## 1. By Country Groups

## Source details

**Canonical URL:** [1. By Country Groups](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020019-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020019-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020019-print-pdf.pdf.json)

---

### Robustness: Alternative definitions of informality (share of self-employed workers)
- Purpose: re-estimate regressions using the share of self-employed workers (World Bank WDI) as dependent variable to assess robustness and broaden country coverage.
- Main finding: countercyclicality of informality is broadly confirmed across specifications, with estimated coefficients smaller in absolute value than those in Table 1, indicating smaller sensitivity of the self-employment measure to output changes.
- Regional variation: the share of self-employed workers seems to vary more with the cycle in LAC countries compared to the full set of countries.
- Selected regression coefficients (Table 2):
  - ∆ GDP_t: -0.0227*** (FE, All Countries); -0.0596*** (FE, LAC Only); -0.0427*** (CCE, All Countries); -0.0597*** (CCE, LAC Only); -0.0416*** (CCE, All Countries, alt); -0.0740*** (CCE, LAC Only, alt).
  - ∆ GDP_t-1: -0.00671 (FE, All Countries); 0.0424 (FE, LAC Only).
  - Constant examples: -0.0391; 0.517**; 0.0241; -0.0284; 0.0261; -0.0782.
  - Observations: 3,651 (All Countries specifications); 546 (LAC Only specifications).
  - Countries: 142 (All Countries); 21 (LAC Only).
- Notes: CCE = Common Correlated Effects Estimation. Standard errors reported in parentheses. *** p<0.01, ** p<0.05, * p<0.1.

### Informality and labor market characteristics (cross-country correlations)
- Variables analyzed (Figure 2): ratio of informal employment over total non-agricultural employment versus:
  - severance payments (redundancy costs) measured in weeks of salary.
  - ratio of the minimum wage to value added per worker (minimum wage ratio).
- Data source: World Bank, Doing Business Indicators.
- Qualitative finding: positive correlations observed between informality rates and:
  - regulations that increase job security (redundancy costs).
  - higher minimum wage ratio.
- Country labeling: ISO country codes used in figures.

### De jure versus de facto labor market regulations and enforcement
- Concept: similar (stringent) labor regulations may be less binding where government enforcement is weak; analysis follows Caballero et al. (2013) by interacting regulations with a proxy for government enforcement (government effectiveness).
- Key empirical results (Table 3 summary):
  - Positive correlation between regulations that increase job security and informality rates (Table 3, Columns 1–5), especially when these regulations are more strictly enforced.
  - High minimum wage is positively correlated with informality:
    - High minimum wage coefficients: 0.349***; 0.324***; 0.141**; 0.0713*; 0.0755*; 0.0595 (across columns).
  - Interaction terms with Government effectiveness show that stricter enforcement can amplify the effect of regulation on informality:
    - Job security * Government effectiveness: 0.0847*; 0.0253 (selected columns).
    - High minimum wage * Government effectiveness: 0.218*; 0.0755 (selected columns).
  - Government effectiveness main effect is negative:
    - Examples: -0.488***; -0.395***; -0.121**; -0.118***.
  - Inclusion of Log GDP per capita reduces statistical significance but coefficients on regulation remain supportive of a link between regulation and informality:
    - Log GDP per capita coefficients: -0.159***; -0.160***; -0.151***; -0.138***; -0.133***.
  - Observations across specifications: between 104 and 110 observations in various columns; R-squared ranges from 0.144 to 0.784 depending on specification.
- Note: High minimum wage is a dummy equal to one if minimum wage to labor productivity ratio exceeds the sample average. Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1.

