## _wp15228

## Source details

**Canonical URL:** [_wp15228](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15228.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15228.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15228.pdf.json)

---

### I. INTRODUCTION
- Objective: examine to what extent central banks in emerging markets (EMs) that adopted inflation targeting (IT) tend to face conflicts of objectives leading them to manage exchange rates more closely under certain macroeconomic and financial conditions.
- Key claim: although IT countries should on average have relatively flexible exchange rate regimes (ERR), under specific macroeconomic conditions the positive association between IT and exchange rate flexibility disappears.
- Channels discussed:
  - Trade openness and exchange rate pass-through can motivate FX intervention to control inflation.
  - Financial stability concerns (foreign-currency liabilities, foreign-currency public debt, foreign investor share) can motivate reduced exchange-rate flexibility to protect balance sheets and debt sustainability.
  - Policy instrument limitations: the policy interest rate alone may be insufficient to achieve both inflation and financial stability objectives.

### II. DATA AND PRELIMINARY DISCUSSION
- ERR classification:
  - De facto classification with six categories coded 1 to 6 (most fixed to most flexible); categories 5 and 6 are dropped, keeping “freely floating” as the most flexible ERR.
- Sample:
  - 36 EMs in sample, including 16 IT countries.
- Data frequency and coverage:
  - Annual data for the period 1985– (period end as in source).

### III. METHODOLOGY
- Empirical approaches:
  - Random effects ordered probit and ordered logit estimators on panel data with latent variable y*it = X'it β + α ITit + δ ITit zit + φ zit + εit; observed yit ∈ {1,2,3,4} determined by thresholds c1, c2, c3.
  - Propensity score matching (PSM) estimators to tackle self-selection bias; Rosenbaum bounds to assess sensitivity to hidden biases.
- Standard determinants (vector X), lagged one year: Trade openness; Financial openness; Economic development (log real GDP per capita); Growth (annual GDP growth); Financial development (bank credit to private sector as % of GDP); Inflation (annual CPI growth); FX reserves (total reserves in months of imports); Fiscal position (change in total government debt as % of GDP); Politics (political stability).
- Estimation details:
  - Random-effects ordered probit/logit; country-specific effects via random effects.
  - PSM matching algorithms: nearest neighbor (1, 3, 5), radius matching (r=0.1, r=0.05, r=0.02), kernel matching with 0.06 fixed bandwidth and Epanechnikov kernel; bootstrapped standard errors (500 replications).

### IV. KEY FINDINGS — AVERAGE EFFECTS
- On average, IT adoption is associated with more flexible de facto ERR.
- Representative IT coefficients (random-effects ordered probit, Table 1 examples):
  - IT 1.374*** (t-stat 7.503) in column (1).
  - IT 1.535*** (t-stat 7.222) in column (3).
  - IT 0.680** (t-stat 2.378) in column (5).
- Representative IT coefficients (random-effects ordered logit, Table 2 example):
  - IT 2.532*** (t-stat 7.435) in column (1).
- Linear probability model (Appendix Table 6) IT coefficients across specifications:
  - (1) 0.504***  (6.105)
  - (2) 0.529***  (6.419)
  - (3) 0.468***  (5.754)
  - (4) 0.457***  (5.464)
  - (5) 0.262*    (1.931)
  - (6) 0.160     (1.043)
  - (7) 0.0940    (0.726)
  - (8) 0.494***  (6.069)
  - (9) 0.378***  (4.591)
  - (10) 0.489*** (5.208)
  - (11) 0.148    (1.402)

### V. CONDITIONAL EFFECTS — INTERACTION RESULTS (random-effects ordered probit/logit; selected coefficients)
- Interactions that amplify IT → flexibility (positive δ, significant at 1% unless noted):
  - IT*Trade openness 0.0134*** (t-stat 2.800)
  - IT*Financial openness 0.0194*** (t-stat 3.827)
  - IT*Financial development 0.0175*** (t-stat 3.701)
- Interactions that reduce IT → flexibility (negative δ, significant):
  - IT*Bank foreign assets/total assets -0.0331** (t-stat -2.163)
  - IT*Bank foreign liabilities/total assets -0.00846*** (t-stat -2.635)
  - IT*Inflation -0.141*** (t-stat -3.898)
  - IT*Net imports -0.0971*** (t-stat -4.759)
  - IT*External debt -0.0193*** (t-stat -2.677)
- Time and selection effects:
  - IT*Time 0.218*** (t-stat 5.796)
  - IT*Pscore 2.179** (t-stat 2.368)

### VI. PROPENSITY SCORE MATCHING (PSM) — ATT ESTIMATES (selected quantitative outcomes)
- Baseline ATT (full sample):
  - Baseline: 1.029*** (t-statistic (6.244)). Obs. 617.
- Conditional ATTs (Low = IT countries’ observations ≤ median; High = > median):
  - Total external debt (percent GDP)
    - Low: 1.093*** (4.776). Obs. 535.
    - High: 0.984*** (5.409). Obs. 542.
  - Bank foreign assets/total assets
    - Low: 0.984*** (5.088). Obs. 540.
    - High: 0.922*** (4.349). Obs. 555.
  - Bank foreign liabilities/total assets
    - Low: 1.339*** (6.813). Obs. 540.
    - High: 0.792*** (4.519). Obs. 555.
  - Inflation rate
    - Low: 1.444*** (6.502). Obs. 514.
    - High: 0.910*** (4.718). Obs. 545.
  - Net imports (percent GDP)
    - Low: 1.018*** (4.909). Obs. 535.
    - High: 0.734*** (3.968). Obs. 542.
  - Financial openness
    - Low: 0.915*** (4.905). Obs. 549.
    - High: 1.169*** (5.698). Obs. 546.
  - Trade openness
    - Low: 1.368*** (6.852). Obs. 547.
    - High: 0.957*** (4.800). Obs. 548.
  - Financial development
    - Low: 0.944*** (5.369). Obs. 550.
    - High: 0.741*** (3.060). Obs. 548.

