## _wp15159 - Section IV concludes.

## Source details

**Canonical URL:** [_wp15159 - Section IV concludes.](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15159.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15159.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15159.pdf.json)

---

### II. A linear model of the portfolio balance channel
- Purpose:
  - Present a simple model with imperfect substitutability between domestic and external assets to illustrate foreign exchange intervention (FXI) through a portfolio balance channel in the context of (exogenous) capital inflow shocks, and to generate testable predictions for the empirical analysis.
- Key definitions and identities:
  - GPKI = gross (private) inflows; GPKO = gross (private) outflows; FXI = foreign exchange intervention (defined as reserve sales); CA = current account balance.
  - Balance of payments identity: GPKI − GPKO + FXI + CA = 0.
- Behavioral specifications (linear forms allowing departures from UIP):
  - GPKI_{jt} = α_j [i_{jt} − i^*_t + E_{t−1} e_{j,t+1}] + (1) z_t + ...
  - GPKO_{jt} = β_j [i_{jt} − i^*_t + E_{t−1} e_{j,t+1}] + ρ(...) −
  - CA_{jt} = −γ_j e_{jt} with 0 ≤ γ.
- Parameters and economic interpretation:
  - α and β: sensitivity of gross inflows and outflows to rate-of-return differentials (expected positive).
  - ρ: relative responsiveness of domestic investors to global financial shocks (sign ambiguous; negative ρ implies domestic and foreign investors repatriate assets together).
  - d: sensitivity of domestic interest rate to the exchange rate (i_{jt} = i^*_t − d e_{jt}); d may be positive or negative.
  - φ: degree to which FXI offsets the exogenous component of net private capital flows: FXI_{jt} = −φ (1+ρ) z_t.
  - Global shock dynamics: z_t follows AR(1) with coefficient ψ.
- Equilibrium and comparative statics:
  - Combined equilibrium yields expressions for e_{jt}, GPKI_{jt}, GPKO_{jt}, and FXI_{jt} in terms of z_t and parameters.
  - Assumption: γ ≤ γ̄ (ensuring current account elasticity sufficiently low).
  - A positive exogenous global financial (risk appetite) shock leads to appreciation of the domestic currency (for γ ≤ γ̄).
  - Central bank can dampen appreciation by decreasing the interest rate (d positive) or using FXI (φ positive).
- Effects of FXI (portfolio balance channel):
  - Intervention dampens exchange rate response to z_t provided ρ > −1; if ρ = −1 domestic investors perfectly offset foreigners and intervention is unnecessary.
  - Higher φ increases GPKI and reduces GPKO: lower immediate appreciation reduces expected later depreciation, making domestic assets more attractive to foreigners and domestic investors more likely to stay home.
  - In the limiting case of full intervention, exchange rate unchanged and GPKI = z and GPKO = z ρ−, with intervention equal to net private inflows (1+ρ) z.
- Parameterized illustrations (as in source figures):
  - Comparative statics parameterization: α=β=0.5, ψ=0.2, γ=0.2, d=0 (unless indicated otherwise).
  - Dynamic responses benchmark: α=β=0.5, ψ=0.2, γ=0.2, ρ=−0.3, d=0; comparison of φ=0 and φ=0.5.
- Model-generated testable predictions:
  - (i) a positive exogenous global financial (risk appetite) shock leads to an appreciation of the domestic currency;
  - (ii) gross capital inflows increase and, if ρ is negative, gross outflows also increase (but less than inflows if ρ>−1);
  - (iii) FX intervention mitigates the effect of the shock on the exchange rate;
  - (iv) larger interventions are accompanied by larger gross inflows and/or smaller gross outflows.