### Labor market flexibility and the speed of adjustment of employment
- Motivation: a flexible labor market aids aggregate adjustment to shocks and productivity growth; speed of adjustment determines how quickly employment reverts to its long-run relationship with GDP.
- Methodology:
  - Error-correction model (ECM) in a heterogeneous panel following Eberhardt and Presbitero (2015).
  - Long-term equilibrium employment related to GDP; ECM examines response of employment growth to GDP shocks and deviations from the long-run employment-GDP relationship.
  - Heterogeneous parameters: β_i (response of employment growth to GDP shocks), θ_i (long-run elasticity of employment to GDP), α_i (speed of adjustment).
  - Use of cross-sectional averages to capture unobservable common factors; CCE estimator with maximum lag p=2 chosen per Chudik and Pesaran (2015).
  - Data: annual, 171 countries, 1990–2017.
- Main empirical patterns (Table 4, summary):
  - Employment growth positively correlated with contemporaneous GDP growth across specifications:
    - GDP growth coefficients: 0.031*** (FE); 0.11*** (CCE); 0.106*** (CCE alt); 0.125*** (CCE alt).
  - Employment growth negatively correlated with “excess” employment (employment above GDP-predicted levels), indicating reversion.
  - Lagged log employment coefficients (speed element proxies):
    - -0.018*** (FE estimation);
    - -0.189***; -0.199***; -0.214*** (CCE specifications).
  - Lagged log GDP: 0.003 (FE); 0.086***; 0.081***; 0.067*** (CCE).
  - Implied long-run elasticity examples: 0.1535 (FE); 0.4581*** (CCE); 0.4084*** (CCE); 0.3141 (CCE alt).
  - Observations: 4,327 (FE); 4,025; 3,993; 3,881 (CCE/alternatives). Number of countries: 171; 171; 170; 164.
- Key statistics on speed of adjustment (distributional results):
  - Average estimated speed of adjustment coefficient: -0.22.
    - Interpretation: average country takes approximately 3 years to close half the employment gap (half-life).
  - Heterogeneity: approximately 50 percent of countries have a speed of adjustment higher (in absolute value) than the average.
  - 75th percentile speed of adjustment: -0.41 (example: approximately the estimated coefficient for Nicaragua and El Salvador), implying a half-life of 1.3 years.
  - Median by groups:
    - Median LAC speed of adjustment: -0.2.
    - Median advanced economy: -0.28.
    - Median emerging market: -0.26.
  - Within LAC: South America median speed lower than Central America median.
  - Noted caveat: despite point estimate differences, many country differences may not be statistically significant; several large LAC economies (Chile, Brazil, Peru, Mexico) show coefficients ranging from -0.2 to 0.2 and are statistically non-significant.
- Methodological note: omission of common factors (in fixed effects) leads to downward bias; average speed of adjustment in CCE model substantially larger than in standard fixed effects (factor of 10 in comparison of Columns 1 and 2).

*Source: Authors’ calculations based on International Labour Organization, World Bank, World Development Indicators, World Bank Doing Business Indicators, and Inter-American Development Bank SIMS.*

### 2. Selected LAC Economies

### Correlates of the Speed of Adjustment
- Higher informality rates and stricter job security regulations are associated with a lower speed of adjustment of employment.
- The estimated coefficient for informality implies that a 10 percentage point decrease in informality (roughly the difference between the region’s average and the average for EMs in the sample of 110 countries) would raise LAC’s speed of adjustment to roughly the EM average.
- A decrease of 1 point in the job security composite index (for example, by eliminating third party dismissal notification) would raise the region’s speed of adjustment of employment to roughly the estimated average for AEs.
- Informality acts as a buffer that attenuates the impact of GDP shocks on unemployment but makes the adjustment process more protracted.
- Informality is heterogeneous: evidence for Brazil (Ulyssea, 2018) classifies informal establishments into:
  - “survival” firms (not productive enough to formalize),
  - “opportunistic” firms (avoid formalization when possible),
  - productive informal establishments that would formalize if costs were low enough.
  - Roughly half of informal firms in Brazil are “survival”; close to 40 percent are “opportunistic”.
- Labor market segmentation implies large costs of switching from informal to formal jobs (Arias et al., 2018), comparable to switching across sectors.

### Micro-Flexibility and the Speed of Adjustment — Methodology
- Micro model estimated: Δe_{i,t} = α_i + λ_i (e^*_{i,t} − e_{i,t−1}) + ε_{i,t}, where λ_i is the speed of adjustment.
- Two-step econometric approach (Caballero et al. 2013):
  - First stage: construct proxy of employment gap by estimating key variables of the micro-founded model (including sectoral productivity differences).
  - Second stage: estimate the error-correction equation using the employment gap proxy.
- Extensions relative to Caballero et al. (2013):
  - Change period of analysis from 1980–2000 to 2000–2017.
  - Assess role of informality, policies governing worker compensation, and macro-stabilization policies.
  - Compare speed of adjustment using manufacturing-only (UNIDO) data vs. manufacturing + services + construction (UNIDO + Timmers, de Vries and de Vries 10-sector dataset + OECD STAN).

### Empirical Results — Reference Estimates and Dataset Differences
- Speed of adjustment estimates differ by dataset:
  - Manufacturing-only (UNIDO) sample: sectors close approximately 50 percent of the employment gap in each period (Table 5, Column 1: Employment gap coefficient = 0.501***; standard error (0.0427)).
  - Manufacturing + services + construction: coefficient ≈ 45.9 percent (Table 5, Column 2: Employment gap coefficient = 0.459***; standard error (0.0410)).
- The combined dataset yields slower adjustment estimates; concerns about aggregation and coverage lead analysis to focus on the UNIDO dataset.