### VII. ROBUSTNESS AND SENSITIVITY
- Robustness checks:
  - Alternative estimators: ordered logit produces broadly similar conclusions.
  - Linear probability model: Appendix Table 6 shows consistent signs and significance for IT and key interaction terms.
  - Additional controls: crisis dummies (currency, banking, sovereign) inclusion does not alter main results (Appendix Table 4).
  - Central bank independence control: IT main effects remain unchanged; CBI coefficient positive but not always significant (Appendix Table 5).
- Rosenbaum bounds (sensitivity to unobserved heterogeneity):
  - With values of Γ between 1 and 5, the confidence intervals do not include 0.
  - Interpretation provided in source: even if unobserved characteristics increase the odds ratio by a factor of 5, there will be no significant effect of hidden bias on the ATT estimates.

### VIII. INTERPRETATION AND SYNTHESIS
- Core empirical conclusions:
  - (i) EM IT countries on average have a relatively more flexible exchange rate regime than other EMs.
  - (ii) The prevailing macroeconomic environment affects the choice of the exchange rate regime: high import dependency, large shares of public and private assets/liabilities denominated in foreign currencies, high inflation, and high external debt reduce ERR flexibility in IT countries; greater financial development and capital openness increase ERR flexibility in IT countries.
- Dynamics:
  - IT*Time positive and significant implies longer duration since IT adoption raises probability of floating (learning-by-doing).
  - IT*Pscore positive and significant implies countries that better meet preconditions (higher predicted probability of adopting IT) are more likely to float.

### IX. POLICY-RELEVANT IMPLICATIONS
- Recognition of conditional trade-offs:
  - Policymakers in IT regimes should recognize conditional trade-offs between inflation targeting and exchange-rate flexibility, especially in environments with trade dependence or financial vulnerabilities.
- Strengthen complementary frameworks:
  - Strengthen macroprudential frameworks and broaden the policy toolkit if financial-stability objectives are to be pursued alongside inflation targeting.
- Institutional design and sequencing:
  - Consider institutional arrangements that preserve credibility of the inflation-targeting commitment while allowing appropriate responses to financial-stability risks.
  - Sequencing and strengthening macrofinancial preconditions (financial development, management of external debt, reducing currency mismatches in banking sector, improving inflation track record) can increase the effectiveness of IT and facilitate a move toward exchange rate flexibility.
- Monitoring:
  - Careful monitoring of shifts in de facto exchange rate policies is warranted to avoid inconsistencies between de jure regime declarations and de facto policy.

*Source: IMF staff calculations, _wp15228 (2010) content unit provided.*

### 1. Random Effects Ordered Probit Estimates ..........................................................................16

### 1. Random Effects Ordered Probit Estimates ..........................................................................16

### I. INTRODUCTION
- Inflation targeting (IT) regime: central bank mandate to target a defined medium-term inflation rate compatible with macroeconomic stability; main policy instrument is the official policy interest rate.
- Conditions for effective and credible IT:
  - Main mandate of central bank is to maintain inflation rate close to the official target.
  - Limit pressures that prevent focus on this main objective (e.g., government budget financing or exchange rate policies).
- Main implication: price stability has priority over other goals such as exchange rate stabilization, though many IT central banks remain concerned about exchange rate fluctuations.
- Paper objective: examine to what extent central banks in emerging markets (EMs) that adopted IT tend to face conflicts of objectives leading them to manage exchange rates more closely under certain macroeconomic and financial conditions.
- Key claim: although IT countries should on average have relatively flexible exchange rate regimes, under specific macroeconomic conditions the positive association between IT and exchange rate flexibility disappears.
- Motivations and contributions:
  - Investigate characteristics of the macroeconomic environment that make IT countries prone to deviate from their flexibility commitment, addressing self-selection bias into IT.
  - Use de facto classification of exchange rate regimes (ERR) rather than computed standard deviation of nominal exchange rate.
  - Employ ordered models of limited dependent variables and propensity score matching estimators to identify effects of IT adoption on ERR.
  - Demonstrate that IT countries exhibit more flexible de facto ERR on average, but disagreement between de jure and de facto regimes increases following shifts in macroeconomic conditions.
- Channels for “fear of floating” among IT countries:
  - Trade openness: high dependence on imports and high exchange rate pass-through can make FX intervention useful to control inflation.
  - Financial stability concerns: large foreign currency liabilities in banking system, sizable foreign-currency-denominated public debt, or high share of foreign investors in domestic debt markets can motivate reduced exchange-rate flexibility to protect balance sheets and debt sustainability.
  - Policy instrument limitations: controlling the policy interest rate alone may be insufficient to achieve both inflation and financial stability objectives, pointing to the need for expanded instruments (e.g., macroprudential tools) and raising potential credibility trade-offs.
- Paper organization:
  - Section 2: data and stylized facts.
  - Section 3: baseline empirical analysis using ordered probit and logit estimators on panel data to investigate non-linearities in IT effect on exchange rate flexibility.
  - Section 4: propensity score matching estimators to address self-selection into IT; Rosenbaum bounds used to assess sensitivity to hidden biases.
  - Section 5: conclusion.