### III. Empirical evidence
- Identification and VAR setup:
  - Exploit largely exogenous character of global capital flows (GKF) for each small country; estimate country-specific VARs to obtain responses of FXI, GPKI, GPKO, interest rate differential, and exchange rate to GKF.
  - VAR: Y_{jt} = A_1 L Y_{jt−1} + ... + A_p L^p Y_{jt−p} + ε_{jt}; Y_{jt} = [ER, FXI, GPKI, Y, GPKO, INT, GKF]'.
  - Variables:
    - ER: log nominal exchange rate vis-à-vis the U.S. (baseline); later real ER explored.
    - FXI: net reserve sales plus changes in central bank off-balance-sheet FX position (broad measure); robustness checks use narrow measure (reserves only).
    - GPKI and GPKO: gross private capital inflows and outflows (Financial Flows Analytics), annualized flows normalized by lagged GDP in USD.
    - INT = i_{jt} − i^*_t (short-term interest rate differential with U.S.).
    - GKF: country-specific measure of aggregate global gross capital flows (sum of gross private capital inflows to all non-reserve currency countries excluding country j, divided by sum of corresponding nominal GDPs in USD).
  - Sample:
    - 35 emerging market and advanced economies (reserve currency issuers excluded).
    - Quarterly data 1990Q1-2013Q4 with country-specific start dates; VAR lag length by Akaike; average p = 2.5 lags (implying average of 81 degrees of freedom for countries with data back to 1990).
- Main impulse response results to a one-standard deviation GKF shock:
  - Exchange rate:
    - Unambiguous appreciation of local currency, averaging about 1½ percent over the first 2 quarters, gradually fading away.
  - Gross private inflows and outflows:
    - Gross inflows increase, reaching about 1 percent of GDP on impact and remaining statistically positive for about 3 quarters.
    - Gross outflows also increase, but less than inflows (consistent with ρ > −1).
  - FXI and interest rates:
    - Most countries use FXI in response to increased gross inflows and accumulate reserves.
    - Short-term interest rate differentials fall in response to the shock initially, but the effect is very small on average.
- Cross-section (non-peg economies; cumulative responses at t=0 and t=2):
  - Only one of 27 non-peg economies shows a negative impact on GPKI on impact (Norway); no country shows a negative cumulative impact at a 2-quarter horizon.
  - All countries display positive exchange rate effects on impact and at 2 quarters.
  - Heterogeneity in FXI responses: 20 out of 27 countries show reserve accumulation on impact; 22 out of 27 show reserve accumulation at a 2-quarter horizon.
- De-facto FXI regime classification and group comparisons:
  - Classification rules:
    - Exclude de-facto pegs.
    - Include only economies showing sensitivity to global capital flow shocks (statistical significance of cumulative impulse responses of GPKI or FXI, or impulse response of ER at 2-quarter horizon).
    - Among sensitive flexible-rate economies, split into floaters or interveners by whether cumulative FXI at 2 quarters is smaller or larger than the sample median.
  - Group-level findings (weighted averages):
    - Difference in FXI responses between interveners and floaters is close to 1 percent of quarterly GDP (0.25 percent of annual GDP) on impact.
    - Interveners display smaller appreciation vs floaters: a 1.5 percentage point differential in appreciation between interveners and floaters over the first 3-4 quarters (differential fades afterwards).
    - Quantified effect: a quarterly annualized intervention of 1 percent of GDP (0.25 percent non annualized) leads to about 1.5 percent lower appreciation on impact.
    - No evidence of different interest rate behavior between interveners and floaters on average.
    - Gross capital inflows respond equally or more markedly in intervening countries compared to floaters.
    - Gross outflows increase for both groups, but more in floaters (consistent with domestically-driven offsets; negative ρ).
- Causal interpretation and checks:
  - Two causal possibilities for negative relation between FXI size and GPKO responsiveness:
    - (A) FXI → smaller appreciation → smaller expected depreciation → smaller gross outflows (model-consistent).
    - (B) Lower domestic offsetting (smaller GPKO response) → central bank substitutes with more FXI.
  - Empirical checks find no robust relation between measures of size/depth of domestic investor base and strength of GPKO response; evidence supports causality running from stronger FXI to smaller appreciation to smaller outflows.
- Extensions and additional evidence:
  - Real exchange rate analysis yields results similar to nominal analysis (floaters vs interveners).
  - Pegs vs interveners: pegs show small real depreciation initially and some appreciation afterwards; pegs appear to carry out less intervention than interveners within the managed floater group; pegs display somewhat larger volumes of gross flows (in and out).
- Robustness checks (high-level summary):
  - Narrow FXI (reserves-only) yields qualitatively similar results; some country reclassifications (e.g., Korea and Czech Republic become floaters; New Zealand and Romania become interveners).
  - Including country-specific terms-of-trade shocks leaves results roughly unchanged.
  - Excluding countries with frequent capital control use (above 70th percentile of Quinn-Toyoda index standard deviation) yields similar results; excluding countries with high average capital controls also leaves main results intact and amplifies the differential in GPKO responsiveness.
  - Excluding cases with evidence of unsterilized intervention (Brazil, Hungary, Indonesia, Philippines, Russia) does not alter main results; alternative specifications excluding other countries produce similar results.
  - Alternative (unconditional) FXI classification and using VIX as shock deliver qualitatively similar conclusions: intervention effective in dampening exchange rate response.
  - Other checks (first differences of ER, normalizing flows by trend GDP, using broader net official flows) keep results close to baseline.

