## wp17206

## Source details

**Canonical URL:** [wp17206](https://www.imf.org/-/media/files/publications/wp/2017/wp17206.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2017/wp17206.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2017/wp17206.pdf.json)

---

### Overview and research question
- Research question: "What would have been the level of inflation if the CNB did not announce the exchange rate floor?"
- Context: CNB introduced a one-sided exchange rate floor at CZK 27 to the euro in November 2013 as an additional instrument within its inflation-targeting framework to avoid deflation after policy rates hit the zero lower bound.
- JEL Classification Numbers: F31; E58
- Keywords: Foreign exchange intervention, exchange rate, synthetic control method
- CNB policy rate brought to technical zero: 0.05 percent (November 2012).
- FX floor introduced and communicated target level: CZK 27/EUR (November 7, 2013).
- CNB commitments/extensions:
  - Keep exchange rate close to CZK 27/EUR at least until the start of 2015 (communication on December 17, 2013).
  - Would not discontinue the use of the exchange rate as a monetary policy instrument before 2016 (communication on December 17, 2014).
  - At meeting on February 2, 2017, confirmed it would not discontinue use before the second quarter of 2017 and would intervene to keep the koruna close to CZK 27/EUR.

### Data, sample, and stylized facts
- Sample: 14 countries from 2006Q1 to 2015Q3 (countries: Bulgaria, Croatia, Czech Republic, Denmark, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia, Slovenia, Sweden and the United Kingdom). Sample start for some estimations: 2007Q3.
- All variables expressed in 12-month growth rates; inflation computed using HICP (Eurostat).
- Observed patterns:
  - Headline and core inflation reversed their downward trend after the introduction of the FX floor.
  - Headline inflation has been moving very close to zero since the beginning of 2015.
  - Import price inflation shows a clear upward trend after the introduction of the floor, suggesting the policy might have worked through higher price of imports.
- Key summary statistics (Table 1):
  - Headline inflation: Obs. 585; Mean 0.028; Std. Dev. 0.028; Min -0.039; Max 0.175
  - Core inflation: Obs. 585; Mean 0.019; Std. Dev. 0.020; Min -0.049; Max 0.111
  - Oil price: Obs. 585; Mean 0.009; Std. Dev. 0.045; Min -0.089; Max 0.150
  - Food price: Obs. 585; Mean 0.009; Std. Dev. 0.045; Min -0.089; Max 0.150
  - Unemployment gap: Obs. 585; Mean -0.001; Std. Dev. 0.014; Min -0.049; Max 0.062
  - Administered prices: Obs. 585; Mean 0.007; Std. Dev. 0.007; Min -0.014; Max 0.042
  - Taxes: Obs. 585; Mean 0.005; Std. Dev. 0.009; Min -0.030; Max 0.060
  - Inflation expectations - 1 year ahead: Obs. 495; Mean 0.025; Std. Dev. 0.012; Min -0.023; Max 0.091
  - Core inflation in the EA: Obs. 585; Mean 0.005; Std. Dev. 0.002; Min 0.001; Max 0.013

### Empirical methodologies
- Three complementary strategies to construct counterfactuals and estimate effect on inflation:
  - Event study: estimate open-economy New Keynesian Phillips curve on pre-event sample and use forecasts to measure deviations attributable to the intervention.
  - Difference-in-difference (DD): compare headline and core inflation in the Czech Republic before/after November 2013 relative to a control group of similar European countries; dynamic panel with country and time fixed effects, clustered standard errors.
  - Synthetic control method: build a weighted average of other unaffected countries to closely match pre-treatment characteristics.
- Event-study regression fit: R-squared 0.91 on pre-treatment sample (estimation sample up to 2013Q4; sample start 2007Q3).
- DD specification summary:
  - 휋it = α + 휋t−1 + 휋t−2 + βXit + λi + γt + δTREATi * POSTt + εit.
  - POSTt dummy indicates quarters from 2013Q4 onwards.
  - Standard errors clustered at country level.
- Synthetic control preferred specification includes six quarters of lagged inflation: 2009Q3, 2010Q1, 2011Q2, 2012Q1, 2012Q3, 2013Q3. RMSPE for preferred specification: 0.007.