### II. DATA AND PRELIMINARY DISCUSSION
- ERR classification:
  - Analysis relies on de facto classification shown in Appendix Table 1.
  - Basic classification has six categories coded from 1 to 6, describing the most fixed (hard peg) to the most flexible regimes respectively.
  - The two last categories (5 and 6) are dropped, keeping “freely floating” as the most flexible ERR.
- Sample:
  - 36 EMs in sample, including 16 IT countries, selected on the basis of data availability (see Appendix Table 2).
- Data frequency and coverage:
  - We use annual data for the period 1985–

### METHODOLOGY (as described in source)
- Empirical approaches:
  - Random effects ordered probit and ordered logit estimators applied to panel data to explore conditional and non-linear effects of IT adoption on degree of exchange rate flexibility.
  - Propensity score matching estimators employed to tackle self-selection bias in IT adoption; quality of propensity score estimates assessed via Rosenbaum bounds to measure sensitivity to hidden biases.
- Robustness checks (listed in Appendix Tables):
  - De Facto Exchange Rate Regime Classification (Appendix Table 1).
  - Sample details (Appendix Table 2).
  - Data and Sources (Appendix Table 3).
  - Robustness – Random Effects Ordered Probit Estimates (Controlling for Crisis Dummies) (Appendix Table 4).
  - Robustness – Random Effects Ordered Probit Estimates (Controlling for Central Bank Independence) (Appendix Table 5).
  - Robustness – Linear Probability Model (Appendix Table 6).
  - Probit Model of the Matching Estimates (Appendix Table 7).

### KEY FINDINGS AND INSIGHTS (from the provided text)
- On average, IT adoption is associated with more flexible de facto exchange rate regimes, consistent with prior findings (e.g., Lin, 2010).
- However, the positive association between IT and exchange rate flexibility is conditional:
  - Under limited trade and financial openness, limited financial development, or when financial stability is a concern, IT central banks are more likely to manage the exchange rate and exhibit less flexibility.
  - Shifts in macroeconomic conditions increase the “disagreement” between de jure and de facto regimes among IT countries.
- Rationale for conditional behavior:
  - Trade openness and exchange rate pass-through can prompt FX intervention to control inflation when bilateral exchange rate swings affect imported prices.
  - Financial fragilities (foreign-currency borrowing, foreign-owned banks, unhedged mortgages, foreign-currency public debt, large foreign investor presence) create incentives to limit nominal exchange rate flexibility to prevent balance-sheet and debt-stability risks.
  - Achieving both inflation targeting and financial stability may require additional instruments beyond the policy interest rate (e.g., macroprudential tools), complicating credibility and the institutional design of IT regimes.

### POLICY RECOMMENDATIONS AND IMPLICATIONS (implied by source discussion)
- Policymakers in IT regimes should:
  - Recognize conditional trade-offs between inflation targeting and exchange-rate flexibility, especially in environments with trade dependence or financial vulnerabilities.
  - Strengthen macroprudential frameworks and broaden the policy toolkit if financial-stability objectives are to be pursued alongside inflation targeting.
  - Consider institutional arrangements that preserve credibility of the inflation-targeting commitment while allowing appropriate responses to financial-stability risks.
- Research and policy evaluation:
  - Further assessment of how specific macroeconomic and financial conditions affect the IT–ERR relationship is warranted; empirical identification should address self-selection into IT and test robustness to unobserved confounders.

*Source: Excerpt from the IMF working paper chapter titled "1. Random Effects Ordered Probit Estimates ..........................................................................16"*

### 2010. Table 3 in the Appendix provides detailed information

### _wp15228 - 2010. Table 3 in the Appendix provides detailed information

### Purpose and main research question
- Purpose: assess the extent to which, above and beyond its common determinants, the monetary policy framework can affect the choice of an exchange rate regime (ERR).
- Primary question: whether adoption of the inflation targeting (IT) strategy increases the probability of relying more on a floating exchange rate regime.
- Empirical expectation: IT countries are expected to have a flexible ERR (implying no, or very limited, interventions on the FX market), but emerging markets (EMs) may move toward more floating regimes after IT adoption rather than entering fully flexible regimes beforehand.

### Empirical patterns and descriptive evidence
- Average correlation: positive between IT and flexibility of the de facto exchange rate regime (Figure 1, first panel, left chart).
- Within IT sample: ERR moves significantly toward more flexibility after IT adoption (Figure 1, “Baseline” plot).
- Caveat: the 5th category mostly captures hyperinflationary periods, and the 6th category includes countries or periods that cannot be classified due to lack of data availability.

### Macroeconomic conditions that modify the IT → ERR relationship
- A. Financial instability
  - Higher banking sector vulnerability to external shocks raises the likelihood of central bank FX interventions and reduces ERR flexibility.
  - Conditional variables considered: Foreign assets to total assets; Foreign liabilities to total assets.
  - Empirical correlation: IT countries with higher ratio of bank foreign liabilities/total assets have on average less flexible ERR (Figure 1, first panel, right chart).

- B. External debt
  - Higher total external debt (as a share of GDP) makes exchange rate flexibility undesirable because it increases uncertainty about debt servicing and can undermine debt sustainability.
  - Empirical pattern: IT countries with higher external debt skew toward more rigid ERR; those with lower external debt skew toward flexibility (Figure 1, middle panel, left chart).

- C. Financial development
  - Greater financial development mitigates exchange rate fluctuation risks via hedging instruments and improves monetary transmission, making ERR flexibility more likely.
  - IT countries with better developed financial sectors are expected to meet inflation objectives better and be less prone to FX interventions.

