## 1. Capital Account Liberalization

## Source details

**Canonical URL:** [1. Capital Account Liberalization](https://www.imf.org/-/media/files/publications/wp/2018/wp1883.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2018/wp1883.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2018/wp1883.pdf.json)

---

### Introduction and motivation
- Purpose: reassess aggregate (output) and distributional (inequality and labor share) effects of policies to liberalize international capital flows (financial globalization).
- Two motivations:
  - Efficiency/output benefits claimed for capital account liberalization have often proven elusive in empirical studies.
  - Distributional impacts of financial globalization have received less scrutiny despite trade’s known winners and losers.
- Theoretical backdrop:
  - Neoclassical model predicts resource flows from low-return to high-return countries, reducing cost of capital in recipients, triggering temporary investment and growth increases and a permanent level effect on output.
  - Since 1980 about 150 episodes of surges in capital inflows in more than 50 emerging market economies; about 20% of these episodes ended in a financial crisis, and many associated with large output declines.
- Prior literature: ambiguous or limited growth benefits; only a small fraction of studies find robust positive effects.

### Contributions of this paper
- Identify large and discrete changes in capital account policy that may cause aggregate and distributional effects.
- Trace evolution of output and inequality after these changes.
- Identify factors shaping macro and distributional impacts.
- Provide causal evidence using an IV approach and industry-level data.
- Employ a difference-in-difference identification strategy à la Rajan and Zingales (1998), exploiting three industry-level channels:
  - External financial dependence.
  - Natural layoff rates (propensity to use layoffs as adjustment).
  - Elasticity of substitution between capital and labor.

### Empirical strategy and identification
- Three-dimensional panel: j industries × i countries × t time periods.
- Fixed effects: country–time, country–industry, and industry–time to absorb unobserved macro shocks, cross-country sectoral differences, and common industry shocks.
- Main independent variables: interactions between capital account liberalization episodes and industry-specific factors to reduce reverse causality concerns.
- Causal strategies:
  - Instrumental variables (IV) approach with two instruments: (i) four-year lagged capital account openness (scope of reform) and (ii) weighted-average of reforms in other countries (peer pressure), weights = bilateral trade shares.
  - Industry-level difference-in-difference with rich fixed effects (country–time, country–industry, industry–time).

### Country-level findings (aggregate and distributional)
- On average, capital account liberalization has had a limited effect on output.
- Capital account liberalization has led to an economically and statistically significant increase in inequality.
- Dynamic response (medium term, five years after major liberalization):
  - Output: no significant impact detected.
  - Inequality: sizeable increase of about 4 percentage points five years after liberalization (corresponds to about one standard deviation of the average increase in the Gini in the sample).
- Heterogeneity:
  - Liberalization increases output in countries with high financial depth.
  - Inequality effects are magnified in countries with lower levels of financial depth and inclusion.
  - Liberalization episodes followed by financial crises lead to significant output contractions and increases in inequality; adverse effects are greatly reduced when not followed by crises.

### IV first-stage evidence (preserved coefficients and statistics)
- First-stage regression reported:
  - D_i,t = 0.239 I_i,t + 0.105 I_i,t−1 − 0.010 Kaopene_i,t−4
    - t-statistics in parentheses: (5.62) (2.77) (−6.28).
- Correlation between bilateral trade and capital flow linkages (where available): about 0.7.
- IV results similar to baseline; instruments judged strongly exogenous by Kleibergen–Paap rk Wald F statistic and Hansen J p-value reported in source.

### Industry-level findings and causal interpretation
- Industry-level evidence corroborates country-level findings and supports causal interpretation.
- Output:
  - Output gains associated with liberalization are small and short-lived.
  - Differential medium-term output gain between an industry at the 25th percentile of external financial dependence and the 75th percentile is about 1%.
  - This differential is statistically significant at the 10% confidence level for up to 2 years after liberalization, but vanishes in the medium term.
- Distribution (labor share):
  - Distributional effects (labor share) are economically and statistically significant and long-lasting.
  - Liberalization episodes reduce the share of labor income particularly in industries with:
    - Higher external financial dependence.
    - Higher natural propensity to use layoffs to adjust to idiosyncratic shocks.
    - Higher elasticity of substitution between capital and labor.

### Industry-level channel estimates (medium term: five years after reform)
- External financial dependence:
  - Differential medium-term reduction in labor share between 75th-percentile and 25th-percentile: about 2 percentage points.
  - Conservative average reduction through this channel: about 1¾ percentage points.
- Natural layoff rate:
  - Differential medium-term reduction between 75th-percentile and 25th-percentile: about 2 percentage points.
  - Conservative average reduction through this channel: about 1½ percentage points.
- Elasticity of substitution between capital and labor:
  - Differential medium-term reduction between 75th-percentile and 25th-percentile: about 2.5 percentage points.
  - Effect significant only when elasticity > 1.
  - Using medium-term estimates (−3.3 for elasticity above 1; 0 for below), average reduction through this channel: about 2½ percentage points.

### Data — country-level
- Capital account liberalization measure:
  - De jure indicator from Chinn and Ito (2008) (Kaopen).
  - Coverage: unbalanced panel of 182 countries from 1970 to 2010.
  - Index range: −1.9 (more restricted) to 2.5 (less restricted).
- Identification of liberalization episodes:
  - Episodes defined as large changes in Kaopen: changes that exceed by two standard deviations the average annual change over all observations (i.e. 0.76).
  - This criterion identifies 224 episodes, the majority in the last two decades.
- Output and inequality data:
  - Real GDP growth from IMF WEO database.
  - Gini coefficients from the Standardized World Income Inequality Database (SWIID).
    - SWIID provides comparable estimates for 173 countries for as many years as possible from 1970 to 2010.
    - Gini coefficients in sample range from 18 to 78.
- Country-level sample used for baseline OLS: annual observations 1970 to 2010 for 149 advanced and developing economies.
- Number of lags in baseline: 5.

