## _wp1408

## Source details

**Canonical URL:** [_wp1408](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1408.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1408.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1408.pdf.json)

---

### Overview and research question
- Study focus: Whether, and under what conditions, a tightening of outflow restrictions can be used effectively to reduce net capital outflows.
- Sample: 37 emerging market economies during 1995-2010.
- Methodology: Panel vector autoregression (PVAR) with interaction terms following Towbin and Weber (2011) to allow for endogeneity of capital flow restrictions and to compute conditional impulse response functions.
- Outflow restrictions measure: Fund staff’s narrow de jure restrictiveness index (based on Schindler (2009) for disaggregated outflow/inflow distinctions). Annual Schindler index correlated 92 percent with staff’s narrow index for overlapping period.

### Methodology and data construction
- Net inflows definition: Net assets plus net liabilities, in percent of GDP (IFS definition).
- Gross inflows/gross outflows (crude measures, following Schindler (2009) assumption):
  - gross inflows = max(net assets flows, 0) + max(net liabilities flows, 0)
  - gross outflows = max(-net assets flows, 0) + max(-net liabilities flows, 0)
  - Gross in- and outflows are constructed so they will never be negative.
- Data frequency and period: Quarterly data 1995Q1-2010Q4 required for all major variables.
- Country inclusion criterion: Excluded countries with capital control indices that were zero throughout the sample period.
- Baseline PVAR specification:
  - Endogenous variables: outflow control index, industrial production, the interest rate, inflation, the local currency/USD exchange rate, and net capital flows as a share of GDP.
  - Lag structure: two lags used in estimation.
  - Identification: impulse responses via Choleski ordering with baseline ordering:
    1) output (industrial production)
    2) capital outflow control index
    3) real exchange rate
    4) interest rate
    5) inflation
    6) net capital flows
  - Shock normalization: shocks normalized to represent an increase of 0.25 in the outflow control index (a control tightening).
  - Index properties in sample: ranges between 0 (open) and 1 (closed), with a mean of 0.63 and a standard deviation of 0.35.
- Augmented PVAR: includes interaction terms to let dynamic responses vary with structural country characteristics; conditional impulse responses evaluated at the 25th and 75th percentiles of each indicator’s sample distribution.
- Estimation details: two lags and Runkle’s bootstrapping technique; heterogeneity controlled with country and time fixed effects; confidence intervals cover 90 percent of the estimates from 1,000 simulations.

### Main findings — average sample
- Average PVAR (no interaction terms):
  - A tightening of outflow restrictions on average reduces gross capital outflows.
  - Simultaneously, there is a contraction in gross inflows—mainly driven by non-resident investors—which is larger in size than the contraction in outflows.
  - Net result for the average country: a tightening of outflow restrictions is ineffective as net outflows increase as a result.
- Additional average-sample findings:
  - Gross outflows decline in response to a control tightening (intended effect).
  - Gross inflows also fall, resulting in an increase in net inflows (or a fall in net outflows) that is small and insignificant.
  - Residents repatriate investment; non-residents reduce investments in the domestic economy.
  - Symmetry: tightenings raise net inflows, loosenings lower net inflows; conditioning on tightenings in the initial period confirms findings.

### Conditional effectiveness — key preconditions
- Restrictions are effective (i.e., tightening reduces net capital outflows) if supported by one or more of the following preconditions:
  - Good macroeconomic fundamentals.
  - Strong institutions (government effectiveness).
  - Pre-existing outflow controls are already fairly comprehensive (a high outflow control index at the time of tightening).
- When none of these three conditions is fulfilled, a tightening of restrictions is ineffective.
- Robustness: results for macro fundamentals are robust to various specification changes; results for government effectiveness are positive but slightly less clear-cut and robust.