- D. Inflation and exchange-rate pass-through
  - EM IT countries often miss announced inflation targets; high inflation and high exchange-rate pass-through increase incentives to control the exchange rate.
  - Net imports (as a share of GDP) strengthen pass-through; higher net imports imply greater imported inflation pressure and may reduce ERR flexibility.
  - Preliminary conditional correlations in Figure 1 (lower panel) are inconclusive for inflation and net imports.

- E. Economic openness
  - According to the “impossible trinity,” independent monetary policy, capital mobility, and exchange rate stabilization cannot all be achieved simultaneously.
  - More financially open IT countries (higher capital mobility) have less room to control exchange rates and are more likely to float (Figure 1, middle panel, right chart).
  - Trade openness may operate in the same direction.

### Empirical framework (Ordered probit/logit)
- Latent variable model: y*it = X'it β + α ITit + δ ITit zit + φ zit + εit, with observed ordered choice yit ∈ {1,2,3,4} determined by thresholds c1, c2, c3.
- ITit: dummy = 1 if inflation targeting in country i at time t.
- Conditional variables zit are mean-deviated to reduce collinearity with interaction terms.
- Standard determinants (vector X) included as one-year lags: Trade openness; Financial openness; Economic development (log real GDP per capita); Growth (annual GDP growth); Financial development (bank credit to private sector as % of GDP); Inflation (annual CPI growth); FX reserves (total reserves in months of imports); Fiscal position (change in total government debt as % of GDP); Politics (political stability).
- Estimation: random-effects ordered probit and ordered logit; country-specific effects controlled via random effects.

### Key regression results (random-effects ordered probit — Table 1 highlights)
- IT main effect: positive and often strongly significant, indicating IT adoption increases probability of a more flexible ERR.
  - Example coefficients: IT 1.374*** (t-stat 7.503) in column (1); IT 1.535*** (t-stat 7.222) in column (3); IT 0.680** (t-stat 2.378) in column (5).
- Interaction terms that increase IT’s positive effect on flexibility (positive δ, significant at 1%):
  - IT*Trade openness 0.0134*** (t-stat 2.800)
  - IT*Financial openness 0.0194*** (t-stat 3.827)
  - IT*Financial development 0.0175*** (t-stat 3.701)
- Interaction terms that reduce IT’s positive effect on flexibility (negative δ, significant):
  - IT*Bank foreign assets/total assets -0.0331** (t-stat -2.163)
  - IT*Bank foreign liabilities/total assets -0.00846*** (t-stat -2.635)
  - IT*Inflation -0.141*** (t-stat -3.898)
  - IT*Net imports -0.0971*** (t-stat -4.759)
  - IT*External debt -0.0193*** (t-stat -2.677)
- Additive control variable signs (representative):
  - Trade openness: negative (e.g., -0.00529*; t-stat -1.719)
  - Growth: negative and often significant (e.g., -0.0343**; t-stat -2.309)
  - Inflation: positive and significant (e.g., 0.0177***; t-stat 3.337)
  - Reserves: negative and significant (e.g., -0.0590***; t-stat -3.107)
  - Capital openness: negative and significant in several specs (e.g., -0.139**; t-stat -2.462)
  - Politics: negative effects not always robust (e.g., -0.0176**; t-stat -1.988 in some columns)
- Sample sizes and statistics:
  - Observations: range from 588 to 642 across specifications.
  - Number of id (countries): mostly 36 (one specification reports 35).
  - Wald chi2 statistics: range from 78.09 to 117.5 in Table 1 variants.
- Interpretation: IT increases ERR flexibility on average, but the effect is moderated by country-specific macroeconomic and financial conditions.

### Non-linearities and conditional effects (interpretation)
- IT increases probability of floating on average, but:
  - IT countries with higher external debt, higher bank foreign assets/ liabilities ratios, higher inflation, and higher net imports show less ERR flexibility (strong negative interaction coefficients).
  - IT countries with greater financial development, trade openness, and financial openness show amplified IT → flexibility (positive interaction coefficients).
- Time and Pscore effects:
  - IT*Time 0.218*** (t-stat 5.796) — longer time since IT adoption raises probability of floating (learning-by-doing).
  - IT*Pscore 2.179** (t-stat 2.368) — IT countries that better meet preconditions (higher predicted probability of adopting IT) are more likely to float.

### Robustness checks
- Alternative estimators: random-effects ordered logit (Table 2) produces broadly similar conclusions; example IT coefficients: 2.532*** (t-stat 7.435) in column (1).
- Linear probability model: OLS panel fixed effects deliver consistent signs and significance for IT and interaction terms (Appendix Table 6).
- Additional controls: currency crises, banking crises, sovereign debt crises dummies have no effect on ERR, and inclusion does not alter main results (Appendix Table 4).
- Central bank independence: coefficient positive but not statistically significant; main IT effects remain unchanged (Appendix Table 5).
- Interaction interpretation caveat: Ai and Norton (2003) concerns addressed by complementary linear analyses.

