## 4. Bilateral Exchange Rate Volatility and Exchange Rate Regimes

## Source details

**Canonical URL:** [4. Bilateral Exchange Rate Volatility and Exchange Rate Regimes](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14146.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14146.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14146.pdf.json)

---

### Introduction and research question
- Central question: Does exchange rate flexibility facilitate external adjustment (faster mean reversion of trade imbalances) once bilateral exchange rate relationships with trading partners are taken into account?
- Motivation: Aggregate/composite regime classifications can mask heterogeneous bilateral relationships (example: the US dollar “floats” in aggregate but is effectively non‑flexible vis‑à‑vis major partners such as China).
- Hypothesis (Friedman’s view): The speed of mean reversion of the trade balance is faster under a float than under a peg; indirect pegs allow relatively greater flexibility than direct pegs and therefore faster adjustment than direct pegs but slower than floats.

### Data, bilateral regime construction, and sample
- Coverage: 181 countries over 1980–2011.
- Bilateral regime classification:
  - Three-way coding: fixed = 0, intermediate = 0.5, floating = 1.
  - Distinguishes direct pegs and indirect pegs (common anchor relationships, including first- and second-generation indirect links).
  - Uses IMF de jure and de facto (AREAER) classifications as primary inputs; robustness checks use R&R and LY&S classifications.
  - Basket pegs: top five trading partners used as anchors.
  - SDR pegs: pre-1999 and post-1999 SDR basket currencies used as anchors.
  - ERM/ECU handled via ECU currencies pre-euro and Eurozone anchors thereafter.
- Final bilateral dataset size: about 380,000 observations (missing data reduce theoretical (181×180×32)/2 = 521,280).
- Subsamples: Advanced-EMDC (one advanced, one EM/developing), Advanced (both advanced), EMDC (both EM/developing).

### Stylized facts on bilateral trade balance persistence and regime shares
- External balance measure: bilateral trade balance scaled by the sum of bilateral exports and imports (ratio in [-1, 1]).
- Partitioning TB for transition analysis: four ranges: < -0.5, -0.5 to 0, 0 to 0.5, > 0.5.
- Bilateral regime distribution (de jure and de facto):
  - Direct pegs (fixed and intermediate): about 4 percent of full sample.
  - Indirect pegs: about 23 percent.
  - Floats: about 73 percent.
  - Restricting to top trading partners increases direct fixed/intermediate share to about 15 percent.
- Real exchange rate volatility (annual standard deviation) by bilateral relationship:
  - Direct fixed regime: 3 percent per year.
  - Direct intermediate regime: about 6 percent per year.
  - Indirect fixed regime: 7½ percent per year.
  - Indirect intermediate regime: 9 percent per year.
  - Impure floats: 9 percent per year.
  - Pure floats: 10 percent per year.
- Transition probability benchmarks (aggregate composite classification):
  - Large deficits persistence: about 76 percent under fixed, about 82 percent under intermediate, about 50 percent under floats.
  - Aggregate composite results: no clear, consistent pattern across regimes.
- Transition probabilities using bilateral regimes:
  - Persistence of TB is statistically significantly greater under pegs relative to floats across all TB ranges.
  - For large bilateral deficits and surpluses, probability of staying within same range from one period to next is about 4 percentage points higher under pegged regimes than under floats; for other ranges probability is about 6–27 percentage points higher for pegs.
- Time trends:
  - Persistence of bilateral TB has generally increased from 1980s → 2000s.
  - For fixed regimes, probability of maintaining same balance increased by about 2–8 percentage points from the 1980s to 2000s across TB ranges; for floats the increase was about 1–2 percentage points.
  - Suggested driver: greater financial integration allowing larger imbalances to be sustained longer.

### Estimation strategy and main empirical results
- Baseline AR specification (aggregate/composite regime benchmark):
  - Equation (1) estimated following C&W and others, using de jure and de facto composite classifications.
  - Estimated AR coefficient for full sample: in the range of about 0.7–0.9 (pooled OLS and fixed effects specifications).
  - Interaction term (lagged TB × composite XRR): negative but statistically insignificant across IMF, R&R, LY&S classifications.
- Bilateral specification:
  - Equation (2) estimated on bilateral TBijt with bilateral XRRijt (0/0.5/1), including country-pair fixed effects (CPFE) and clustering standard errors at country-pair level.
  - Main finding: strong support for Friedman’s hypothesis when using bilateral regimes—more flexible bilateral exchange rate relationships are associated with significantly faster mean reversion (less persistent bilateral imbalances).
- Representative results (full sample):
  - Pooled OLS: half-life of bilateral TB under a direct fixed regime ≈ 5 years versus ≈ 2.5 years under floats.
  - With country fixed effects (CFE): half-life under direct fixed ≈ 5 years versus ≈ 2 years under floats.
  - With country-pair fixed effects (CPFE): persistence parameter declines; half-lives ≈ 1.2 years (fixed), 1 year (intermediate), and 0.9 years (float).
- Subsample patterns:
  - Advanced-EMDC: half-life under pegs between 1.2–5.0 years (depending on specification); floats between 1–3 years.
  - EMDC pairs: pooled OLS half-life ≈ 4 years under pegs vs ≈ 2 years under floats; CPFE estimates ≈ 1 year for pegs vs ≈ 0.8 years for floats.
  - Advanced pairs: interaction term significant in pooled OLS and CFE but insignificant with CPFE (limited regime-switch variation among advanced pairs).

### Indirect pegs, impure floats, and corrective movements
- Indirect pegs:
  - Preliminary finding: indirect pegs exhibit lower persistence than direct pegs (faster adjustment than direct pegs) but higher persistence than floats.
  - Difference in half-life between indirect fixed regimes and floats: about 0.1 years (Table 8, cols. [2]-[3]).
- Impure floats:
  - Under CPFE specifications, dynamics under impure floats are not statistically different from pure floats.
  - Under impure floats, exchange rate flexibility is not much lower than under pure floats.
- Corrective movements:
  - Under floats bilateral real exchange rate movements tend to move in the “correct” direction: deficit countries experience real depreciations; surplus countries experience real appreciations (Table 11).
  - Under direct and indirect pegs, bilateral real exchange rate response to the trade balance is statistically insignificant.