### Data — industry-level
- Industry-level labor shares computed from harmonized value added and labor compensation in EUKLEMS 2012 release (O’Mahony and Timmer 2009).
- Industry-specific factors:
  - External financial dependence measured as median across firms in industry of (total capital expenditures − current cash flow) / total capital expenditures (data from Hui Tong / Tong and Wei (2011)).
  - Natural layoff rates from Bassanini, Nunziata and Venn (2009), based on 2004 CPS Displaced Workers Supplement (available for 22 industries ISIC Rev. 3; mapped to ISIC Rev. 4 resulting in layoff data for 21 of the 31 industries).
  - Elasticity of substitution between capital and labor estimated using production function assumptions and industry-level TFP, capital–output ratios, and price data (labor demand elasticity μ set to −0.39 from Hamermesh (1993) average).
- Industry-level sample: unbalanced panel of 23 advanced economies and 25 industries over 1975–2010.

### Empirical methodology — overview and estimation equations
- Country-level baseline (Equation (7)):
  - g_it = a_i + γ_t + Σ_{k=0}^l δ_k D_{i,t−k} + Σ_{k=0}^l Σ_{l} ϖ_k X_{i,t−k} + ε_it.
  - g: annual change in log output (or Gini); D: dummy for start of liberalization episode; controls X include trade, current account, product market, labor market reforms, employment protection legislation.
  - Estimation: OLS with clustered (country) robust standard errors; impulse response functions for up to five years.
- Threshold specification (Equation (8)):
  - Smooth transition function G(z_it) = exp(−γ z_it) / (1 + exp(−γ z_it)), γ > 0, where z is normalized indicator of financial development.
- Industry-level specification (Equation (9)):
  - g_jit = a_ij + γ_it + ρ_jt + Σ_{k=0}^l δ_k S_j D_{i,t−k} + ε_jit,
  - a_ij country–industry fixed effects; γ_it country–year fixed effects; ρ_jt industry–time fixed effects; S_j sector-specific channel measures.
- Fixed effects rationale: isolate within-industry effects and mitigate confounding from comparative advantage, country macro shocks, and common industry trends.

### Robustness checks (selected)
- Results robust to inclusion of additional macro and sectoral controls interacted with industry channels:
  - Trade liberalization, current account liberalization, domestic financial liberalization, product market deregulation, Employment Protection Legislation (EPL), union density, technology (relative price of investment), and sectoral growth.
- Controlling for political orientation, changes in redistributive policies, level and square of log GDP per capita, changes in government expenditures/GDP, and changes in industry and agriculture shares in value added does not alter main country-level results.
- Controlling for past growth and expected growth (t−1 IMF WEO forecasts) yields results very close to baseline.
- Inclusion of two lags of dependent variable yields similar results.
- Union density found to have a positive and statistically significant effect on the labor share through the natural layoff rate; union density in advanced economies has, on average, declined following major capital account liberalization episodes (Annex figure referenced).

### Key quantitative results (selected preserved figures and table coefficients)
- Aggregate medium-term (five years after reform) effects (Table 1, estimates based on equation (7)):
  - Medium-term effect on Output: 0.665
  - Medium-term effect on Gini: 4.018***
  - N: Output 2,001; Gini 1,789
  - R2: Output 0.38; Gini 0.13
- Threshold effects (Table 2, equation (8)):
  - Medium-term effect*G(Z): −2.558 (1.95) for Output in column I; 4.341** (4.47) for Gini in column IV; other Gini coefficients: 3.959** (4.34) and 4.288** *(10.21).
  - Medium-term effect*[1−G(Z)]: 3.924* (3.03) in Output column I.
  - N: 2,001 (columns I–III); 1,789 (columns IV–VI). R2: 0.38 and 0.13 accordingly.
- Sectoral results (Table 3 and Table 4, equation (9), N: 16,616):
  - Capital account reform it*Sj (t=0) on Output: 1.802** (1.95)
  - Medium-term differential effect on Output: 0.520 (F-test 0.21)
  - Labor share short-run coefficients at t=0:
    - External Financial Dependence: −1.835*** (−2.82)
    - Layoff Rate: −0.023 (−0.15)
    - Elasticity of Substitution: −0.208*** (−3.59)
  - Medium-term differential effects on labor share:
    - External Financial Dependence: −2.230***
    - Layoff Rate: −2.078*
    - Elasticity of Substitution: −2.580*

### Policy-relevant implications and recommendations
- Aggregate benefits of capital account liberalization are limited on average; distributional costs (higher inequality, lower labor share) are significant and persistent in many settings.
- Distributional and macro effects depend crucially on:
  - Country-level institutions (financial depth and inclusion).
  - Whether liberalization is followed by financial crises.
- Policy design considerations:
  - Strengthen financial institutions and inclusion to enhance potential output benefits and mitigate inequality effects.
  - Implement prudential and macroeconomic safeguards to reduce the probability and cost of crisis episodes following liberalization.
  - Consider restricting certain types of flows that generate adverse equity–efficiency trade-offs (e.g., carry-trade flows or flows that give rise to unhealthy asset price or credit booms).
  - Encourage flows that give durable increases in investment and growth (such as greenfield investments).
  - Use fiscal redistribution to mitigate adverse distributional consequences (acknowledged limited effect on efficiency unless extreme).
  - Pre-distribution policies: increased spending on education and training to foster greater equality of opportunity.
- Interpretation caveat: findings do not imply countries should not liberalize, but suggest caution and careful design given weak average efficiency gains and notable distributional consequences.

*Source: IMF Working Paper (content unit: "1. Capital Account Liberalization", wp1883).*

### 1. Capital Account Liberalization ___________________________________8

### 1. Capital Account Liberalization

### Introduction and motivation
- Purpose: reassess aggregate (output) and distributional (inequality and labor share) effects of policies to liberalize international capital flows (financial globalization).
- Two motivations:
  - Efficiency/output benefits claimed for capital account liberalization have often proven elusive in empirical studies.
  - Distributional impacts of financial globalization have received less scrutiny despite trade’s known winners and losers.
- Theoretical backdrop:
  - Neoclassical model predicts resource flows from low-return to high-return countries, reducing cost of capital in recipients, triggering temporary investment and growth increases and a permanent level effect on output.
  - Risks: since 1980 about 150 episodes of surges in capital inflows in more than 50 emerging market economies; about 20% of these episodes ended in a financial crisis, and many associated with large output declines.
- Prior literature finds ambiguous or limited growth benefits from liberalization (surveys and studies cited): only a small fraction of studies find robust positive effects.