### Main quantitative findings
- Overall effect on headline inflation:
  - Introduction of the FX floor prevented headline inflation from going into negative territory.
  - Estimated magnitude of the impact varies between 0.5 and 1.5 percentage points across specifications.
- Difference-in-difference interaction-term coefficients (FX floor effect across six specifications in Table 2):
  - (1) 0.005* (standard error 0.003)
  - (2) 0.007*** (standard error 0.002)
  - (3) 0.009*** (standard error 0.002)
  - (4) 0.015*** (standard error 0.003)
  - (5) 0.003 (standard error 0.002)
  - (6) 0.005 (standard error 0.003)
- Lag coefficients (selected):
  - Inflation t-1: (1) -0.327*** (0.040); (2) -0.287** (0.087); (3) -0.318*** (0.061); (4) -0.275*** (0.077); (5) -0.356*** (0.059); (6) -0.250*** (0.067)
  - Inflation t-2: (1) 0.937*** (0.048); (2) 0.849*** (0.128); (3) 0.908*** (0.089); (4) 0.798*** (0.109); (5) 0.836*** (0.097); (6) 0.671*** (0.107)
- Selected control variable coefficients (Column 1 example):
  - Unemployment gap: -0.098*** (0.037)
  - Price of food: 0.062** (0.030)
  - Inflation expectation: 0.386*** (0.055)
  - Contribution of admin. prices: 0.488*** (0.082)
  - Contribution of taxes: 0.200*** (0.043)
  - EA inflation: -0.209 (0.654)
- Observations and fit examples:
  - Observations (Column 1): 462; Number of id: 14; R-squared: 0.952
  - Observations (Column 2): 297; Number of id: 9; R-squared: 0.937
  - R-squared up to 0.960 in some specifications.
- Synthetic control headline-inflation increase: between 0.4 and 1.1 percentage points (time-horizon dependent).
- Core inflation:
  - Effect of interaction term similar in magnitude to headline estimates.
  - Coefficient on interaction term significant across the first five specifications with values ranging from 0.1 to 0.6 percentage points.
- Robustness and sensitivity:
  - Results robust across event study, DD, and synthetic control and several robustness checks (anticipatory-effect tests, leads and lags, placebo tests, alternate samples such as non-EMU).
  - Some DD specifications with full country-specific time trends (Columns 5 and 6) show positive but not significant interaction, indicating sensitivity to specification and potential exposure to different shocks.
  - GMM estimator with robust standard errors yields coefficient on interaction equal to 0.07, consistent with fixed effects estimation.

### Transmission channels, dynamics, and interpretation
- Transmission evidence:
  - Import price inflation increased after the floor, suggesting part of the mechanism operated through higher import prices.
  - Inflation expectations channel supported; CNB communication presented FX tool as temporary to guide expectations.
- Dynamics and anticipatory effects:
  - Four lead variables and four lag variables used to test anticipation and dynamics.
  - No anticipatory effects when controlling for specific time trends (Figure 9 – Column 5 and 6 of Table 3).
  - Less conservative specifications show some signs of anticipation; CNB started communicating potential use as early as September 2012.
- Market reaction and operations:
  - On announcement day, the koruna moved from 25.8 to close to 27, trading at end of day almost exactly 27.
  - Interventions limited since November 2013, but koruna strengthened and moved close to the floor in recent years; from summer 2015 a strengthening coincided with higher interventions.
  - CNB improved communications: written statement by the Bank Board published after press conferences, summary of Inflation report published the day after Board meeting, dedicated exchange rate commitment page on CNB website.

### Comparative and contextual notes
- CNB had prior FX intervention episodes: February–July 1998; October 1999–March 2000; October 2001–September 2002.
- Comparable international examples referenced:
  - Swiss National Bank minimum exchange rate of 1.20 francs to the euro (September 2011).
  - Central Bank of Brazil announced FX intervention program (August 2013).
  - Bank of Israel has intervened since 2008 to smooth excess volatility.
- Literature context: empirical evidence on FX intervention effectiveness is mixed; recent literature emphasizes macroeconomic management motives in addition to precautionary motives.

### Conclusions and suggested further research
- Common conclusion across methods: introduction of the FX floor prevented Czech inflation from going into negative territory and helped fight deflationary pressures.
- Estimated effects on headline and core inflation: values ranging between 0.5 to 1.5 percentage points (synthetic control: 0.4 to 1.1 percentage points depending on horizon).
- Suggested avenues for further research:
  - Explicitly investigate transmission channels, e.g., test the impact of the FX floor on import and export prices.

