerpt from the supplied content unit.*

### 6.1  Linear model

### 6.1  Linear model

### Baseline estimates
- Sample: full sample of emerging markets (28 countries).
- Exchange rate pass-through (ERPT) on consumer prices:
  - around 22 percent after 12 months after the initial shock
  - reaches 25 percent after two years
- Interpretation example: "for a depreciation of 100 percent yoy inflation would increase by 22 percent."
- Consistency with literature:
  - Choudhri and Hakura (2006): ERPT coefficient of 14 percent after 1 quarter, 24 percent after 4 quarters and 27 percent after 20 quarters; higher-inflation countries around 50 percent after 4 quarters.
  - Bussière and Peltonen (2008): pass-through on import prices is 35 percent after 1 quarter.
  - Kohlscheen (2010): pass through equal to 5 percent, 17 percent, 20 percent, 24 percent (after 3, 6, 9, 12 months).

### Key statistical/formal elements
- Baseline model estimated for 28 emerging markets.
- The model includes lags of ∆CPI, contemporaneous ∆NEER, crisis controls, terms for foreign CPI lags, and interaction terms for episode dummies (see source equations).

---

### 6.2  Non-linearities

- Goal: test whether price responses during periods of high depreciation are more than proportional relative to normal times.
- Episode definitions (two alternative criteria):
  - 1) monthly yoy percent change > 10 percent
  - 2) monthly yoy percent change > 20 percent
- Dummy variable depisode defined as:
  - depisode = 1 if ∆NEER > 0 and ∆NEER > Ψ; 0 otherwise (Ψ per above thresholds)
- Main findings:
  - High depreciation episodes have been quite frequent since 1990; frequency of >20 percent episodes has been reduced during the last 10 years.
  - Responses of inflation during episodes of large depreciation are statistically different from responses during normal times.
  - For 10 percent depreciation episodes:
    - effect of prices is significant for almost a year
    - ERPT = 40 percent after 6 months
    - ERPT = 57 percent after 12 months
    - in normal times ERPT = 10 percent
    - implication: yoy inflation increases by 4 percentage points in the 6 months after a 10 percent depreciation episode
  - For 20 percent depreciation episodes:
    - ERPT = 44 percent after 6 months
    - ERPT = 45 percent after 12 months
  - Dynamics:
    - During big depreciation episodes ERPT is faster: after 1 month ERPT is almost 20 percent and reaches 40 percent after 6 months.
    - Depreciations of 20 percent and more have an even faster effect.
  - Interpretation: large movements may reduce the wedge between importers and consumers, forcing importers to pass through more to consumers.
- Related literature: Frankel et al. (2012) find depreciation of 25 percent have proportionately larger pass-through.

---

### 6.3  Asymmetries

- Objective: test whether depreciation effects are symmetric to appreciation effects of the same size.
- Specification adds interaction with ddepr, where:
  - ddepr = 1 if ∆NEER > 0; 0 otherwise
- Main findings:
  - Significant evidence of asymmetries in the first 8 months after the initial shock.
  - Appreciation episodes generate a positive but not significant reaction in inflation.
  - Depreciation episodes:
    - about 38 percent pass-through after 12 months
  - Appreciation episodes:
    - less than 10 percent after 12 months

---

### 6.4  Targeters vs non-targeters

- Hypothesis tested: inflation targeting reduces ERPT.
- Specification includes IT dummy (IT = 1 if country is an inflation targeter).
- Main finding:
  - Pass-through for inflation targeters is considerably lower than for non-inflation targeters.
  - Non-inflation targeters display more than 20 percent pass-through after 12 months.

---

### 7  Robustness

### 7.1  Addressing endogeneity
- Concern: potential endogeneity / reverse causality between inflation and depreciation.
- Analytical approach: derive direction of bias using simple bivariate model:
  - ∆CPI = α + β ∆NEER + ε
  - ∆NEER = δ + γ_i ∆CPI + ν
  - OLS estimator and bias expressions derived (see source equations (7) and (8)).
- Key reasoning and implications:
  - Expect γ_IT > γ_NON−IT, implying targeters should suffer a higher upward bias than non-targeters (B_IT > B_NON−IT).
  - If endogeneity were driving results, targeters would display higher coefficients than non-targeters.
  - Empirical result: targeters do not display higher coefficients than non-targeters (Figure 9), which attenuates concerns about reverse causality bias.

### 7.2  Permanent vs temporary episodes
- Definition: permanent episode when exchange rate depreciates by 20 percent for more than 3 consecutive months.
- Results:
  - In baseline permanent episode case ERPT > 40 percent after 6 months and then becomes insignificant.
  - Additional combinations tested: 20 percent for more than 6 months; 10 percent for more than 3 months; 10 percent for more than 6 months.
  - Turkey has a higher number of episodes than average; re-estimation without Turkey yields slightly lower coefficients but results remain in line with baseline.