### Conditional effectiveness — macroeconomic fundamentals (quantified)
- Macroeconomic fundamentals index: constructed by ranking countries along growth, inflation, fiscal balance, and current account balance; fundamentals rank = simple average of four ranks.
- Sample distribution statistics (75th vs 25th percentiles of fundamentals):
  - Average growth rate: 6.4 (75th) and 1.1 (25th) percent
  - Average inflation rate: 8.9 (75th) and 2.3 (25th) percent
  - Average current account balance: 0.7 (75th) and -1.1 (25th) percent of GDP
  - Average fiscal balance: -0.5 (75th) and -4.8 (25th) percent of GDP
- Findings for strong fundamentals (75th percentile):
  - A control tightening increases net inflows at the peak by 4 percent of GDP, keeping them elevated for more than one and a half years.
  - Small and statistically insignificant decline in gross inflows; substantial fall in gross outflows dominates.
  - Real exchange rate appreciates; interest rate falls (not always statistically significant).
  - Industrial production shows a substantial boost.
- Interpretation: in countries with good fundamentals, restrictions can increase net inflows, halt exchange rate depreciation, allow for lower interest rates, and boost activity — possibly via investor sentiment and expectations.

### Conditional effectiveness — institutional quality (quantified)
- Institutional quality measured using the World Bank’s government effectiveness index. Evaluated at 75th and 25th percentiles (values 0.7 and -0.2 respectively).
- Findings:
  - Restrictions are effective at countering capital flight when institutional quality is high.
  - Net inflows increase by about 3 percent of GDP on impact and by about 2 percent thereafter in response to the shock (high government effectiveness).
  - Restrictions are ineffective when institutional quality is low.
  - Similar results obtained using alternative governance indicators (political stability, regulatory quality, control of corruption).
- Note: responses of interest rates and the real exchange rate are relatively muted compared with the macro fundamentals case.

### Conditional effectiveness — pre-existing restriction intensity
- Impulse responses evaluated at different starting levels of the capital control index:
  - high level: 75th percentile of outflow control intensity (value of 1);
  - low level: 25th percentile of outflow control intensity (value of 0.3).
- Where pre-existing restrictions are fairly comprehensive:
  - net inflows increase significantly by about 5 percent of GDP on impact;
  - net inflows increase by 2.5 percent for the following 6 quarters.
  - the tightening leads to a larger decline in gross outflows than in gross inflows.
  - the exchange rate surprisingly depreciates in response to the tightening.
  - interest rates show a fairly muted response.
- Where restrictions are tightened but do not attain a minimum level of comprehensiveness:
  - a control tightening shows little effect on gross or net flows.
- Sensitivity to ordering:
  - In some robustness-ordering variants (allowing outflow restrictions to respond contemporaneously to capital flows) the response can be significantly more muted: no longer positive on impact and peaks at only about 0.5 percent of GDP.
  - When changing ordering but still positive across horizon, the response can peak at about 1.2 percent of GDP.

### Selected country episodes (summaries)
- Thailand (May 1997):
  - Measures: Limits on forward transactions, export surrender requirements, requirements to convert baht proceeds onshore.
  - Outcomes: Exchange rate depreciation 50 percent six months after implementation; reserves fell 20 percent six months after implementation; interest rates continued to rise; restrictions abandoned in January 1998.
- Malaysia (September 1998):
  - Measures: Comprehensive restrictions including a 12-month waiting period for non-residents to convert ringgit proceeds from sale of Malaysian securities; closed legal channels for offshore trading of the ringgit.
  - Outcomes: Capital flight halted; reserves accumulated; exchange rate stabilized; interest rates reduced to aid economic revitalization.
- Iceland (November 2008):
  - Context: Collapse of three largest commercial banks with very large short-term foreign liabilities (600% of GDP).
  - Measures: Comprehensive outflow restrictions including bans on movement of capital with exceptions below ISK 10 million and bans on conversion of offshore krona to foreign exchange; foreign exchange transactions for external trade remained unrestricted.
  - Outcomes: Exchange rate stabilized shortly after restrictions; restrictions provided room for easing monetary policy; intent to lift restrictions gradually.
- Ukraine (2008-09):
  - Measures: Exchange restrictions, 5-day waiting period for non-residents to convert hryvnia proceeds from investment transactions.
  - Outcomes: Some temporary alleviation of exchange rate pressures, but failed to avoid massive central bank intervention; reserves plummeted 30 percent six months after tightening; policy slippages undermined effectiveness.