### Contributions of this paper
- Identify large and discrete changes in capital account policy that may cause aggregate and distributional effects.
- Trace evolution of output and inequality after these changes.
- Identify factors shaping macro and distributional impacts.
- Provide causal evidence using an IV approach and industry-level data.
- Employ a difference-in-difference identification strategy à la Rajan and Zingales (1998), exploiting three industry-level channels:
  - External financial dependence.
  - Natural layoff rates (propensity to use layoffs as adjustment).
  - Elasticity of substitution between capital and labor.

### Empirical strategy and advantages of industry-level approach
- Use three-dimensional data (j industries, i countries, t time periods).
- Controls include country–time, country–industry, and industry–time fixed effects.
  - Country–time fixed effect absorbs unobserved macro shocks affecting countries’ output and income distribution.
  - Mitigates reverse causality concerns: more plausible that liberalization affects cross-industry differences than vice versa.
- Main independent variables are interactions between capital account liberalization and industry-specific factors, further reducing reverse causality plausibility.

### Main country-level findings (aggregate and distributional)
- On average, capital account liberalization has had a limited effect on output.
- Capital account liberalization has led to an economically and statistically significant increase in inequality.
- Heterogeneity:
  - Effects vary with strength of financial institutions and timing relative to crises.
  - Liberalization increases output in countries with high financial depth.
  - Inequality effects are magnified in countries with lower levels of financial depth and inclusion.
  - Liberalization episodes followed by financial crises lead to significant output contractions and increases in inequality; adverse effects are greatly reduced when not followed by crises.

### Main industry-level findings and causal interpretation
- Industry-level evidence corroborates country-level findings and supports causal interpretation.
- Output gains associated with liberalization are small and short-lived.
- Distributional effects (labor share) are economically and statistically significant and long-lasting.
- Specific industry-level results:
  - Liberalization episodes reduce the share of labor income particularly in industries with:
    - Higher external financial dependence.
    - Higher natural propensity to use layoffs to adjust to idiosyncratic shocks.
    - Higher elasticity of substitution between capital and labor.

### Data: country-level
- Capital account liberalization measure:
  - De jure indicator from Chinn and Ito (2008) (Kaopen).
  - Coverage: unbalanced panel of 182 countries from 1970 to 2010.
  - Index range: -1.9 (more restricted) to 2.5 (less restricted).
- Identification of liberalization episodes:
  - Episodes defined as large changes in Kaopen: changes that exceed by two standard deviations the average annual change over all observations (i.e. 0.76).
  - This criterion identifies 224 episodes, the majority in the last two decades.
- Output and inequality data:
  - Real GDP growth from IMF WEO database.
  - Gini coefficients from the Standardized World Income Inequality Database (SWIID).
    - SWIID provides comparable estimates for 173 countries for as many years as possible from 1970 to 2010.
    - Gini coefficients in sample range from 18 to 78.

### Data: industry-level
- Industry-level labor shares computed from harmonized value added and labor compensation in EUKLEMS 2012 release (O’Mahony and Timmer 2009).
- Industry-specific factors used to identify channels:
  - External financial dependence.
  - Natural layoff rates.
  - Elasticity of substitution between capital and labor.
- These industry-specific factors are weakly correlated among each other (see Table A4), supporting the independence of identification channels.

### Empirical methodology (overview)
- Country-level approach:
  - Baseline specifications and methods to address differential and threshold effects are discussed in the paper.
- Industry-level approach:
  - Difference-in-difference framework with rich fixed effects structure (country–time, country–industry, industry–time).
- Causal identification:
  - Instrumental variables (IV) approach and industry-level heterogeneity exploited to strengthen causal claims.

### Policy-relevant implications (as emphasized in text)
- Aggregate benefits of capital account liberalization are limited on average; distributional costs (higher inequality, lower labor share) are significant and persistent in many settings.
- The distributional and macro effects depend crucially on country-level institutions (financial depth and inclusion) and on whether liberalization is followed by financial crises.
- Policy design should consider:
  - Strengthening financial institutions and inclusion to enhance potential output benefits and mitigate inequality effects.
  - Prudential and macroeconomic safeguards to reduce the probability and cost of crisis episodes following liberalization.

*Source: IMF Working Paper (content unit: "1. Capital Account Liberalization")*

### 1. External Financial Dependence

### 1. External Financial Dependence

### External financial dependence — definition and measurement
- External financial dependence for each industry is measured as the median across firms in the industry of:
  - (total capital expenditures − current cash flow) / total capital expenditures.
- Figure A1 (in source) presents industry-specific measures of external financial dependence.
- Data source note: Hui Tong kindly provided the data (see Tong and Wei (2011) in source).

### Natural layoff rate — data and mapping
- Layoff rates used as a proxy for job destruction are from Bassanini, Nunziata and Venn (2009), computed using US layoff rates data from the 2004 CPS Displaced Workers Supplement.
- US layoff rate data are available for 22 industries categorized according to the ISIC Rev. 3 classification.
- The EU KLEMS database follows ISIC Rev. 4; the authors map ISIC Rev. 3 layoff rates to ISIC Rev. 4 using the many-to-one method of O’Mahony and Timmer (2009).
- Post and telecommunications were a single industry under ISIC Rev. 3 and two industries under ISIC Rev. 4; the same layoff rate is imposed for postal and courier and for telecommunications.
- After matching, layoff data are available for 21 of the 31 industries in the sample.
- Figure A2 (in source) shows the US layoff rates by industry.

