## wp18186

## Source details

**Canonical URL:** [wp18186](https://www.imf.org/-/media/files/publications/wp/2018/wp18186.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2018/wp18186.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2018/wp18186.pdf.json)

---

### Scope, data and identification
- Sample: 26 advanced economies over the period 1970-2015.
- Country-industry-time panel: unbalanced panel comprising 32 industries in 22 advanced economies from 1970 to 2015.
- Main policy variable: narrative dataset of major reforms of employment protection legislation (EPL) for regular workers (value 0 in non-reform years, 1 in liberalizing reform years, and -1 in tightening reform years).
- Reform identification criteria:
  - (i) OECD Economic Survey uses strong normative language;
  - (ii) action mentioned repeatedly across Survey editions;
  - (iii) OECD EPL indicator is in the 5th percentile of absolute changes (with supplementary domestic source verification when needed).
- Empirical strategies:
  - Country-time analysis using local projection method (Jordà, 2005).
  - Country-industry-time analysis using differences-in-differences à la Rajan and Zingales (1998) with two industry interaction assumptions:
    - Industries with higher “natural” layoff rates (U.S. layoff rates used as proxy) are more affected by stringent dismissal regulations.
    - Industries with lower elasticity of substitution between capital and labor are more likely to experience labor share declines after deregulation.
- Fixed effects: country-time (j,t), industry-time (i,t) and country-industry (j,i) fixed effects.
- Controls and robustness: include past and expected GDP growth, proxies for technological progress and international trade, major reforms for temporary workers, trade union density, and alternative EOS measures.

### Key empirical findings and magnitudes
- Aggregate country-level effect:
  - A major liberalizing EPL reform reduces the aggregate labor share by 0.6 to 0.8 percentage point, on average, over the medium term.
  - Stylized fact: in the five years after major reforms, the aggregate labor share declined by more than seven tenth of a percentage point in reforming countries, on average, compared to status quo countries.
- Country-industry-level differentials:
  - The medium-term effect of a major reform is about 0.9 percentage point higher in high layoff-rate industries (75th percentile of cross-industry U.S. layoff rates) than in low layoff-rate industries (25th percentile).
  - The differential medium-term effect between industries with low and high elasticity of substitution (25th vs 75th percentiles of cross-industry EOS) is 1.5 percentage point.
  - Interaction (preferred specification): joint effect of moving from 25th to 75th percentile in layoff rate and from 75th to 25th percentile in EOS ≈ -1.7 percentage point 5 years after a liberalizing EPL reform.
- Channel evidence:
  - Effects are mainly driven by a decline in the real wage; differential effect between 75th and 25th percentiles reaches about -1.5 percent four years after the reform (significant at the 5% level).
  - Employment: positive differential response becoming significant two years after reform in some specifications (e.g., 1y 0.41; 3y 0.83).
  - Capital-to-output ratio: negative medium-term response but not always significant; examples include 1y 2.84; 4y -4.66; 5y -3.87 in one extension.
- Back-of-the-envelope aggregate contribution:
  - Illustrative calculation suggests job protection deregulation may have contributed about 15 percent to the overall decline in the labor share in advanced economies, reflecting primarily the deregulation wave of the 1990s and 2000s.
  - Country-level back-of-the-envelope statements: deregulation may have accounted for about 14 percentage points of the overall labor share decline in advanced economies over 1970-2015 (text phrase: "about 14 percentage points of the overall labor share decline"); alternative periods: over 1970-2007 changes in EPL contributed about 14 percent; over 1990-2015 contributed about 15 percent.

### Data construction and key statistics
- EPL reforms dataset: covers 26 countries with episode dating from 1970-2013 based on OECD Economic Surveys and country-specific sources.
- Industry layoff rates:
  - Constructed as percentage ratio of laid-off workers over total wage and salary employment using the 2014 Displaced Workers Survey (IPUMS-CPS).
  - U.S. data used as proxy for “natural” layoff rates.
  - Industry with highest layoff rate: “Electrical & Optical” (Average = 5.62).
  - Industry with lowest layoff rate: “Coke & Refined Petroleum” (Average = 1.07).
  - Average layoff rates by year: 2011 = 2.50; 2012 = 2.81; 2013 = 3.74; Average = 3.02.
- Elasticities of substitution (EOS):
  - Industry-specific EOS inferred via structural estimation from firm profit maximization using EU KLEMS capital services and Jorgensonian rental rates as baseline.
  - Average EOS ≈ 0.71 (EOS1); EOS range in baseline: 0.30 (“Construction”) to 1.48 (“Telecommunications”); selected industry baseline EOS1 examples: Textiles 0.38; Electrical & Optical 0.58; Telecommunications 1.48; Real Estate 1.20; Transport & Storage 1.36.
  - Alternative EOS sets yield averages EOS2 = 0.74; EOS3 = 0.85; EOS4 = 0.68; correlation among EOS sets ranges from 0.4 to 0.9.
- Labor share and industry data:
  - Country-level labor shares from OECD Analytical Database (1970–2015).
  - Country-industry labor shares constructed from EU KLEMS (ISIC Rev. 4) value added and labor compensation (2017 and 2012 releases) with two industries dropped.
  - Baseline industry sample excludes public-sector-type industries, agriculture, construction, Coke/Refined Petroleum/Nuclear Fuel, and Other Manufacturing for measurement and EPL applicability reasons; sensitivity checks confirm baseline robustness.
- Summary industry statistics (selected mean values from Table A4.1):
  - Average share in value added = 3.20; Average labor share = 65.73; Average elasticity of substitution = 0.71; Average natural layoff rate = 3.02.
  - Examples: Food, Beverages & Tobacco: Share in value added 2.66 | Labor share 57.25 | Elasticity of substitution 0.69 | Natural layoff rate 3.03; Construction: Share in value added 6.71 | Labor share 78.85 | Elasticity of substitution 0.30 | Natural layoff rate 6.37; Real Estate: Share in value added 8.95 | Labor share 6.83 | Elasticity of substitution 1.20 | Natural layoff rate 1.90.

### Theoretical guidance and mechanisms
- Two stylized wage-bargaining models used for guidance:
  - Right-to-Manage model:
    - Wage set by bargaining, employer then sets employment; labor paid marginal product; labor share LS = 1 − α k^ε.
    - EPL deregulation lowers bargaining power (θ↓), lowers wage, induces substitution of labor for capital; sign of LS change depends on ε.
  - Efficient Bargaining model:
    - Bargaining over employment and wages; employment set efficiently; wage is weighted average: w/p = θ A f(l)/l + (1−θ) A f′(l).
    - Labor share LS = 1 − α (1−θ) k^ε; EPL deregulation unambiguously lowers LS (employment unaffected).
- Empirical implication: deregulation is more likely to lower the labor share in industries where capital and labor are less substitutable (lower EOS).

### Econometric framework and identification strengths
- Country-level specification: local projection estimating dynamic responses for horizons k = 0,..,5 with OLS; controls include forward reform dummies, recession dummies, temporary-contract reform dummies, two lags of 1-period labor share change and of the EPL dummy; 90 percent confidence bands reported.
- Country-industry specification: differences-in-differences estimating y i,j,t+k − y i,j,t−1 = αj,t + γi,j + μi,t + βk ϑi Rj,t + θX i,j,t + … with standard errors clustered at country-industry level (bootstrapping 500 replications for EOS and interaction identifications).
- Identification strengths:
  - Three-dimensional panel with country-time and industry-time fixed effects absorbs unobserved aggregate macroeconomic conditions and industry-specific technological changes.
  - Interaction-based identification (reforms × industry layoff rates or × industry EOS) mitigates reverse causality concerns because country-time fixed effects control for aggregate labor share influences on reform timing.
  - Results robust to inclusion of additional controls for technological change, globalization, GDP growth expectations, and EPL reforms for temporary workers.

### Robustness checks and sensitivity (selected numerical results)
- Country-level baseline coefficients (Table 1, coefficients in percentage points):
  - Baseline: Impact -0.27; 1y -0.63; 2y -0.78; 3y -0.53; 4y -0.47; 5y -0.55.
  - Other labor share drivers: Impact -0.26; 1y -0.58; 2y -0.85; 3y -0.69; 4y -0.62; 5y -0.69.
- Country-industry identification through layoff rates (Table 2, Panel A baseline, coefficients in percentage points):
  - Baseline (2 lags): Impact 0.01; 1y -0.50; 2y -0.42; 3y -0.76; 4y -0.83; 5y -0.93.
- Country-industry identification through elasticities (Table 2, Panel B baseline):
  - Baseline (2 lags): Impact -0.44; 1y -0.86; 2y -1.28; 3y -0.93; 4y -1.24; 5y -1.51.
- Combined identification (layoff rates and elasticities) (Table 2, Panel C baseline):
  - Baseline (2 lags): Impact -0.47; 1y -1.22; 2y -1.56; 3y -1.35; 4y -1.42; 5y -1.70.
- Robustness to lags, sample composition, alternative layoff and EOS measures, and omitted variables generally confirm the negative labor share response, though magnitudes vary across specifications (see reported tables for exact coefficients).

### Stylized facts (1970-2015) and decomposition
- Long-run trend:
  - Labor shares have generally been on a declining trend since the mid-1970s, with the decline accelerating in the 1990s.
- Cross-country and cross-industry heterogeneity:
  - Estimated linear trends in country labor shares: negative and significant in 15 out of 22 countries.
  - Of 32 industries, 23 display a negative and statistically significant trend, and 4 have a significant positive trend.
- Within vs between decomposition:
  - Decomposition ∆LS_j = ∑_i ω̅_ij ∆LS_ij + ∑_i LS̅_ij ∆ω_ij.
  - Empirical finding: about 70 percent of the decline in country-level labor shares can be accounted by within-industry changes.