### Propensity score matching (PSM) estimates (average treatment effects on the treated, ATT)
- Motivation: address potential self-selection bias in IT adoption; propensity scores estimated via probit including lagged inflation, trade openness, GDP growth, FX reserves, fiscal deficit, economic development, financial development, central bank independence.
- Matching algorithms used: nearest neighbor (1, 3, 5 neighbors), radius matching (r=0.1, r=0.05, r=0.02), kernel matching.
- Main PSM findings (Table 3 summary):
  - Baseline ATT: IT has a positive and significant effect on exchange rate flexibility (IT countries more flexible vs. non-IT).
  - Conditional ATT results:
    - IT countries with lower external debt, lower bank foreign assets, and lower bank foreign liabilities (relative to total bank assets) float relatively more than counterparts above the median of these variables.
    - IT countries with better inflation performance (lower inflation) float relatively more than IT countries with higher inflation.
    - IT countries less import-dependent (lower net imports) float relatively more than those with higher net imports.
    - More financially open IT countries float relatively more; ATT is lower for less financially open IT countries.
    - Mixed or no significant differences for ATT conditional on financial development; trade openness results suggest less flexibility for IT countries that trade more.

### Policy-relevant implications derived from the analysis
- Adoption of IT generally raises the likelihood of moving to a more flexible ERR in emerging markets, but:
  - Countries with high external vulnerability (external debt, bank foreign asset/liability exposure) may face a “fear of floating” even after IT adoption and choose interventions to stabilize the exchange rate.
  - High inflation and high exchange-rate pass-through (e.g., high net imports) reduce IT countries’ propensity to float.
  - Greater financial development, trade openness, and capital mobility reinforce IT’s association with increased ERR flexibility.
  - The process is dynamic: longer duration since IT adoption and stronger preconditions for IT increase probability of floating (learning-by-doing and selection effects).
- Policy consideration: sequencing and strengthening macrofinancial preconditions (financial development, management of external debt, reducing currency mismatches in banking sector, improving inflation track record) can increase the effectiveness of IT and facilitate a move toward exchange rate flexibility.

*Source: IMF staff calculations, _wp15228 (2010) content unit provided.*

### Conclusions are broadly in line with this finding when estimating the ATT conditional to the growth rates of

### Conclusions are broadly in line with this finding when estimating the ATT conditional to the growth rates of

### Matching estimates (PSM) — main quantitative outcomes
- Baseline average treatment effect on treated (ATT) for the full sample:
  - Baseline: 1.029*** (t-statistic reported as (6.244)). Obs. 617.
- Total external debt (percent GDP)
  - Low: 1.093*** (4.776). Obs. 535.
  - High: 0.984*** (5.409). Obs. 542.
- Bank foreign assets/total assets
  - Low: 0.984*** (5.088). Obs. 540.
  - High: 0.922*** (4.349). Obs. 555.
- Bank foreign liabilities/total assets
  - Low: 1.339*** (6.813). Obs. 540.
  - High: 0.792*** (4.519). Obs. 555.
- Inflation rate
  - Low: 1.444*** (6.502). Obs. 514.
  - High: 0.910*** (4.718). Obs. 545.
- Net imports (percent GDP)
  - Low: 1.018*** (4.909). Obs. 535.
  - High: 0.734*** (3.968). Obs. 542.
- Financial openness
  - Low: 0.915*** (4.905). Obs. 549.
  - High: 1.169*** (5.698). Obs. 546.
- Trade openness
  - Low: 1.368*** (6.852). Obs. 547.
  - High: 0.957*** (4.800). Obs. 548.
- Financial development
  - Low: 0.944*** (5.369). Obs. 550.
  - High: 0.741*** (3.060). Obs. 548.

Notes on matching implementation reported in the source:
- Neighbour matching, radius matching, and kernel matching are reported (columns labelled Nearest neighbour, 3 nearest neighbours, 5 nearest neighbours, r=0.1, r=0.05, r=0.02, Kernel).
- A 0.06 fixed bandwidth and an Epanechnikov kernel are used for kernel regression matching.
- T-statistics based on bootstrapped standard errors are reported in parentheses (500 replications).
- ***, **, and * indicate statistical significance at the 1, 5, and 10 percent levels, respectively.
- For the conditional variable considered, “Low” and “High” indicate that IT countries’ observations have been restricted to values lower and higher than the median respectively, the control group remaining unchanged.

### Robustness: Rosenbaum bounds (sensitivity to unobserved heterogeneity)
- Rosenbaum (2002) sensitivity analysis summary:
  - Parameter Γ (Gamma) assesses extent to which treatment effect may be affected by unobserved factors.
  - An odds ratio equal to 1 (Γ=1) suggests no hidden bias.
- Main robustness finding:
  - With values of Γ between 1 and 5, the confidence intervals do not include 0.
  - Interpretation: even if unobserved characteristics increase the odds ratio by a factor of 5, there will be no significant effect of hidden bias on the ATT estimates.
  - Conclusion: average treatment effect of inflation targeting on the exchange rate regime shows very little sensitivity to countries’ unobserved characteristics.