### UNIDO Sample — LAC vs. Non-LAC and Role of Informality and Regulations
- LAC exhibits slower speed of adjustment than other countries:
  - Column 3 (UNIDO): the speed of adjustment for the average LAC country is approximately 2.5 percentage points lower compared to countries outside the region (Employment gap coefficient in Column 3 = 0.502***; interaction Employment gap * LAC = −0.0255*; standard error (0.0143)).
- Informality associated with slower adjustment:
  - Employment gap * Informality = −0.0596*** (standard error (0.0216)) (Table 5, Column 4).
- Job security and minimum wages relative to labor productivity reduce microeconomic flexibility:
  - Employment gap * Job security = −0.0374*** (standard error (0.00561)) and −0.0223*** (standard error (0.00654)) in related specifications.
  - Employment gap * (Minimum wage/Labor productivity) = −0.131*** (standard error (0.0175)) and −0.0542*** (standard error (0.0185)) in related specifications.
- Enforcement (government effectiveness) amplifies these effects:
  - Employment gap * Job security * High government effectiveness = −0.0678*** (standard error (0.0134)).
  - Employment gap * (Minimum wage/Labor productivity) * High government effectiveness = −0.335*** (standard error (0.0532)).
- Positive association of high government effectiveness with speed:
  - Employment gap * High government effectiveness = 0.0469*** (standard error (0.0157)) in some specifications.
- Table 5 selected statistics:
  - Constant terms examples: 0.000832; 0.00460*** (standard error (0.000675)).
  - Observations vary by specification: 27988; 30895; 27647; 20123; 27056; 26942; 76472; 7585.
  - Number of groups reported across specifications: 1604; 1693; 1586; 1141; 1553; 1549; 1586; 1582.
  - Significance notation: *** p<0.01, ** p<0.05, * p<0.1.

### Economic Magnitude — Implied Speed of Adjustment and Half-Life (Table 6)
- Average and subgroup implied speeds and half-lives (selected):
  - Average speed of adjustment = 0.50; half life = 12.00 (months).
  - LAC speed of adjustment = 0.48; half life = 12.91 (months).
  - Low informality: speed = 0.51; half life = 11.79.
  - High informality: speed = 0.47; half life = 13.26; implied growth differential (low-high) = 0.17pp.
  - Low, High government effectiveness: speed = 0.54; half life = 10.56.
  - High, High government effectiveness: speed = 0.43; half life = 14.61; implied growth differential = 0.47pp.
  - Low, Low government effectiveness: speed = 0.52; half life = 11.39.
  - High, Low government effectiveness: speed = 0.49; half life = 12.32; implied growth differential = 0.11pp.
  - For minimum wage/Labor productivity dimension: Low, High government effectiveness: speed = 0.55; half life = 10.56. High, High government effectiveness: speed = 0.39; half life = 16.85; implied growth differential = 0.74pp.
- Additional magnitudes:
  - Informality: difference between 80th and 20th percentile implies ≈4 percentage points lower speed of adjustment for high-informality country relative to low-informality country; translates into 1.5 additional months to close half the employment gap.
  - Difference in speed between low and high employment protection is 10 percentage points when government effectiveness is high → ≈4 additional months to close half the employment gap.
  - Difference in speed between low and high relative minimum wages is 16 percentage points when enforcement is high → ≈6 additional months to close half the employment gap.
- Half-life calculation used: Half-life (months) = 12*(log(0.5)/log(1 − speed of adjustment)).

### Sectoral Heterogeneity — Labor Intensity and Regulations (Table 7)
- Labor market regulations have heterogeneous effects across sectors, amplified in labor-intensive sectors.
- Sectoral interaction results (selected):
  - Employment gap * Labor intensity = 0.485*** (standard error (0.160)) and 0.625*** (standard error (0.186)) in alternative specifications.
  - Employment gap * Job security * Labor intensity = −0.174 (standard error (0.115)) (not significant at high levels reported).
  - Employment gap * (Minimum wage/Labor productivity) * Labor intensity = −0.997*** (standard error (0.320)) — indicating strong negative impact of high minimum wages on adjustment in labor-intensive sectors.
- Interactions with exchange rate flexibility:
  - Employment gap * Exchange Rate Flexibility = 0.0802*** (standard error (0.0117)) and 0.0889*** (standard error (0.0201)).
  - Employment gap * (Minimum wage/Labor productivity) * Exchange Rate Flexibility = 0.199*** (standard error (0.0671)).
  - Employment gap * Job security * Exchange Rate Flexibility = −0.0107 (standard error (0.0149)) (not significant).