*Source: wp17206 - 1.5 percentage points. The results are robust to different econometric specifications.*

### 1.5 percentage points. The results are robust to different econometric specifications.

### wp17206 - 1.5 percentage points. The results are robust to different econometric specifications.

### Overview and research question
- Research question: "What would have been the level of inflation if the CNB did not announce the exchange rate floor?"
- Context: CNB introduced a one-sided exchange rate floor at CZK 27 to the euro in November 2013 as an additional instrument within its inflation-targeting framework to avoid deflation after policy rates hit the zero lower bound.
- JEL Classification Numbers: F31; E58
- Keywords: Foreign exchange intervention, exchange rate, synthetic control method
- Author’s E-mail Address: fcaselli@imf.org
- Working Paper disclaimer: This Working Paper should not be reported as representing the views of the IMF. The views expressed ... do not necessarily represent those of the IMF or IMF policy.

### Key institutional and policy facts
- CNB policy rate brought to technical zero: 0.05 percent (November 2012).
- FX floor introduced and communicated target level: CZK 27/EUR (November 7, 2013).
- CNB commitments/extensions:
  - Keep exchange rate close to CZK 27/EUR at least until the start of 2015 (communication on December 17, 2013).
  - Would not discontinue the use of the exchange rate as a monetary policy instrument before 2016 (communication on December 17, 2014).
  - At meeting on February 2, 2017, confirmed it would not discontinue use before the second quarter of 2017 and would intervene to keep the koruna close to CZK 27/EUR.

### Data and stylized environment (summary)
- CNB had implemented FX interventions in three earlier episodes: February–July 1998; October 1999–March 2000; October 2001–September 2002.
- Comparable international examples referenced:
  - Swiss National Bank set minimum exchange rate of 1.20 francs to the euro (September 2011).
  - Central Bank of Brazil announced FX intervention program (August 2013).
  - Bank of Israel has intervened since 2008 to smooth excess volatility.

### Empirical methodologies employed
- Three complementary strategies to construct counterfactuals and estimate the effect on inflation:
  - Event study:
    - Estimate a standard New Keynesian Phillips curve for the Czech Republic on the pre-event sample.
    - Use forecasts to measure deviations attributable to the intervention.
  - Difference-in-difference:
    - Compare headline and core inflation in the Czech Republic before/after November 2013 relative to a control group of similar European countries.
    - Test for anticipatory effects and dynamic patterns.
    - Estimate multiple specifications with multiple sets of fixed effects to control omitted variables.
  - Synthetic control method (Abadie et al. (2010)):
    - Build a synthetic control as a weighted average of other unaffected countries to closely match pre-treatment characteristics.

### Main findings and quantitative estimates
- Overall effect on headline inflation:
  - Introduction of the FX floor prevented headline inflation from going into negative territory.
  - Estimated magnitude of the impact varies between 0.5 and 1.5 percentage points.
- Core inflation:
  - Some evidence that the exchange rate floor had an impact on core inflation.
  - Brůha and Tonner (2017) find introduction of the floor prevented core inflation from falling into negative territory; effects on other macro variables were positive but not statistically significant.
- Robustness:
  - Results are robust across the three different methodologies (event study, difference-in-difference, synthetic control) and several robustness checks.
  - The paper implements multiple specifications, anticipatory-effect tests, leads and lags analyses, placebo tests, and alternate samples (e.g., non-EMU).

### Policy interpretation and mechanisms
- Rationale for FX floor:
  - With policy rate at 0.05 percent and limited scope for quantitative/qualitative easing, CNB used an exchange rate floor to depreciate the koruna and offset deflationary risks by affecting import prices and inflation expectations.
- Communication strategy:
  - Key to success was presenting the FX tool as temporary to guide inflation expectations without de-anchoring long-term expectations.
- Comparative literature:
  - Empirical evidence on FX intervention effectiveness is mixed; recent literature emphasizes a macroeconomic management motive in addition to precautionary motives.
  - Prior studies find varying impacts depending on methodology, frequency of interventions, and country context.