### Robustness: panel ordered probit/logit interaction results (selected coefficients)
- Appendix Table 4 — Random effects ordered probit estimates (controlling for crisis dummies), selected coefficients and statistics:
  - IT coefficient range across specifications: 1.380***, 1.384***, 1.556***, 1.298***, 0.695**, 0.449, 0.438, 1.486***, 1.230***, 1.414***, 0.375.
  - IT*Trade openness: 0.0142*** (2.941).
  - IT*Financial openness: 0.0198*** (3.889).
  - IT*Financial development: 0.0183*** (3.820).
  - IT*Banks foreign assets/total assets: -0.0334** (-2.157).
  - IT*Banks foreign liabilities/total assets: -0.00877*** (-2.690).
  - IT*Inflation: -0.142*** (-3.894).
  - IT*Net imports: -0.0984*** (-4.816).
  - IT*External debt: -0.0193*** (-2.641).
  - IT*Pscore: 2.226** (2.415).
  - IT*Time: 0.223*** (5.828).
  - Observations (varied by column): 640, 640, 642, 640, 594, 594, 640, 588, 624, 602, 640.
  - Number of id: 36, 36, 36, 36, 35, 35, 36, 36, 36, 35, 36.
  - Wald chi2 stat examples: 90.83, 98.79, 92.60, 100.3, 81.27, 82.20, 102.8, 121.1, 96.47, 92.00, 115.6.
- Appendix Table 5 — Random effects logit estimates (controlling for central bank independence), selected coefficients mirror Table 4:
  - IT: 1.390***, 1.388***, 1.536***, 1.309***, 0.674**, 0.465, 0.453, 1.502***, 1.239***, 1.391***, 0.396.
  - IT*Trade openness: 0.0128*** (2.694).
  - IT*Financial openness: 0.0189*** (3.738).
  - IT*Financial development: 0.0177*** (3.687).
  - IT*Banks foreign assets/total assets: -0.0349** (-2.265).
  - IT*Banks foreign liabilities/total assets: -0.00863*** (-2.683).
  - IT*Inflation: -0.140*** (-3.872).
  - IT*Net imports: -0.0977*** (-4.792).
  - IT*External debt: -0.0193*** (-2.677).
  - IT*Pscore: 2.175** (2.356).
  - IT*Time: 0.218*** (5.822).
  - Observations (examples): 628, 628, 630, 628, 583, 583, 628, 576, 614, 602, 628.
  - Number of id examples: 35, 35, 35, 35, 34, 34, 35, 35, 35, 35, 35.
  - Wald chi2 stat examples: 89.16, 95.97, 89.73, 97.86, 78.72, 79.00, 101.2, 118.1, 96.37, 90.24, 114.7.

### Overall conclusions (synthesis of empirical findings)
- Core empirical findings:
  - (i) EM IT countries on average have a relatively more flexible exchange rate regime than other EMs.
  - (ii) The prevailing macroeconomic environment affects the choice of the exchange rate regime.
- Specific conditional effects:
  - Macroeconomic characteristics such as high import dependency and large share of public and private assets/liabilities denominated in foreign currencies reduce the degree of exchange rate flexibility in EM IT countries.
  - A more-developed domestic financial system and credible financial openness policy contribute to greater exchange rate flexibility in IT countries.
- Robustness statement:
  - Results are robust across panel limited dependent variable models (ordered logit and probit) and propensity-score matching techniques; qualitatively and quantitatively similar to baseline estimates.

### Policy implications and recommendations
- Heterogeneity and monitoring:
  - Careful monitoring of shifts in de facto exchange rate policies is warranted, particularly when macroeconomic conditions change significantly, to avoid inconsistencies between de jure exchange rate regime and de facto policy.
- Maturity of IT arrangements:
  - Inflation targeting arrangements in EMs are not yet mature: the positive association between IT and exchange rate flexibility is stronger in countries that have been more successful in controlling inflation, and weaker where inflation control is more difficult.
- Complementary policies to improve IT outcomes:
  - Policies that promote financial development and financial openness can improve the marginal benefits of adopting an IT regime and help reduce trade-offs between competing objectives.
- Managing the trade-off between inflation targeting and financial stability:
  - The multiplicity of central bank objectives requires additional policy tools.
  - Recommendation: retain inflation objective within monetary policy while using effective macroprudential measures to manage financial stability risks stemming from exchange rate fluctuations.

*Source: IMF staff calculations.*

### Appendix Table 6. Robustness–Linear Probability Model

### _wp15228 - Appendix Table 6. Robustness–Linear Probability Model

### Appendix Table 6 — Robustness: Linear Probability Model (dependent variable: de facto exchange rate regime)
- IT coefficients across specifications (columns (1) to (11)):
  - (1) 0.504***  (6.105)
  - (2) 0.529***  (6.419)
  - (3) 0.468***  (5.754)
  - (4) 0.457***  (5.464)
  - (5) 0.262*    (1.931)
  - (6) 0.160     (1.043)
  - (7) 0.0940    (0.726)
  - (8) 0.494***  (6.069)
  - (9) 0.378***  (4.591)
  - (10) 0.489*** (5.208)
  - (11) 0.148    (1.402)

- Interaction terms (reported in the column where included):
  - IT*Trade openness: 0.00689***  (3.115)
  - IT* Financial openness: 0.00251**  (2.032)
  - IT* Financial development: 0.00546***  (2.981)
  - IT*Banks foreign assets/total assets: -0.0124*  (-1.782)
  - IT*Banks foreign liabilities/total assets: -0.00336**  (-2.259)
  - IT*Inflation: -0.0619***  (-4.079)
  - IT* Net imports: -0.0361***  (-4.515)
  - IT*External debt: -0.00861***  (-2.956)
  - IT*Pscore: 0.984**  (2.335)
  - IT*Time: 0.0764***  (5.221)

- Constants by specification:
  - (1) 0.860  (0.727)
  - (2) 0.761  (0.648)
  - (3) 1.007  (0.849)
  - (4) 0.601  (0.510)
  - (5) 2.010  (1.532)
  - (6) 2.193* (1.672)
  - (7) 0.954  (0.817)
  - (8) 1.973* (1.728)
  - (9) -0.732  (-0.612)
  - (10) -0.921 (-0.725)
  - (11) 1.711  (1.464)

- Model diagnostics and sample:
  - Controls and additive terms included? yes / Yes / yes / yes / yes / yes / Yes / yes / yes / yes / yes
  - Observations: 640, 640, 642, 640, 594, 594, 640, 588, 624, 602, 640
  - R-squared: 0.118, 0.133, 0.115, 0.131, 0.110, 0.112, 0.142, 0.208, 0.173, 0.129, 0.157
  - Number of id: 36, 36, 36, 36, 35, 35, 36, 36, 36, 35, 36
  - F stat: 7.976, 8.239, 7.019, 8.155, 6.179, 6.274, 8.954, 12.91, 10.01, 6.854, 10.05

- Estimation details (notes):
  - OLS panel fixed effects estimates; all the control variables as well as additive terms forming the interaction variables (not reported) are the same as in Table 1; control variables (except IT) are included with 1 year lag; robust T-statistics in parentheses; ***, **, * indicate the statistical significance at 1, 5, and 10 percent respectively.
  - Source: IMF staff estimates.