### Selected summary statistics (1995-2010) — values preserved exactly as reported
- Outflow control index: Mean 0.6, Standard Deviation 0.4, 10th percentile 0.0, 25th percentile 0.3, 50th percentile 0.8, 75th percentile 1.0, 90th percentile 1.0
- Net inflows/GDP: Mean 3.11, Standard Deviation 11.1, 10th percentile -5.0, 25th percentile -0.7, 50th percentile 1.4, 75th percentile 6.6, 90th percentile 12.8
- Gross inflows/GDP: Mean 10.1, Standard Deviation 19.3, 10th percentile 0.0, 25th percentile 1.1, 50th percentile 4.8, 75th percentile 12.3, 90th percentile 21.1
- Gross outflows/GDP: Mean 7.0, Standard Deviation 17.4, 10th percentile 0.0, 25th percentile 0.3, 50th percentile 2.0, 75th percentile 6.7, 90th percentile 12.8
- Net assets/GDP: Mean -5.4, Standard Deviation 18.6, 10th percentile -1.6, 25th percentile 0.1, 50th percentile 1.7, 75th percentile 5.9, 90th percentile 13.7
- Net liabilities/GDP: Mean 8.5, Standard Deviation 19.5, 10th percentile -1.0, 25th percentile 0.8, 50th percentile 4.1, 75th percentile 11.2, 90th percentile 20.0
- Macro fundamentals index: Mean 16.0, Standard Deviation 5.3, 10th percentile 9.0, 25th percentile 12.3, 50th percentile 16.3, 75th percentile 19.8, 90th percentile 22.5
  - GDP growth (y/y percent change): Mean 3.5, Standard Deviation 4.8, 10th percentile -2.7, 25th percentile 1.1, 50th percentile 4.2, 75th percentile 6.4, 90th percentile 8.9
  - Inflation (y/y percent change): Mean 6.5, Standard Deviation 6.3, 10th percentile 0.1, 25th percentile 2.3, 50th percentile 6.5, 75th percentile 8.9, 90th percentile 10.9
  - Current account surplus (as a share of GDP): Mean -0.1, Standard Deviation 1.9, 10th percentile -2.0, 25th percentile -1.1, 50th percentile -0.3, 75th percentile 0.7, 90th percentile 2.2
  - Fiscal surplus (as a share of GDP): Mean -2.4, Standard Deviation 4.4, 10th percentile -7.3, 25th percentile -4.8, 50th percentile -2.8, 75th percentile -0.5, 90th percentile 2.6
- Government effectiveness index: Mean 0.3, Standard Deviation 0.7, 10th percentile -0.6, 25th percentile -0.2, 50th percentile 0.1, 75th percentile 0.7, 90th percentile 1.2

### Robustness, scope, and limitations
- Robustness checks: alternative lag structures, alternative Choleski orderings, alternative specifications replacing net inflows with gross flows and net assets/liabilities, substitution of industrial production by real GDP, augmenting PVAR to include an inflow control index, excluding China and India.
- Key robustness outcomes:
  - The finding that restrictions are effective in countries with strong fundamentals is robust to all checks (only when reducing lag structure to 1 is there a somewhat pronounced drop in magnitude).
  - The response for countries with strong institutions is robust to most checks; changing ordering to allow contemporaneous response of outflow restrictions can reduce impact markedly (peak about 0.5 percent of GDP).
  - The finding that outflow restrictions need to be comprehensive to be effective is relatively robust; changing ordering reduces magnitude but response remains positive and can peak at about 1.2 percent of GDP.
- Empirical scope limitations:
  - The paper does not study the impact of long-standing (persistent) restrictions on capital flows, nor does it assess the broader benefits and costs of capital flow management measures.
  - Potential inference complications acknowledged: restrictions are often reactive to rising capital flows; treating restrictions as endogenous via PVAR addresses this concern.
  - Caveat about prior literature: treating controls as exogenous likely biases effectiveness estimates downward because restrictions tend to be introduced in response to rising outflows.

### Policy implications and conclusions
- Main criterion for effectiveness: impact of tightening outflow restrictions on net capital flows; additional criteria include exchange rate and interest stabilization and regaining monetary policy independence.
- Policy-relevant conclusions:
  - Tightening outflow restrictions can reduce net capital outflows when accompanied by good macroeconomic fundamentals, strong institutions, or when pre-existing restrictions are comprehensive enough to slow a “rush to the exit.”
  - For the average country without these preconditions, tightening outflow restrictions can be ineffective or counterproductive for net inflows because gross inflows may contract by more than gross outflows.
  - Policymakers should consider country-specific fundamentals, institutional quality, and the comprehensiveness of existing controls before relying on outflow restrictions as a stabilization tool.