### Natural experiments: causal evidence from exogenous regime changes
- CFA franc zone and euro adoption (post-1999):
  - Pooled OLS implies increase in half-life of CFA countries’ bilateral trade balance with Eurozone countries (excluding France) of about 1.3 years under fixed system relative to previous floating regime (col. [1]).
  - Difference in half-lives drops to about 0.1 years when CPFE and year effects are included, and remains statistically significant at the 10 percent level (col. [2]).
  - Excluding France and Eurozone countries with largest ECU weight (Germany, Netherlands, Belgium) increases interaction coefficient magnitude and makes it significant at the 5 percent level with CPFE (col. [3]).
- Lithuania anchor switch (2002):
  - Pre-2002 AR coefficient against the US: 0.85, implying a half-life of about 4 years when the anchor for the currency board was the US dollar (Table 9, col. [4]).
  - After the anchor switch (effective float against the dollar), a large and statistically significant negative interaction coefficient implies virtually instantaneous adjustment of the trade balance with the US.
  - Against Eurozone partners, Lithuania’s external adjustment slowed significantly after adopting the euro as anchor; half-life almost doubled (Table 9, col. [5]).
- Eurozone formation (1999):
  - Pooled OLS suggests euro adoption increased half-life of bilateral trade balance adjustment among Eurozone countries by about 1 year (col. [6]).
  - Including CPFE and year effects leaves coefficients similar but statistical significance of interaction term disappears (limited time variation).

### Role of capital mobility and financial openness
- Evidence that greater capital mobility weakens relationship between exchange rate flexibility and external adjustment.
- Time variation in differential:
  - Interaction coefficient capturing regime differential in adjustment speed fell from 1980s to 2000s.
  - In CPFE estimation, differential between speed under fixed regimes and floats:
    - is about 0.2 years for the 1980s,
    - drops to 0.1 years in the 1990s and 2000s.
- Openness heterogeneity (OLS vs CPFE):
  - Based on OLS, half-life differential between fixed and floating regimes is about 0.9 years greater in relatively closed trading partners (both in bottom quartile of Chinn-Ito index) than in more open pairs.
  - CPFE estimates give about 0.1 years greater.

### Robustness checks and sensitivity analyses
- Robust to alternate model specifications, estimation methods (including dynamic panel checks for Nickell bias), samples, measures of bilateral trade balance, and alternate bilateral regime classifications (R&R, LY&S).
- Robust to excluding regime switch years and to restricting to top partners.
- Sensitivity table highlights:
  - Baseline/benchmarks (Table 13):
    - TB_t-1 = 0.507*** (0.013); Regime_t = 0.063*** (0.009); TB_t-1 × Regime_t = -0.136*** (0.013); Observations = 258,075; R-squared = 0.149; No. of trading pairs = 12,660.
  - Alternative specs show TB_t-1 ranging from 0.507*** to 0.734*** with various interaction coefficients (examples preserved exactly in source).
- Using alternate bilateral trade balance measures (absolute value, exports-to-imports ratio, scaled by individual country’s GDP) does not materially change results. With CPFE and time effects, estimated AR coefficient is about 0.5–0.6.
- When TB scaled by GDP, half-life under direct fixed exchange rates is about 0.7 years longer than under flexible rates; with other measures it is about 0.3 years longer.
- CPFE differences in half-life under fixed vs floats across robustness checks: about 0.3–0.5 years (cols. [7]-[14]).

### Key econometric coefficients and sample statistics (selected exact entries)
- Table 6 (Composite regimes, All countries):
  - TB_t-1 = 0.932*** (0.011); Regime_t = 0.001 (0.003); TB_t-1 * Regime_t = -0.002 (0.022); Observations = 4,460; R-squared = 0.872; No. of countries = 180.
- Table 7 (Bilateral regimes, All countries):
  - TB_t-1 = 0.886*** (0.010); Regime_t = 0.001 (0.004); TB_t-1 × Regime_t = -0.128*** (0.010); Observations = 258,075; R-squared = 0.585; No. of trading pairs = 12,660.
- Table 9 (Natural experiments, selected):
  - TB_t-1 = 0.888*** (0.027); Direct regime_t = 0.009** (0.004); TB_t-1 × Direct regime_t = -0.123*** (0.010); Indirect regime_t = 0.039*** (0.004); TB_t-1 × Indirect regime_t = -0.009 (0.007); Observations = 258,075; R-squared = 0.585.
- Table 12 (Financial openness, example):
  - TB_t-1 = 0.897*** (0.033); TB_t-1 × Regime_t = -0.112*** (0.022) in one specification (observations vary by openness split: e.g., 50,943; 78,312; 128,820).
- Autoregressive coefficients on aggregate external balance: New Zealand = 0.4; US = 0.8; Greece ≈ 0.9; Ecuador ≈ 0.5.

### Main empirical findings on persistence and regimes (summary of magnitudes)
- Floats are associated with faster external adjustment than pegs when bilateral exchange rate relationships and currency anchors are taken into account.
- When both cross-sectional and time variation in bilateral exchange rate regimes is allowed (either with or without country fixed effects), the half-life of the bilateral trade balance:
  - is 5 years under a direct peg, and
  - is 2.5 years under a float.
- When only time variation is considered (CPFE), half-life under a direct peg is about 0.3 years higher than under floats.
- Under impure floats, exchange rate flexibility is not much lower than under pure floats; dynamics under impure floats are not statistically different from pure floats in the CPFE specifications.
- Countries with greater trade shares with partners to which they peg have higher persistence in their current accounts (positive association between AR coefficients on aggregate external balance and average trade share with pegged partners).

### Policy implications and interpretation
- Findings support Friedman’s hypothesis that greater exchange rate flexibility facilitates external adjustment by allowing corrective movements before imbalances build up.
- Detailed bilateral regime information and currency anchor identification matter for assessing the relationship between exchange rate regimes and external adjustment.
- Indirect pegs (pegging to a common anchor) have non‑negligible effects on persistence—smaller than direct pegs but significantly larger than floats.
- Financial globalization and increased capital flows have reduced, but not eliminated, the differential in adjustment speed between fixed regimes and floats.
- Policy relevance:
  - Greater exchange rate flexibility could help reduce global imbalances where implicit or explicit fixed exchange rates of surplus countries contribute to persistence.
  - Countries seeking to regain competitiveness and restore external balances face formidable challenges under fixed exchange rate regimes (for example, Eurozone countries).

### Appendix A — Data and summary statistics (selected)
- Definition of indirect pegs:
  - First generation: Countries A and C peg to anchor B → A and C indirect peg.
  - Second generation: D pegs to A which pegs to B; or D pegs to A and E pegs to C producing second-generation links.
  - Cases beyond second generation are very few (only 4 country-pairs).
- AR(1) coefficient estimation for individual countries: OLS of current account balance to GDP on lagged current account balance to GDP; trade share is average trade share with countries against which country has direct fixed/intermediate regime; AR(1) coefficient = 0.001** Trade share with direct pegs + const. (** indicates significance at 5 percent).
- Variables and data sources (selected exact definitions preserved): Age dependency ratio — World Bank's WDI; Bilateral trade balance definition — IMF's Direction of Trade Statistics; Capital account openness — Chinn-Ito (2006); Exchange rate regime classifications — IMF de jure and de facto; GDP, fiscal balance, real GDP — IMF WEO, World Bank WDI as specified.
- Sample country lists include Advanced, Emerging, and Developing groups as enumerated in source.

_Italicized source attribution provided in original unit._

### References .............................................................................................................