### Elasticity of substitution between capital and labor — theory and estimation
- Production assumed to follow a multiplicative production function with capital, labor, and labor- and capital-augmenting technical progress.
- Relationship used (Bentolila and Saint-Paul 2003): ds_i/dk_i = −(1 + σ_KL.i) / ( (k_i / μ_i) )  (Equation (1) in source).
  - s: labor share of value added.
  - σ_KL: elasticity of substitution between labor and capital (holding input prices constant).
  - k: capital to value added ratio.
  - μ: elasticity of labor demand with respect to wages (holding capital and real price of inputs constant).
- Labor share modeled multiplicatively (Equation (2) in source):
  - s_ijt = A_ijt^β0 (k_ijt)^β1i ((q_ijt / p_ijt))^β2i ϑ_ijt^β3, with β3 = 1.
- Log-linear specification (Equation (3) in source):
  - ln s_ijt = β0 ln A_ijt + β1i d_i ln k_ijt + β2i d_i ln(q_ijt / p_ijt) + ϑ_ijt.
- Error term decomposition (Equation (4) in source):
  - ϑ_ijt = γ_it + δ_jt + θ_it + ε_ijt, with γ industry–time, δ country–time, θ country–industry effects.
- Estimation equation (Equation (5) in source) includes TFP proxy (Â) and the fixed effects above.
- Data used:
  - Labor share from EUKLEMS (labor compensation and value-added).
  - Capital–output ratio from OECD STAN (gross capital stock and value added, volumes).
  - Real price of inputs from OECD STAN (price deflators of intermediate inputs and gross output).
  - TFP from EUKLEMS.
- Elasticities of substitution derived via rearranged expression (Equation (6) in source):
  - −σ_KL.i = 1 + ( (ds_i/dk_i) * k_i / μ_i ).
- Labor demand elasticity with respect to wages (μ) assumed constant across industries and set to the average estimate from Hamermesh (1993): −0.39.
- Figure A3 (in source) presents industry-specific estimates of the elasticities of substitution.

### Empirical methodology — overview
- Two broad approaches:
  - Country-level (aggregate) analysis.
  - Industry-level (three-way industry–country–time panel) analysis.
- Identification at industry level relies on three sectoral channels through which capital account liberalization may affect outcomes:
  1. Dependence on external finance — for output and the labor share.
  2. Natural layoff rate (job turnover) — for the labor share.
  3. Elasticity of substitution between capital and labor — for the labor share.

### Country-level approach — baseline specification and estimation
- Baseline regression (Equation (7) in source):
  - g_it = a_i + γ_t + Σ_{k=0}^l δ_k D_{i,t−k} + Σ_{k=0}^l Σ_{l} ϖ_k X_{i,t−k} + ε_it,
  - where g is annual change in log output (or Gini); D is a dummy equal to one at start of a capital account liberalization episode; a_i country fixed effects; γ_t time fixed effects; X vector of controls (trade, current account, product market, labor market reforms including employment protection legislation).
- Estimation details:
  - OLS on an unbalanced panel of annual observations from 1970 to 2010 for 149 advanced and developing economies.
  - Number of lags chosen to capture medium effect of reforms: 5.
  - Impulse response functions (IRFs) produced for output and inequality following liberalization; confidence bands based on clustered (country) robust standard errors.

### Country-level approach — accounting for differential and threshold effects
- Hypothesis: benefits of financial globalization depend on the quality of financial institutions (thresholds of financial development).
- Interaction and smooth transition specification (Equation (8) in source) uses G(z_it) = exp(−γ z_it) / (1 + exp(−γ z_it)), γ > 0, where z is normalized indicator of financial development (zero mean, unit variance).
- G(z) takes value 1 (0) when z → −∞ (+∞); also test replacing G(z) with a crisis dummy.
- Use of STAR-like smooth transition function enables use of more observations and easier clustering at country level.

### Industry-level approach — specification and sample
- Industry-level regression (Equation (9) in source):
  - g_jit = a_ij + γ_it + ρ_jt + Σ_{k=0}^l δ_k S_j D_{i,t−k} + ε_jit,
  - where a_ij country–industry fixed effects; γ_it country–year fixed effects; ρ_jt industry–time fixed effects; S sector-specific channels discussed above.
- Estimated for an unbalanced panel of 23 advanced economies and 25 industries over the period 1975–2010.

### Identification and fixed effects rationale
- Country–industry fixed effects control for industry-specific factors and cross-country differences in sectoral growth (comparative advantage).
- Country–year fixed effects control for reforms and macro shocks common to a country’s sectors.
- Industry–time fixed effects control for common factors affecting specific industries across countries (sectoral reallocation drivers).
- The specification isolates within-industry effects and does not capture between-industry composition effects (noting within-industry changes are argued to be more important for labor share movements).

### Results — country-level analysis (aggregate findings)
- Dynamic response (Figure 1 in source) over five years after major capital account liberalization:
  - Output: no significant impact detected.
  - Inequality: sizeable and statistically significant increase of about 4 percentage points five years after liberalization.
    - This increase corresponds to about one standard deviation of the average increase in the Gini coefficient in the sample.
- Robustness checks and addressing endogeneity:
  - Expanded controls added: political orientation (left/center/right), changes in redistributive policies (proxy: difference between gross and net Gini), level and square of log GDP per capita, changes in government expenditures/GDP, changes in industry and agriculture shares in value added. Inclusion of these does not affect results (Figure 2 in source).
  - Controlled for past growth and expected growth (t−1 IMF WEO forecasts) over horizon t to t+5; results very close to baseline (Figure 3 in source).
  - Inclusion of two lags of dependent variable yields similar results (Figure 4 in source). Footnote: looking at Gini before/after liberalization suggests Gini broadly stable up to 5 years before liberalization, and increased by about 0.8 percentage point five years after (see Furceri and Loungani, 2018).
- Instrumental variables (IV) approach to address endogeneity:
  - Two instruments:
    1. Scope of reform proxied by four-year lagged value of capital account openness indicator (initial stance).
    2. Peer pressure proxied by a weighted-average of current and lagged capital account liberalization episodes in other countries, weights = strength of trade linkages (share of exports + imports with partner j in total exports + imports of i).
  - Rationale: lower lagged openness → more scope to reform; reforms in partners induce peer pressure.
  - For observations with bilateral capital flow data, correlation between bilateral trade and capital flow linkages is about 0.7 and statistically significant.
  - Instrument construction formula (Equation (10) in source): I_i,t = Σ_{j=1,..,n; j≠i} D_j,t w_i,j,t, with w_i,j,t = (Export_i,j,t + Import_i,j,t) / (Export_i,t + Import_i,t).
  - First-stage IV estimates (reported in source):
    - D_i,t = 0.239 I_i,t + 0.105 I_i,t−1 − 0.010 Kaopene_i,t−4
      - t-statistics in parentheses: (5.62) (2.77) (−6.28).
  - Kleibergen–Paap rk Wald F statistic and Hansen J p-value for overidentification indicate instruments can be considered strongly exogenous; instrument effects on output are not statistically significant once controlled for liberalization episodes.
  - IV results similar to baseline (Figure 5 in source).