### Extensions, mechanism checks and policy implications
- Extensions and mechanism checks (selected numerical examples from Table 7 and Table 8):
  - Real wage (using layoff rates identification): Impact 0.22; 1y -0.96; 2y -1.22; 3y -1.38; 4y -1.47; 5y -1.30 (percentage points or percent changes as reported).
  - Labor share by EOS split (Table 8): Elasticity above 1: Impact 0.96; 1y 1.30; 2y 1.51; 3y 1.83; 4y 1.39; 5y 0.86. Elasticity below 1: Impact 0.01; 1y -0.61; 2y -0.47; 3y -0.92; 4y -0.92; 5y -0.98.
- Interpretation: evidence supports a bargaining-power mechanism in which EPL deregulation reduces worker bargaining power, lowering real wages and thereby labor shares in affected industries; heterogeneous sign of effects across industries is consistent with Right-to-Manage model predictions (sign depends on EOS relative to 1).
- Policy takeaways:
  - Policy-makers should assess labor market reform plans against a wide range of macroeconomic outcomes, including productivity, employment, output, wages and labor shares.
  - When designing reforms, trade-offs between efficiency and equity need to be carefully considered.
- Research implication: further work warranted on the role of labor market deregulation, alongside technology and globalization, in explaining labor share declines.

*Source: wp18186 - IMF Working Paper content provided.*

### 1. Cumulative Changes in Country Labor Shares Around Reform Years ...............................34

### 1. Cumulative Changes in Country Labor Shares Around Reform Years ...............................34

### Sections included
- 1. Cumulative Changes in Country Labor Shares Around Reform Years ...............................34
- 2. Cumulative Changes in Industry Labor Shares Around Reform Years by Industry ...........35
- 3. Country-level Analysis—Baseline Results ..........................................................................36
- 4. Country-industry-level Analysis—Baseline Results ...........................................................37
- 5. Country-industry-level Analysis: Robustness to Excluding Individual Countries ..............38
- 6. Country-industry-level Analysis: Robustness to Excluding Individual Industries ..............39

### Tables
- 1. Country-level Analysis: Robustness Checks .......................................................................40
- 2. Country-industry-level Analysis: Robustness Checks on Lag Specification ......................40
- 3. Country-industry-level Analysis: Robustness Checks on Sample Composition .................41
- 4. Country-industry Analysis: Robustness Checks on the Layoff Rates .................................42
- 5. Country-industry Analysis: Robustness Checks on the Elasticities of Substitution ............42
- 6. Country-industry-level Analysis: Robustness Checks on Potential Omitted ......................43
- 7. Country-industry-level Analysis: Extension on Labor Share Drivers .................................43
- 8. Country-industry-level Analysis: Extension on Sample Split According to Elasticity of ...44

### Appendixes
- 1. Dataset of Reforms ..............................................................................................................45
- 2. Layoff rates ..........................................................................................................................53
- 3. Elasticities of Substitution ...................................................................................................55
- 4. Summary Statistics of Industry Data ...................................................................................60
- 5. Stylized Facts. ......................................................................................................................61

*Source: wp18186 - 1. Cumulative Changes in Country Labor Shares Around Reform Years ...............................34 (PDF).*

### 6. Back-to-the-envelope Calculation Using Country-industry-level Estimates .......................71

### 6. Back-to-the-envelope Calculation Using Country-industry-level Estimates

### Scope, data and identification
- Sample: 26 advanced economies over the period 1970-2015.
- Country-industry-time panel: unbalanced panel comprising 32 industries in 22 advanced economies from 1970 to 2015.
- Main policy variable: narrative dataset of major reforms of employment protection legislation (EPL) for regular workers (value 0 in non-reform years, 1 in liberalizing reform years, and -1 in tightening reform years).
- Reform identification: actions classified as major liberalizing or tightening reforms if one of three criteria is met: (i) OECD Economic Survey uses strong normative language; (ii) action mentioned repeatedly across Survey editions; or (iii) OECD EPL indicator is in the 5th percentile of absolute changes (with supplementary domestic source verification when needed).
- Empirical strategies:
  - Country-time analysis using local projection method (Jordà, 2005) to trace labor share responses.
  - Country-industry-time analysis using differences-in-differences à la Rajan and Zingales (1998) with two theory-grounded industry interaction assumptions:
    - Industries with higher “natural” layoff rates (U.S. layoff rates used as proxy) are more affected by stringent dismissal regulations (Micco and Pagés, 2006; Bassanini et al., 2009).
    - Industries with lower elasticity of substitution between capital and labor are more likely to experience labor share declines after deregulation (Blanchard and Giavazzi, 2003; Bentolila and Saint-Paul, 2003).
- Fixed effects: country-time (j,t), industry-time (i,t) and country-industry (j,i) fixed effects to control for unobserved time-varying and time-invariant heterogeneity.
- Controls and robustness: include past and expected GDP growth, proxies for technological progress and international trade, major reforms for temporary workers, trade union density, and alternative EOS measures.

### Key empirical findings and magnitudes
- Aggregate country-level effect:
  - A major liberalizing EPL reform reduces the aggregate labor share by 0.6 to 0.8 percentage point, on average, over the medium term.
  - Stylized fact: in the five years after major reforms, the aggregate labor share declined by more than seven tenth of a percentage point in reforming countries, on average, compared to status quo countries.
- Country-industry-level differentials:
  - The medium-term effect of a major reform is about 0.9 percentage point higher in high layoff-rate industries (75th percentile of cross-industry U.S. layoff rates) than in low layoff-rate industries (25th percentile).
  - The differential medium-term effect between industries with low and high elasticity of substitution (25th vs 75th percentiles of cross-industry EOS) is 1.5 percentage point.
- Channel evidence:
  - Effects are mainly driven by a decline in the real wage, supporting interpretation that weaker worker bargaining power is the key channel through which EPL deregulation lowers labor shares.
- Back-of-the-envelope aggregate contribution:
  - Illustrative calculation using country- and industry-level estimates suggests job protection deregulation may have contributed about 15 percent to the overall decline in the labor share in advanced economies, reflecting primarily the deregulation wave of the 1990s and 2000s.

### Data construction and key statistics
- EPL reforms dataset: covers 26 countries with episode dating from 1970-2013 based on OECD Economic Surveys and country-specific sources.
- Industry layoff rates:
  - Constructed as percentage ratio of laid-off workers over total wage and salary employment using the 2014 Displaced Workers Survey (IPUMS-CPS).
  - U.S. data used as proxy for “natural” layoff rates because employment protection is essentially non-existent there.
  - Industry with highest layoff rate: “Electrical & Optical”.
  - Industry with lowest layoff rate: “Coke & Refined Petroleum”.
- Elasticities of substitution (EOS):
  - Industry-specific EOS inferred via structural estimation from firm profit maximization (estimation equation summarized in source).
  - Average EOS ≈ 0.7.
  - EOS range: 0.3 (“Construction”) to 1.5 (“Telecommunications”).
- Labor share and industry data:
  - Country-level labor shares from OECD Analytical Database (1970–2015).
  - Country-industry labor shares constructed from EU KLEMS (ISIC Rev. 4) value added and labor compensation (2017 and 2012 releases) with two industries dropped (households as employers; extraterritorial organizations and bodies).
  - Baseline industry sample excludes public-sector-type industries, agriculture, construction, Coke/Refined Petroleum/Nuclear Fuel, and Other Manufacturing for measurement and EPL applicability reasons; sensitivity checks reported in source confirm baseline robustness.
- Additional controls:
  - Openness to trade proxied by imports and exports to GDP.
  - Technological change proxied by price of investment goods relative to output (Penn World Tables 9.0).
  - Real wage constructed from average hourly earnings, hours worked (EU KLEMS) and price level (IMF WEO).
  - Trade union density from OECD and ICTWSS (Visser, 2016).

### Theoretical guidance for identification
- Two stylized wage-bargaining models motivate empirical interactions:
  - Right-to-Manage model:
    - Wage set by bargaining, employer then sets employment; labor paid marginal product; labor share LS = 1 − α k^ε.
    - EPL deregulation lowers bargaining power (θ↓), lowers wage, induces substitution of labor for capital; sign of LS change depends on ε (capital-labor complementarity/substitutability).
  - Efficient Bargaining model:
    - Bargaining over both employment and wages; employment set efficiently; wage is weighted average of average and marginal products: w/p = θ A f(l)/l + (1−θ) A f′(l).
    - Labor share LS = 1 − α (1−θ) k^ε; EPL deregulation unambiguously lowers LS (employment unaffected).
- Empirical implication: deregulation is more likely to lower the labor share in industries where capital and labor are less substitutable.

### Robustness and identification strengths
- Three-dimensional panel with country-time and industry-time fixed effects absorbs unobserved aggregate macroeconomic conditions and industry-specific technological changes, respectively.
- Interaction-based identification (reforms × industry layoff rates or × industry EOS) mitigates reverse causality concerns because country-time fixed effects control for aggregate labor share influences on reform timing.
- Results robust to inclusion of additional controls for technological change, globalization, GDP growth expectations, and EPL reforms for temporary workers.

*Source: wp18186 - 6. Back-to-the-envelope Calculation Using Country-industry-level Estimates*

### Appendix 5 discusses some stylized facts about the evolution of labor shares over the period

### wp18186 - Appendix 5: Stylized facts and empirical analysis of labor share evolution (1970-2015)

### Stylized facts on labor shares (1970-2015)
- Labor shares have generally been on a declining trend since the mid-1970s, with the decline accelerating in the 1990s.
- There exist significant heterogeneities both across countries and industries.
- About 70 percent of the decline in country-level labor shares can be accounted by within-industry changes.
- The decline in the labor share has been typically larger in periods following EPL reforms; liberalizing reforms were predominantly implemented during the 1990s and the 2000s, coinciding with the steepest labor share decline.