### _wp14146 - References .............................................................................................................

### References
- References ............................................................................................................................... 26

### Appendix A
- Appendix A: Data and Summary Statistics ............................................................................ 38

### Tables
- 1. Bilateral Real Exchange Rate Volatility and Exchange Rate Regimes, 1980–2011 ...........30
- 2. Distribution of Bilateral Exchange Rate Regimes with IMF Classification, 1980–2011 ....30
- 3. Transition Probabilities: Composite Regime Classification, 19802011 ............................31
- 4. Transition Probabilities: Bilateral Regime Classification, 19802011 ...............................31
- 5. Transition Probabilities by Decade: Bilateral Regime Classification, 19802011 .............32
- 6. External Dynamics: Composite Exchange Rate Regimes Classifications, 19802011 ........33
- 7. External Dynamics: Bilateral Exchange Rate Regime Classification, 19802011 .............34
- 8. Bilateral Direct and Indirect Pegs, 19802011 ....................................................................35
- 9. External Dynamics: Natural Experiments ...........................................................................35
- 10. External Dynamics: Threshold Effects, 19802011 ..........................................................36
- 11. Real Exchange Rate Corrective Movement, 19802011 ...................................................36
- 12. External Dynamics and Financial Openness, 19802011 .................................................36
- 13. External Dynamics: Sensitivity Analysis, 19802011 ......................................................37

### Figures
- 1. US Trade Balance and Nominal Exchange Rate Volatility with Key Trading Partners .....28
- 2. External Balance: US and New Zealand ..............................................................................28
- 3. Classifying Bilateral Exchange Rate Regime Relationships ...............................................29

*Source: _wp14146 - References (PDF).*

### 4. Bilateral Exchange Rate Volatility and Exchange Rate Regimes .......................................29

### 4. Bilateral Exchange Rate Volatility and Exchange Rate Regimes .......................................29

### Introduction and research question
- Central question: Does exchange rate flexibility facilitate external adjustment (faster mean reversion of trade imbalances) once bilateral exchange rate relationships with trading partners are taken into account?
- Motivation: Aggregate/composite regime classifications can mask heterogeneous bilateral relationships (example: the US dollar “floats” in aggregate but is effectively non‑flexible vis-à-vis major partners such as China).
- Hypothesis (Friedman’s view): The speed of mean reversion of the trade balance is faster under a float than under a peg; indirect pegs allow relatively greater flexibility than direct pegs and therefore faster adjustment than direct pegs but slower than floats.

### Data, bilateral regime construction, and sample
- Coverage: 181 countries over 1980–2011.
- Bilateral regime classification:
  - Three-way coding: fixed = 0, intermediate = 0.5, floating = 1.
  - Distinguishes direct pegs and indirect pegs (common anchor relationships, including first- and second-generation indirect links).
  - Uses IMF de jure and de facto (AREAER) classifications as primary inputs; robustness checks use R&R and LY&S classifications.
  - Treatment of special cases:
    - Basket pegs: top five trading partners used as anchors.
    - SDR pegs: pre-1999 and post-1999 SDR basket currencies used as anchors.
    - ERM/ECU handled via ECU currencies pre-euro and Eurozone anchors thereafter.
- Final bilateral dataset size: about 380,000 observations (instead of (181×180×32)/2 = 521,280 because of missing data in earlier years).
- Subsample definitions: Advanced-EMDC (one advanced, one EM/developing), Advanced (both advanced), EMDC (both EM/developing).

### Stylized facts on bilateral trade balance persistence and regime shares
- Measure of external balance: bilateral trade balance scaled by the sum of bilateral exports and imports (ratio lies between -1 and 1).
- Partitioning TB for transition analysis: four ranges defined as less than -0.5, -0.5 to 0, 0 to 0.5, and greater than 0.5.
- Bilateral regime distribution (de jure and de facto):
  - Direct pegs (fixed and intermediate): about 4 percent of the full sample.
  - Indirect pegs: about 23 percent.
  - Floats: about 73 percent.
  - Restricting to top trading partners increases direct fixed/intermediate share to about 15 percent.
- Real exchange rate volatility (annual standard deviation) by bilateral relationship (sample results):
  - Direct fixed regime: 3 percent per year.
  - Direct intermediate regime: about 6 percent per year.
  - Indirect fixed regime: 7½ percent per year.
  - Indirect intermediate regime: 9 percent per year.
  - Impure floats: 9 percent per year.
  - Pure floats: 10 percent per year.
- Transition probability benchmarks (aggregate composite classification):
  - Large deficits persistence: about 76 percent under fixed, about 82 percent under intermediate, about 50 percent under floats.
  - Aggregate composite results: no clear, consistent pattern across regimes.
- Transition probabilities using bilateral regimes:
  - Persistence of TB is statistically significantly greater under pegs relative to floats across all TB ranges.
  - For large bilateral deficits and surpluses, the probability of staying within the same range from one period to the next is about 4 percentage points higher under pegged regimes than under floats; for other ranges the probability is about 6–27 percentage points higher for pegs.
- Time trends:
  - Persistence of bilateral TB has generally increased over time (1980s → 2000s).
  - For fixed regimes, the probability of maintaining the same balance increased by about 2–8 percentage points from the 1980s to 2000s across TB ranges; for floats the increase was about 1–2 percentage points.
  - Suggested driver: greater financial integration (greater capital mobility) allowing larger imbalances to be sustained longer.

### Estimation strategy and main empirical results
- Baseline AR specification (aggregate/composite regime benchmark):
  - Equation (1) estimated following C&W and others, using de jure and de facto composite classifications.
  - Estimated AR coefficient for full sample: in the range of about 0.7–0.9 (pooled OLS and fixed effects specifications).
  - Interaction term (lagged TB × composite XRR): negative but statistically insignificant across IMF, R&R, LY&S classifications → consistent with earlier findings that composite regime classification shows no robust association between regime and external adjustment.
- Bilateral specification:
  - Equation (2) estimated on bilateral TBijt with bilateral XRRijt (0/0.5/1), including country-pair fixed effects (CPFE) and clustering standard errors at country-pair level.
  - Main finding: strong support for Friedman’s hypothesis when using bilateral regimes—more flexible bilateral exchange rate relationships are associated with significantly faster mean reversion (less persistent bilateral imbalances).
  - Representative results (full sample):
    - Pooled OLS: half-life of bilateral TB under a direct fixed regime ≈ 5 years versus ≈ 2.5 years under floats.
    - With country fixed effects (CFE): half-life under direct fixed ≈ 5 years versus ≈ 2 years under floats.
    - With country-pair fixed effects (CPFE): persistence parameter declines; half-lives ≈ 1.2 years (fixed), 1 year (intermediate), and 0.9 years (float).
  - Subsample patterns:
    - Advanced-EMDC: half-life under pegs between 1.2–5.0 years (depending on specification); floats between 1–3 years.
    - EMDC pairs: pooled OLS half-life ≈ 4 years under pegs vs ≈ 2 years under floats; CPFE estimates ≈ 1 year for pegs vs ≈ 0.8 years for floats.
    - Advanced pairs: interaction term significant in pooled OLS and CFE but insignificant with CPFE (likely limited regime-switch variation among advanced pairs).
- Indirect pegs and impure floats:
  - Preliminary finding: indirect pegs exhibit lower persistence than direct pegs (faster adjustment than direct pegs) but higher persistence than floats, consistent with the ranking by realized real exchange rate volatility.