### Asymmetric Effects of Labor Market Regulations
- Effects differ by sign of employment gap:
  - Dummy = 1 if employment gap is negative (employment above equilibrium).
  - EPLs slow adjustment more when employment is above equilibrium (negative gap):
    - Employment gap * Job security * Negative gap dummy = −0.0414** (standard error (0.0166)).
  - Binding minimum wages affect adjustment more when employment is below equilibrium (positive gap):
    - Employment gap * (Minimum wage/Labor productivity) * Negative gap dummy = 0.0927* (standard error (0.0484)).
  - Baseline asymmetry example: Employment gap * Negative gap dummy = 0.0690*** (standard error (0.0212)) in some specifications.

### Labor Market Regulation and Growth
- Slower labor market responsiveness can lower productivity growth by hampering reallocation of factors across sectors and firms.
- Simple correlation: GDP per worker growth is positively correlated with the country’s speed of adjustment (Figure 5 referenced).
- Back-of-envelope: changes in employment protection regulations moving a country from the 80th percentile to the 20th percentile of the speed of adjustment distribution are associated with an increase in medium-term labor productivity growth (Caballero et al. 2013 framework referenced).

### Main findings (summary and policy-relevant magnitudes)
- Changing the minimum wage from the 80th to the 20th percentile of the distribution when government effectiveness is high can increase growth by approximately 0.75 pp per annum.
- A change equivalent to 0.5 percentage points (pp) per annum is reported (Table 5, last column).
- Labor market regulations (employment protection legislation, EPL, and high minimum wages relative to labor productivity) lower labor productivity growth, with larger effects in labor-intensive sectors.
- Informality:
  - Dampens effects of macroeconomic shocks on employment (provides a buffer for low-skilled workers in downturns) but makes employment adjustment more sluggish.
  - Is linked to strict labor market regulations (certain dimensions of stricter EPL increase informality).
  - High minimum wages relative to labor productivity increase informality, lower the speed of adjustment to shocks, and hamper growth.

### Methods and empirical strategy (selected)
- Difference-in-difference (DID) approach inspired by Rajan and Zingales (1998) to exploit heterogeneous sectoral labor intensity:
  - Estimated equation (4): g_ijt = α_jt + β * (reg_i * α_jt-1) + γ X_ijt + θ_i + ε_ijt (notation as in source).
  - ijt g is sector j’s labor productivity growth in country i at time t.
  - α_jt is the labor share in value added of sector j at time t (median labor share across countries).
  - reg_i is either the proxy for employment protection legislations or the minimum wage over labor productivity ratio.
  - X_ijt controls include sector’s capital share, sector’s initial share in total value added, and country-year fixed effects.
- Sectoral CCE-ECM estimated for agriculture, industry, services to obtain speed of adjustment coefficients.
- Caballero et al. (2013) two-step framework used to link microeconomic flexibility to long-term growth differences.

### Quantitative results and robustness (Table 8 and appendix notes)
- Sectoral and interaction coefficients (selected exact entries):
  - Sector's share in country's total value added, t-1: -0.156*** (0.0174); -0.175*** (0.0180); -0.176*** (0.0183); -0.174*** (0.0180); -0.154*** (0.0178); -0.153*** (0.0174).
  - Sectoral labor share, t-1: -0.132*** (0.0327); -0.0466 (0.0519); -0.0366 (0.0546).
  - Country's job security * Sectoral labor share, t-1: -0.0884** (0.0423).
  - Country's rel. minimum wage * Sectoral labor share, t-1: -0.267** (0.122).
  - Country's job security * Sectoral capital share, t-1: -0.0249 (0.119).
  - Country's rel. minimum wage * Sectoral capital share, t-1: -0.00348 (0.377).
  - Constant terms: 0.0455*** (0.00159); 0.0628*** (0.00459); 0.0629*** (0.00462); 0.0624*** (0.00459); 0.0568*** (0.00359); 0.0564*** (0.00350).
  - Country-Year Fixed Effects: YES.
  - Observations: 26,552; 26,539; 25,977; 26,539; 25,856; 26,418.
  - R-squared: 0.196; 0.197; 0.199; 0.197; 0.199; 0.197.
- Appendix C: industry shows highest average estimated speed of adjustment; services the lowest. Informality reduces the speed parameter for services and agriculture but not for industry.
- Estimation detail: Murphy-Topel standard errors used in second stage of Caballero et al. (2013) two-step estimation to account for φ estimation error.