### Empirical strategy and econometric framework
- Country-level specification: local projection method (Jordà, 2005) estimating dynamic responses for horizons k = 0,..,5 with OLS; includes country fixed effects (αj), time fixed effects (γt), EPL reform variable Rj,t (0 in non-reform years, 1 in liberalizing years, -1 in tightening years), control vector Xj,t, forward reform dummies, recession dummies, temporary-contract reform dummies, two lags of 1-period labor share change and of the EPL dummy.
- Confidence bands: 90 percent confidence bands around βk, using clustered robust standard errors.
- Country-industry specification: differences-in-differences (Rajan and Zingales, 1998) estimating
  - y i,j,t+k − y i,j,t−1 = αj,t + γi,j + μi,t + βk ϑi Rj,t + θX i,j,t + … with forward reform dummies and lags.
  - Fixed effects: country-time (αj,t), country-industry (γi,j), industry-time (μi,t).
  - Industry-specific characteristics ϑi used for identification: (i) natural layoff rate; (ii) 1−ε = 1/σ (inverse of EOS); (iii) interaction between layoff rate and 1−ε.
  - Standard errors clustered at country-industry level; for EOS and interaction identifications, standard errors obtained via bootstrapping (500 replications).

### Main empirical findings — country-level
- Major liberalizing EPL reforms have a statistically significant and persistent negative effect on the labor share.
- Effect magnitude: reaches 0.8 percentage point two years after the reform, then declines marginally to 0.6 percentage point; it eventually levels off at about -0.6 percentage point (baseline reports stabilization at about -0.8 percentage point after 8 years in an extended horizon).
- Robustness: results unchanged when adding controls for technological progress, international trade, trade union density, past and expected GDP growth; results robust to different lag specifications and inclusion of forward reform dummies.

### Main empirical findings — country-industry
- Effects concentrated in industries with:
  - higher natural layoff rates, and
  - lower elasticity of substitution (EOS) between capital and labor (i.e., higher relative complementarity).
- Quantitative differentials (medium-term, 5 years after reform):
  - Layoff-rate identification (Panel A): differential reduction between 75th and 25th percentile industries ≈ 0.9 percentage point.
  - EOS identification (Panel B): differential reduction between 25th (low EOS) and 75th (high EOS) percentile industries ≈ 1.5 percentage point.
  - Interaction (preferred, Panel C): joint effect of moving from 25th to 75th percentile in layoff rate and from 75th to 25th percentile in EOS ≈ -1.7 percentage point 5 years after a liberalizing EPL reform; effect statistically significant over almost the entire horizon except upon impact.
- Robustness checks:
  - Results robust to varying lag structures (1, 3, 4 lags), exclusion of forward reform dummies, leave-one-country/industry-out analyses.
  - Robust to alternative labor-share vintage (EU KLEMS versions), exclusion of specific industries, and using public sector/agriculture/construction as control group (which yields even larger differential effects).
  - Alternative layoff measures (Displaced Workers Survey dummy, 2013 layoff rates) yield qualitatively similar impulse responses; when using the conservative dummy, coefficients are quantitatively lower but not directly comparable.
  - Alternative EOS measures (Appendix 3 variants) yield very similar results, strongest significance when using interaction identification.
  - Adding interactions capturing change in union density, relative price of investment, and trade openness with industry characteristics leaves estimated effects close to baseline.

### Channels and auxiliary results
- Wage channel:
  - Re-estimating Equation (9) for log hourly real wage (identification via layoff rate) shows EPL deregulation leads to a relative fall in real wages in high-layoff-rate industries versus low-layoff-rate industries.
  - Differential effect between 75th and 25th percentiles reaches about -1.5 percent four years after the reform (significant at the 5% level).
- Employment and capital-to-output:
  - Employment growth: positive differential response, becoming significant two years after reform.
  - Capital-to-output ratio: negative medium-term response but not significant at conventional levels.
- Interpretation: findings support a bargaining-power mechanism in which EPL deregulation reduces worker bargaining power, lowering real wages and thereby the labor share in affected industries.

### EOS sign test and heterogeneous sign of effects
- Splitting sample by EOS > 1 versus EOS < 1 (identification via layoff rates) yields results consistent with the Right-to-Manage model predictions:
  - For industries with EOS above 1 (substitutability), moving from 25th to 75th percentile of layoff rate yields an increase in labor share, significant upon impact and reaching about +1.8 percentage point after 3 years.
  - For industries with EOS below 1 (complementarity), the corresponding effect is negative and significant, about -0.9 percentage point.
  - At longer horizons, the positive effect in EOS>1 group becomes statistically insignificant (possibly due to small sample), while the negative effect in EOS<1 remains significant.

### Back-of-the-envelope macro contribution calculations
- Country-level estimate:
  - A major liberalizing EPL reform causes an approximate labor-share decline of 0.6 percentage point over the four years following the reform.
  - Averaging across countries, deregulation may have accounted for about 14 percentage points of the overall labor share decline in advanced economies over 1970-2015 (text uses the phrase "about 14 percentage points of the overall labor share decline").
- Alternative period calculations:
  - Over 1970-2007 (excluding the Great Recession): changes in EPL contributed about 14 percent to the overall labor share decline.
  - Over 1990-2015 (period of steepest decline): changes in EPL contributed about 15 percent to the overall labor share decline.
- Country-industry back-of-the-envelope calculations under simplifying assumptions (Appendix 6) yield reassuringly similar implications.

### Policy implications and concluding points
- Job protection deregulation has a statistically and economically significant negative effect on labor shares, concentrated in industries with higher propensity to adjust workforce and lower elasticity of substitution.
- Effects likely driven by reductions in wage rents (bargaining power channel).
- Policy takeaways:
  - Policy-makers should assess labor market reform plans against a wide range of macroeconomic outcomes, including productivity, employment, output, wages and labor shares.
  - When designing reforms, trade-offs between efficiency and equity need to be carefully considered.
- Research implication: further work is warranted on the role of labor market deregulation, alongside technology and globalization, in explaining labor share declines.

*Source: wp18186 - Appendix 5 and associated sections (1970-2015) — IMF working paper content provided.*