### Results — liberalization vs. capital flows (heterogeneity)
- Focus on de jure measures of capital account liberalization to isolate policy changes.
- Re-estimation by interacting liberalization episodes with extent of change in capital flows over five years after liberalization (same horizon as IRFs):
  - Output effects remain not statistically significant regardless of flow size.
  - Impact on inequality is much stronger and statistically significant in cases with higher flows (Figure 6 in source).

### Key numeric values and sample details (preserved from source)
- US layoff rate data available for 22 industries (ISIC Rev. 3).
- After mapping, layoff data available for 21 of the 31 industries in the sample.
- Labor demand elasticity (μ) used: −0.39 (Hamermesh (1993) average across 70 studies).
- Country-level sample: annual observations 1970 to 2010 for 149 advanced and developing economies.
- Number of lags for baseline: 5.
- Industry-level sample: unbalanced panel of 23 advanced economies and 25 industries over 1975–2010.
- Estimated inequality increase following liberalization: about 4 percentage points five years after reform.
- Footnote finding: Gini increased by about 0.8 percentage point five years after liberalization (Furceri and Loungani, 2018).
- Correlation between bilateral trade and capital flow linkages (where available): about 0.7.
- First-stage IV regression coefficients (with t-statistics in parentheses): D_i,t = 0.239 I_i,t (5.62) + 0.105 I_i,t−1 (2.77) − 0.010 Kaopene_i,t−4 (−6.28).

*Italic: Source — IMF Working Paper "1. External Financial Dependence" (wp1883 chapter content provided).*

### 3. Threshold Effects

### Threshold Effects

### Interaction with Domestic Financial Institutions
- The strength of financial institutions influences both output gains from financial globalization and the distributional effects of capital account liberalization.
- Findings from Figure 7 and Table 2:
  - Positive output effects when the domestic financial market is highly liberalized.
  - Negative (but not significant) output effects when the domestic financial market remains largely restricted.
  - Output effects are positive (but not statistically significant) for liberalizations not followed by a crisis, and sharply negative when a crisis follows liberalization.
  - The effect of capital account liberalization on inequality is magnified in:
    - countries with largely restricted domestic financial markets,
    - countries with limited financial inclusion, and
    - episodes followed by a crisis.

### Industry-Level Analysis — Overview
- Industry-level results corroborate country-level evidence.
- While output gains associated with capital account liberalization are small and short-lived, distributional effects (effects on the labor share of income) are economically and statistically significant and long-lasting.

### Industry Heterogeneity in Output Effects
- Differential medium-term output gain associated with liberalizing the capital account between:
  - an industry at the 25th percentile of external financial dependence (e.g., transport equipment), and
  - an industry at the 75th percentile (e.g., construction),
  - is about 1%.
- Statistical significance:
  - This differential effect is statistically significant at the 10% confidence level for up to 2 years after liberalization, but it vanishes in the medium term.
- Estimates for labor productivity and employment are imprecise, with point estimates suggesting:
  - a positive differential effect for productivity,
  - a negative differential effect for employment.

### Differential Effects on Industry Labor Share (Three Identification Strategies)
- Identification strategies rely on industry heterogeneity in:
  - (i) external financial dependence;
  - (ii) the natural layoff rate;
  - (iii) the elasticity of substitution between capital and labor.

- Panel A (External Financial Dependence):
  - Over the medium term (five years after reform), capital account liberalization tends to reduce the labor share in industries with higher external financial dependence.
  - Differential medium-term reduction in labor share between 75th-percentile external dependence and 25th-percentile is about 2 percentage points.
  - Under the conservative assumption that liberalization had no impact on sectors at the 25th percentile, the results suggest capital account liberalization episodes have on average reduced, through the external dependence channel, the labor share in a reform country by about 1¾ percentage points.

- Panel B (Natural Layoff Rate):
  - Capital account liberalization tends to reduce the labor share in sectors with a higher natural layoff rate.
  - Differential medium-term reduction between 75th-percentile layoff rate (e.g., textiles) and 25th-percentile (e.g., chemicals) is about 2 percentage points.
  - Under similar assumptions as above, the medium-term reduction in the labor share in a reform country through the natural layoff rate is about 1½ percentage points.

- Panel C (Elasticity of Substitution between Capital and Labor):
  - The effect on the labor share is larger (in absolute value) in industries with higher elasticity of substitution.
  - Differential medium-term reduction between 75th-percentile elasticity (e.g., machinery and equipment) and 25th-percentile (e.g., accommodation) is about 2.5 percentage points.
  - The effect is only significant in industries with elasticity of substitution between capital and labor greater than 1 (Panels C1 and C2).
  - Using medium-term estimates of the differential effect in these figures—that is, -3.3  (0) for industries with elasticity of substitution above (below) 1—the results suggest that capital account liberalization episodes have on average reduced the labor share in a reform country by about 2½ percentage points.

### Computation Note
- Footnote calculation (as presented in the source):
  - This effect is computed as ∑ 훿5 퐼푖 푤푖푖, where 훿5 is the medium-term coefficient estimates in equation (9), 퐼푖 is an indicator variable which takes value 1 for industries with a level of external financial dependence above the 25th percentile of distribution, and 푤푖 is average (across countries and time) share of value added in total value added for industry i.