### 7.3  Dealing with outliers
- Procedure:
  - Dropped hyper-inflation episodes as in Fischer et al. (2002).
  - Implemented median regression (minimizes absolute residuals) with bootstrapped standard errors.
- Findings:
  - Results align with baseline, ERPT highly significant but slightly slower.
  - Non-linear interaction results hold and show higher significance than baseline.
  - For 20 percent episodes median regression yields ERPT almost 100 percent after 20 months.
  - Advanced economies comparison (sample of 27): ERPT = 6 percent after 1 year.

### 7.4  Dealing with omitted variables
- Controls added:
  - Banking and sovereign crises variables (Laeven and Valencia (2012)).
  - Country fixed effects to control for country-specific time trends.
- Findings:
  - ERPT coefficients robust to inclusion of crisis controls and country fixed effects (Figure 18).
  - Comparison of targeters vs non-targeters robust to these inclusions (Figure 19).

---

### 8  Conclusion and policy implications

- Summary of main empirical findings:
  - During depreciations greater than 10 percent and 20 percent ERPT is equal to 40 percent compared to 20 percent during normal times.
  - Depreciations elicit faster and more pronounced responses in domestic prices.
  - Evidence of asymmetries: depreciations have larger pass-through than appreciations.
  - Adoption of inflation targeting reduces the degree of ERPT.
  - Results robust to tests for reverse causality, permanent vs temporary episodes, outliers, and omitted variables.
- Policy implications:
  - Policymakers should acknowledge non-linear and state-dependent dynamics when designing policy responses to exchange rate shocks.
  - Monetary policy responses should not assume time-invariant or state-independent parameters for the transmission of exchange rate shocks to inflation.
  - Caution is warranted in policy-rate adjustments after/during depreciation episodes and/or changes in the economy’s nominal anchor to avoid harming price stability and anchoring of inflation expectations.