### Policy implications and trade-offs
- Two key trade-offs when designing labor market regulations:
  1. Between employment protection and labor market dynamism & job quality:
     - Stringent EPL may limit employment losses in downturns but can increase informal employment (typically lower paying) and lead to sluggish recoveries.
     - Possible partial resolution: complement less-stringent EPL with a modern safety net (robust unemployment insurance and other benefits).
  2. Between labor market legislation and effectiveness of macro-stabilization tools:
     - EPL can undermine the employment benefits of macro-stabilization tools such as exchange rate flexibility.
     - Policy design could consider contingency mechanisms (examples provided):
       - Severance payments or minimum wage adjustments contingent on the state of the business cycle.
- Structural reforms that increase labor productivity are recommended as the best way to address the informality determinant related to high minimum wages relative to labor productivity (as suggested in IMF (2019b)).

### Data sources and variable definitions (selected)
- Real GDP growth: Change in log of real GDP (in %), 178 countries, 1990-2017, IMF WEO.
- Total employment: Share of employment to population 15+ (in %), 140 countries, 1990-2017, World Bank WDI.
- Informal employment (ILO): Share of informal employment on total non-agricultural employment (in %), 119 countries, 2016, ILO (2018).
- Informal employment (IDB): Share of total active workers that do not contribute to social security (in %), 17 LAC countries, 1990-2017, IDB SIMS.
- Informal employment (WB): Share of self-employed workers (in %), 142 countries, 1991-2017, World Bank WDI.
- Minimum wage ratio: Ratio between the national minimum wage and GDP per worker (labor productivity), 188 countries, 2014-2018, Doing Business indicators.
- Redundancy costs: Dismissal costs in weeks of salary, 189 countries, 2014-2018, Doing Business indicators.
- Government effectiveness: Dummy = 1 if government effectiveness estimate in 1996 above global median in that year, 214 countries, 1996, Kaufmann, Kraay, and Mastruzzi (2010).
- Exchange rate flexibility: Dummy = 1 if IRR index values 3 and 4 (flexible), zero if 1 and 2 (non-flexible), 194 countries, 1990-2016, Ilzetzki, Reinhart and Rogoff (2019).
- Job security indicator: Index following Botero et al. (2004), normalized 0–1, 186 countries, average for 2014-2018, authors’ calculations based on Doing Business indicators.
- Employment, output, and wages by sector: 173 countries, 1963-2017, INDSTAT and STAN databases.

_Italic: Source — wpiea2020019-print-pdf._

### 1.  By Country Groups

### 1. By Country Groups

### Robustness: Alternative definitions of informality (share of self-employed workers)
- Purpose: re-estimate regressions using the share of self-employed workers (World Bank WDI) as dependent variable to assess robustness and broaden country coverage.
- Main finding: countercyclicality of informality is broadly confirmed across specifications, with estimated coefficients smaller in absolute value than those in Table 1, indicating smaller sensitivity of the self-employment measure to output changes.
- Regional variation: the share of self-employed workers seems to vary more with the cycle in LAC countries compared to the full set of countries.
- Selected regression coefficients (Table 2):
  - ∆ GDP_t: -0.0227*** (FE, All Countries); -0.0596*** (FE, LAC Only); -0.0427*** (CCE, All Countries); -0.0597*** (CCE, LAC Only); -0.0416*** (CCE, All Countries, alt); -0.0740*** (CCE, LAC Only, alt).
  - ∆ GDP_t-1: -0.00671 (FE, All Countries); 0.0424 (FE, LAC Only).
  - Constant examples: -0.0391; 0.517**; 0.0241; -0.0284; 0.0261; -0.0782.
  - Observations: 3,651 (All Countries specifications); 546 (LAC Only specifications).
  - Countries: 142 (All Countries); 21 (LAC Only).
- Notes: CCE = Common Correlated Effects Estimation. Standard errors reported in parentheses. *** p<0.01, ** p<0.05, * p<0.1.

### Informality and labor market characteristics (cross-country correlations)
- Variables analyzed (Figure 2): ratio of informal employment over total non-agricultural employment versus:
  - severance payments (redundancy costs) measured in weeks of salary.
  - ratio of the minimum wage to value added per worker (minimum wage ratio).
- Data source: World Bank, Doing Business Indicators.
- Qualitative finding: positive correlations observed between informality rates and:
  - regulations that increase job security (redundancy costs).
  - higher minimum wage ratio.
- Country labeling: ISO country codes used in figures.