### Chapter 7. The MIT Press,Cambridge, Massachussets.

### wp18186 - Chapter 7. The MIT Press,Cambridge, Massachussets.

### Literature and datasets cited
- Checchi, Daniele and Cecilia Garcia Penalosa. 2008. “Labour Market Institutions and Income Inequality.” Economic Policy, Vol. 56: 601-649.
- 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.
- Devries, Pete, Jaime Guajardo, Daniel Leigh and Andrea Pescatori. 2011. “A New Action-based Dataset of Fiscal Consolidation.” IMF Working Paper, No. 11/128, June. Washington.
- Drazen, Allan, and William Easterly. 2001. “Do Crises Induce Reform? Simple Empirical Tests of Conventional Wisdom.” Economics and Politics, 13 (2), 129–157.
- Duval, Romain, Davide Furceri, Bingjie Hu, Joao Jalles and Huy Nguyen. 2018. “A New Narrative Database of Product and Labor Market Reforms in Advanced Economies.” IMF Working Paper, 18/19, January. Washington.
- Eden, Maya and Paul Gaggl. 2015 “On the Welfare Implications of Automation.” World Bank Policy Research Working Paper, No. 7487, November. Washington.
- 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.
- Feenstra, Robert C., Robert Inklaar and Marcel P. Timmer. 2015. "The Next Generation of the Penn World Table." American Economic Review, Vol. 105, No. 10: 3150-3182.
- Flood, Sarah, Miriam King, Steven Ruggles, and J. Robert Warren. 2017. “Integrated Public Use Microdata Series, Current Population Survey: Version 5.0.” Database. Minneapolis: University of Minnesota.
- Furceri, Davide, Prakash Loungani and Jontahn D. Ostry. 2017. “The Aggregate and Distributional Effects of Financial Globalization: Evidence from Macro and Sectoral Data.” Working Paper.
- Hummels, David, Jakob Munch and Chong Xiang. 2016. “Offshoring and Labor Markets.” IZA Discussion Paper No. 9741, February.
- International Monetary Fund (IMF). 2017. “Understanding the Downward Trend in Labor Income Shares.” Chapter 3. World Economic Outlook April 2017: 121-172. Washington.
- Jäger, Kirsten. 2017. “EU KLEMS Growth and Productivity Accounts 2017 Release, Statistical Module.” The Conference Board.
- Jordà, Òscar. 2005. “Estimation and Inference of Impulse Responses by Local Projections.” American Economic Review. Vol. 95. No. 1: 161–82.
- Jorgenson, Dale. 1963. “Capital Theory and Investment Behavior.” American Economic Review, Vol. 53, No. 2, may: 247–259.
- Kaldor, Nicholas. 1957. “A Model of Economic Growth.” The Economic Journal, Vol. 67. No. 268, December: 591-624.
- Kaldor, Nicholas. 1957. “Capital Accumulation and Economic Growth.” In The Theory of Capital, ed. F. A. Lutz and D. C. Hague, 177-222. New York.
- Karabarbounis, Loukas and Brent Neiman. 2014. “The Global Decline of the Labor Share.” The Quarterly Journal of Economics, Vol. 129, Issue 1: 61-103.
- Kehrig, Matthias and Nicolas Vincent. 2017. “Growing Productivity without Growing Wages: The Micro-Level Anatormy of the Aggregate Labor Share Decline.” CESifo Working Paper, No. 6554. Munich.
- Koh, Dongya, Raul Santaeulalia-Llopis and Yu Zheng. 2016. “Labor Share Decline and Intellectual Property Products Capital.” Working Paper.
- Kramarz, Francis. 2016. “Offshoring, Wages and Employment: Evidence from Data Matching Imports, Firms and Workers.” Chapter 9 in Fontagne, Lionel and Ann Harrison (eds.), The Factory-Free Economy: Outsourcing, Servitization, and the Future of Industry, Oxford University Press, Oxford, United Kingdom.
- Layard, Richard, Stephen Nickell and Richard Jackman. 1991. “Unemployment. Macroeconomic Performance and the Labour Market.” Oxford University Press. Oxford.
- Lawrence, Robert Z. 2015. “Recent Declines in Labor’s Share in US Income: A Preliminary Neoclassical Account.” Peterson Institute Working Paper, No. 15-10. Washington.
- Manning, Alan. 1987. “An Integration of Trade Union Models in a Sequential Bargaining Framework.” The Economic Journal, Vol. 97, Issue 385: 121-139.
- Micco, Alejandro and Pagés, Carmen. 2006. “The Economic Effects of Employment Protection: Evidence from International Industry-Level Data.” IZA Discussion Paper No. 2433.
- O’Mahony, Mary and Marcel P. Timmer. 2009. “Output, Input and Productivity Measures at the Industry Level: the EU KLEMS Database.” The Economic Journal, Vol. 119, Issue 538: 374-403.
- Oberfield, Ezra, and Devesh Raval. 2014. “Micro Data and Macro Technology.” NBER Working Paper, September, No. 20452.
- Organisation for Economic Cooperation and Development (OECD). 2013. “Protecting Jobs, Enhancing Flexibility: A New Look at Employment Protection Legislation.” Chapter 2, OECD Employment Outlook 2013, Paris.
- Organisation for Economic Cooperation and Development (OECD). 2012. “Understanding the Downward Trend in Labor Income Shares.” Chapter 3, OECD Employment Outlook 2012, Paris.
- Rajan, Raghuram G. and Luigi Zingales. 1998. “Financial Dependence and Growth.” American Economic Review, Vol. 88, No. 3, June: 559-586.
- Rios-Rull, Jose-Victor and Raul Santaeulalia-Llopis. 2010. “Redistributive shocks and productivity shocks.” Journal of Monetary Economics. Vol. 57: 931-948.
- Romer, Christina D., and David Romer. 2017 “New Evidence on the Impact of Financial Crises in Advanced Countries.” American Economic Review. Vol. 107, No. 110, October 2017: 3072-3118.
- Romer, Christina D., and David Romer. 2010. “The Macroeconomic Effects of Tax Changes: Estimates Based on a New Measure of Fiscal Shocks.” American Economic Review, Vol. 100, No.3, June: 763– 801.
- Romer, Christina D., and David Romer. 2004. “A New Measure of Monetary Shocks: Derivation and Implications.” American Economic Review. Vol. 94, No. 4, September: 1055–1084.
- Romer, Christina D., and David Romer. 1989. “Does Monetary Policy Matter? A New Test in the Spirit of Friedman and Schwartz.” NBER Macroeconomics Annual, 4: 121–170.
- Rognlie, Matthew. 2015. “Decipering the Fall and Rise in the Net Capital Share: Accumulation or Scarcity.” Brookings Papers on Economic Activity, Spring: 1-69.
- Teulings, Coen N., Nikolay Zubanov. 2014. “Is Economic Recovery a Myth? Robust Estimates of Impulse Responses.” Journal of Applied Econometrics, Vol. 29: 497-514.
- Tommasi, Mariano, and Andres Velasco. 1996. “Where Are We in the Political Economy of Reform?” Journal of Economic Policy Reform, Vol. 1, Issue 2: 187-238.
- Visser, Jelle. 2016. “ICTWSS Data base. Version 5.1.” Database. Amsterdam: Amsterdam Institute for Advanced Labour Studies (AIAS), University of Amsterdam.

### Main empirical findings (country-level and country-industry-level)
- Figure 3 notes: estimates based on Equation (8). Solid line denotes the percentage point response of labor share to EPL reforms. Dotted lines indicate 90 percent confidence interval based on clustered standard errors. X-axis: horizon with 0 indicating the reform year. Y-axis: magnitude of estimated coefficients (in percentage points).
- Table 1. Country-level Analysis: Robustness Checks (coefficients are in percentage points; bold indicates significance at the 90 percent confidence interval).
  - Baseline: Impact -0.27; 1y -0.63; 2y -0.78; 3y -0.53; 4y -0.47; 5y -0.55
  - Other labor share drivers: Impact -0.26; 1y -0.58; 2y -0.85; 3y -0.69; 4y -0.62; 5y -0.69
  - (Exp.) GDP: Impact -0.28; 1y -0.61; 2y -0.73; 3y -0.47; 4y -0.47; 5y -0.49
- Table 2. Country-industry-level Analysis: Robustness Checks on Lag Specification (coefficients in percentage points).
  - Panel A) Identification through layoff rates
    - Baseline (2 lags): Impact 0.01; 1y -0.50; 2y -0.42; 3y -0.76; 4y -0.83; 5y -0.93
    - 1 lag: Impact 0.01; 1y -0.50; 2y -0.63; 3y -0.75; 4y -0.79; 5y -0.94
    - 3 lags: Impact 0.01; 1y -0.35; 2y -0.48; 3y -0.85; 4y -0.87; 5y -0.98
    - 4 lags: Impact 0.03; 1y -0.40; 2y -0.54; 3y -1.00; 4y -1.10; 5y -1.25
    - No forward dummies: Impact 0.01; 1y -0.48; 2y -0.37; 3y -0.58; 4y -0.70; 5y -0.74
  - Panel B) Identification through elasticities of substitution
    - Baseline (2 lags): Impact -0.44; 1y -0.86; 2y -1.28; 3y -0.93; 4y -1.24; 5y -1.51
    - 1 lag: Impact -0.39; 1y -0.81; 2y -1.45; 3y -0.85; 4y -1.13; 5y -1.43
    - 3 lags: Impact -0.48; 1y -0.74; 2y -1.39; 3y -1.08; 4y -1.35; 5y -1.63
    - 4 lags: Impact -0.53; 1y -0.90; 2y -1.52; 3y -1.35; 4y -1.72; 5y -2.08
    - No forward dummies: Impact -0.44; 1y -0.79; 2y -0.86; 3y -0.60; 4y -1.28; 5y -1.12
  - Panel C) Identification through layoff rates and elasticities of substitution
    - Baseline (2 lags): Impact -0.47; 1y -1.22; 2y -1.56; 3y -1.35; 4y -1.42; 5y -1.70
    - 1 lag: Impact -0.45; 1y -1.21; 2y -1.94; 3y -1.32; 4y -1.37; 5y -1.72
    - 3 lags: Impact -0.50; 1y -0.97; 2y -1.68; 3y -1.52; 4y -1.54; 5y -1.83
    - 4 lags: Impact -0.51; 1y -1.15; 2y -1.83; 3y -1.88; 4y -2.06; 5y -2.48
    - No forward dummies: Impact -0.47; 1y -1.13; 2y -1.12; 3y -0.95; 4y -1.44; 5y -1.30
- Table 3. Country-industry-level Analysis: Robustness Checks on Sample Composition (coefficients in percentage points).
  - Panel A) Layoff rates (Baseline and variants)
    - Baseline: Impact 0.01; 1y -0.50; 2y -0.42; 3y -0.76; 4y -0.83; 5y -0.93
    - All manufacturing: Impact -0.06; 1y -0.33; 2y -0.22; 3y -0.31; 4y -0.30; 5y -0.49
    - Control group: Impact -0.25; 1y -0.56; 2y -0.54; 3y -1.04; 4y -0.97; 5y -0.62
    - KLEMS 2017 database: Impact 0.06; 1y -0.44; 2y -0.48; 3y -0.94; 4y -0.89; 5y -0.96
  - Panel B) Elasticities of substitution (Baseline and variants)
    - Baseline: Impact -0.44; 1y -0.86; 2y -1.28; 3y -0.93; 4y -1.24; 5y -1.51
    - All manufacturing: Impact -0.46; 1y -0.89; 2y -1.25; 3y -0.86; 4y -1.15; 5y -1.50
    - Control group: Impact -0.73; 1y -1.15; 2y -1.53; 3y -1.64; 4y -1.74; 5y -1.52
    - KLEMS 2017 database: Impact -0.35; 1y -0.66; 2y -1.20; 3y -1.08; 4y -1.29; 5y -1.51
  - Panel C) Layoff rates and elasticities combined (Baseline and variants)
    - Baseline: Impact -0.47; 1y -1.22; 2y -1.56; 3y -1.35; 4y -1.42; 5y -1.70
    - All manufacturing: Impact -0.39; 1y -0.96; 2y -1.05; 3y -0.80; 4y -0.77; 5y -1.21
    - Control group: Impact -0.77; 1y -1.50; 2y -1.75; 3y -2.02; 4y -1.95; 5y -1.66
    - KLEMS 2017 database: Impact -0.40; 1y -1.02; 2y -1.57; 3y -1.59; 4y -1.52; 5y -1.77
- Table 4. Robustness Checks on the Layoff Rates (coefficients in percentage points).
  - Panel A) Layoff rates only
    - Baseline: Impact 0.01; 1y -0.50; 2y -0.42; 3y -0.76; 4y -0.83; 5y -0.93
    - Qualitative measure: Impact 0.01; 1y -0.34; 2y -0.61; 3y -1.48; 4y -1.05; 5y -1.46
    - 2013 layoff rate: Impact 0.11; 1y -0.74; 2y -0.46; 3y -0.68; 4y -0.69; 5y -0.77
  - Panel B) Layoff rates and elasticities combined
    - Baseline: Impact -0.47; 1y -1.22; 2y -1.56; 3y -1.35; 4y -1.42; 5y -1.70
    - Qualitative measure: Impact 0.02; 1y -0.07; 2y -0.10; 3y -0.20; 4y -0.22; 5y -0.26
    - 2013 layoff rate: Impact -0.40; 1y -1.17; 2y -1.37; 3y -1.09; 4y -1.13; 5y -1.27
- Table 5. Robustness Checks on the Elasticities of Substitution (coefficients in percentage points).
  - Panel A) Elasticities only
    - Baseline: Impact -0.44; 1y -0.86; 2y -1.28; 3y -0.93; 4y -1.24; 5y -1.51
    - Stock: Impact -0.53; 1y -0.52; 2y -1.49; 3y -0.78; 4y -0.93; 5y -1.45
    - Rental rate: Impact -0.30; 1y -0.83; 2y -0.82; 3y -0.70; 4y -0.99; 5y -1.28
    - Technical change: Impact -0.38; 1y -0.72; 2y -0.86; 3y -0.65; 4y -1.10; 5y -1.19
  - Panel B) Layoff rates and elasticities combined
    - Baseline: Impact -0.47; 1y -1.22; 2y -1.56; 3y -1.35; 4y -1.42; 5y -1.70
    - Stock: Impact -0.50; 1y -1.05; 2y -1.92; 3y -1.43; 4y -1.41; 5y -1.86
    - Rental rate: Impact -0.38; 1y -1.19; 2y -1.11; 3y -1.08; 4y -1.17; 5y -1.53
    - Technical change: Impact -0.41; 1y -1.09; 2y -1.17; 3y -1.10; 4y -1.26; 5y -1.40
- Table 6. Robustness Checks on Potential Omitted Variables (coefficients in percentage points).
  - Panel A) Layoff rates
    - Baseline: Impact 0.01; 1y -0.50; 2y -0.42; 3y -0.76; 4y -0.83; 5y -0.93
    - Relative investment price: Impact 0.03; 1y -0.42; 2y -0.48; 3y -0.79; 4y -0.84; 5y -1.10
    - Trade openness: Impact 0.01; 1y -0.45; 2y -0.50; 3y -0.83; 4y -0.88; 5y -1.13
    - Trade union density: Impact 0.02; 1y -0.43; 2y -0.48; 3y -0.81; 4y -0.78; 5y -1.00
  - Panel B) Elasticities of substitution
    - Baseline: Impact -0.44; 1y -0.86; 2y -1.28; 3y -0.93; 4y -1.24; 5y -1.51
    - Relative investment price: Impact -0.15; 1y -0.60; 2y -1.34; 3y -1.04; 4y -1.24; 5y -1.57
    - Trade openness: Impact -0.18; 1y -0.63; 2y -1.36; 3y -1.08; 4y -1.31; 5y -1.67
    - Trade union density: Impact -0.16; 1y -0.60; 2y -1.39; 3y -1.11; 4y -1.19; 5y -1.47
  - Panel C) Layoff rates and elasticities combined
    - Baseline: Impact -0.47; 1y -1.22; 2y -1.56; 3y -1.35; 4y -1.42; 5y -1.70
    - Relative investment price: Impact -0.19; 1y -0.90; 2y -1.70; 3y -1.51; 4y -1.48; 5y -1.93
    - Trade openness: Impact -0.23; 1y -0.94; 2y -1.72; 3y -1.56; 4y -1.56; 5y -2.02
    - Trade union density: Impact -0.20; 1y -0.91; 2y -1.73; 3y -1.59; 4y -1.41; 5y -1.75