### IV. Conclusions
- Core evaluation:
  - Sterilized intervention is assessed as an instrument to insulate countries from exchange rate pressures stemming from capital flow shocks using country-specific VARs and cross-country variation in FXI responses.
- Core empirical conclusion:
  - Consistent with the portfolio balance channel, larger sterilized intervention leads to less exchange rate appreciation in response to gross inflows.
  - The magnitude is macroeconomically relevant: FXI can be a valid policy tool for macroeconomic management.
- Key quantitative takeaways reiterated:
  - Average appreciation about 1½ percent over first 2 quarters after a positive global gross capital flow shock.
  - Gross private inflows reach about 1 percent of GDP on impact.
  - 20 out of 27 non-peg countries show reserve accumulation on impact; 22 out of 27 at 2-quarter horizon.
  - Difference in FXI between interveners and floaters: close to 1 percent of quarterly GDP (0.25 percent of annual GDP) on impact.
  - Interveners show about 1.5 percentage point lower appreciation than floaters over first 3-4 quarters.
  - A quarterly annualized intervention of 1 percent of GDP (0.25 percent non annualized) associated with about 1.5 percent lower appreciation on impact.

*Source: _wp15159 - Section IV concludes.*

### Section IV concludes.

### _wp15159 - Section IV concludes.

### II. A linear model of the portfolio balance channel
- Purpose: present a simple model with imperfect substitutability between domestic and external assets to illustrate foreign exchange intervention (FXI) through a portfolio balance channel in the context of (exogenous) capital inflow shocks, and to generate testable predictions for the empirical analysis.
- Key definitions and identities:
  - GPKI = gross (private) inflows; GPKO = gross (private) outflows; FXI = foreign exchange intervention (defined as reserve sales); CA = current account balance.
  - Balance of payments identity: GPKI − GPKO + FXI + CA = 0 (equation (1)).
- Behavioral specifications (linear functional forms allowing departures from UIP):
  - GPKI: GPKI_{jt} = α_j [i_{jt} − i^*_t + E_{t−1} e_{j,t+1}] + (1) z_t + ... (equation (2), notation as in source).
  - GPKO: GPKO_{jt} = β_j [i_{jt} − i^*_t + E_{t−1} e_{j,t+1}] + ρ(...) − (equation (3)).
  - Current account: CA_{jt} = −γ_j e_{jt} with 0 ≤ γ.
- Parameters and economic interpretation:
  - α and β: sensitivity of gross inflows and outflows to rate-of-return differentials (expected positive).
  - ρ: relative responsiveness of domestic investors to global financial shocks (sign ambiguous; negative ρ implies domestic and foreign investors repatriate assets together).
  - d: sensitivity of domestic interest rate to the exchange rate (i_{jt} = i^*_t − d e_{jt} in equation (4); d may be positive or negative).
  - φ: degree to which FXI offsets the exogenous component of net private capital flows (equation (5)): FXI_{jt} = −φ (1+ρ) z_t.
  - Global shock dynamics: z_t follows AR(1) with coefficient ψ.
- Equilibrium (combined equations (1)–(5) yield):
  - Expressions for exchange rate e_{jt}, GPKI_{jt}, GPKO_{jt}, and FXI_{jt} in terms of z_t and parameters (equations (6)–(9) in source).
  - Assumption: γ ≤ γ̄, with γ̄ defined in source (ensuring current account elasticity sufficiently low).
- Comparative statics and implications:
  - Positive exogenous global financial (risk appetite) shock leads to appreciation of domestic currency (for γ ≤ γ̄).
  - Central bank can dampen appreciation by decreasing the interest rate (d positive) or using FXI (φ positive).
  - Effect of FXI:
    - Intervention dampens exchange rate response to z_t provided ρ > −1; if ρ = −1 domestic investors perfectly offset foreigners and intervention is unnecessary.
    - Higher φ increases GPKI and reduces GPKO through a portfolio balance channel: lower immediate appreciation reduces expected later depreciation, making domestic assets more attractive to foreigners and domestic investors more likely to stay home.
    - In the limiting case of full intervention, exchange rate unchanged and GPKI = z and GPKO = z ρ−, with intervention equal to net private inflows (1+ρ) z.
- Parameterized illustrations (figures in source):
  - Comparative statics parameterization: α=β=0.5, ψ=0.2, γ=0.2, d=0 (unless indicated otherwise).
  - Dynamic responses benchmark: α=β=0.5, ψ=0.2, γ=0.2, ρ=−0.3, d=0; comparison of φ=0 and φ=0.5.
- Model-generated testable predictions summarized:
  - (i) a positive exogenous global financial (risk appetite) shock leads to an appreciation of the domestic currency;
  - (ii) gross capital inflows increase and, if ρ is negative, gross outflows also increase (but less than inflows if ρ>−1);
  - (iii) FX intervention mitigates the effect of the shock on the exchange rate;
  - (iv) larger interventions are accompanied by larger gross inflows and/or smaller gross outflows.