*Source: _wp1408 - References (selected content excerpt).*

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

### _wp1408 - References ................................................................................................

### Overview and research question
- Study focus: Whether, and under what conditions, a tightening of outflow restrictions can be used effectively to reduce net capital outflows.
- Sample: 37 emerging market economies during 1995-2010.
- Methodology: Panel vector autoregression (PVAR) with interaction terms following Towbin and Weber (2011) to allow for endogeneity of capital flow restrictions and to compute conditional impulse response functions.
- Outflow restrictions measure: Fund staff’s narrow de jure restrictiveness index (based on Schindler (2009) for disaggregated outflow/inflow distinctions). Annual Schindler index correlated 92 percent with staff’s narrow index for overlapping period.

### Key empirical definitions and data construction
- Net inflows: Net assets plus net liabilities, in percent of GDP (IFS definition).
- Gross inflows/gross outflows (crude measures, following Schindler (2009) assumption):
  - gross inflows = max(net assets flows, 0) + max(net liabilities flows, 0)
  - gross outflows = max(-net assets flows, 0) + max(-net liabilities flows, 0)
  - Gross in- and outflows are constructed so they will never be negative.
- Data frequency and period: Quarterly data 1995Q1-2010Q4 required for all major variables.
- Country inclusion criterion: Excluded countries with capital control indices that were zero throughout the sample period.

### Main findings
- Average PVAR (no interaction terms):
  - A tightening of outflow restrictions on average reduces gross capital outflows.
  - Simultaneously, there is a contraction in gross inflows—mainly driven by non-resident investors—which is larger in size than the contraction in outflows.
  - Net result for the average country: a tightening of outflow restrictions is ineffective as net outflows increase as a result.
- Conditional effectiveness (with interaction terms):
  - Restrictions are effective (i.e., tightening reduces net capital outflows) if supported by one or more of the following preconditions:
    - Good macroeconomic fundamentals.
    - Strong institutions (government effectiveness).
    - Pre-existing outflow controls are already fairly comprehensive (a high outflow control index at the time of tightening).
  - When none of these three conditions is fulfilled, a tightening of restrictions is ineffective.
- Robustness: Results for macro fundamentals are robust to various specification changes; results for government effectiveness are positive but slightly less clear-cut and robust.

### Candidate conditions and rationale
- Candidate conditions tested via interaction terms include:
  - Institutional quality (e.g., government effectiveness).
  - Macroeconomic fundamentals (composite index and components such as GDP growth, inflation, fiscal balance, current account).
  - Tightness/intensity of pre-existing controls.
- Motivation: Historical cases (Malaysia, Iceland, Thailand, Ukraine) and literature suggesting initial conditions and institutional quality influence the success of capital control measures.

### Selected country experiences (Box 1 summaries)
- Thailand (May 1997):
  - Measures: Limits on forward transactions, export surrender requirements, requirements to convert baht proceeds onshore.
  - Outcomes: Exchange rate depreciation 50 percent six months after implementation; reserves fell 20 percent six months after implementation; interest rates continued to rise; restrictions abandoned in January 1998.
- Malaysia (September 1998):
  - Measures: Comprehensive restrictions including a 12-month waiting period for non-residents to convert ringgit proceeds from sale of Malaysian securities; closed legal channels for offshore trading of the ringgit.
  - Outcomes: Capital flight halted; reserves accumulated; exchange rate stabilized; interest rates reduced to aid economic revitalization.
- Iceland (November 2008):
  - Context: Collapse of three largest commercial banks with very large short-term foreign liabilities (600% of GDP).
  - Measures: Comprehensive outflow restrictions including bans on movement of capital with exceptions below ISK 10 million and bans on conversion of offshore krona to foreign exchange; foreign exchange transactions for external trade remained unrestricted.
  - Outcomes: Exchange rate stabilized shortly after restrictions; restrictions provided room for easing monetary policy; intent to lift restrictions gradually.
- Ukraine (2008-09):
  - Measures: Exchange restrictions, 5-day waiting period for non-residents to convert hryvnia proceeds from investment transactions.
  - Outcomes: Some temporary alleviation of exchange rate pressures, but failed to avoid massive central bank intervention; reserves plummeted 30 percent six months after tightening; policy slippages undermined effectiveness.