### Extensions and mechanism checks
- Table 7. Extensions on labor share drivers (estimates based on Equation (9); using layoff rates for identification; coefficients in percentage points).
  - Labor share: Impact 0.01; 1y -0.50; 2y -0.42; 3y -0.76; 4y -0.83; 5y -0.93
  - Real wage: Impact 0.22; 1y -0.96; 2y -1.22; 3y -1.38; 4y -1.47; 5y -1.30
  - Employment: Impact 0.11; 1y 0.41; 2y 0.42; 3y 0.83; 4y 0.66; 5y 0.19
  - Capital-to-output ratio: Impact 1.45; 1y 2.84; 2y 0.49; 3y -2.30; 4y -4.66; 5y -3.87
- Table 8. Extension on sample split according to elasticity of substitution (coefficients in percentage points).
  - Full sample: Impact 0.01; 1y -0.56; 2y -0.48; 3y -0.85; 4y -0.93; 5y -1.05
  - Elasticity above 1: Impact 0.96; 1y 1.30; 2y 1.51; 3y 1.83; 4y 1.39; 5y 0.86
  - Elasticity below 1: Impact 0.01; 1y -0.61; 2y -0.47; 3y -0.92; 4y -0.92; 5y -0.98

### Figures and identification notes
- Figure 1: Compares mean cumulative change in country labor shares relative to reform years between reforming countries (green bars) and status quo countries (red bars). Panel A: raw data; Panel B: de-meaned and de-trended data. Y-axis: mean cumulative change (in percentage points). X-axis: years relative to base year (0).
- Figure 2: Compares mean cumulative change in country-industry labor shares around EPL reforms between reforming and status quo countries by industry quartiles of layoff rates and elasticities of substitution. Y-axis: labor share change (in percentage points). X-axis: years relative to base year (0).
- Figure 4: Country-industry baseline estimates based on Equation (9). Solid lines show average differential labor share effect of EPL reforms between industry percentiles; dotted lines indicate 90 percent confidence intervals clustered at country-industry level. For Panels B and C standard errors are via bootstrapping (500 replications).
- Figures 5–6: Robustness to excluding individual countries (Figure 5) and industries (Figure 6); each solid line is from excluding one unit at a time; comparison with baseline 90 percent confidence interval is shown.

### Key takeaways (from reported estimates)
- EPL reforms are associated with statistically and economically meaningful declines in labor shares in many specifications: country-level baseline shows negative responses up to -0.78 (2y) and persistent negative effects in country-industry specifications.
- Effects are heterogeneous across identification strategies and industry characteristics:
  - Identification through layoff rates alone yields modest negative impacts (e.g., 1y -0.50; 5y -0.93 in baseline).
  - Identification through elasticities of substitution yields larger negative impacts (e.g., 2y -1.28; 5y -1.51 in baseline).
  - Combined identification (layoff rates and elasticities) shows larger and more persistent negative impacts (e.g., 2y -1.56; 5y -1.70 in baseline).
- Robustness checks (lags, sample composition, alternative layoff and elasticity measures, omitted variables) generally confirm the negative labor share response, though magnitudes vary (see tables for exact coefficients).
- Mechanism checks:
  - Real wage responses are negative and sizable in several horizons (e.g., 1y -0.96; 2y -1.22; 3y -1.38).
  - Employment dynamics differ: positive short-run changes in employment in some horizons (e.g., 1y 0.41; 3y 0.83) when using employment as dependent variable.
  - Capital-to-output ratio shows large positive short-run impacts and negative medium-term adjustments (e.g., 1y 2.84; 4y -4.66).