### III. Empirical evidence
- Methodological approach:
  - Identification strategy: exploit largely exogenous character of global capital flows (GKF) from the perspective of each small country; estimate country-specific VARs to obtain responses of FXI, GPKI, GPKO, interest rate differential, and exchange rate to GKF.
  - VAR specification: Y_{jt} = A_1 L Y_{jt−1} + ... + A_p L^p Y_{jt−p} + ε_{jt} (equation (10)); endogenous vector Y_{jt} = [ER, FXI, GPKI, Y, GPKO, INT, GKF]' (equation (11) in source notation).
  - Variables:
    - ER: log nominal exchange rate vis-à-vis the U.S. in baseline (also later real exchange rate explored).
    - FXI: net reserve sales plus changes in central bank off-balance-sheet FX position (broad measure); robustness checks use narrow measure (reserves only).
    - GPKI and GPKO: gross private capital inflows and outflows (Financial Flows Analytics database, annualized flows normalized by lagged GDP in USD).
    - INT: short-term interest rate differential with respect to the U.S. (INT = i_{jt} − i^*_t).
    - GKF: country-specific measure of aggregate global gross capital flows = sum of gross private capital inflows to all non-reserve currency countries excluding country j, divided by sum of corresponding nominal GDPs in USD (equation (12)).
  - Sample:
    - 35 emerging market and advanced economies (listed in source).
    - Reserve currency issuers excluded (United States, United Kingdom, Switzerland, Japan, Euro Area members).
    - Quarterly data 1990Q1-2013Q4, with country-specific restrictions; VAR lag length chosen by Akaike information criterion, average p = 2.5 lags (implying average of 81 degrees of freedom for countries with data back to 1990).
- Main empirical results (impulse response evidence to a one-standard deviation GKF shock):
  - Exchange rate:
    - Unambiguous appreciation of local currency, averaging about 1½ percent over the first 2 quarters, gradually fading away.
  - Gross private inflows and outflows:
    - Gross inflows increase, reaching about 1 percent of GDP on impact and remaining statistically positive for about 3 quarters.
    - Gross outflows also increase, but less than inflows (consistent with ρ > −1).
  - FXI and interest rates:
    - Most countries use FXI in response to increased gross inflows and accumulate reserves.
    - Short-term interest rate differentials fall in response to the shock initially, but the effect is very small on average.
  - Cross-section evidence (non-peg economies; cumulative responses at t=0 and t=2):
    - Only one of 27 non-peg economies shows a negative impact on GPKI on impact (Norway); no country shows a negative cumulative impact at a 2-quarter horizon.
    - All countries display positive exchange rate effects on impact and at 2 quarters.
    - Heterogeneity in FXI responses: 20 out of 27 countries show reserve accumulation on impact; 22 out of 27 show reserve accumulation at a 2-quarter horizon.
- De-facto FXI regime classification and group comparisons:
  - Classification rules:
    - Exclude de-facto pegs (Ilzetzki et al (2011)).
    - Include only economies showing sensitivity to global capital flow shocks (statistical significance of cumulative impulse responses of GPKI or FXI, or impulse response of ER at 2-quarter horizon).
    - Among sensitive flexible-rate economies, split into floaters or interveners by whether cumulative FXI at 2 quarters is smaller or larger than the sample median.
  - Table 1 provides group membership (as in source).
  - Group-level empirical findings (weighted averages):
    - Difference in FXI responses between interveners and floaters is sizeable: close to 1 percent of quarterly GDP (0.25 percent of annual GDP) on impact.
    - Interveners display smaller appreciation of their currencies vs floaters: a 1.5 percentage point differential in appreciation between interveners and floaters over the first 3-4 quarters (differential fades afterwards). This is statistically and economically significant.