### De jure versus de facto labor market regulations and enforcement
- Concept: similar (stringent) labor regulations may be less binding where government enforcement is weak; analysis follows Caballero et al. (2013) by interacting regulations with a proxy for government enforcement (government effectiveness).
- Key empirical results (Table 3 summary):
  - Positive correlation between regulations that increase job security and informality rates (Table 3, Columns 1–5), especially when these regulations are more strictly enforced.
  - High minimum wage is positively correlated with informality:
    - High minimum wage coefficients: 0.349***; 0.324***; 0.141**; 0.0713*; 0.0755*; 0.0595 (across columns).
  - Interaction terms with Government effectiveness show that stricter enforcement can amplify the effect of regulation on informality:
    - Job security * Government effectiveness: 0.0847*; 0.0253 (selected columns).
    - High minimum wage * Government effectiveness: 0.218*; 0.0755 (selected columns).
  - Government effectiveness main effect is negative:
    - Examples: -0.488***; -0.395***; -0.121**; -0.118***.
  - Inclusion of Log GDP per capita reduces statistical significance but coefficients on regulation remain supportive of a link between regulation and informality:
    - Log GDP per capita coefficients: -0.159***; -0.160***; -0.151***; -0.138***; -0.133***.
  - Observations across specifications: between 104 and 110 observations in various columns; R-squared ranges from 0.144 to 0.784 depending on specification.
- Note: High minimum wage is a dummy equal to one if minimum wage to labor productivity ratio exceeds the sample average. Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1.

### Labor market flexibility and the speed of adjustment of employment
- Motivation: a flexible labor market aids aggregate adjustment to shocks and productivity growth; speed of adjustment determines how quickly employment reverts to its long-run relationship with GDP.
- Methodology:
  - Error-correction model (ECM) in a heterogeneous panel following Eberhardt and Presbitero (2015).
  - Long-term equilibrium employment related to GDP; ECM examines response of employment growth to GDP shocks and deviations from the long-run employment-GDP relationship.
  - Heterogeneous parameters: β_i (response of employment growth to GDP shocks), θ_i (long-run elasticity of employment to GDP), α_i (speed of adjustment).
  - Use of cross-sectional averages to capture unobservable common factors; CCE estimator with maximum lag p=2 chosen per Chudik and Pesaran (2015).
  - Data: annual, 171 countries, 1990–2017.
- Main empirical patterns (Table 4, summary):
  - Employment growth positively correlated with contemporaneous GDP growth across specifications:
    - GDP growth coefficients: 0.031*** (FE); 0.11*** (CCE); 0.106*** (CCE alt); 0.125*** (CCE alt).
  - Employment growth negatively correlated with “excess” employment (employment above GDP-predicted levels), indicating reversion.
  - Lagged log employment coefficients (speed element proxies):
    - -0.018*** (FE estimation);
    - -0.189***; -0.199***; -0.214*** (CCE specifications).
  - Lagged log GDP: 0.003 (FE); 0.086***; 0.081***; 0.067*** (CCE).
  - Implied long-run elasticity examples: 0.1535 (FE); 0.4581*** (CCE); 0.4084*** (CCE); 0.3141 (CCE alt).
  - Observations: 4,327 (FE); 4,025; 3,993; 3,881 (CCE/alternatives). Number of countries: 171; 171; 170; 164.
- Key statistics on speed of adjustment (distributional results):
  - Average estimated speed of adjustment coefficient: -0.22.
    - Interpretation: average country takes approximately 3 years to close half the employment gap (half-life).
  - Heterogeneity: approximately 50 percent of countries have a speed of adjustment higher (in absolute value) than the average.
  - 75th percentile speed of adjustment: -0.41 (example: approximately the estimated coefficient for Nicaragua and El Salvador), implying a half-life of 1.3 years.
  - Median by groups:
    - Median LAC speed of adjustment: -0.2.
    - Median advanced economy: -0.28.
    - Median emerging market: -0.26.
  - Within LAC: South America median speed lower than Central America median.
  - Noted caveat: despite point estimate differences, many country differences may not be statistically significant; several large LAC economies (Chile, Brazil, Peru, Mexico) show coefficients ranging from -0.2 to 0.2 and are statistically non-significant.
- Methodological note: omission of common factors (in fixed effects) leads to downward bias; average speed of adjustment in CCE model substantially larger than in standard fixed effects (factor of 10 in comparison of Columns 1 and 2).