*Source: wp1883 - 3. Threshold Effects*

### 1. Robustness Checks

### 1. Robustness Checks

### Robustness of Equation (9) to Additional Macroeconomic Controls
- Concern: estimates of Equation (9) may be biased by omitted macroeconomic variables that affect output and the labor share through dependence on external finance (or natural layoff rates or elasticity of substitution) and are correlated with capital account liberalization episodes.
- Trade liberalization:
  - Major trade liberalization in many countries occurred in the 1970s, often before capital account liberalization in the 1990s.
  - Re-estimating Equation (9) with interaction between the index of external finance (natural layoff, elasticity of substitutions) and trade reforms yields results similar to the baseline specification (Figure 10).
- Current account liberalization:
  - Current and capital account liberalization have proceeded in parallel (Quinn and Toyoda 2008).
  - Results are robust to inclusion of current account liberalization interacted with industry-specific channels (Figure 11).
- Domestic financial liberalization:
  - Domestic financial liberalization increases financial depth and may affect output through sectoral external financial dependence.
  - Augmenting Equation (9) with interaction between domestic financial liberalization and dependence on external finance produces results similar to baseline (Figure 12).
- Product market deregulation:
  - Deregulation affects output and the labor share, particularly privatization of large public network monopolies.
  - Adding interaction between external finance indices and product market regulation does not change results (Figure 13).
- Employment Protection Legislation (EPL):
  - EPL may affect the labor share via layoff rates and elasticity of substitution.
  - Accounting for differential effect of EPL on the labor share yields results similar to baseline (Figure 14).
- Union density:
  - Union density can influence labor bargaining power and be correlated with capital account liberalization.
  - Adding interaction between external finance indices and union density does not change results (Figure 15).
  - Estimated finding: union density has a positive and statistically significant effect on the labor share through the natural layoff rate.
- Technology (relative price of investment):
  - Technology, proxied by the relative price of investment, can reduce the cost of capital, increasing output and reducing the labor share in industries with elasticity of substitution greater than one.
  - Including interaction between external financial dependence indicators and the relative price of investment does not change results (Figure 16).
- Sectoral growth (control for cyclical labor share):
  - Labor share is typically countercyclical (Kehrig and Vincent 2017).
  - Controlling for sectoral growth: results remain statistically significant and close to baseline (Figure 17).
- Annex result on union density:
  - Union density in advanced economies has, on average, declined following major capital account liberalization episodes (see Figure C in the Annex).

### Main Conclusions of the Paper (Section V)
- Scope: uses aggregate and sector-level data to re-examine effects of capital account liberalization policies on output and inequality and how these depend on strength of financial institutions.
- Aggregate and distributional summary findings:
  - Capital account liberalization reforms on average have led to limited output gains but contributed to significant increases in inequality.
  - Heterogeneity: output increases in countries with well-liberalized domestic financial sectors; adverse macroeconomic effects on output when domestic financial markets remain largely restricted or when liberalization episodes are followed by a crisis.
  - Inequality: salient adverse effects on average, particularly when domestic financial liberalization is low and not-inclusive or when a crisis follows liberalization.
  - Crises: analysis controls for direct effect of financial crises, which reduce output and increase inequality.
- Sectoral labor share effects:
  - Capital account liberalization episodes reduce the share of labor income especially for industries with:
    - high external financial dependence, or
    - higher natural propensity to adjust workforce (natural layoff rate), or
    - elasticity of substitution between capital and labor relatively high (and greater than unity).
- Policy interpretation:
  - Findings do not imply countries should not liberalize capital accounts, but distributional impacts suggest caution given weak efficiency gains.
  - If reduction in inequality is an important policy goal, liberalization design should balance equity impact against other effects.
  - Potential policy measures:
    - Restrict certain types of flows that generate adverse equity–efficiency trade-offs (e.g., carry-trade flows or flows that give rise to unhealthy asset price or credit booms).
    - Encourage flows that give durable increases in investment and growth (such as greenfield investments).
    - Develop domestic financial institutions, depth, and inclusion.
    - Use fiscal redistribution to mitigate adverse distributional consequences (with limited effect on efficiency unless extreme).
    - Pre-distribution policies: increased spending on education and training to foster greater equality of opportunity.

### Key Quantitative Results (selected tables)
- Table 1 — The Aggregate and Distributional Effects of Capital Account Liberalization (medium-term effect five years after reform; estimates based on equation (7)):
  - Medium-term effect on Output: 0.665
  - Medium-term effect on Gini: 4.018***
  - N: Output 2,001; Gini 1,789
  - R2: Output 0.38; Gini 0.13
- Table 2 — Role of Financial Institutions and Crises (estimates based on equation (8); G(Z)=1(0) for low(high) levels of financial liberalization and financial inclusion and when reforms are (not) followed by crises):
  - Selected medium-term effects (coefficients with F-statistics in parentheses as in source):
    - Medium-term effect*G(Z): -2.558 (1.95) for Output in column I; 4.341** (4.47) for Gini in column IV; 3.959** (4.34) and 4.288** *(10.21) in other Gini specifications.
    - Medium-term effect*[1- G(Z)]: 3.924* (3.03) in Output column I; other values shown in table.
  - N: 2,001 (columns I–III); 1,789 (columns IV–VI)
  - R2: 0.38 (columns I–III); 0.13 (columns IV–VI)
- Table 3 — Effect on Sectoral Output and Components (estimates based on equation (9); differential between sector at 75th vs 25th percentile of external financial dependence):
  - Capital account reform it*Sj (t=0) on Output: 1.802** (1.95)
  - Medium-term differential effect on Output: 0.520 (F-test 0.21)
  - N: 16,616
- Table 4 — Effect on Sectoral Labor Share (estimates based on equation (9); differential between sector at 75th vs 25th percentile):
  - Capital account reform it*Sj (t=0):
    - External Financial Dependence: -1.835*** (-2.82)
    - Layoff Rate: -0.023 (-0.15)
    - Elasticity of Substitution: -0.208*** (-3.59)
  - Medium-term differential effects:
    - External Financial Dependence: -2.230***
    - Layoff Rate: -2.078*
    - Elasticity of Substitution: -2.580*
  - N: 16,616