### Structure of the paper (content inventory of main sections)
- I. Introduction
- II. The Exchange Rate Tool in the CNB’s IT Framework
- III. Data and Stylized Facts
- IV. Empirical Methodologies
  - A. An Event Study Approach
  - B. A Difference-in-Difference Approach
  - C. A Synthetic Method Approach
- V. Conclusion

*Source: wp17206 - 1.5 percentage points. The results are robust to different econometric specifications.*

### introduction of the floor was mostly critical. The change in policy was seen as unexpected

### wp17206 - introduction of the floor was mostly critical. The change in policy was seen as unexpected

### CNB FX operations and immediate market reaction
- On the day of the announcement, the exchange rate of the Czech koruna with the euro moved from 25.8 to close to 27, trading at the end of the day at almost exactly 27.
- Interventions in the FX market have been limited since November 2013, but the koruna has been strengthening and moving close to the floor in recent years.
- Starting in the summer of 2015, a strengthening of the koruna coincided with higher interventions.
- The CNB changed its communications strategy, providing details on monetary policy decisions to the public on the day following the monetary policy meeting; communications improvements include a written statement by the Bank Board published after the press conference on the day of the monetary policy meeting, a summary of the Inflation report published the day after the Board meeting, and a dedicated page to the exchange rate commitment on the CNB website.

### Data and stylized facts (sample and variables)
- Sample: 14 countries from 2006Q1 to 2015Q3 (countries: Bulgaria, Croatia, Czech Republic, Denmark, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia, Slovenia, Sweden and the United Kingdom). Sample start for some estimations: 2007Q3.
- All variables are expressed in 12-month growth rates.
- Inflation is computed using the Harmonized Indices of Consumer Prices (HICP) published by Eurostat.
- Commodity price changes: world oil and food price indices in US dollars multiplied by the weights of energy and food in the consumer baskets.
- Cyclical unemployment rate: Baxter-King bandpass filter.
- Impact of taxes and administered prices: their contribution to headline inflation.
- Inflation expectations: one-year horizon.
- Price pressures from the euro area: euro area core inflation multiplied by the share of foreign value added in domestic demand.
- Observed patterns:
  - Headline and core inflation reversed their downward trend after the introduction of the FX floor.
  - Headline inflation has been moving very close to zero since the beginning of 2015.
  - Import price inflation shows a clear upward trend after the introduction of the floor, suggesting the policy might have worked through higher price of imports.

### Key summary statistics (Table 1)
- Headline inflation: Obs. 585; Mean 0.028; Std. Dev. 0.028; Min -0.039; Max 0.175
- Core inflation: Obs. 585; Mean 0.019; Std. Dev. 0.020; Min -0.049; Max 0.111
- Oil price: Obs. 585; Mean 0.009; Std. Dev. 0.045; Min -0.089; Max 0.150
- Food price: Obs. 585; Mean 0.009; Std. Dev. 0.045; Min -0.089; Max 0.150
- Unemployment gap: Obs. 585; Mean -0.001; Std. Dev. 0.014; Min -0.049; Max 0.062
- Administered prices: Obs. 585; Mean 0.007; Std. Dev. 0.007; Min -0.014; Max 0.042
- Taxes: Obs. 585; Mean 0.005; Std. Dev. 0.009; Min -0.030; Max 0.060
- Inflation expectations - 1 year ahead: Obs. 495; Mean 0.025; Std. Dev. 0.012; Min -0.023; Max 0.091
- Core inflation in the EA: Obs. 585; Mean 0.005; Std. Dev. 0.002; Min 0.001; Max 0.013

### Empirical methodologies (research question and approaches)
- Research question: What would have been the level of Czech inflation in the absence of the FX floor?
- Challenges: counterfactual not observed; difficulty building a good control group for macro policies.
- Three empirical strategies adopted:
  - An event study that uses the pre-intervention sample to build a model to forecast inflation in the absence of the FX floor.
  - A difference-in-difference regression analysis that compares the evolution of inflation in the Czech Republic with a control group of similar countries.
  - A synthetic control method that chooses a control group that best matches pre-treatment characteristics of the treated country.

### Event study approach (model specification and results)
- Estimated open-economy new Keynesian Phillips curve on pre-intervention sample (up to 2013Q4 excluded):
  - Dependent variable: Czech headline inflation (휋t).
  - Regressors: lagged inflation (휋t−1, 휋t−2), oil (표il_t) and food (food_t) world prices adjusted by weights, unemployment gap (u_t), administered prices contribution (admin_t), taxes contribution (tax_t), inflation expectations (πexp_t), euro-area core inflation weighted by foreign value added (EAπ_t). The nominal exchange rate omitted in baseline because directly impacted by policy.
- Estimation sample: up to 2013Q4; sample start 2007Q3.
- Regression fit: R-squared 0.91.
- Findings:
  - Predicted inflation from pre-treatment regression trends into negative territory after the floor introduction.
  - Actual inflation remains above predicted values; difference becomes statistically significant towards beginning of 2015 (with a 95 percent confidence interval).
  - Robustness: augmenting baseline with the nominal exchange rate yields consistent results—predicted inflation in absence of FX floor trends into negative territory.