### Selected summary statistics (1995-2010) — values preserved exactly as reported
- Outflow control index: Mean 0.6, Standard Deviation 0.4, 10th percentile 0.0, 25th percentile 0.3, 50th percentile 0.8, 75th percentile 1.0, 90th percentile 1.0
- Net inflows/GDP: Mean 3.11, Standard Deviation 11.1, 10th percentile -5.0, 25th percentile -0.7, 50th percentile 1.4, 75th percentile 6.6, 90th percentile 12.8
- Gross inflows/GDP: Mean 10.1, Standard Deviation 19.3, 10th percentile 0.0, 25th percentile 1.1, 50th percentile 4.8, 75th percentile 12.3, 90th percentile 21.1
- Gross outflows/GDP: Mean 7.0, Standard Deviation 17.4, 10th percentile 0.0, 25th percentile 0.3, 50th percentile 2.0, 75th percentile 6.7, 90th percentile 12.8
- Net assets/GDP: Mean -5.4, Standard Deviation 18.6, 10th percentile -1.6, 25th percentile 0.1, 50th percentile 1.7, 75th percentile 5.9, 90th percentile 13.7
- Net liabilities/GDP: Mean 8.5, Standard Deviation 19.5, 10th percentile -1.0, 25th percentile 0.8, 50th percentile 4.1, 75th percentile 11.2, 90th percentile 20.0
- Macro fundamentals index: Mean 16.0, Standard Deviation 5.3, 10th percentile 9.0, 25th percentile 12.3, 50th percentile 16.3, 75th percentile 19.8, 90th percentile 22.5
  - GDP growth (y/y percent change): Mean 3.5, Standard Deviation 4.8, 10th percentile -2.7, 25th percentile 1.1, 50th percentile 4.2, 75th percentile 6.4, 90th percentile 8.9
  - Inflation (y/y percent change): Mean 6.5, Standard Deviation 6.3, 10th percentile 0.1, 25th percentile 2.3, 50th percentile 6.5, 75th percentile 8.9, 90th percentile 10.9
  - Current account surplus (as a share of GDP): Mean -0.1, Standard Deviation 1.9, 10th percentile -2.0, 25th percentile -1.1, 50th percentile -0.3, 75th percentile 0.7, 90th percentile 2.2
  - Fiscal surplus (as a share of GDP): Mean -2.4, Standard Deviation 4.4, 10th percentile -7.3, 25th percentile -4.8, 50th percentile -2.8, 75th percentile -0.5, 90th percentile 2.6
- Government effectiveness index: Mean 0.3, Standard Deviation 0.7, 10th percentile -0.6, 25th percentile -0.2, 50th percentile 0.1, 75th percentile 0.7, 90th percentile 1.2

### Empirical scope and limitations
- The paper does not study the impact of long-standing (persistent) restrictions on capital flows, nor does it assess the broader benefits and costs of capital flow management measures.
- The authors note potential inference complications in event studies because restrictions are often reactive to rising capital flows; they treat restrictions as endogenous via PVAR to address this concern.
- A caveat raised about prior literature: treating controls as exogenous likely biases effectiveness estimates downward because restrictions tend to be introduced in response to rising outflows.

*Source: _wp1408 - References (selected content excerpt).*

### 12.      A deterioration of

### _wp1408 - 12.      A deterioration of

### Episodes and descriptive findings
- Table 5 lists 24 periods of significant outflow control tightening in the sample.
- About two thirds of these are associated with deterioration in macroeconomic conditions and/or net capital outflows.
- Control-tightening episodes tend to cluster around crisis periods such as the Asian crisis, the Argentinean economic crisis and the global financial crisis.
- Figure 1: Number of Emerging Market Countries Tightening Capital Outflow Restrictions, 1996-2010 — spikes correspond to crisis episodes.

### Empirical strategy (baseline PVAR)
- Baseline model: a panel VAR (PVAR) that includes the capital control (outflow) index alongside controls such as a variety of capital flow variables; all variables treated as potentially endogenous.
- Endogenous variables in the benchmark specification: the outflow control index, industrial production, the interest rate, inflation, the local currency/USD exchange rate, and a variable capturing net capital flows as a share of GDP.
- Lag structure: number of lags = 1, ..., p (the study uses two lags in estimation).
- Structural form (ignoring constant): system of structural equations with shocks assumed uncorrelated.
- Alternative capital flow measures used in robustness checks: gross inflows, gross outflows, net flows of assets, net flows of liabilities (replacing net capital flow variable selectively).