### Additional empirical patterns and robustness elements reported
- Large imbalances behavior:
  - Under floating regimes, large deficits and surpluses (bottom and top quartiles of bilateral TB distribution) adjust significantly faster than smaller imbalances.
  - Under pegs (direct and indirect) no such size-dependent faster correction is observed.
  - Under floats the direction of bilateral real exchange rate movement is consistent with adjustment: deficit countries experience real depreciations; surplus countries experience real appreciations. Under pegs the bilateral real exchange rate response is statistically insignificant.
- Natural experiments supporting main results:
  - CFA Franc zone’s peg to Eurozone currencies when France adopted the euro in 1999.
  - Lithuania’s switch from US dollar to euro anchor in 2002 for its currency board.
  - Euro adoption: shift from EMS/ERM and ERMII (more flexible) to completely rigid rates with euro adoption.
  - In each case, trade balance adjustment against the corresponding partner was significantly slower under the less flexible arrangement.
- Role of capital mobility:
  - Some evidence that greater capital mobility weakens the relationship between exchange rate flexibility and external adjustment.
- Robustness checks (summary of approaches; detailed results in paper):
  - Results robust to alternate model specifications, estimation methods (including concern about “Nickell bias” and checks with dynamic panel methods), samples, measures of bilateral trade balance, and to bilateral regime classifications constructed using other composite exchange rate classifications (R&R, LY&S).
  - Results robust when excluding regime switch years and when including top partner restrictions.

*Source: IMF Working Paper — “4. Bilateral Exchange Rate Volatility and Exchange Rate Regimes” (excerpt).*

### 0.3 years, it is about 0.1 years between indirect fixed regimes and floats (Table 8, cols. [2]-

### _wp14146 - 0.3 years, it is about 0.1 years between indirect fixed regimes and floats (Table 8, cols. [2]-[3])

### Main empirical findings on persistence and regimes
- Floats are associated with faster external adjustment than pegs when bilateral exchange rate relationships and currency anchors are taken into account.
- When both cross-sectional and time variation in bilateral exchange rate regimes is allowed (either with or without country fixed effects), the half-life of the bilateral trade balance:
  - is 5 years under a direct peg, and
  - is 2.5 years under a float.
- When only time variation is considered (i.e., when country-pair fixed effects are included), the half-life under a direct peg is about 0.3 years higher than under floats.
- The difference in half-life between indirect pegged relationships and floats is about 0.1 years (Table 8, cols. [2]-[3]).
- Under impure floats, exchange rate flexibility is not much lower than under pure floats; dynamics under impure floats are not statistically different from pure floats in the CPFE specifications (cols. [5]-[6]).
- Countries with greater trade shares with partners to which they peg have higher persistence in their current accounts (positive association between AR coefficients on aggregate external balance and average trade share with pegged partners).

### Natural experiments: causal evidence from exogenous regime changes
- CFA franc zone and euro adoption (post-1999):
  - Pooled OLS implies an increase in the half-life of CFA countries’ bilateral trade balance with Eurozone countries (excluding France) of about 1.3 years under the fixed exchange rate system relative to the previous floating regime (col. [1]).
  - The difference in half-lives drops to about 0.1 years when CPFE and year effects are included, and remains statistically significant at the 10 percent level (col. [2]).
  - Excluding France and Eurozone countries with largest ECU weight (Germany, Netherlands, Belgium) increases the interaction coefficient magnitude and makes it statistically significant at the 5 percent level with CPFE (col. [3]).
- Lithuania anchor switch (2002):
  - Pre-2002 AR coefficient against the US: 0.85, implying a half-life of about 4 years when the anchor for the currency board was the US dollar (Table 9, col. [4]).
  - After the anchor switch (effective float against the dollar), a large and statistically significant negative interaction coefficient implies virtually instantaneous adjustment of the trade balance with the US.
  - Against Eurozone partners, Lithuania’s external adjustment slowed significantly after adopting the euro as anchor; the half-life of the trade balance almost doubled (Table 9, col. [5]).
- Eurozone formation (1999):
  - Pooled OLS suggests euro adoption increased the half-life of bilateral trade balance adjustment among Eurozone countries by about 1 year (col. [6]).
  - Including CPFE and year effects leaves coefficients similar but statistical significance of the interaction term disappears, likely due to limited time variation in bilateral regimes (col. [7]).

### Extensions and robustness checks
- Threshold effects:
  - Under floats, persistence of large surpluses/deficits (top/bottom quartiles) is significantly lower than persistence of smaller imbalances, indicating floats correct large bilateral imbalances (Table 10).
  - In OLS, evidence that large deficits adjust more abruptly under direct pegs appears but becomes insignificant in CPFE specifications.
  - Threshold effects are significant for both impure and pure floats; insignificant when fixed and intermediate regimes (direct pegs) are estimated separately.
- Corrective movements in exchange rates:
  - Bilateral real exchange rate movements under floats move in the “correct” direction: countries with trade surpluses experience statistically significant real exchange rate appreciations under floats (Table 11).
  - Under direct and indirect pegged regimes, the response of the real exchange rate to the trade balance is statistically insignificant (cols. [1]-[6]).
  - Real exchange rates move in the correct direction under both pure and impure floats when estimated separately.
- Financial openness and time variation:
  - The interaction coefficient capturing regime differential in adjustment speed fell from the 1980s to the 2000s, consistent with financial globalization weakening the differential.
  - In CPFE estimation, the differential between speed of adjustment under fixed regimes and floats:
    - is about 0.2 years for the 1980s,
    - drops to 0.1 years in the 1990s and 2000s.
  - Based on OLS, the half-life differential between fixed and floating regimes is about 0.9 years greater in relatively closed trading partners (both in bottom quartile of Chinn-Ito index) than in more open pairs; CPFE estimates give about 0.1 years greater.
- Sensitivity analyses (Table 13 and related checks):
  - Alternative bilateral trade balance measures (absolute value, exports-to-imports ratio, scaled by individual country’s GDP) do not materially change results. With CPFE and time effects, estimated AR coefficient is about 0.5–0.6.
  - When trade balance is scaled by GDP, the half-life under direct fixed exchange rates is about 0.7 years longer than under flexible rates; with other measures it is about 0.3 years longer.
  - Using trade balance over goods and services yields a difference in half-lives a little over 0.1 years despite smaller sample (col. [4]).
  - Augmented specifications with controls (economic size, age dependency ratio, fiscal balances) yield qualitatively similar results—difference in half-life about 0.2 years (col. [5]).
  - Using lagged regime variables and system GMM, separating direct fixed and intermediate regimes with interactions, excluding small states (population < 2 million) and oil exporters (oil exports ≥ 50 percent of total exports), excluding extreme TB observations, or restricting to top trading partners all leave CPFE differences in half-life under fixed vs floats at about 0.3–0.5 years (cols. [7]-[14]).
  - Using alternate aggregate regime classifications (R&R and LY&S) with anchor information from IMF AREAER produces similar conclusions.