*Source: Authors’ calculations based on International Labour Organization, World Bank, World Development Indicators, World Bank Doing Business Indicators, and Inter-American Development Bank SIMS.*

### 2.  Selected LAC Economies

### 2.  Selected LAC Economies

### Correlates of the Speed of Adjustment
- Higher informality rates and stricter job security regulations are associated with a lower speed of adjustment of employment.
- The estimated coefficient for informality implies that a 10 percentage point decrease in informality (roughly the difference between the region’s average and the average for EMs in the sample of 110 countries) would raise LAC’s speed of adjustment to roughly the EM average.
- A decrease of 1 point in the job security composite index (for example, by eliminating third party dismissal notification) would raise the region’s speed of adjustment of employment to roughly the estimated average for AEs.
- Informality acts as a buffer that attenuates the impact of GDP shocks on unemployment but makes the adjustment process more protracted.
- Informality is heterogeneous: evidence for Brazil (Ulyssea, 2018) classifies informal establishments into:
  - “survival” firms (not productive enough to formalize),
  - “opportunistic” firms (avoid formalization when possible),
  - productive informal establishments that would formalize if costs were low enough.
  - Roughly half of informal firms in Brazil are “survival”; close to 40 percent are “opportunistic”.
- Labor market segmentation implies large costs of switching from informal to formal jobs (Arias et al., 2018), comparable to switching across sectors.

*Sources cited in text: Figure 4; ILOSTAT; WDI; ISO country codes; AE, CA, EM, LAC, SA groupings.*

### Micro-Flexibility and the Speed of Adjustment — Methodology
- The paper estimates a microeconomic model of labor market adjustment using the equation:
  - Δe_{i,t} = α_i + λ_i (e^*_{i,t} − e_{i,t−1}) + ε_{i,t}
  - where Δ is the first difference operator, e_t is ln(employment), e^* is (log) equilibrium employment, (e^* − e_{t−1}) is the employment gap in t−1, and λ_i is the speed of adjustment.
- Two-step econometric approach (Caballero et al. 2013):
  - First stage: construct proxy of employment gap by estimating key variables of the micro-founded model (including sectoral productivity differences).
  - Second stage: estimate the error-correction equation using the employment gap proxy.
- Extensions relative to Caballero et al. (2013):
  - Change period of analysis from 1980–2000 to 2000–2017.
  - Assess role of informality, policies governing worker compensation, and macro-stabilization policies.
  - Compare speed of adjustment using manufacturing-only (UNIDO) data vs. manufacturing + services + construction (UNIDO + Timmers, de Vries and de Vries 10-sector dataset + OECD STAN).

### Empirical Results — Reference Estimates and Dataset Differences
- Speed of adjustment estimates differ by dataset:
  - Manufacturing-only (UNIDO) sample: sectors close approximately 50 percent of the employment gap in each period (Table 5, Column 1: Employment gap coefficient = 0.501***; standard error (0.0427)).
  - Manufacturing + services + construction: coefficient ≈ 45.9 percent (Table 5, Column 2: Employment gap coefficient = 0.459***; standard error (0.0410)).
- The combined dataset yields slower adjustment estimates; concerns:
  - Differences in country and time coverage across datasets complicate interpretation.
  - Services subsectors and construction are more aggregated than 2-digit manufacturing sectors in UNIDO, possibly attenuating speed estimates.
- For this reason, the analysis focuses on the UNIDO dataset.

### UNIDO Sample — LAC vs. Non-LAC and Role of Informality and Regulations
- LAC exhibits slower speed of adjustment than other countries:
  - Column 3 (UNIDO): the speed of adjustment for the average LAC country is approximately 2.5 percentage points lower compared to countries outside the region (Employment gap coefficient in Column 3 = 0.502***; interaction Employment gap * LAC = −0.0255*; standard error (0.0143)).
- Informality associated with slower adjustment:
  - Employment gap * Informality = −0.0596*** (standard error (0.0216)) (Table 5, Column 4).
- Job security and minimum wages relative to labor productivity reduce microeconomic flexibility:
  - Employment gap * Job security = −0.0374*** (standard error (0.00561)) and −0.0223*** (standard error (0.00654)) in related specifications (Table 5, Columns shown).
  - Employment gap * (Minimum wage/Labor productivity) = −0.131*** (standard error (0.0175)) and −0.0542*** (standard error (0.0185)) in related specifications.
- Enforcement (government effectiveness) amplifies these effects:
  - Employment gap * Job security * High government effectiveness = −0.0678*** (standard error (0.0134)).
  - Employment gap * (Minimum wage/Labor productivity) * High government effectiveness = −0.335*** (standard error (0.0532)).
- Positive association of high government effectiveness with speed:
  - Employment gap * High government effectiveness = 0.0469*** (standard error (0.0157)) in some specifications.