    - Quantified effect: a quarterly annualized intervention of 1 percent of GDP (0.25 percent non annualized) leads to about 1.5 percent lower appreciation on impact.
    - No evidence of different interest rate behavior between interveners and floaters on average.
    - Gross capital inflows respond equally or more markedly in intervening countries compared to floaters.
    - Gross outflows increase for both groups, but more in floaters (consistent with domestically-driven offsets; negative ρ).
  - Causal interpretation considerations:
    - Two possible causal directions for negative relation between FXI size and GPKO responsiveness:
      - (A) FXI → smaller appreciation → smaller expected depreciation → smaller gross outflows (model-consistent).
      - (B) Lower domestic offsetting (smaller GPKO response) → central bank substitutes with more FXI.
    - Empirical checks: no robust relation found between measures of size/depth of domestic investor base and strength of GPKO response; evidence suggests causality runs from stronger FXI to smaller appreciation to smaller outflows.
  - Cross-country dispersion:
    - Figure 7 and 8 in source show interveners generally have lower ER appreciation responses, with notable outliers (e.g., Brazil).
- Extensions and additional evidence:
  - Real exchange rate analysis:
    - Using real effective exchange rates produces results similar to nominal analysis (floaters vs interveners).
    - Pegs vs interveners: pegs show small real depreciation initially and some appreciation afterwards; pegs appear to carry out less intervention than interveners within the managed floater group; pegs display somewhat larger volumes of gross flows (in and out).
- Robustness checks (summary):
  - Proxy for FXI:
    - Narrow FXI (reserves-only) yields qualitatively similar results; a few country classifications change (Korea and Czech Republic become floaters; New Zealand and Romania become interveners).
  - Terms of trade: including country-specific terms-of-trade shocks leaves results roughly unchanged.
  - Capital controls:
    - Excluding countries with frequent capital control use (above 70th percentile of Quinn-Toyoda index standard deviation) yields results similar to baseline.
    - Excluding countries with high average levels of capital controls (1990-2012) also leaves main results intact; differential in GPKO response between interveners and floaters is greater in this restricted sample.
  - Sterilization:
    - Excluding cases with evidence of unsterilized intervention (statistically significant negative response of interest rate differential to shocks: Brazil, Hungary, Indonesia, Philippines, Russia) does not alter main results.
    - Alternative specification with domestic and foreign rates separately excludes Hungary, Indonesia, Russia, Thailand under this criterion; results remain similar.
  - Alternative classifications:
    - Relaxing sensitivity criteria does not change main results.
    - Unconditional FXI regime classification (based on degree of intervention relative to gross inflows over sample) leads to some country reassignments (Guatemala, Israel, Norway become interveners; New Zealand and Poland become floaters) but leaves main results broadly similar (except weaker differentiated pattern of gross flows under unconditional classification).
  - Alternative shock measure:
    - Using VIX as exogenous shock (instead of GKF) delivers qualitatively similar conclusion: intervention effective in dampening exchange rate response.
  - Other robustness checks (not reported in main text): first differences of exchange rate, normalizing flows by trend GDP, using broader net official flows instead of central bank FXI, etc. — results remain close to baseline.