### Key numerical parameters and illustrative AR coefficients
- Half-life under direct peg: 5 years (when cross-sectional and time variation allowed).
- Half-life under float: 2.5 years (same specification).
- Difference in half-life when only time variation (CPFE): about 0.3 years (direct peg vs float).
- Difference between indirect fixed regimes and floats: about 0.1 years.
- CFA pooled OLS change in half-life: about 1.3 years (col. [1]); falls to about 0.1 years with CPFE and year effects (col. [2]).
- Lithuania pre-switch AR: 0.85 → half-life about 4 years (col. [4]).
- Euro adoption pooled OLS implied half-life increase: about 1 year (col. [6]).
- Differential in half-life across decades (CPFE): 0.2 years (1980s) → 0.1 years (1990s and 2000s).
- CPFE estimated AR coefficient range (alternative measures): about 0.5–0.6.
- CPFE difference in half-life under fixed vs floats across robustness checks: about 0.3–0.5 years.
- Autoregressive coefficients on aggregate external balance: New Zealand = 0.4; US = 0.8; Greece ≈ 0.9; Ecuador ≈ 0.5.

### Policy implications and interpretation
- Findings support Friedman’s hypothesis that greater exchange rate flexibility facilitates external adjustment by allowing corrective movements before imbalances build up.
- Detailed bilateral regime information and currency anchor identification matter for assessing the relationship between exchange rate regimes and external adjustment.
- Indirect pegs (pegging to a common anchor) have non-negligible effects on persistence—smaller than direct pegs but significantly larger than floats.
- Financial globalization and increased capital flows have reduced, but not eliminated, the differential in adjustment speed between fixed regimes and floats.
- Policy relevance:
  - Greater exchange rate flexibility could help reduce global imbalances where implicit or explicit fixed exchange rates of surplus countries contribute to persistence.
  - Countries seeking to regain competitiveness and restore external balances face formidable challenges under fixed exchange rate regimes (for example, Eurozone countries).

*Source: _wp14146 - 0.3 years, it is about 0.1 years between indirect fixed regimes and floats (Table 8, cols. [2]-[3])*

### REFERENCES

### _wp14146 - REFERENCES

### References (selected entries from the source)
- Anderson, H., 2008, “Exchange Policies before Widespread Floating (1945–89),” mimeo (Washington DC: International Monetary Fund).
- Berger, H., and V. Nitsch, 2014, “Wearing Corset, Losing Shape: The Euro’s Effects on Trade Imbalances,” Journal of Policy Modeling, 36(1), pp. 136-155.
- Berka, M., M. Devereux, and C. Engel, 2012, “Real Exchange Rate Adjustment In and Out of the Eurozone,” American Economic Review: Papers & Proceedings, 102(3), pp. 179-185.
- Bubula, A., and I. Ötker, 2003, “Are Pegged and Intermediate Regimes More Crisis Prone?” IMF Working Paper No. WP/03/223 (Washington DC: International Monetary Fund).
- Catão, L., and S. Solomou, 2005, “Effective Exchange Rates and the Classical Gold Standard Adjustment,” American Economic Review, 95(4), pp. 1259-1275.
- Chinn, M., and S.-J. Wei, 2013, “A Faith-based Initiative Meets the Evidence: Does a Flexible Exchange Rate Regime Really Facilitate Current Account Adjustment?” Review of Economics and Statistics, 95(1), pp. 168-184.
- Clark, P., N. Tamirisa, S. Wei, A. Sadikov, and L. Zeng, 2004, “A New Look at Exchange Rate Volatility and Trade Flows,” IMF Occasional Paper 235 (Washington DC: IMF).
- Clower, E., and H. Ito, 2012, “The Persistence of Current Account Balances and its Determinants: The Implications for Global Rebalancing,” ADBI Working Paper No. 400 (Tokyo: Asian Development Bank Institute).
- Dong, W., 2012, “The Role of Expenditure Switching in the Global Imbalance Adjustment,” Journal of International Economics, 86(2), pp. 237-251.
- Faruqee, H., and J. Lee, 2009, “Global Dispersion of Current Accounts: Is the Universe Expanding?” IMF Staff Papers, 56(3), pp. 574-595.
- Frankel, J., 2008, “The Estimated Effects of the Euro on Trade: Why are they Below Historical Effects of Monetary Unions among Smaller Countries?” NBER Working Paper 14542 (Cambridge, MA: National Bureau of Economic Research).
- Friedman, M., 1953, “The Case for Flexible Exchange Rates,” in M. Friedman, Essays in Positive Economics (Chicago: University of Chicago Press), pp. 157-203.
- Ghosh, A., M. Terrones, and J. Zettelmeyer, 2010, “Exchange Rate Regimes and External Adjustment: New Answers to an Old Debate,” in C. Wyplosz (ed.), The New International Monetary System: Essays in Honor of Alexander Swoboda (UK: Routledge).
- Ghosh, A., M. Qureshi, and C. Tsangarides, 2013, “Is Exchange Rate Regime Really Irrelevant for External Adjustment?” Economics Letters 118(1), pp. 104-109.
- Herrmann, S., 2009, “Do We Really Know that Flexible Exchange Rates Facilitate Current Account Adjustment? Some New Empirical Evidence for CEE Countries,” Applied Economics Quarterly, 55(4), pp. 295-312.
- IMF, 2008, Annual Report on Exchange Rate Arrangements and Exchange Rate Restrictions. (Washington DC: International Monetary Fund).
- Judson, R., and A. Owen, 1999, “Estimating Dynamic Panel Data Models: A Guide for Macroeconomists,” Economic Letters, 65(1), pp. 9-15.
- Levy-Yeyati, E., and F. Sturzenegger, 2003, “To Float or to Fix: Evidence on the Impact of Exchange Rate Regimes on Growth,” American Economic Review, 93(4), pp. 1173-1193.
- Levy-Yeyati, E., F. Sturzenegger, and I. Reggio, 2010, “On the Endogeneity of Exchange Rate Regimes,” European Economic Review, 54(5), pp. 659-677.
- Reinhart, C., and K. Rogoff, 2004, “The Modern History of Exchange Rate Arrangements: A Reinterpretation,” Quarterly Journal of Economics, 119(1), pp. 1-48.
- Shambaugh, J., 2004, “The Effects of Fixed Exchange Rates on Monetary Policy,” Quarterly Journal of Economics, 119(1), pp. 301-352.
- Tippkötter, M., 2010, “Global Imbalances and the Current Account Adjustment Process: An Empirical Analysis,” DIW Discussion Paper No. 1011 (Berlin: DIW).