*Source: _wp1601 - 6.1  Linear model*

### References

### _wp1601 - References

### References (bibliography)
- Aron, J., Macdonald, R., Muellbauer, J., 2014.  Exchange rate pass-through in developing and emerging markets:  A survey of conceptual, methodological and policy issues, and selected empirical findings.  Journal of Development Studies 50(1), 101–143.
- Auerbach, A. J. & Gorodnichenko, Y., 2013.  Fiscal multipliers in recession and expansion, in:  Fiscal Policy After the Financial Crisis. Alberto Alesina and Francesco Giavazzi.
- Berman, N., Martin, P., Mayer, T., 2012.  How do different exporters react to exchange rate changes?  The Quarterly Journal of Economics 127(1), 437–492.
- Burstein, A., Eichenbaum, M., Rebelo, S., 2005.  Large devaluations and the real exchange rate.  Journal of Political Economy.
- Bussière, M., 2013.  Exchange rate pass-through to trade prices:  The role of nonlinearities and asymmetries*.  Oxford Bulletin of Economics and Statistics 75(5), 731–758.
- Bussière, M., Delle Chiaie, S., Peltonen, T.A., 2014.  Exchange rate pass-through in the global economy:  The role of emerging market economies.  IMF Economic Review 62(1), 146–178.
- Bussière, M., Peltonen, T.A., 2008.  Exchange rate pass-through in the global economy:  the role of emerging market economies.
- Calvo, G.A., Reinhart, C.M., 2002. Fear of floating. Quarterly Journal of Economics 117(2), 379–408.
- Canova, F., Ciccarelli, M., 2009.  Estimating multicountry var models*.  International Economic Review 50(3), 929–959.
- Carranza, L., Galdon-Sanchez, J.E., Gomez-Biscarri, J., 2009.  Exchange rate and inflation dynamics in dollarized economies.  Journal of Development Economics 89(1), 98–108.
- Ca’Zorzi, M., Hahn, E., S ́anchez, M., 2007. Exchange rate pass-through in emerging markets. ECB Working Paper No 739, March 2007.
- Choudhri, E.U., Hakura, D.S., 2006.  Exchange rate pass-through to domestic prices:  does the inflationary environment matter?  Journal of International Money and Finance 25(4), 614–639.
- Corsetti, G., Dedola, L., 2005. A macroeconomic model of international price discrimination. Journal of International Economics 67(1), 129–155.
- Coulibaly, D., Kempf, H., 2010. Does inflation targeting decrease exchange rate pass-through in emerging countries?
- Driscoll, J.C., Kraay, A.C., 1998.  Consistent covariance matrix estimation with spatially dependent panel data.  Review of economics and statistics 80(4), 549–560.
- Edwards, S., 2006. The relationship between exchange rates and inflation targeting revisited. Technical Report. National Bureau of Economic Research.
- Fischer, S., Sahay, R., V ́egh, C.A., 2002.  Modern hyper-and high inflations. Technical Report. National Bureau of Economic Research.
- Frankel, J., Parsley, D., Wei, S.J., 2012. Slow pass-through around the world:  a new import for developing countries?  Open Economies Review 23(2), 213–251.
- Gagnon, J.E., Ihrig, J., 2004. Monetary policy and exchange rate pass-through. International Journal of Finance & Economics 9(4), 315–338.
- Hausmann, R., Panizza, U., Rigobon, R., 2006.  The long-run volatility puzzle of the real exchange rate.  Journal of International Money and Finance 25(1), 93–124.
- Im, K.S., Pesaran, M.H., Shin, Y., 2003.  Testing for unit roots in heterogeneous panels. Journal of econometrics 115(1), 53–74.
- Jarotschkin, A., Kraay, A., 2013.  Aid, disbursement delays, and the real exchange rate. World Bank Policy Research Working Paper (6501).
- Jord`a, `O., 2005. Estimation and inference of impulse responses by local projections. American economic review pp. 161–182.
- Jord`a, `O., Schularick, M., Taylor, A.M., 2013.  When credit bites back.  Journal of Money, Credit and Banking 45(s2), 3–28.
- Jord`a, `O., Taylor, A.M., 2013.  The time for austerity:  estimating the average treatment effect of fiscal policy. Technical Report. National Bureau of Economic Research.
- Kilian, L., Kim, Y.J., 2011. How reliable are local projection estimators of impulse responses? Review of Economics and Statistics 93(4), 1460–1466.
- Kohlscheen, E., 2010. Emerging floaters: pass-throughs and (some) new commodity currencies. Journal of International Money and Finance 29(8), 1580–1595.
- Laeven, L., Valencia, F., 2012.  Systemic banking crises database:  An update.
- Leduc, S., Wilson, D., 2012. Roads to prosperity or bridges to nowhere? theory and evidence on the impact of public infrastructure investment. Technical Report. National Bureau of Economic Research.
- Maddala, G.S., Wu, S., 1999.  A comparative study of unit root tests with panel data and a new simple test.  Oxford Bulletin of Economics and statistics 61(S1), 631–652.
- Meese, R.A., Rogoff, K., 1983.  Empirical exchange rate models of the seventies:  Do they fit out of sample?  Journal of international economics 14(1), 3–24.
- Mishkin, F.S., Schmidt-Hebbel, K., 2007.  Does inflation targeting make a difference? Technical Report. National Bureau of Economic Research.
- Panizza, U., Presbitero, A.F., 2014.  Public debt and economic growth:  is there a causal effect?  Journal of Macroeconomics 41, 21–41.
- Pesaran, M.H., 2006. Estimation and inference in large heterogeneous panels with a multifactor error structure. Econometrica 74(4), 967–1012.
- Pollard, P.S., Coughlin, C.C., 2004. Size matters: asymmetric exchange rate pass-through at the industry level. University of Nottingham Research Paper (2004/13).
- Ramey, V.A., 2012. Comment on roads to prosperity or bridges to nowhere? theory and evidence on the impact of public infrastructure investment, National Bureau of Economic Research.
- Ramey, V.A., Zubairy, S., 2014. Government spending multipliers in good times and in bad: Evidence from us historical data. Technical Report. National Bureau of Economic Research.
- Ronayne, D., 2011. Which impulse response function? Warwick Economic Research Paper No. 971.
- Taylor, J.B., 2000. Low inflation, pass-through, and the pricing power of firms. European economic review 44(7), 1389–1408.

### Figures and Tables (list of figures and select table contents)
- Figures (titles as listed)
  - Figure 1: Exchange rate movements after the tapering announcement
  - Figure 2: Depreciation and Inflation in EMs and AEs
  - Figure 3: Stylezed Facts - Non-linearities
  - Figure 4: Exchange rate pass-through - Linear model
  - Figure 5: Depreciation and Inflation in EMs and AEs
  - Figure 6: Exchange rate pass-through during 20 percent depreciation episodes
  - Figure 7: Exchange rate pass-through during 10 percent depreciation episodes
  - Figure 8: Exchange rate pass-through during appreciation vs depreciation
  - Figure 9: Exchange rate pass-through: targeters vs non-targeters
  - Figure 10: Distribution of permanent episodes
  - Figure 11: Exchange rate pass-through: permanent vs temporary (3 percent episodes)
  - Figure 12: Exchange rate pass-through: permanent vs temporary (all episodes)
  - Figure 13: Distribution of the inflation variable
  - Figure 14: Median regression - linear specification
  - Figure 15: Median regression - 10 percent episodes
  - Figure 16: Median regression - 20 percent episodes
  - Figure 17: Exchange rate pass-through - Advanced Economies
  - Figure 18: Exchange rate pass-through - dealing with omitted variables
  - Figure 19: Targeters vs non-targeters - dealing with omitted variables
  - Figure 20: Depreciation and Inflation in EMs and AEs (country panels)
  - Figure 21: Depreciation and Inflation (country panels)
  - Figure 22: Depreciation and Inflation (country panels)