### IV. Conclusions
- Main evaluation: sterilized intervention is assessed as an instrument to insulate countries from exchange rate pressures stemming from capital flow shocks using country-specific VARs and cross-country variation in FXI responses.
- Core empirical conclusion:
  - Consistent with the portfolio balance channel, larger sterilized intervention leads to less exchange rate appreciation in response to gross inflows.
  - The magnitude is macroeconomically relevant: FXI can be a valid policy tool for macroeconomic management.
- Key quantitative takeaways reiterated:
  - Average appreciation about 1½ percent over first 2 quarters after a positive global gross capital flow shock.
  - Gross private inflows reach about 1 percent of GDP on impact.
  - 20 out of 27 non-peg countries show reserve accumulation on impact; 22 out of 27 at 2-quarter horizon.
  - Difference in FXI between interveners and floaters: close to 1 percent of quarterly GDP (0.25 percent of annual GDP) on impact.
  - Interveners show about 1.5 percentage point lower appreciation than floaters over first 3-4 quarters.
  - A quarterly annualized intervention of 1 percent of GDP (0.25 percent non annualized) associated with about 1.5 percent lower appreciation on impact.

*Source: _wp15159 - Section IV concludes.*

### REFERENCES

### _wp15159 - REFERENCES

### References list
- Comprehensive bibliography on foreign exchange intervention, capital flows, reserve accumulation, and related empirical and theoretical work. Includes journal articles, working papers, mimeos, IMF Working Papers, NBER Working Papers, BIS Papers, and policy briefs by authors such as Adler, Aizenman, Bayoumi, Benes, Bluedorn, Broner, Bruno and Shin, Calvo, Cavallo, De Bock, Dominguez, Forbes, Gagnon, Ghosh, Jeanne, Kumhof, Menkhoff, Milesi-Ferretti, Neely, Quinn, Reinhart, Rey, Sarno and Taylor, Schindler, Stone, Tapia and Tokman, and many others as listed in the unit.

### Table A1 — Country sample period
- Brazil: 1998Q4 onwards
- Bulgaria: 1996Q4 onwards
- Canada: 2002Q2 onwards
- Chile: 1992Q2 onwards
- Croatia: 1994Q4 onwards
- Estonia: 1992Q3 to 2010Q4
- Indonesia: 1992Q2 onwards
- Korea: 1998Q3 onwards
- Latvia: 1994Q4 onwards
- Lithuania: 1995Q2 onwards
- Peru: 1994Q4 onwards
- Philippines: 1998Q1 onwards
- Poland: 1995Q3 onwards
- Romania: 2001Q2 onwards
- Russian Federation: 2000Q1 onwards
- South Africa: 1995Q2 onwards
- Sweden: 1993Q1 onwards
- Thailand: 1999Q1 onwards
- Turkey: 2003Q2 onwards