### Figures and Tables (selected numeric entries and definitions)
- Figure note: Exchange rate (V) volatility between countries i and j at horizon n is defined as |[Eijt/Eijt-n]1/n - 1|; where E is the bilateral exchange rate.
- Figure 1 source: Authors’ estimates based on IMF’s WEO and INS databases.
- Figure 2 source: Authors’ estimates based on IMF’s WEO database.

### Key econometric coefficients and sample statistics (extracted from tables)
- Table 6 (External Dynamics: Composite Exchange Rate Regime Classifications, 1980–2011) — All countries, selected coefficients:
  - TB_t-1 = 0.932*** (0.011)
  - Regime_t = 0.001 (0.003)
  - TB_t-1 * Regime_t = -0.002 (0.022)
  - Observations = 4,460
  - R-squared = 0.872
  - No. of countries = 180
- Table 6 — Advanced countries, selected coefficients:
  - TB_t-1 = 0.933*** (0.023)
  - Regime_t = -0.001 (0.003)
  - Observations = 798
  - R-squared = 0.886
- Table 6 — Emerging markets, selected coefficients:
  - TB_t-1 = 0.873*** (0.027)
  - Regime_t = 0.007 (0.008)
  - Observations = 1,192
  - R-squared = 0.809
- Table 7 (External Dynamics: Bilateral Exchange Rate Regime Classification, 1980–2011) — All countries, selected coefficients:
  - TB_t-1 = 0.886*** (0.010)
  - Regime_t = 0.001 (0.004)
  - TB_t-1 × Regime_t = -0.128*** (0.010)
  - Observations = 258,075
  - R-squared = 0.585
  - No. of trading pairs = 12,660
- Table 9 (External Dynamics: Natural Experiments) — selected coefficients:
  - TB_t-1 = 0.888*** (0.027)
  - Direct regime_t = 0.009** (0.004)
  - TB_t-1 × Direct regime_t = -0.123*** (0.010)
  - Indirect regime_t = 0.039*** (0.004)
  - TB_t-1 × Indirect regime_t = -0.009 (0.007)
  - Observations = 258,075
  - R-squared = 0.585
- Table 10 and Table 11: definitions and dependent variables
  - Dependent variable in Table 10 and Table 11 notes: bilateral trade balance to the sum of bilateral exports and imports (TB) and percentage change in bilateral real exchange rate (larger values indicating a currency depreciation), respectively.
- Table 12 (External Dynamics and Financial Openness) — example coefficients (by openness):
  - TB_t-1 = 0.897*** (0.033) in one specification
  - Regime_t = 0.011 (0.007) in one specification; Regime_t = 0.022*** (0.008) in another
  - TB_t-1 × Regime_t = -0.112*** (0.022) in one specification
  - Observations vary by specification (e.g., 50,943; 78,312; 128,820)
- Table 13 (External Dynamics: Sensitivity Analysis) — selected benchmark and robustness results:
  - Baseline/benchmarks:
    - TB_t-1 = 0.507*** (0.013)
    - Regime_t = 0.063*** (0.009)
    - TB_t-1 × Regime_t = -0.136*** (0.013)
    - Observations = 258,075
    - R-squared = 0.149
    - No. of trading pairs = 12,660
  - Alternative specifications (examples):
    - Column (2): TB_t-1 = 0.573*** (0.015); Regime_t = 0.031*** (0.010); TB_t-1 × Regime_t = -0.086*** (0.015); Observations = 217,769
    - Column (8): TB_t-1 = 0.734*** (0.035); Regime_t = 0.009 (0.013); TB_t-1 × Regime_t = -0.255*** (0.035); Observations = 260,774
    - Column (14): TB_t-1 = 0.666*** (0.013); Regime_t = -0.030*** (0.011); TB_t-1 × Regime_t = -0.120*** (0.019); Observations = 147,376
  - Notes on Table 13:
    - Regime in most specifications is direct bilateral de facto exchange rate regime (fixed=0, intermediate=0.5, float=1).
    - Standard errors clustered at country-pair level; ***,**, * indicate significance at 1, 5, and 10 percent levels, respectively.

### Additional methodological and definitional notes (from figures/tables)
- Bilateral relationship classifications:
  - Peg*: A pegs to B and/or B pegs to A or both peg to common anchor or both peg to anchors that peg to a common anchor.
  - Nonpeg/Float: Neither A nor B pegs to each other, nor do they share a common anchor or peg to anchors that share a common anchor.
  - Direct peg: A pegs to B and/or B pegs to A.
  - Indirect peg: A and B related by common anchor(s).
  - Impure Float: A and/or B peg to another country.
  - Pure Float: Neither A nor B pegs to another country.
  - Direct Fixed Exchange Rate: Monetary union, currency board, dollarization, conventional peg.
  - Direct Intermediate Exchange Rate: Basket peg, crawls, bands, managed float.
  - Indirect Fixed Exchange Rate: Monetary union, currency board, dollarization, conventional peg.
  - Indirect Intermediate Exchange Rate: Basket peg, crawls, bands, managed float.
- Notes on regime codings used across tables:
  - IMF de jure classification regime (0=fixed; 0.5=intermediate; 1=float).
  - IMF de facto classification regime (0=fixed; 0.5=intermediate; 1=float).
  - Reinhart and Rogoff (2004) de facto classification regime (0=fixed; 0.5=intermediate; 1=float).
  - Levy-Yeyati and Sturzenegger (2003) de facto classification regime (0=fixed; 0.5=intermediate; 1=float).
- Sample periods, subsamples, and grouping labels appear across tables (e.g., 1980–2011; 1980-89; 1990-99; 2000-11; All countries; Advanced; EMDC; Advanced-EMDC; Top trading partners).

*Italicized source attribution provided in original unit.*

### APPENDIX A: DATA AND SUMMARY STATISTICS

### APPENDIX A: DATA AND SUMMARY STATISTICS

### Bilateral Direct and Indirect Pegs (Figure A1)
- Definition of indirect pegs:
  - Countries A and C peg to anchor country B → “first generation” indirect peg.
  - Country D pegs to A, which pegs to B → D and B considered “first generation” indirect peg.
  - Country D pegs to A and E pegs to C → D’s relationship with C and E (and E’s relationship with A and D) is a “second generation” indirect peg.
  - Cases beyond second generation indirect pegs are very few in the dataset (only 4 country-pairs in the estimated sample).