### Robustness Figures (selected descriptions)
- Figure 1: Aggregate and distributional responses of Output and Gini to capital account liberalization episodes; solid lines are responses, dotted lines 90% confidence bands; t=0 is year of reform (estimates based on equation (7)).
- Figures 2–6: Alternative controls and IV estimation show baseline effects broadly preserved (solid black lines denote baseline).
- Figure 6: Medium-term effects five years after reform for High flow vs Low flow; flows defined as cumulative five-year change in total asset and liabilities as a percentage of GDP after the reform. Chart indicates statistical significance with **, * markers.
- Figures 7–9: Medium-term role of institutions and crises; differential sectoral effects on output, employment, productivity, and labor share by external financial dependence, layoff rate, and elasticity of substitution (estimates based on equation (9)).
- Figures 10–17: Robustness of sectoral differential effects controlling for trade reforms, current account reforms, domestic financial liberalization, product market reforms, EPL, union density, technology (relative price of investment), and sectoral growth. Across these controls, differential effects on output and labor share remain similar to baseline.

### Appendix — Selected descriptive statistics and coverage
- Table A1 — Descriptive statistics for Kaopen and D.Kaopen (by income groups):
  - Panel A. All Countries:
    - Kaopen N 6023 Average -0.002 SD 1.529 Min. -1.856 Max. 2.456
    - D.Kaopen N 5829 Average 0.024 SD 0.370 Min. -3.253 Max. 3.253
  - Panel B. High Income:
    - Kaopen N 1667 Average 1.036 SD 1.516 Min. -1.856 Max. 2.456
    - D.Kaopen N 1618 Average 0.044 SD 0.299 Min. -2.292 Max. 2.292
  - Panel C. Upper-Middle Income, Panel D. Lower-Middle Income, Panel D. Low Income: full values in appendix table.
- Table A2 — Number of capital account liberalization reforms by decade (1970s–2000s; 1970–2010 totals):
  - All: 1970s 38; 1980s 25; 1990s 100; 2000s 61; 1970‒2010 total 224
  - High income total 58; Upper-middle income total 79; Lower-middle income total 54; Lower income total 33
- Table A3 — Country coverage lists per-income-group Kaopen range (years) for many countries (1970‒2010 and country-specific ranges).

*Source: wp1883 - 1. Robustness Checks (IMF working paper content provided in source PDF).*

### REFERENCES

### wp1883 - REFERENCES

### Labor share, automation, and firm dynamics
- Acemoglu, Daron and Pascual Restrepo. 2016. “The Race Between Machine and Man: Implications of Technology for Growth, Factor Shares and Employment.” NBER Working Paper, May, No. 22252.
- Autor, David, David Dorn, Lawrence F. Katz, Christina Patterson and John Van Reenen. 2017b. “The Fall of the Labor Share and the Rise of Superstar Firms.” NBER Working Paper, No. 23396, May.
- Barkai, Simcha. 2017 “Declining Labor and Capital Shares.” Working Paper.
- Bentolila, Samuel, and Saint-Paul Gilles. 2003. “Explaining Movements in the Labor Share.” B.E. Journal of Macroeconomics, De Gruyter, 3, no. 1 (October): 1–33.
- Bridgman, Benjamin. 2017. “Is Labor's Loss Capital's Gain? Gross versus Net Labor Shares.” Macroeconomic Dynamics. July.
- Elsby, Michael W. L., Bart Hobijn and Aysegul Sahin. 2013. “The Decline of the U.S. Labor Share.” Brookings Papers on Economic Activity, Fall: 1-63.
- Karabarbounis, Loukas and Brent Neiman. 2014. “The Global Decline of the Labor Share.” The Quarterly Journal of Economics, Vol. 129, Issue 1: 61-103.
- Koh, Dongya, Raul Santaeulalia-Llopis and Yu Zheng. 2016. “Labor Share Decline and Intellectual Property Products Capital.” Working Paper
- Rognlie, Matthew. 2015. “Decipering the Fall and Rise in the Net Capital Share: Accumulation or Scarcity.” Brookings Papers on Economic Activity, Spring: 1-69.
- O’Mahony, M., and M. P. Timmer. 2009. “Output, Input and Productivity Measures at the Industry Level: The EU KLEMS Database.” Economic Journal 119, no. 538: F374–403.
- Alvarez-Cuadrado, Francisco, Ngo Van Long and Markus Poschke. 2015. “Capital-Labor Substitution, Structural Change and the Labor Income Share.” IZA Discussion Paper, No. 8941, March.
- Boehm, Cristoph E., Aaron Flaaen and Nitya Pandalai-Nayar. 2017. “Multinationals, Offshoring, and the Decline of U.S. Manufacturing.” Center for Economic Studies Research Paper, No. 17-22, February.
- Eden, Maya and Paul Gaggl. 2015 “On the Welfare Implications of Automation.” World Bank Policy Research Working Paper, No. 7487, November.