---

### Appendix Table 7. Probit Model of the Matching Estimates

### Table structure and conditional splits
- Columns correspond to baseline and conditional samples where IT countries’ observations are restricted to values lower ("Low") and higher ("High") than the median for the conditional variable considered. Conditional variables (column groups): External debt, Banks foreign assets/total assets, Banks foreign liabilities/total assets, Inflation, Net imports, Financial openness, Trade openness, Financial development.
- For each reported coefficient the T-statistic is in parentheses. ***, **, * indicate statistical significance at the 1, 5, and 10 percent levels, respectively.

### Coefficients and t-statistics (selected rows shown as in the source)
- Trade openness (Baseline, then Low/High across conditional variables):
  - Baseline: -0.0065***  (-4.172)
  - External debt Low: -0.0095***  (-4.455)
  - External debt High: -0.0021  (-1.118)
  - Banks foreign assets Low: -0.0153***  (-4.501)
  - Banks foreign assets High: -0.00307*  (-1.942)
  - Banks foreign liabilities Low: -0.0122***  (-4.300)
  - Banks foreign liabilities High: -0.0033**  (-2.022)
  - Inflation Low: -0.0056***  (-2.902)
  - Inflation High: -0.0066***  (-3.523)
  - Net imports Low: -0.0069***  (-3.640)
  - Net imports High: -0.0057***  (-2.898)
  - Financial openness Low: -0.0091***  (-2.958)
  - Financial openness High: -0.006***  (-3.640)
  - Trade openness Low (conditional on financial development): -0.0384***  (-6.600)
  - Trade openness High (conditional on financial development): -0.0012  (-0.756)
  - Financial development Low: -0.0014  (-0.470)
  - Financial development High: -0.0087***  (-4.881)

- Growth:
  - Baseline: -0.0385**  (-2.034)
  - External debt Low: -0.0292  (-1.251)
  - External debt High: -0.0401*  (-1.819)
  - Banks foreign assets Low: -0.0108  (-0.448)
  - Banks foreign assets High: -0.0569***  (-2.595)
  - Banks foreign liabilities Low: -0.0242  (-0.977)
  - Banks foreign liabilities High: -0.0461**  (-2.188)
  - Inflation Low: -0.0188  (-0.763)
  - Inflation High: -0.0456**  (-2.174)
  - Net imports Low: -0.0494**  (-2.193)
  - Net imports High: -0.0222  (-0.980)
  - Financial openness Low: -0.0199  (-0.888)
  - Financial openness High: -0.0454*  (-1.902)
  - Trade openness Low: -0.0477*  (-1.920)
  - Trade openness High: -0.0214  (-0.903)
  - Financial development Low: 0.00566  (0.244)
  - Financial development High: -0.0666***  (-2.618)

- Economic development:
  - Baseline: 0.479***  (7.463)
  - External debt Low: 0.401***  (4.936)
  - External debt High: 0.476***  (6.335)
  - Banks foreign assets Low: 0.412***  (5.100)
  - Banks foreign assets High: 0.453***  (6.006)
  - Banks foreign liabilities Low: 0.365***  (4.530)
  - Banks foreign liabilities High: 0.485***  (6.470)
  - Inflation Low: 0.457***  (5.356)
  - Inflation High: 0.414***  (5.702)
  - Net imports Low: 0.360***  (4.666)
  - Net imports High: 0.497***  (6.327)
  - Financial openness Low: 0.375***  (5.161)
  - Financial openness High: 0.608***  (6.293)
  - Trade openness Low: 0.576***  (5.944)
  - Trade openness High: 0.438***  (5.600)
  - Financial development Low: 0.397***  (5.111)
  - Financial development High: 0.663***  (6.127)

- Financial development:
  - Baseline: 0.00203  (1.033)
  - External debt Low: 0.00840***  (3.714)
  - External debt High: -0.00662**  (-2.120)
  - Banks foreign assets Low: 0.00423  (1.564)
  - Banks foreign assets High: 0.00261  (1.143)
  - Banks foreign liabilities Low: 0.00678***  (2.614)
  - Banks foreign liabilities High: -0.000260  (-0.109)
  - Inflation Low: 0.00555**  (2.134)
  - Inflation High: 0.000769  (0.329)
  - Net imports Low: 0.00809***  (3.284)
  - Net imports High: -0.00218  (-0.873)
  - Financial openness Low: -0.00661**  (-2.156)
  - Financial openness High: 0.0107***  (4.426)
  - Trade openness Low: 0.00398  (1.532)
  - Trade openness High: 0.00280  (1.112)
  - Financial development Low: -0.0351***  (-5.683)
  - Financial development High: 0.0191***  (6.805)