### Difference-in-difference (DD) approach (specification and results)
- DD estimator setup:
  - 푌CZE denotes Czech inflation (treated); 푌i denotes inflation in control country.
  - DD estimate: 훿DD = (Y_CZE_post − Y_CZE_pre) − (Y_control_post − Y_control_pre).
  - Regression (dynamic panel): 휋it = α + 휋t−1 + 휋t−2 + βXit + λi + γt + δTREATi * POSTt + εit.
  - TREATi dummy indicates Czech Republic; POSTt dummy indicates quarters from 2013Q4 onwards. TREAT and POST absorbed by fixed effects.
  - Controls Xit: same controls as in event-study model.
  - Standard errors clustered at country level.
- Parallel trends assessment:
  - Inflation in the Czech Republic and control group followed parallel trends before introduction of the FX floor (including post-global financial crisis restriction).
  - After 2013Q4, average inflation in control group is lower than in Czech Republic.
- Baseline Table 2 results (six specifications reported; note exact numbers preserved):
  - FX floor (interaction term) coefficients across columns:
    - (1) 0.005* (standard error 0.003)
    - (2) 0.007*** (standard error 0.002)
    - (3) 0.009*** (standard error 0.002)
    - (4) 0.015*** (standard error 0.003)
    - (5) 0.003 (standard error 0.002)
    - (6) 0.005 (standard error 0.003)
  - Lag coefficients:
    - Inflation t-1: (1) -0.327*** (0.040); (2) -0.287** (0.087); (3) -0.318*** (0.061); (4) -0.275*** (0.077); (5) -0.356*** (0.059); (6) -0.250*** (0.067)
    - Inflation t-2: (1) 0.937*** (0.048); (2) 0.849*** (0.128); (3) 0.908*** (0.089); (4) 0.798*** (0.109); (5) 0.836*** (0.097); (6) 0.671*** (0.107)
  - Selected control variable coefficients (Column 1 example):
    - Unemployment gap: -0.098*** (0.037)
    - Price of food: 0.062** (0.030)
    - Inflation expectation: 0.386*** (0.055)
    - Contribution of admin. prices: 0.488*** (0.082)
    - Contribution of taxes: 0.200*** (0.043)
    - EA inflation: -0.209 (0.654)
  - Observations and fit:
    - Observations (Column 1): 462; Number of id: 14; R-squared: 0.952
    - Observations (Column 2): 297; Number of id: 9; R-squared: 0.937
    - Observations (Columns 3–6 vary as reported in Table 2); R-squared up to 0.960 in some specifications.
  - Fixed effects and sample choices:
    - Country FE: YES in all columns.
    - Time FE: YES in all columns.
    - Pre country specific time trends included in Columns 3 and 4.
    - Country specific time trends included in Columns 5 and 6.
    - NO-EMU sub-sample excludes Slovakia, Estonia, Lithuania, Latvia and Slovenia.
  - Statistical significance notation: *** p<0.01, ** p<0.05, * p<0.1.
- Interpretation of DD results:
  - The interaction term is positive and significant across the first four specifications.
  - Magnitude of estimated effect varies between 0.5 to 1.5 percentage points across specifications; baseline model (Column 1) implies the introduction of the floor brought inflation up by 0.5 percentage points in Czech Republic relative to the control group.
  - Suggests that in absence of the floor, inflation would have been between 0.5 and 1.5 percentage points lower.
  - Results remain significant when restricting sample to non-EMU countries (Column 2) and when including pre-treatment country-specific time trends (Columns 3 and 4).
  - Specifications with full country-specific time trends (Columns 5 and 6) show positive but not significant interaction, suggesting possible exposure to different shocks after the floor or that the specification may be too demanding for country-level data.
- Robustness checks:
  - GMM estimator with robust standard errors yields coefficient on interaction equal to 0.07, consistent with fixed effects estimation (noting dynamic panel fixed-effects bias discussion).