- Tables (titles and reported numerical entries)
  - Table 1: Stationarity tests
    - Inflation / Exchange rate change
    - Fischer - PP
      - Inverse chi-squared(56) 591.59 0.00 33.80 0.00
      - Inverse normal -17.11 0.00 -13.89 0.00
      - Inverse logit t(144) -30.61 0.00 -17.36 0.00
      - Modified inv. chi-squared 50.61 0.00 26.25 0.00
    - Fischer - ADF
      - Inverse chi-squared(56) 306.27 0.00 416.17 0.00
      - Inverse normal -11.50 0.00 -15.24 0.00
      - Inverse logit t(144) -14.84 0.00 -21.48 0.00
      - Modified inv. chi-squared 23.65 0.00 34.03 0.00
    - Im-Pesaran-Shin
      - Wtbar -10.92 0.00 -14.82 0.00
    - Pesaran CADF
      - Ztbar -9.41 0.00 -13.39 -2.08

  - Table 2: ERPT coefficient - Linear model
    - Horizon - months / ERTP / Std. Error
      - 1 0.07 0.00
      - 12 0.22 0.05
      - 14 0.26 0.10

  - Table 3: IT adoption dates
    - Country / Start of IT
      - Brazil 1999m6
      - Chile 1999m9
      - Colombia 1999m10
      - Guatemala 2005m1
      - Hungary 2001m6
      - Indonesia 2005m7
      - Mexico 2001m1
      - Peru 2002m1
      - Philippines 2002m1
      - Poland 1998m10
      - Serbia 2006m9
      - South Africa 2000m2
      - Thailand 2000m5

  - Table 4: Data sources (Appendix)
    - Variable / Data source / Description
      - CPI / IMF - IFS / Consumer price index (2010=100)
      - NEER / Bruegel / Nominal effective exchange rate
      - RPI / IMF - INS / Domestic CPI / weighted averages of the CPI of partner countries

  - Table 5: Summary statistics (Depreciation means and sd; two columns labeled "Depreciation")
    - Albania 3.15 22.38 7.51 10.62
    - Algeria 6.25 14.86 9.69 10.60
    - Brazil 18.34 36.45 7.83 7.47
    - Chile -3.84 5.51 2.39 1.14
    - China 0.36 9.30 3.18 4.54
    - Colombia 1.42 11.13 11.58 8.75
    - Egypt 3.90 14.26 8.78 5.41
    - Guatemala 1.13 5.96 8.57 7.22
    - Hungary 4.80 8.15 11.77 9.14
    - Indonesia 4.33 18.79 10.73 12.89
    - Jordan -1.33 6.34 4.05 3.53
    - Latvia -1.92 9.48 9.13 12.47
    - Lithuania -1.73 6.49 8.34 14.32
    - Malaysia -0.23 10.44 2.81 1.57
    - Mexico 5.26 11.99 10.71 10.04
    - Pakistan 5.64 4.88 9.19 4.47
    - Peru 6.40 19.66 9.41 15.81
    - Philippines 1.93 8.71 6.04 3.84
    - Poland 4.13 10.86 13.76 18.16
    - Russian Federation 13.27 24.59 16.09 15.01
    - Serbia 13.28 24.43 20.35 19.08
    - Seychelles 1.60 13.26 4.71 11.48
    - South Africa 4.36 12.16 7.28 4.38
    - Thailand 0.21 8.16 3.51 2.38
    - Tunisia 2.02 2.25 4.07 1.60
    - Turkey 21.75 20.71 39.20 31.12
    - Ukraine 12.04 24.52 13.91 14.42
    - Uruguay 7.51 15.11 18.96 20.02

### Appendix (figures, distributions, and additional panels)
- Figures 20–22: Country-level panels showing "Depreciation and Inflation" and separate panels per country (ALB, BRA, CHL, CHN, COL, DZA, EGY, GTM, HUN, IDN, JOR, LTU, LVA, MEX, MYS, PAK, PER, PHL, POL, RUS, SRB, SYC, THA, TUN, TUR, UKR, URY, ZAF) with plotted ranges as indicated (axes showing values such as -100 to 100 and other country-specific ranges).
- Various figures compare responses during "normal time" and during episodes (10% episode, 20% episode), appreciation vs depreciation, targeters vs non-targeters, permanent vs temporary episodes, median regressions, and robustness checks (omitted variables, no Turkey).

*Source: _wp1601 - References (PDF).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp1601.pdf_