*Source: wp18186 - Chapter 7. The MIT Press,Cambridge, Massachussets. (PDF).*

### Appendix 1. Dataset of Reforms

### Appendix 1. Dataset of Reforms

### Reform events by country and year (selected entries)
- United Kingdom, 2000 — area: severance pay. "Quadrupling maximum compensation for unfair dismissals from October 1999 (pg. 116, 2000)"; Mention in other reports: yes for 2000; Score: -1.
- Austria, 2003 — area: severance pay. Excerpt: "...the system underwent thorough reform. In the new system, which became effective in January 2003, the management of severance pay is attributed to retirement accounts, which are legally independent from the employers and funded by employers via a monthly untaxed payment of some 1.5 per cent of gross wages. Accumulated entitlements rest in the employee’s account until retirement, unless the work contract has been terminated by the employer, which makes cash payments admissible under certain conditions... (pg. 66, 2003)"; Mention in other reports: yes for 2003; Score: 1.
- Belgium, 1970 — area: notice for individual dismissal. "In November 1970, the notice period...was increased to three months. The possibility of a further extension to five months...the five months' period was applied in most cases. (pg. 27, 1971)"; Mention in other reports: no data but would qualify if scoring applied; Score: -1.
- Belgium, 1971 — area: notice for individual dismissal. "In April 1971, the period of prior notice was reduced to two months (pg. 27, 1971)"; Mention in other reports: no data but would qualify if scoring applied; Score: 1.
- Belgium, 1985 — area: severance pay. Measures include authorisation for ailing businesses to pay severance allowances in monthly instalments; incentives for fixed-term contracts; lengthening of probation periods from 3 and 6 months to 6 and 12 months. (pg. 47, 1985). Mention in other reports: no; Score: 1.
- France, 1987 — area: procedural inconvenience. "Checks on the genuineness of redundancies in firms with fewer than 10 employees to be discontinued (and from 1st January 1987, official authorisation for layoffs no longer necessary). (pg. 76, 1987)"; Mention in other reports: pg. 33 or 44, 1989; pg. 59, 1990; Large change in OECD indicator: yes for 1987; Score: 1.
- France, 2003 — area: collective dismissal. Social Modernisation Law (2002) tightened constraints on dismissal of more than 10 employees; government suspended some provisions in 2003; later law in 2004 moderated some aspects while increasing obligation to try to find alternative jobs; Law permits “economic” dismissal only if necessary to preserve competitiveness; group financial position to be taken into account. (pg. 105, 2005). Mention in other reports: yes for 2003; Score: -1.
- France, 2009 — area: procedural inconvenience. "Layoff law has been simplified by introducing the possibility of mutually agreed termination (rupture conventionnelle) of the CDI. (pg. 52, 2009)"; Mention in other reports: yes for 2009; Score: 1.
- Germany, 1994 — area: notice for individual dismissal. "Notice period for blue-collar workers extended to four weeks, thereby aligning it with that of white-collar workers"; Mention in other reports: yes for 1994; Score: -1.
- Germany, 1997 — area: procedural inconvenience. Legislation easing employment protection: employment ceiling raised from five to ten employees per firm (increasing number of firms not subject to general job protection by some 15 percent; these companies employ some 30 per cent of all employees); general requirement to consider social criteria relaxed for large scale redundancies. (pg. 132, 1997); Mention in other reports: no; Score: 1.
- Germany, 2004 — area: procedural inconvenience. Protection against Dismissal Act (PaDA) explicit listing of four social criteria in § 1(3); prior regulations applied to firms with more than five permanent employees; since 2004 criteria explicitly listed. Mention in other reports: yes for 2004; Score: -1.
- Italy, 1970 — area: procedural inconvenience. The Act of 1970 ("workers' statute") mechanism for reinstatement after unlawful dismissal laid down by Article 18. Mention in other reports: No data but would qualify if scoring applied; Score: -1.
- Italy, 1991 — area: procedural inconvenience. "The job allocation scheme was abolished in June 1991 (pg. 54, 1991)"; Mention in other reports: pg. 19, 1994; pg. 11, 1995; pg. 134, 1999; no; Score: 1.
- Italy, 2013 — area: procedural inconvenience. Comprehensive labour market reform including relaxation of employment protection rules, reduced incentives to hire on non-permanent contracts, and limiting possibility of reinstatement following unfair dismissal. (pg. 42, 2013; pg. 27, 2015). Mention in other reports: pg. 27, 2015; yes for 2013; Score: 1.
- Netherlands, 1976 — area: collective dismissal. "Compulsory 3-month advance notification to employment exchange and trade unions required for the intended dismissal of 20 or more employees (pg. 47, 1977)." Mention in other reports: no data but would qualify if scoring applied; Score: -1.
- Netherlands, 1996 — area: procedural inconvenience. Government decides to shorten dismissal procedures; employer can dismiss at the same time or before asking permission from director of Public Employment Service. (pg. 122, 1996). Mention in other reports: yes in 1995; Score: 1.
- Norway, 1977 — area: procedural inconvenience. Main legislation on employment protection is the law on worker protection and the working environment dating back to 1977 (pg. 164, 2004). Mention in other reports: pg. 164, 2004; no data; Score: -1.
- Sweden, 1975 — area: notice for individual dismissal. "Introduction of the employment security act in July 1974, stipulating that employers are to give 6 months' warning in advance of layoffs... (pg. 21, 1976)"; Mention in other reports: pg. 36-37, 1980; no data but would qualify if scoring applied; Score: -1.
- Sweden, 1997 — area: notice for individual dismissal. Revised Employment Protection legislation modifies notice length criteria (tenure-based), rehiring obligation shortened from twelve to nine months, introduction of twelve-month fixed-term contracts with up to five persons and new establishments allowed to extend to eighteen months. (pg. 81-82, 1998). Mention in other reports: pg. 105, 1999; Large change in OECD indicator: yes in 1997 and 1999; Score: 1.
- Japan, 2007 — area: procedural inconvenience. "Labor Contract Act of 2007." Mention in other reports: yes in 2007; Score: 1.
- Finland, 1989 — area: notice for individual dismissal. "Protection of workers is improved. Periods of notice will be extended from 1989. Dismissal for economic reasons will be possible only if work has decreased significantly and permanently and if employees cannot be transferred or trained for new tasks. (pg. 120, 1989)"; no data but would qualify if scoring applied; Score: -1.
- Finland, 1997 — area: notice for individual dismissal. Employers' period of notice shortened to one month (from two months) and employees to fourteen days (from one month). (pg. 63, 1997). Mention in other reports: pg. 63, 1997; yes in 1997; Score: 1.
- Greece, 2011 — areas: notice for individual dismissal, severance pay, collective dismissal, reallocation. Measures introduced in 2010 (Laws 3863/2010 and 3899/2010) include reduction in notice period (example: for 28+ years tenure, notification reduced to 6 from 24 months); severance payment instalments when exceeding 2 months’ pay; redefinition of collective dismissal thresholds (new thresholds: 6 employees for enterprises with 20-150 employees and 5% or 30 employees for those with more than 150 employees, compared to previous thresholds of 4 employees per month for enterprises with 20-200 employees and 2-3% or 30 employees for enterprises with more than 200 employees); extension of probationary period from 2 months to 1 year. (pg. 123, 2011). Mention in other reports: yes in 2011; Score: 1.
- Greece, 2012 — area: severance pay. "The length of prior notice of dismissal has been shortened to a maximum of four months, compared to 24 months for white-collar workers previously. The severance pay for white-collar workers has been reduced and subjected to a ceiling of 12 months’ salary. (pg. 50, 2013)"; Mention in other reports: yes for 2012; Score: 1.
- Ireland, 1973 — area: notice for individual dismissal. Minimum Notice and Terms of Employment Act, 1973 introduces and defines minimum notice period for dismissal. Mention in other reports: no data but would qualify if scoring applied; Score: -1.
- Ireland, 1977 — areas: procedural inconvenience, notice for individual dismissal. Protection of Employment Act (1977), Unfair Dismissals Act (1977), Employment Equality Act (1977) enacted. Mention in other reports: (pg. 89, 1987); no data; Score: -1.
- Ireland, 2006 — area: notice for individual dismissal. Revision of the 1973 Minimum Notice and Terms of Employment Act. Mention in other reports: yes in 2006; Score: 1.
- Ireland, 2012 — area: severance pay. Government rebate to employers for redundancy payouts: up until 1 January 2012 rebate amounted to 60%; between 1 January 2012 and 1 January 2013 rebate was 15%; from 2013 onwards rebate was abolished. Mention in other reports: yes in 2012; Score: -1.
- Portugal, 1975 — area: collective dismissal. "Collective dismissal procedures become subject to regulation. (pg. 43, 1976)"; Mention in other reports: pg. 12, 1979; pg. 67, 1989; no data but would qualify if scoring applied; Score: -1.
- Portugal, 1976 — area: procedural inconvenience. Legislation virtually prohibiting all dismissals to combat unemployment and return of expatriates. (pg. 9, 1976). Mention in other reports: pg. 12, 1979; pg. 67, 1989; no data but would qualify if scoring applied; Score: -1.
- Portugal, 1978 — area: procedural inconvenience. August 29: Authorisation for firms to suspend work contracts on account of economic difficulties. (pg. 40, 1977). no data but would qualify if scoring applied; Score: 1.
- Portugal, 1990 — area: procedural inconvenience. "the possibility of dismissal for failure to fulfill job requirements (pg. 19, 1992)"; Mention in other reports: pg. 94, 1996; yes in 1990; Score: 1.
- Portugal, 1992 — area: procedural inconvenience. Changes in layoff legislation and collective labour contracts aimed at making labour markets more flexible. (pg. 92, 1993); yes in 1992; Score: 1.
- Portugal, 2004 — area: procedural inconvenience. New Labour Code (Código do Trabalho) came into force in December 2003; employers right to oppose reinstatement under certain conditions; eased procedures for collective dismissal (shortened deadlines; eliminated priority to trade union representatives and workers councils). (pg. 78-79, 2004; pg. 128, 2008); yes in 2004; Score: 1.
- Portugal, 2010 — areas: procedural inconvenience, notice for individual dismissal, severance pay. Introduction of new labour code in 2009 reducing EPL for regular contracts to reduce labour market dualism (pg. 42, 2010); Mention in other reports: pg. 33, 2012; yes in 2010; Score: 1.
- Spain, 1978 — area: procedural inconvenience. Decree-Law of 4th March, 1977 made dismissal regulations considerably more flexible (pg. 13, 1977); Mention in other reports: pg. 34, 1978; pg. 27, 1982; no data; Score: 1.
- Spain, 1981 — areas: procedural inconvenience, collective dismissals. New Workers Statute liberalised dismissals (pg. 27, 1981); Mention in other reports: pg. 27, 1982; no data; Score: 1.
- Spain, mid-1994/1995 — areas: procedural inconvenience, collective dismissals. Draft law simplifies layoff procedures; small-number dismissals no longer require prior consultation or administrative authorization; described as far-reaching reforms expected to increase labour market flexibility considerably. Mention in other reports: yes for 1995; Score: 1.
- Spain, 1998 — area: severance pay. Employers and trade unions agree on labour market reform encouraging creation of indefinite-term jobs. (entry truncated in source).

### Types of reforms and common measures (themes extracted from entries)
- Procedural inconvenience changes: simplifying layoff procedures; abolishing requirements for official authorisation to lay off workers; shortening dismissal procedures; revising Protection against Dismissal Act provisions; unified labour codes clarifying procedures.
- Notice for individual dismissal: extensions or reductions of notice periods (examples: increases to three months, reductions to two months, introductions of six-month warnings, tenure-based notice rules, shortening to one month or fourteen days).
- Severance pay reforms: introduction of retirement accounts funded by monthly untaxed payments of 1.5 per cent of gross wages (Austria, 2003); ability to pay in instalments; caps and ceilings (e.g., Greece ceiling of 12 months’ salary; Ireland rebate changes from 60% to 15% to abolition).
- Collective dismissal rules: changes in thresholds and notification requirements (e.g., Greece redefinition of thresholds; Netherlands compulsory 3-month advance notification for 20+ employees; Spain liberalisation of collective dismissal procedures).
- Labour market flexibility measures: promotion of fixed-term contracts, part-time work, temporary work; raising firm-size thresholds for EPL applicability (Germany: from five to ten employees, increasing exempt firms by some 15 percent employing some 30 per cent of employees).