### Identification and ordering
- Impulse response functions identified via Choleski ordering with contemporaneous effects allowed from capital flow restrictions to all variables except output.
- Baseline Choleski ordering:
  1) output (industrial production)
  2) capital outflow control index
  3) real exchange rate
  4) interest rate
  5) inflation
  6) net capital flows
- Rationale: restrictions can affect flows contemporaneously, but restrictions respond to flows with a lag.
- Alternative ordering (tested for robustness): capital flows contemporaneously affect restrictions and restrictions affect flows with a lag.

### Interaction terms and conditional analysis
- Augmented PVAR includes interaction terms to let dynamic responses vary with structural country characteristics.
- Structural characteristics matrix includes indicators of institutional quality (qi) and macroeconomic fundamentals (mf) and others (time-invariant country-specific averages).
- Coefficient matrix decomposed into baseline coefficients plus coefficients on interactions with each structural characteristic.
- Conditional impulse responses evaluated at the 25th and 75th percentiles of each indicator’s sample distribution.

### Main empirical results — average sample
- On average, no evidence that a tightening of outflow restrictions is effective in the full sample.
- Impulse responses show:
  - Gross outflows decline in response to a control tightening (intended effect).
  - Gross inflows also fall, resulting in an increase in net inflows (or a fall in net outflows) that is small and insignificant.
  - Residents repatriate investment; non-residents reduce investments in the domestic economy.
- Symmetry: results are symmetric — tightenings raise net inflows, loosenings lower net inflows; conditioning on tightenings in the initial period confirms findings.

### Conditional effectiveness — macroeconomic fundamentals
- Constructed a time-varying composite index of macroeconomic fundamentals by ranking countries along growth, inflation, fiscal balance, and current account balance; country’s fundamentals rank = simple average of four ranks.
- Sample distribution statistics (75th vs 25th percentiles of fundamentals):
  - Average growth rate: 6.4 (75th) and 1.1 (25th) percent
  - Average inflation rate: 8.9 (75th) and 2.3 (25th)
  - Average current account balance: 0.7 (75th) and -1.1 (25th) percent of GDP
  - Average fiscal balance: -0.5 (75th) and -4.8 (25th) percent of GDP
- Findings for strong fundamentals (75th percentile):
  - A control tightening increases net inflows at the peak by 4 percent of GDP, keeping them elevated for more than one and a half years.
  - Small and statistically insignificant decline in gross inflows; substantial fall in gross outflows dominates.
  - Real exchange rate appreciates; interest rate falls (not always statistically significant).
  - Industrial production shows a substantial boost.
- Interpretation: in countries with good fundamentals, restrictions can increase net inflows, halt exchange rate depreciation, allow for lower interest rates, and boost activity — possibly via investor sentiment and expectations.

### Conditional effectiveness — institutional quality
- Institutional quality measured using the World Bank’s government effectiveness index.
- Impulse responses evaluated at high and low levels of government effectiveness (75th and 25th percentiles; values 0.7 and -0.2 respectively).
- Findings:
  - Restrictions are effective at countering capital flight when institutional quality is high.
  - Net inflows increase by about 3 percent of GDP on impact and by about 2 percent thereafter in response to the shock (high government effectiveness).
  - Restrictions are ineffective when institutional quality is low.
  - Similar results obtained using alternative governance indicators (political stability, regulatory quality, control of corruption).
- Note: responses of interest rates and the real exchange rate are relatively muted compared with the macro fundamentals case.

### Robustness and additional notes
- Robustness checks include alternative lag structures, alternative Choleski orderings, alternative specifications replacing net inflows with gross flows and net assets/liabilities.
- The analysis uses industrial production (not GDP) because quarterly GDP unavailable for some countries.
- The OCI index interpretation: in the sample the index mean = 0.63, sd = 0.35; shock = 0.25.