### External Balance Persistence and Trade Share with Direct Pegs (Figure A2)
- AR(1) coefficient computed for individual countries by estimating OLS regressions of current account balance to GDP on lagged current account balance to GDP over the sample period.
- Trade share is the average trade share (over the sample period) with countries against which the country has a direct fixed or intermediate regime.
- Countries with fewer than 20 observations to compute the AR(1) coefficient are dropped.
- Estimated relationship shown:
  - AR(1) coefficient = 0.001** Trade share with direct pegs + const.
  - ** indicates statistical significance of the estimated coefficient of trade share with direct pegs at the 5 percent level (based on robust standard errors).

### Sample: List of Countries (Table A1)
- The sample includes countries across Advanced, Emerging markets, and Developing groups. Examples listed (not exhaustive here due to source structure) include:
  - Advanced: Australia, Austria, Belgium, Canada, Hong Kong, Cyprus, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Israel, Italy, Japan, Luxembourg, Malta, Netherlands, New Zealand, Norway, Portugal, Singapore, Slovenia, Spain, Sweden, Switzerland, United Kingdom, United States.
  - Emerging markets: Albania, Algeria, Argentina, Armenia, Belarus, Bosnia & Herzegovina, Brazil, Bulgaria, Chile, China, Colombia, Costa Rica, Croatia, Czech Rep., Dominican Rep., Ecuador, Egypt, El Salvador, Estonia, Georgia, Guatemala, Hungary, India, Indonesia, Jamaica, Jordan, Kazakhstan, Korea, Latvia, Lebanon, Lithuania, Macedonia, Malaysia, Mexico, Morocco, Pakistan, Panama, Peru, Philippines, Poland, Romania, Russia, Serbia, Slovak Rep., South Africa, Thailand, Tunisia, Turkey, Ukraine, Uruguay, Venezuela, Vietnam.
  - Developing: Afghanistan, Angola, Antigua and Barbuda, Azerbaijan, Bahamas, Bahrain, Bangladesh, Barbados, Belize, Benin, Bhutan, Bolivia, Botswana, Burkina Faso, Burundi, Cambodia, Cameroon, Cape Verde, Central African Rep., Chad, Comoros, Congo, Dem. Rep. of, Congo, Rep. of, Cote d'Ivoire, Djibouti, Dominica, Eritrea, Ethiopia, Fiji, Gabon, Gambia, Ghana, Grenada, Guatemala, Guinea, Guinea-Bissau, Guyana, Haiti, Honduras, Iran, Iraq, Kenya, Kiribati, Kuwait, Kyrgyz Rep., Laos, Lesotho, Liberia, Libya, Madagascar, Malawi, Maldives, Mali, Mauritania, Mauritius, Mexico (listed earlier under Emerging), Moldova, Mongolia, Morocco (listed earlier), Mozambique, Myanmar, Namibia, Nepal, Niger, Nigeria, North …, Oman, Pakistan (listed earlier), Papua New Guinea, Paraguay, Peru (listed earlier), Philippines (listed earlier), Qatar, Rwanda, Samoa, Saudi Arabia, Senegal, Seychelles, Sierra Leone, Solomon Islands, Somalia, South ... , Sri Lanka, St. Kitts and Nevis, St. Lucia, St. Vincent & Grens., Sudan, Suriname, Swaziland, Syrian Arab Rep., Tajikistan, Tanzania, Togo, Tonga, Trinidad and Tobago, Turkmenistan, Uganda, United Arab Emirates, Uzbekistan, Vanuatu, Yemen, Zambia, Zimbabwe.
- Total observations and country counts appear in regression tables (see Table A5).

### Variables and Data Sources (Table A2)
- Variable definitions and sources (selected entries preserved exactly as presented):
  - Age dependency ratio: In percent of working age population — World Bank's WDI.
  - Bilateral trade balance: Difference between exports of country i to country j and imports of country i from j (In billions of USD). Bilateral exports and imports are computed as the average of i’s exports to j, and j’s imports from i, and the average of i’s imports from j, and j’s exports to i, respectively — IMF's Direction of Trade Statistics.
  - Capital account openness: Index — Chinn-Ito (2006).
  - Current account: In billions of USD — IMF's WEO database.
  - Exchange rate regime classifications: IMF's de jure and de facto; Fixed (no separate legal tender/currency board; peg to single currency); Intermediate (basket currency peg; horizontal band; crawling peg/band; managed float); Float (independent float) — Based on Anderson (2008); IMF's AREAER (various issues); Levy-Yeyati and Sturzenegger (2003); Reinhart and Rogoff; Ilzetzki et al. (2010) (available online at http://personal.lse.ac.uk/ilzetzki/data.htm).
  - Fiscal balance: In billions of local currency (LC) — IMF's WEO database.
  - Gross domestic product (GDP): In billions of USD (or LC) — IMF's WEO database.
  - Real GDP and Real GDP per capita: In constant 2000 USD — World Bank's WDI.

### List of Anchor Currencies (Table A3)
- Country — Currency (as listed):
  - Australia — Australian Dollar
  - Belgium — Belgian Franc
  - Eurozone countries — Euro
  - France — French Franc
  - Germany — Deutsche Mark
  - India — Indian Rupee
  - Japan — Japanese Yen
  - New Zealand — New Zealand Dollar
  - Portugal — Portuguese escudo
  - Russia — Russian Ruble
  - Singapore — Singapore Dollar
  - Spain — Spanish Peseta
  - South Africa — South African Rand

### Transition Probabilities of Current Account Balance: Composite Regime Classification (Table A4)
- Note: CA is current account balance to GDP (in percent). Extreme values (top and bottom 0.5th percentile) and exchange rate regime switch years are excluded. Threshold of -15 percent of GDP corresponds to the bottom 10th percentile of the CA distribution for the full sample. Exchange rate regime is the IMF’s de facto aggregate classification. ***,**, * indicate statistically significant difference between probabilities of floats and fixed/intermediate regime across the diagonal at the 1, 5 and 10 percent levels, respectively.
- Transition matrices (percent) are reported for regime = Fixed, Float, Intermediate. The matrices as presented (rows = origin CA bin; columns = destination CA bin) include these exact values:
  - For Fixed regime:
    - CA ≤ -15: 63.3, 29.2, 4.2, 1.7, 1.3, 0.4
    - -15 > CA ≤ -5: 1.7, 64.0, 19.0, 3.6, 1.5, 0.2
    - -5 > CA < 0: 1.4, 22.6, 59.6***, 12.8, 3.5, 0.2
    - 0 ≥ CA < 5: 2.4, 6.5, 23.9, 49.8***, 16.7, 0.7
    - 5 ≥ CA < 15: 1.0, 2.0, 7.9, 23.8, 54.5***, 10.9
    - CA ≥ 15: 1.2, 3.7, 0.0, 0.6, 28.0, 61.0
  - For Float regime:
    - CA ≤ -15: 58.1, 36.8, 3.7, 0.7, 0.7, 0.0
    - -15 > CA ≤ -5: 6.9, 64.6**, 25.0, 3.1, 0.3, 0.0
    - -5 > CA < 0: 0.1, 18.6, 65.9*, 13.5, 1.8, 0.0
    - 0 ≥ CA < 5: 1.2, 5.0, 28.8, 50.7***, 12.8, 1.5
    - 5 ≥ CA < 15: 1.5, 2.2, 9.6, 25.9, 50.4***, 10.4
    - CA ≥ 15: 1.7, 0.0, 0.0, 0.8, 6.8, 69.0*
  - For Intermediate regime:
    - CA ≤ -15: 33.3, 44.4, 11.1, 0.0, 11.1, 0.0
    - -15 > CA ≤ -5: 5.5, 53.8, 38.5, 2.2, 0.0, 0.0
    - -5 > CA < 0: 1.5, 15.1, 72.7, 10.2, 0.5, 0.0
    - 0 ≥ CA < 5: 0.0, 1.8, 18.9, 76.6, 2.7, 0.0
    - 5 ≥ CA < 15: 0.0, 0.0, 2.7, 10.8, 78.4, 8.1
    - CA ≥ 15: 0.0, 0.0, 0.0, 25.0, 50.0, 25.0