### Scoring and indicator flags (as recorded in the table)
- Scores recorded in the dataset include values: -1 and 1 corresponding to direction of reform as coded in the source.
- Flags for "Mention in other reports" and "Large change in OECD indicator" appear as: explicit page citations (e.g., "pg. 27, 1971"), "yes" or "no", "no data but would qualify if scoring applied", and explicit "yes for [year]" markers (examples throughout table).

*Source: Appendix 1. Dataset of Reforms, wp18186 - Appendix 1. Dataset of Reforms.*

### introduction of a new type of indefinite-term

### introduction of a new type of indefinite-term

### Major labor market reforms and features (Spain)
- 1997 reform: introduced a new permanent contract with reduced redundancy costs for certain groups of workers; new definition of grounds for economic redundancies; proposals for improving the collective bargaining process. (pg. 179, 1998)
- Early 2001 measures (deepening of 1997 reform) include: (pg. 65-66, 2001)
  - Extension of the new permanent contract introduced in the 1997 labour market reform beyond May 2001.
  - The permanent contract with reduced firing costs will continue to apply to specific groups (workers aged 18-29, workers with a temporary contract, workers aged over 45, workers that have been unemployed for more than one year, women in some professions).
  - Extension to young workers now defined as those aged between 16 and 30, long-term unemployed (for more than 6 months), unemployed women in sectors where they are underrepresented (most of them) and disabled workers. (pg. 66, 2003)
- September 2010 labour market reform aims to reduce upper range of dismissal costs and smooth differences between temporary and permanent contracts (pg. 101, 2010; pg. 103, 2010):
  - If courts more readily accept firm dismissals as justified, severance payments could be reduced from the current practice of 45 days’ wages to 20 days’ wages.
  - Broadens the base for which the permanent contract with reduced severance payment of 33 days’ wages can be applied and guarantees reduced severance pay applies in cases where firms would prefer to declare the dismissal upfront as “unjustified”.
  - Introduction of a capital-funded component, similar to the Austrian severance pay reform, further reduces the one-time costs of dismissal.
  - The reform adopted by Parliament in September 2010 is characterized as representing significant progress. (pg. 101, 2010)
- 2012 labour market reform further reduces duality and reforms employment protection legislation (pg. 34, 2012; pg. 98, 2012):
  - Redefines economic reasons for dismissal; employer pays 20 days’ wages of severance pay per year of seniority for justified objective dismissals.
  - If dismissal is judged unjustified, maximum severance pay reduced to 33 days’ wages per year of seniority up to a maximum of 24 months (compared with 45 days and a maximum of 42 months on the regular permanent contract before). Applies to all new contracts and future years of service on existing contracts.
  - Eliminates need for administrative authorisation of collective dismissal.
  - Removes option of express dismissal (firms declaring dismissal upfront as “unjustified” and paying 45 days’ wages per year of seniority).
  - Additional changes: firms no longer obliged to pay interim wages during adjudication to avoid litigation; introduces a new type of permanent contract for companies with fewer than 50 employees with an extended trial period of one year (previous maximum six months) and various tax credits. (pg. 98, 2012)
- Assessments in source: “these reforms are a substantial step in the right direction...” and “A potentially important part of the reform is clarifying what justified dismissal means...” (pg. 34, 2012)

### Other country reforms and procedural changes (selected)
- Australia:
  - Workplace Relations Amendment (Work Choices) Act 2005 took effect in Q1 2006, reinforcing employers’ prerogatives at the expense of employees. (pg. 81, 2012)
  - Work Choices removed unfair dismissal protections for employees of firms with fewer than 100 employees. The Fair Work Act restored protections subject to qualifying periods of one-year service for workers in firms with fewer than 15 employees and six months’ service for workers in firms with 15 or more employees; broadened protections against discrimination and adverse actions. (pg. 83-84, 2012)
- New Zealand:
  - Employment Relations Act 2000 (ERA) modifies ECA provisions: mandatory mediation as first step to avoid undue litigation; establishment of the Employment Relations Authority as an investigative body; appeals and transfers to an Employment Court with possible redirection back to mediation or Authority. (pg. 78-79, 2000)
  - Amendment extending trial period provisions up to 90 days from firms with fewer than 20 employees to all firms on 1 April 2011. (pg. 56, 2011)
- Korea:
  - March 1997 labour law reform eased restrictions on layoffs by allowing dismissals for “urgent managerial reasons” with specified management requirements; Tripartite Commission agreed implementation in February 1998 to aid firm restructuring post-crisis. (pg. 166, 1998)
- Czech Republic:
  - New labour code passed early 2006 allowing a wider scope of employment contracts by adopting an “anglo-saxon” legal form (pg. 36, 2006); 2012 revision effective January 1st 2012 introduced wider possibilities for employers to terminate employment and extended probationary period to 6 months for executive employees.
- Slovak Republic:
  - Major amendments to the Labour Code adopted June 2003 effective 1 July 2003: increased flexibility for employer’s right to terminate; statutory notice period reduced to two months regardless of reason; employee with same employer more than five years given 3-months notice. (pg. 121-122, 2004)
  - 2012 amendments easing legislation on regular contracts (shortening of length of notice period). 

### Layoff rates (Appendix 2) — computation and key statistics
- Data source and method:
  - Uses 2014 Displaced Workers Survey (DWS) within IPUMS-CPS; survey covers around ninety thousand individuals (see Flood et al., 2017).
  - For each industry, counts workers displaced for: (a) plant or company closed or moved, (b) work was insufficient, (c) position or shift abolished; computed for 2011, 2012, 2013.
  - Industry-level employment changes for 2014 relative to 2011–2013 obtained from U.S. Bureau of Labor Statistics (BLS); multiplied by 2014 employment level to obtain employment levels for each year covered.
- Table A2.1 layoff rates (selected rows and average):
  - Average layoff rates by year: 2011 = 2.50, 2012 = 2.81, 2013 = 3.74, Average = 3.02.
  - Selected industry layoff rates (2011 / 2012 / 2013 / Average):
    - Food, Beverages & Tobacco: 3.04 / 1.99 / 4.07 / 3.03
    - Textiles: 2.18 / 3.10 / 5.92 / 3.73
    - Electrical & Optical: 4.67 / 5.96 / 6.25 / 5.62
    - Others Manufacturing: 8.54 / 7.48 / 5.92 / 7.31
    - Construction: 4.51 / 5.63 / 8.98 / 6.37
    - Accommodation & Food Services: 1.96 / 2.86 / 4.43 / 3.08
    - Education: 0.90 / 1.43 / 1.72 / 1.35
    - Health & Social: 1.31 / 1.50 / 2.54 / 1.78
- Notes: Layoff rates are the ratio of displaced workers for (i) company/plant closing, (ii) insufficient work, (iii) position or shift abolished over wage and salary employment, computed using individual-level data in the 2014 Displaced Workers Survey of the IPUMS-CPS.

### Elasticities of substitution (Appendix 3) — methodology and findings
- Production function and estimation strategy:
  - Aggregate production function framework: Y = H(K, L, X_i, A); assuming Hicks-neutral technical change and weak separability of K and L from other inputs, aggregate input F(K,L) characterized as constant elasticity of substitution (CES): F(K,L) = (αK^ε + (1−α)L^ε)^{1/ε}.
  - First-order condition equates price of capital to its marginal product leading to an estimable log equation (A3.1):
    - ln(P_{j,t}^K / P_{j,t}) = ln(α_j) + (1/σ) ln(F_{j,t}(K_{j,t},L_{j,t}) / K_{j,t}) + ε_{j,t}
  - Estimated separately for each industry using OLS; baseline uses EU KLEMS 2017 database for capital services and value added data.
  - Two approaches for rental price of capital: (i) nominal capital stock divided by volume of capital services; (ii) Jorgensonian rental rate accounting for depreciation and opportunity cost—preferred approach (Jorgenson 1963). Rental price formula used:
    - P_{i,j,t}^k = q_{i,j,t-1}^k i_{j,t-1} + δ_i^k q_{i,j,t}^k − (q_{i,j,t}^k − q_{i,j,t-1}^k)
  - Internal nominal interest rate i_{i,j,t} computed as a residual of capital compensation, depreciation and capital gains per O’Mahony and Timmer (2009).
  - Estimation restricted to a sample of 13 countries: Austria, Czech Republic, Denmark, Finland, France, Germany, Italy, Luxembourg, the Netherlands, Slovak Republic, Spain, Sweden and the United Kingdom. (footnote)
  - To address potential non-stationarity, Equation (A3.1) is estimated in first differences. Also estimate alternative specification allowing labor- and capital-augmenting technical change growing at constant rates τ_l and τ_k, leading to Equation (A3.2) which effectively adds a linear trend.
  - Baseline set EOS1: assumes Hicks-neutral technical change, uses capital services data and Jorgenson (1963) rental rates. Other EOS sets used for sensitivity analyses.
- Key empirical results and interpretation:
  - Summary across four alternative EOS sets (Table A3.1):
    - Average EOS across industries:
      - EOS1 = 0.71
      - EOS2 = 0.74
      - EOS3 = 0.85
      - EOS4 = 0.68
    - On average, estimated EOS are below one in all four cases, ranging from 0.68 (when allowing for labor- and capital- augmenting technical change) to 0.85 (when dividing nominal capital stock by capital services to obtain the rental rate of capital).
    - EOS is estimated to exceed 1 for just 4 to 6 industries depending on specification.
  - Selected industry EOS1 (baseline) estimates (industry code in source):
    - Food, Beverages & Tobacco: 0.69 (10t12)
    - Textiles: 0.38 (13t15)
    - Wood, Paper & Reproduction: 0.49 (16t18)
    - Coke & Refined Petroleum: 0.88 (19)
    - Chemicals: 0.90 (20t21)
    - Electrical & Optical: 0.58 (26t27)
    - Machinery & Equipment: 0.49 (28)
    - Transport Equipment: 0.47 (29t30)
    - Others Manufacturing: 0.44 (31t33)
    - Transport & Storage: 1.36 (49t52) — note exceeds 1 in EOS1
    - Telecommunications: 1.48 (61) — note exceeds 1 in EOS1
    - Construction: 0.30 (F)
    - Real Estate: 1.20 (L)
    - Other Service Activities: 1.31 (S)
  - Correlation among the four EOS sets ranges from 0.4 to 0.9.
  - Baseline interpretation: labor and capital are, on average, imperfect substitutes (EOS < 1) across industries in the sample under the baseline assumptions and preferred rental rate construction.