*Source: _wp1408 - 12.      A deterioration of*

### 23.      Further, the results suggest that further restrictions are effective in countries

### 23.      Further, the results suggest that further restrictions are effective in countries

### Findings on effectiveness of outflow restrictions
- A tightening of outflow restrictions can be effective if supported by either:
  - strong macroeconomic fundamentals, or
  - good institutions, or
  - existing restrictions that are already fairly comprehensive.
- When estimating the PVAR without interaction terms:
  - a tightening of outflow restrictions on average reduces gross capital outflows;
  - there is also a contraction in gross inflows (mainly driven by non-resident investors) that is larger in size than the contraction in outflows;
  - for the average country, a tightening is ineffective as net outflows increase as a result.
- Adding interaction terms with structural country characteristics reveals effectiveness conditional on the indicators listed above.

### Impulse response magnitudes and dynamics
- Impulse responses evaluated at different starting levels of the capital control index:
  - high level: 75th percentile of outflow control intensity (value of 1);
  - low level: 25th percentile of outflow control intensity (value of 0.3).
- Where pre-existing restrictions are fairly comprehensive:
  - net inflows increase significantly by about 5 percent of GDP on impact;
  - net inflows increase by 2.5 percent for the following 6 quarters.
  - the tightening leads to a larger decline in gross outflows than in gross inflows.
  - the exchange rate surprisingly depreciates in response to the tightening (opposite of expected effect).
  - interest rates show a fairly muted response.
- Where restrictions are tightened but do not attain a minimum level of comprehensiveness:
  - a control tightening shows little effect on gross or net flows.
- In some robustness-ordering variants (allowing outflow restrictions to respond contemporaneously to capital flows):
  - the response can be significantly more muted: no longer positive on impact and peaks at only about 0.5 percent of GDP.
- When changing ordering but still positive across horizon, the response can peak at about 1.2 percent of GDP.

### Robustness checks and sensitivity
- Robustness dimensions tested:
  - substitution of industrial production by real GDP (quarterly GDP extrapolated where necessary; for some countries GDP assumed equal in every quarter of a given year where quarterly data not available);
  - augmenting PVAR (6) to PVAR (7) by including an inflow control index;
  - different lag specifications (PVAR (6), PVAR (7); lag variations including 1 lag, 3 lags, 4 lags);
  - different orderings of variables in the VAR (including allowing outflow restrictions to respond contemporaneously to capital flows).
- Key robustness outcomes:
  - The finding that restrictions are effective in countries with strong fundamentals is robust to all checks. Only when reducing lag structure to 1 is there a somewhat pronounced drop in magnitude.
  - The response of net flows to restrictions in countries with strong institutions is robust to most checks; however, changing the ordering to allow contemporaneous response of outflow restrictions can make the impact significantly more muted (impact no longer positive; peak about 0.5 percent of GDP).
  - The finding that outflow restrictions need to be comprehensive to be effective is relatively robust:
    - substitution of industrial production with GDP and inclusion of inflow control index do not notably impact results;
    - increasing number of lags affects path after about three quarters but response remains positive across the six-quarter horizon in all cases;
    - changing ordering to allow outflow restrictions to be affected contemporaneously by all variables reduces the magnitude but response remains positive and can peak at about 1.2 percent of GDP.
  - Results are robust to excluding China and India from the analysis.

### Methodology and scope
- Sample: 37 emerging market economies during 1995-2010.
- Method: panel vector autoregression (PVAR) methodology allowing for endogeneity of capital flow restrictions.
- Interaction terms computed as products between right-hand-side variables and structural characteristics (indicators of institutional quality, macroeconomic fundamentals, and pre-existing restriction tightness) to compute conditional impulse response functions evaluated at different percentiles.

### Conclusion and policy implications
- Main criterion for effectiveness: impact of tightening outflow restrictions on net capital flows; additional criteria include exchange rate and interest stabilization and regaining monetary policy independence.
- Policy-relevant conclusions:
  - Tightening outflow restrictions can reduce net capital outflows when accompanied by good macroeconomic fundamentals, strong institutions, or when pre-existing restrictions are comprehensive enough to slow a “rush to the exit.”
  - For the average country without these preconditions, tightening outflow restrictions can be ineffective or counterproductive for net inflows because gross inflows may contract by more than gross outflows.
  - Policymakers should consider country-specific fundamentals, institutional quality, and the comprehensiveness of existing controls before relying on outflow restrictions as a stabilization tool.

*Source: IMF working paper text (excerpt).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1408.pdf_