### Summary of empirical inference
- Event study: Predicted inflation without FX floor trends into negative territory; actual inflation diverged positively from prediction, with significance emerging by early 2015 at 95 percent CI. Regression R-squared 0.91 for pre-treatment fit.
- Difference-in-difference: Positive and largely significant effect of FX floor on inflation in baseline specifications; baseline effect ~0.5 percentage points relative to control group, with alternative specifications implying effects up to 1.5 percentage points. Some specifications with country-specific trends render effect statistically insignificant, implying possible confounding shocks or specification sensitivity.
- Import price inflation increase after the floor suggests part of the mechanism could be through higher import prices.

*Source: wp17206 - introduction of the floor was mostly critical. The change in policy was seen as unexpected (IMF Working Paper content provided).*

### introduction of the floor and the treated country (a dummy equal to one for the Czech

### wp17206 - introduction of the floor and the treated country (a dummy equal to one for the Czech Republic)

### Leads and lags: testing for anticipatory effects and dynamic impact
- Four lead variables coded to take value 1 only in the four quarters before the FX floor introduction, and 0 otherwise.
- Four lag variables coded to take value 1 from one quarter after the FX floor introduction to four quarters after the introduction, and 0 otherwise.
- No anticipatory effects when controlling for specific time trends (Figure 9 – Column 5 and 6 of Table 3).
- Less conservative specifications show some signs that the introduction of the FX floor has been anticipated.
- The CNB started communicating the potential use of the FX floor as early as September 2012, which can explain instability in anticipatory-effect estimates.

### Core inflation: parsimonious augmented New Keynesian Phillips curve
- Dependent variable: core inflation (excludes world oil and food prices, administered prices and the contribution of taxes).
- Core inflation data source: Eurostat, downloaded through Haver.
- The effect of the interaction term (FX floor) is similar in magnitude to headline inflation estimates.
- Coefficient on interaction term:
  - Significant across the first five specifications.
  - Values ranging from 0.1 to 0.6 percentage points.

### Synthetic control method: construction and fit
- Synthetic control purpose: build an alternative control group to validate common-trend assumption for difference-in-difference.
- Excluded nominal exchange rate from estimation (instrument of the policy).
- Best-fit specification includes six quarters of lagged inflation: 2009Q3, 2010Q1, 2011Q2, 2012Q1, 2012Q3, 2013Q3.
- Root mean squared error (RMSPE) for the preferred specification: 0.007.
- Pre-treatment inflation is best replicated by a combination of Denmark, Slovenia and Slovakia (Table 5).
- Comparison of actual Czech Republic and synthetic series shows accurate pre-treatment replication (Figure 11).
- Estimated increase in headline inflation due to FX floor (synthetic control): between 0.4 and 1.1 percentage points, depending on time horizon.

### Placebo tests and robustness
- Placebo exercise: estimate the same model allowing all other countries to be “treated” in 2013Q4.
- Placebo rationale: test whether the estimated impact could be driven by chance.
- Findings:
  - Pre-treatment synthetic fit is relatively good (difference close to zero).
  - Positive but small effect of FX floor on inflation for the Czech Republic.
  - Divergence relative to other countries is not extreme; similar magnitude fluctuations present in pre-treatment sample.
  - Overall placebo results consistent with a small effect of the FX floor on inflation, in line with event study and difference-in-difference results.
- Robustness excluding countries that joined the euro during the estimation period:
  - Similar pattern emerges, but poorer fit.
  - RMSPE for non-EMU sample: 0.009.
  - Pre-intervention inflation best reproduced by a combination of Bulgaria, Croatia and Denmark.

### Overall empirical findings
- All estimation strategies used: event study, difference-in-difference, synthetic control.
- Common conclusion: introduction of the FX floor prevented Czech inflation from going into negative territory and helped fight deflationary pressures.
- Estimated effects on inflation:
  - Headline and core inflation: values ranging between 0.5 to 1.5 percentage points.
  - Synthetic control headline-inflation increase: between 0.4 and 1.1 percentage points (time-horizon dependent).
- Evidence supports a gradual impact of the FX floor on inflation.
- Transmission channels: evidence suggests effect operated through higher import prices and inflation expectations (following Svensson (2001) proposal).

### Suggested avenues for further research
- Explicitly investigate transmission channels, e.g., test the impact of the FX floor on import and export prices.

*Source: IMF Working Paper wp17206 (section on introduction of the floor and treated country, synthetic control and robustness analyses).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2017/wp17206.pdf_