### Methodological caveats flagged in source
- Potential non-stationarity of variables could bias OLS; authors estimate in first differences to address this.
- Possible endogeneity of the regressor (firm demand for capital); an IV strategy is used in other literature (Antràs 2004) but is problematic here due to industry-country panel; authors proceed with OLS.
- EOS estimates may be biased if Hicks-neutral technical change assumption fails; authors relax assumption and estimate alternative specification with labor- and capital-augmenting technical change.
- Limited availability of capital stock and services data in EU KLEMS restricts estimation to 13 countries.

*Source: wp18186 - introduction of a new type of indefinite-term (PDF chapter/section).*

### Appendix 4. Summary Statistics of Industry Data

### Appendix 4. Summary Statistics of Industry Data

### Table A4.1 — Mean values of relevant industry characteristics
- Columns: Share in value added | Labor share | Elasticity of substitution | Natural layoff rate | Industry code
- Food, Beverages & Tobacco: 2.66 | 57.25 | 0.69 | 3.03 | 10t12
- Textiles: 1.25 | 77.51 | 0.38 | 3.73 | 13t15
- Wood, Paper & Reproduction: 1.86 | 69.14 | 0.49 | 3.81 | 16t18
- Coke & Refined Petroleum: 0.44 | 41.42 | 0.88 | 1.07 | 19
- Chemicals: 2.19 | 47.82 | 0.90 | 2.58 | 20t21
- Rubber & Plastics: 1.75 | 65.16 | 0.53 | 2.41 | 22t23
- Basic Metals: 2.84 | 68.34 | 0.45 | 3.24 | 24t25
- Electrical & Optical: 2.33 | 62.48 | 0.58 | 5.62 | 26t27
- Machinery & Equipment: 1.67 | 71.09 | 0.49 | 2.86 | 28
- Transport Equipment: 1.85 | 68.92 | 0.47 | 2.77 | 29t30
- Others Manufacturing: 1.34 | 77.38 | 0.44 | 7.31 | 31t33
- Wholesale & Retail, Motor vehicles: 1.43 | 69.20 | 0.47 | 2.32 | 45
- Wholesale ex. Motor Vehicles: 5.69 | 62.13 | 0.52 | 2.69 | 46
- Retail ex. Motor Vehicles: 5.00 | 76.44 | 0.39 | 2.57 | 47
- Transport & Storage: 4.75 | 67.43 | 1.36 | 2.94 | 49t52
- Postal & Courier: 0.89 | 82.18 | 0.73 | 1.44 | 53
- Publishing & Audiovisual: 1.36 | 65.86 | 0.54 | 3.21 | 58t60
- Telecommunications: 1.73 | 39.35 | 1.48 | 1.98 | 61
- IT & Others: 1.30 | 78.76 | 0.37 | 3.10 | 62t63
- Agriculture: 3.35 | 79.65 | 0.90 | 3.61 | A
- Mining & Quarrying: 1.53 | 41.27 | 0.79 | / | B
- Utilities: 2.84 | 34.35 | 0.95 | 1.59 | DtE
- Construction: 6.71 | 78.85 | 0.30 | 6.37 | F
- Accommodation & Food Services: 2.69 | 76.86 | 0.47 | 3.08 | I
- Financial & Insurance: 5.95 | 57.42 | 0.88 | 2.34 | K
- Real Estate: 8.95 | 6.83 | 1.20 | 1.90 | L
- Professional & Support Activities: 7.20 | 72.87 | 0.56 | 3.68 | MtN
- PA, Defense & SS: 7.21 | 76.15 | 0.86 | / | O
- Education: 4.90 | 90.20 | 0.58 | 1.35 | P
- Health & Social: 6.01 | 84.21 | 0.63 | 1.78 | Q
- Arts & Recreation: 1.12 | 72.34 | 1.11 | 3.40 | R
- Other Service Activities: 1.55 | 84.49 | 1.31 | 2.73 | S
- Average: 3.20 | 65.73 | 0.71 | 3.02 | /

Notes:
- "share in value added" and "labor share" are averages across countries and years, computed from the EU KLEMS database.
- Elasticities of substitution are estimated according to Equation (A3.1).
- The natural layoff rate is calculated as the average ratio of displaced over wage and salary employment across the years 2011-2013, computed using individual level data contained in the 2014 Displaced Workers Survey of the IPUMS-CPS.

### Appendix 5 — Stylized Facts
- Long-run trend:
  - The labor share has been on a declining trend since the mid-1970s, with decline accelerating somewhat in the 1990s.
  - Exceptions: global recessions of the early 1990s and 2009 saw labor share increases due to a very small decline in labor compensation relative to value added.
- Cross-country and cross-industry heterogeneity:
  - Estimated linear trends in country labor shares: negative and significant in 15 out of 22 countries.
  - Estimated linear trends in industry labor shares: of 32 industries, 23 display a negative and statistically significant coefficient, and 4 have a significant positive coefficient.
  - No single sector emerges as an outlier; decline is broad based but heterogeneous.
- Within vs. between decomposition (Equation (A5.1)):
  - Decomposition formula: ∆LS_j = ∑_i ω̅_ij ∆LS_ij + ∑_i LS̅_ij ∆ω_ij
    - First term = within-industry component
    - Second term = between-industry (composition) component
  - Empirical finding: plotting aggregate country trends against the within-industry component yields a linear regression that explains about 70 percent of the country variation.
  - Interpretation: within-industry changes are more important than changes in industrial composition in explaining movements at the country level.

Notes on figures and estimation:
- Figure A5.1: coefficients of year fixed effects from regression LS_jit = α + ρ_t + γ_ji + ε_jit; estimates interpreted as average labor share change in percentage points relative to 1970.
- Figure A5.2: estimated linear trends in country labor shares from LS_tj = α_j + τ_tj + ε_tj; trends interpreted as average yearly change over period (mostly 1970-2015).
- Figure A5.3: estimated linear trends in aggregate industry labor shares from LS_ti = α_i + τ_ti + ε_ti.
- Figure A5.4: plots country aggregate labor share trends over within-industry components computed as y_j = ∑ ω̅_ij ∆LS_ij.

### Appendix 6 — Back-of-the-envelope Calculation Using Country-industry-level Estimates
- Aggregate labor share as weighted sum:
  - LS_tj ≡ ∑_i LS_i,t_j θ_i,t_j where θ_i,t_j denotes added value share of industry i.
- Marginal impact of an EPL reform (assuming EPL does not affect value-added shares):
  - ∂LS_t / ∂EPL_t = ∑_i (∂LS_i,t / ∂EPL_t) θ_i,t
- Two-group simplification (groups 1 and 2) with identical impacts within each group:
  - ∂LS_t / ∂EPL_t = ∂(LS_i,t^1 − LS_i,t^2)/∂EPL_t · θ_i,t^1 + ∂LS_i,t^2/∂EPL_t
  - If EPL reform has negligible effect on group 2, then:
    - ∂LS_t / ∂EPL_t ≈ ∂(LS_i,t^1 − LS_i,t^2)/∂EPL_t · θ_i,t^1
- Identification and empirical application:
  - Industries split into two groups: group 1 = natural layoff rate above the median and EOS below 1; group 2 = remaining industries.
  - Equation (9) estimated using industry identification variable ϑ_i = 1 for group 1, 0 for group 2.
  - Assumption: EPL reforms do not have effects in group 2 (consistent with statistically insignificant coefficient at five-year horizon in Table 8).
- Empirical result:
  - Applying the five-year-ahead coefficient estimate to the number of net liberalizing reforms in each country over the period considered, changes in EPL may explain about 15 percent of the overall labor share decline. This is roughly similar to the figure obtained using country-level estimates.

Table A6.1 — Country-industry analysis: extension on back-of-the-envelope calculation of aggregate effects
- Impact horizons: 1y | 2y | 3y | 4y | 5y
- Identification through qualitative 0/1 dummy variable
- Labor share: -0.22 | -1.39 | -1.54 | -1.58 | -1.33 | -1.48
- Value added: 0.10 | 0.64 | 0.79 | 1.55 | 1.38 | 1.26

Notes:
- Estimates based on Equation (9) using for identification a dummy taking value 1 for industries with elasticity of substitution below 1 and layoff rate above median, and 0 otherwise.
- The rows "Labor share” and “Value added” report estimates obtained using, respectively, the labor share and value added as dependent variables.
- Bold numbers indicate significance at the 90 percent confidence interval, based on clustered standard errors at the country-industry level.
- Additional remark: re-running Equation (9) with (log) value added as dependent variable finds no significant effects of EPL reform, supporting the assumption of negligible reallocation effects across the two groups.

*Source: wp18186 - Appendix 4. Summary Statistics of Industry Data.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2018/wp18186.pdf_