### Capital account liberalization, financial globalization, and growth
- Eichengreen, Barry J. 2001. “Capital Account Liberalization: What Do the Cross-Country Studies Tell Us?” World Bank Economic Review 15: 341–66.
- Edison, H. J., M. W. Klein, L. Ricci, and T. Slok. 2004. “Capital Account Liberalization and Economic Performance: Survey and Synthesis.” International Monetary Staff Paper, 02/120.
- Henry, P. B. 2007. “Capital Account Liberalization: Theory, Evidence, and Speculation.” Journal of Economic Literature XLV (December): 887–935.
- IMF. 2012. “The Liberalization and Management of Capital Flows: An Institutional View.” IMF Policy Paper.
- Ghosh, Atish R., Jonathan Ostry, and Mahvash S. Qureshi. 2016. “When Do Capital Inflow Surges End in Tears?” American Economic Review 106, no. 5 (May): 581–85. doi:10.1257/aer.p20161015.
- Kose, Ayhan M., E. Prasad, K. Rogoff, and S. J. Wei. 2009. “Financial Globalization: A Reappraisal.” IMF Staff Papers 56: 8–62. doi:10.1057/imfsp.2008.36
- Kose, Ayhan M., Eswar S. Prasad, and Ashley D. Taylor. 2011. “Thresholds in the Process of International Financial Integration.” Journal of International Money and Finance, Elsevier, 30, no. 1 (February): 147–79.
- Rodrik, Dani, and Arvind Subramanian. 2009. “Why Did Financial Globalization Disappoint?” IMF Staff Papers, Palgrave Macmillan, 56, no. 1 (April): 112–38.
- Prasad, Eswar, Kenneth Rogoff, Shang-Jin Wei, and M. Ayhan Kose. 2003. “Effects of Financial Globalization on Developing Countries: Some Empirical Evidence.” IMF Occasional Paper No. 220.
- Quinn, Dennis. 1997. “The Correlates of Change in International Financial Regulation.” American Political Science Review 91, no. 3: 531–51. doi:10.2307/2952073.
- Quinn, Dennis P., and A. Maria Toyoda. 2008. “Does Capital Account Liberalization Lead to Economic Growth?” Review of Financial Studies 21, no. 3: 1403–49.
- Chinn, M., and H. Ito. 2008. “A New Measure of Financial Openness.” Journal of Comparative Policy Analysis 10, no. 3 (September): 309–22.
- Ostry, Jonathan D., Atish Ghosh, Marcos Chamon, and Mahvash Qureshi. 2012. “Tools for Managing Financial-Stability Risks from Capital Inflows.” Journal of International Economics 88(2): 407-21.
- Edison, H. J., M. W. Klein, L. Ricci, and T. Slok. 2004. “Capital Account Liberalization and Economic Performance: Survey and Synthesis.” International Monetary Staff Paper, 02/120.

### Inequality, redistribution, and labor market institutions
- Ostry, Jonathan D., Andrew Berg and Haris Tsangarides. 2014. “Redistribution, Inequality and Growth.” IMF Staff Discussion Note 14/02. Washington: International Monetary Fund. Forthcoming, Journal of Economic Growth, 2018.
- Dao, Mai Chi, Mitali Das, Zsoka Koczan, Weicheng Lian. 2017. “Why is Labor Receiving a Smaller Share of Global Income? Theory and Empirical Evidence.” IMF Working Paper, No. 17/169, July. Washington.
- Das, M., and S. Mohapatra. 2003. “Income Inequality: The Aftermath of Stock Market Liberalization in Emerging Markets.” Journal of Empirical Finance 10: 217–48.
- De Haan, J., and J. E. Sturm. Forthcoming. “Finance and Income Inequality: A Review and New Evidence.” European Journal of Political Economy http://dx.doi.org/10.1016/j.ejpoleco.2017.04.007.
- Larrain, Mauricio. 2015. “Capital Account Opening and Wage Inequality.” Review of Financial Studies, Society for Financial Studies, 28, no. 6: 1555–87.
- Jayadev, Arjun. 2007. “Capital Account Openness and the Labour Share of Income.” Cambridge Journal of Economics, Oxford University Press, 31, no. 3 (May): 423–43.
- Scruggs, Lyle, and Peter Lange. 2002. “Where Have All the Members Gone? Globalization, Institutions, and Union Density”, the Journal of Politics, 64(1),126-153.
- Solt, Frederick. 2016. “The Standardized World Income Inequality Database.” Social Science Quarterly 97(5):1267-1281.
- Furceri, Davide, and Prakash Loungani. Forthcoming. “Capital Account Liberalization and Inequality.” Journal of Development Economics.

### Corporate transparency, taxation, and policy tools
- Tong, Hui, and Shang-Jin Wei. 2011. “Does Trade Globalization Induce or Inhibit Corporate Transparency? Unbundling the Growth Potential and Product Market Competition Channels. NBER Working Paper No. 17631.
- Romer, Christina D., and David H. Romer. 2010. “The Macroeconomic Effects of Tax Changes: Estimates Based on a New Measure of Fiscal Shocks.” American Economic Review 100, no. 3: 763–801.
- Ghosh, Atish R., Jonathan Ostry, and Mahvash S. Qureshi. 2016. “When Do Capital Inflow Surges End in Tears?” American Economic Review 106, no. 5 (May): 581–85. doi:10.1257/aer.p20161015.
- Collins, S. M. 2007. “Comments on ‘Financial Globalization, Growth, and Volatility in Developing Countries’ by Eswar Prasad, Kenneth Rogoff, Shang-Jin Wei, and M. Ayhan Kose.” In Globalization and Poverty, edited by A. Harrison. National Bureau of Economic Research Conference Report. Chicago: University of Chicago Press.
- Rajan, Raghuram G., and Luigi Zingales. 1998. “Financial Dependence and Growth.” American Economic Review 88, no. 3: 559–86.
- Prasad, Eswar, Kenneth Rogoff, Shang-Jin Wei, and M. Ayhan Kose. 2003. “Effects of Financial Globalization on Developing Countries: Some Empirical Evidence.” IMF Occasional Paper No. 220.

### Methods, econometrics, and databases
- Granger, Clive W. J., and Timo Teräsvirta. 1993. Modelling Nonlinear Economic Relationships. New York: Oxford University Press.
- Demirguc-Kunt, Asli, Leora Klapper, Dorothe Singer, and Peter Van Oudheusden. 2015. “The Global Findex Database 2014: Measuring Financial Inclusion around the World.” Policy Research Working Paper 7255. Washington, DC: World Bank.
- Gupta, N., and K. Yuan. 2009. “On the Growth Effect of Stock Market Liberalizations.” Review of Financial Studies 22: 4715–52.
- Barkai, Simcha. 2017 “Declining Labor and Capital Shares.” Working Paper.
- Ciminelli, Gabriele, Romain Duval and Davide Furceri. (forthcoming). “Employment Protection Deregulation and Labor Shares in Advanced Economies” IMF Working Paper.

*From: wp1883 - REFERENCES*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2018/wp1883.pdf_