- Inflation, lagged:
  - Baseline: -0.0597***  (-5.130)
  - External debt Low: -0.0521***  (-3.358)
  - External debt High: -0.0540***  (-4.256)
  - Banks foreign assets Low: -0.0586***  (-3.921)
  - Banks foreign assets High: -0.0496***  (-3.628)
  - Banks foreign liabilities Low: -0.0618***  (-3.811)
  - Banks foreign liabilities High: -0.047***  (-3.791)
  - Inflation Low: -0.0863***  (-4.026)
  - Inflation High: -0.0423***  (-3.828)
  - Net imports Low: -0.0580***  (-3.679)
  - Net imports High: -0.0484***  (-3.870)
  - Financial openness Low: -0.0487***  (-4.244)
  - Financial openness High: -0.0743***  (-3.579)
  - Trade openness Low: -0.0509***  (-3.960)
  - Trade openness High: -0.0721***  (-3.881)
  - Financial development Low: -0.0597***  (-5.160)
  - Financial development High: -0.0709***  (-3.024)

- Reserves:
  - Baseline: -0.0377**  (-2.295)
  - External debt Low: -0.0294  (-1.478)
  - External debt High: -0.0369*  (-1.907)
  - Banks foreign assets Low: -0.0179  (-0.983)
  - Banks foreign assets High: -0.0538**  (-2.296)
  - Banks foreign liabilities Low: -0.0303  (-1.503)
  - Banks foreign liabilities High: -0.0364*  (-1.857)
  - Inflation Low: -0.0164  (-0.844)
  - Inflation High: -0.0525**  (-2.461)
  - Net imports Low: -0.0137  (-0.767)
  - Net imports High: -0.0596**  (-2.454)
  - Financial openness Low: -0.0271  (-1.528)
  - Financial openness High: -0.0430*  (-1.778)
  - Trade openness Low: -0.0178  (-0.902)
  - Trade openness High: -0.0623**  (-2.302)
  - Financial development Low: -0.0500***  (-2.609)
  - Financial development High: -0.0308  (-1.180)

- Fiscal deficit:
  - Baseline: 0.00121  (0.148)
  - External debt Low: -0.00232  (-0.207)
  - External debt High: 0.00318  (0.367)
  - Banks foreign assets Low: 0.000295  (0.0283)
  - Banks foreign assets High: -0.00161  (-0.170)
  - Banks foreign liabilities Low: -0.00653  (-0.572)
  - Banks foreign liabilities High: 0.00397  (0.447)
  - Inflation Low: 0.00438  (0.440)
  - Inflation High: -0.00194  (-0.213)
  - Net imports Low: -0.0117  (-1.059)
  - Net imports High: 0.0101  (1.120)
  - Financial openness Low: 0.00381  (0.431)
  - Financial openness High: -0.000673  (-0.0624)
  - Trade openness Low: -0.0131  (-1.037)
  - Trade openness High: 0.00362  (0.374)
  - Financial development Low: 0.00501  (0.602)
  - Financial development High: -0.00219  (-0.175)

- Inverse of CBI:
  - Baseline: -0.907**  (-2.339)
  - External debt Low: -1.268**  (-2.426)
  - External debt High: -0.472  (-1.085)
  - Banks foreign assets Low: -0.939**  (-2.030)
  - Banks foreign assets High: -0.738  (-1.549)
  - Banks foreign liabilities Low: -0.901*  (-1.867)
  - Banks foreign liabilities High: -0.743  (-1.640)
  - Inflation Low: -0.217  (-0.422)
  - Inflation High: -1.159***  (-2.655)
  - Net imports Low: -0.417  (-0.891)
  - Net imports High: -1.183**  (-2.499)
  - Financial openness Low: -1.115**  (-2.527)
  - Financial openness High: -0.405  (-0.752)
  - Trade openness Low: -1.595***  (-3.262)
  - Trade openness High: -0.344  (-0.682)
  - Financial development Low: -0.980**  (-2.221)
  - Financial development High: -0.619  (-1.014)

- Constant terms by column group (selected):
  - Baseline: -3.387***  (-5.936)
  - External debt Low: -3.392***  (-4.612)
  - External debt High: -3.735***  (-5.662)
  - Banks foreign assets Low: -3.045***  (-4.269)
  - Banks foreign assets High: -3.782***  (-5.527)
  - Banks foreign liabilities Low: -2.851***  (-3.910)
  - Banks foreign liabilities High: -3.992***  (-5.979)
  - Inflation Low: -4.223***  (-5.304)
  - Inflation High: -3.016***  (-4.732)
  - Net imports Low: -3.323***  (-4.747)
  - Net imports High: -3.738***  (-5.378)
  - Financial openness Low: -2.466***  (-3.917)
  - Financial openness High: -5.542***  (-6.036)
  - Trade openness Low: -2.918***  (-3.810)
  - Trade openness High: -3.949***  (-5.410)
  - Financial development Low: -2.080***  (-3.173)
  - Financial development High: -6.464***  (-6.118)

- Sample sizes and fit:
  - Observations by column: 691, 618, 624, 614, 628, 613, 629, 607, 635, 619, 623, 623, 619, 621, 621, 624, 618
  - Pseudo R2 by column: 0.240, 0.242, 0.224, 0.225, 0.243, 0.219, 0.233, 0.267, 0.201, 0.226, 0.231, 0.195, 0.336, 0.325, 0.266, 0.281, 0.432

- Estimation details (notes):
  - T-statistics are reported in parentheses. ***, **, and * indicate statistical significance at the 1, 5, and 10 percent levels, respectively. For the conditional variables considered, “Low” and “High” indicate that IT countries’ observations have been restricted to values lower and higher than the median, respectively.
  - Source: IMF staff estimates.

*Source: IMF staff estimates.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15228.pdf_