### Table A2 — De facto FXI Regime — Classification across specifications
- Presents country-by-country de facto FXI regime classifications across multiple specifications (Baseline; Excluding Off-BS FXI; Terms of trade; Capital Controls; Capital Controls (2); Sterilized FXI; Sterilized FXI (2); Sensitivity Criterion; Unconditional Classif.; VIX).
- Examples of categorical entries preserved exactly as in source (IFS code and labels shown):
  - 128 Denmark: Peg (Pe g) across specifications ending with "g"
  - 142 Norway: Floater / Floater ... Intervener / Intervener
  - 156 Canada: Floater repeated across specifications
  - 218 Bolivia: Intervener across specifications
  - 223 Brazil: Intervener across most specifications, with one "Floater" entry in list
  - 228 Chile: Floater across listed specifications
  - 293 Peru: Intervener across specifications and "Not Sensitive" in one column
  - 524 Sri Lanka: Intervener across specifications and "Not Sensitive" in one column
  - 534 India: Intervener across specifications with one "Floater" entry
  - 542 Korea, Republic of: Intervener / Floater mixed entries, many Intervener entries
  - 548 Malaysia: Intervener across specifications
  - 566 Philippines: Intervener across specifications with a punctuation mark in one column (Intervener.)
  - Several countries listed with "Pe g" (peg) entries: China, P.R.: Mainland (IFS code 924), Estonia (939), Latvia (941), Lithuania (946), Croatia (960), etc.
- Source note: "Source: Authors' estimations."

### Figures and notes — Responses to global capital flow shocks and robustness checks
- Figure A1. Responses to a Global Capital Flow Shock
  - Steps shown: Step t=0 and Step t=2
  - Panels:
    - Panel (a). INT responses to a global capital flow shock (INT is non-cumulative; other impulse responses are cumulative)
    - Panel (b). GPKO responses to a global capital flow shock
  - Note: "Cumulative impulse responses from individually estimated VAR models, at t=0 and t=2; except for INT, which is non-cumulative. One standard deviation bands are reported."
  - Vertical scales and country-ordered listings appear in the figure (numerical axis markers such as -.02, -.015, -.01, -.005, 0, .005; and separate panels with values e.g., -.06 to .04, -.05 to .15).

- Figure A2. Robustness Check — Responses of Floaters and Interveners to a Global Capital Flow Shock (non-peg economies)
  - Steps shown: Step t=0, Step t=2, Step=4
  - Panels:
    - Panel (a). ER responses
    - Panel (b). GPKI cumulative responses
    - Panel (c). GPKO cumulative responses
    - Panel (d). FXI cumulative responses
    - Panel (e). INT responses
  - Note: "The figure reports results from alternative specifications, as described in the main text, for floaters (in blue) and interveners (in orange). For each of them, the estimated (weighted) average response for floaters and interveners are shown, at t={0,2,4}."
  - Specification labels (numbered) shown on axes:
    - 1. Baseline
    - 2. Exclude Off-BS FXI
    - 3. Terms of trade
    - 4. Capital Controls
    - 5. Capital Controls(2)
    - 6. Sterilized FXI
    - 7. Sterilized FXI(2)
    - 8. Sensitivity Crit.
    - 9. Unconditional FXI
    - 10. VIX
  - Axis ranges and tick values preserved in figures (examples: 0 to .03, -.01 to .05, -.04 to .01, -.01 to .01).

- Figure A3. Impulse Responses to a Global Capital Flow Shock — Unconditional FXI Regime Classification
  - Note: "Impulse response functions to a one standard deviation shock to the global capital flow variable. For each group, weighted averages of the impulse responses and confidence bands are reported, with weights that are inversely proportional to the standard deviation of each impulse response."

- Figure A4. Cross section of Exchange Rate Responses to a Global Capital Flows Shock — Unconditional FXI Regime Classification
  - Steps shown: Step t=0 and Step t=2
  - Note: "Exchange rate impulse responses from individually estimated VAR models, at t=0 and t=2; grouped by floaters and interveners. One standard deviation bands are reported."
  - Chart components reference series labels: ER, FXI, GPKI, GPKO, INT and time axis label "Quarters" with numeric ticks "0 4 8 12 0 4 8 12".
  - Cross-sectional country orderings for floaters and interveners are shown (country lists such as Guatemala, Peru, Bolivia, Philippines, Sri Lanka, Malaysia, Thailand, Russian Federation, Israel, Indonesia, Korea, Republic of, Romania, India, Norway, Brazil, Poland, Czech Republic, Turkey, Sweden, Mexico, Canada, Hungary, Chile, Australia, Colombia, New Zealand, South Africa).

*Source: _wp15159 - REFERENCES (figures, tables and bibliography as provided in the source unit)*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15159.pdf_