### Current Account Balance and Composite Exchange Rate Regime Classifications — Regression Results (Table A5)
- Dependent variable: current account to GDP (CA_t). CA_{t-1} is lagged current account to GDP. Extreme observations of CA_t (top and bottom 0.5th percentile) and years with exchange rate regime switches are excluded. Constant included. Standard errors clustered at the country level. ***,**, * indicate significance at the 1, 5, and 10 percent levels, respectively.
- Selected coefficient estimates and statistics (exact values preserved as in source):
  - All countries (various specifications):
    - CA_{t-1}: 0.649*** (0.093) in OLS; other specifications include 0.475*** (0.092), 0.471*** (0.092), 0.674*** (0.080), 0.506*** (0.081), 0.504*** (0.080), 0.681*** (0.084), 0.527*** (0.087), 0.524*** (0.085), 0.527*** (0.104), 0.351*** (0.097), 0.348*** (0.098).
    - Regime_t coefficients: 0.241 (0.805), 0.564 (0.871), 0.287 (0.861), 0.200 (0.746), 0.407 (0.903), 0.078 (0.880), 0.308 (0.954), -0.795 (0.812), -0.392 (0.796), 0.694 (0.704), 0.998 (0.792), 0.824 (0.767).
    - CA_{t-1} * Regime_t interaction: -0.023 (0.181), -0.090 (0.186), -0.088 (0.183), -0.146 (0.255), -0.254 (0.242), -0.261 (0.238), -0.215 (0.279), -0.412* (0.227), -0.432* (0.220), 0.101 (0.243), -0.022 (0.245), -0.040 (0.242).
    - Observations: 4,436 (varies by spec); R-squared: 0.547, 0.323, 0.336, 0.550, 0.330, 0.344, 0.539, 0.320, 0.339, 0.460, 0.250, 0.267. No. of countries: 181 (or 178/172 depending on spec).
  - Advanced countries:
    - CA_{t-1}: 0.925*** (0.030), 0.806*** (0.041), 0.806*** (0.042), 0.892*** (0.033), 0.736*** (0.058), 0.742*** (0.055), 0.935*** (0.021), 0.831*** (0.036), 0.843*** (0.036), 0.900*** (0.037), 0.737*** (0.037), 0.732*** (0.037).
    - Regime_t and interaction coefficients reported with standard errors as in table (examples): Regime_t: -0.062 (0.249), 0.538 (0.411), 0.874 (0.583), -0.070 (0.252), 0.876* (0.454), 1.044* (0.512), etc. CA_{t-1} * Regime_t: -0.033 (0.059), -0.086 (0.061), -0.102* (0.052), etc.
    - Observations: 795 (varies); R-squared examples: 0.834, 0.645, 0.665, 0.834, 0.645, 0.664, 0.834, 0.650, 0.674, 0.836, 0.638, 0.656. No. of countries: 29 or 28 depending on spec.
  - Emerging markets:
    - CA_{t-1}: 0.841*** (0.060), 0.762*** (0.080), 0.736*** (0.068), 0.734*** (0.036), 0.542*** (0.096), 0.534*** (0.086), 0.807*** (0.042), 0.660*** (0.062), 0.650*** (0.051), 0.695*** (0.045), 0.389*** (0.092), 0.393*** (0.071).
    - CA_{t-1} * Regime_t interactions include statistically significant negative coefficients in several specs: -0.273* (0.144), -0.501** (0.209), -0.483** (0.197), -0.124 (0.144), -0.139 (0.094), -0.178 (0.113), -0.321 (0.256), -0.436 (0.273), -0.463 (0.290), -0.503 (0.310), -0.403 (0.244), -0.438* (0.239).
    - Observations: 1,188 (varies); R-squared examples: 0.526, 0.329, 0.383, 0.517, 0.285, 0.348, 0.516, 0.284, 0.347, 0.319, 0.093, 0.150. No. of countries: 53 or 51 depending on spec.
  - Developing countries:
    - CA_{t-1}: 0.627*** (0.102), 0.449*** (0.095), 0.448*** (0.094), 0.669*** (0.090), 0.508*** (0.089), 0.506*** (0.088), 0.648*** (0.092), 0.508*** (0.093), 0.510*** (0.092), 0.500*** (0.106), 0.342*** (0.100), 0.343*** (0.101).
    - CA_{t-1} * Regime_t interactions: examples include -0.062 (0.236), -0.072 (0.233), -0.071 (0.231), -0.309 (0.299), -0.370 (0.273), -0.367 (0.270), -0.265 (0.318), -0.471** (0.232), -0.491** (0.225), 0.307* (0.176), 0.261* (0.141), 0.252* (0.139).
    - Observations: 2,453 (varies); R-squared examples: 0.521, 0.314, 0.331, 0.529, 0.331, 0.348, 0.508, 0.318, 0.341, 0.454, 0.272, 0.297. No. of countries: 99 or 96/93 depending on spec.
- Notes on regime variables:
  - a/ Regime is IMF's de jure classification regime (0=fixed; 0.5=intermediate; 1=float).
  - b/ Regime is IMF's de facto classification regime (0=fixed; 0.5=intermediate; 1=float).
  - c/ Regime is Reinhart and Rogoff's (2004) de facto classification regime (0=fixed; 0.5=intermediate; 1=float).
  - d/ Regime is Levy-Yeyati and Sturzenegger's (2003) de facto classification regime (0=fixed; 0.5=intermediate; 1=float).
- Country and year effects: specifications vary; table indicates whether country and year effects are included (No/Yes as reported).

*Source: _wp14146 - APPENDIX A: DATA AND SUMMARY STATISTICS*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14146.pdf_