- Table 5 summary statistics (selected):
  - Constant terms range from 0.000832 to 0.00460 with significance levels reported (e.g., Constant = 0.00460***, standard error (0.000675)).
  - Observations vary by specification (e.g., 27988, 30895, 27647, 20123, 27056, 26942, 76472, 7585).
  - Number of groups reported across specifications (e.g., 1604, 1693, 1586, 1141, 1553, 1549, 1586, 1582).
  - Significance notation: *** p<0.01, ** p<0.05, * p<0.1.

### Economic Magnitude — Implied Speed of Adjustment and Half-Life (Table 6)
- Average implied speed of adjustment and half-life (selected entries):
  - Average speed of adjustment = 0.50; half life = 12.00 (months).
  - LAC speed of adjustment = 0.48; half life = 12.91 (months).
  - Low informality: speed = 0.51; half life = 11.79.
  - High informality: speed = 0.47; half life = 13.26; implied growth differential (low-high) = 0.17pp.
  - Low, High government effectiveness: speed = 0.54; half life = 10.56.
  - High, High government effectiveness: speed = 0.43; half life = 14.61; implied growth differential = 0.47pp.
  - Low, Low government effectiveness: speed = 0.52; half life = 11.39.
  - High, Low government effectiveness: speed = 0.49; half life = 12.32; implied growth differential = 0.11pp.
  - For minimum wage/Labor productivity dimension: Low, High government effectiveness: speed = 0.55; half life = 10.56. High, High government effectiveness: speed = 0.39; half life = 16.85; implied growth differential = 0.74pp.
  - Informality: difference between 80th and 20th percentile implies ≈4 percentage points lower speed of adjustment for high-informality country relative to low-informality country; translates into 1.5 additional months to close half the employment gap.
  - Difference in speed between low and high employment protection is 10 percentage points when government effectiveness is high → ≈4 additional months to close half the employment gap.
  - Difference in speed between low and high relative minimum wages is 16 percentage points when enforcement is high → ≈6 additional months to close half the employment gap.

- Note on half-life calculation:
  - Half-life (months) = 12*(log(0.5)/log(1 − speed of adjustment)).

### Sectoral Heterogeneity — Labor Intensity and Regulations (Table 7)
- Labor market regulations have heterogeneous effects across sectors, amplified in labor-intensive sectors.
- Key sectoral interaction results:
  - Employment gap * Labor intensity = 0.485*** (standard error (0.160)) and 0.625*** (standard error (0.186)) in alternative specifications — implying higher baseline adjustment in labor-intensive sectors absent regulation interactions.
  - Employment gap * Job security * Labor intensity = −0.174 (standard error (0.115)) (not significant at high levels reported).
  - Employment gap * (Minimum wage/Labor productivity) * Labor intensity = −0.997*** (standard error (0.320)) — high minimum wages have a major negative impact on the speed of employment adjustment in labor-intensive sectors.
- Interactions with exchange rate flexibility:
  - Employment gap * Exchange Rate Flexibility = 0.0802*** (standard error (0.0117)) and 0.0889*** (standard error (0.0201)) in related specifications.
  - Employment gap * (Minimum wage/Labor productivity) * Exchange Rate Flexibility = 0.199*** (standard error (0.0671)).
  - Employment gap * Job security * Exchange Rate Flexibility coefficient reported as −0.0107 (standard error (0.0149)) (not significant).

### Asymmetric Effects of Labor Market Regulations
- The effect of labor market regulations differs depending on whether employment is above or below equilibrium:
  - Construct a dummy that equals 1 if employment gap is negative (employment above equilibrium) and 0 otherwise; interact this dummy with regulations.
  - EPLs slow adjustment more when employment is above equilibrium (negative gap) — greater impact on net job destruction margin:
    - Employment gap * Job security * Negative gap dummy = −0.0414** (standard error (0.0166)).
  - Binding minimum wages affect adjustment more when employment is below equilibrium (positive gap) — greater impact on net job creation margin:
    - Employment gap * (Minimum wage/Labor productivity) * Negative gap dummy = 0.0927* (standard error (0.0484)).
  - Employment gap * Negative gap dummy = 0.0690*** (standard error (0.0212)) in some specifications, indicating asymmetry in baseline adjustment rates by gap sign.

### Labor Market Regulation and Growth
- Slower labor market responsiveness can lower productivity growth by hampering reallocation of factors across sectors and firms.
- Simple correlation: GDP per worker growth is positively correlated with the country’s speed of adjustment (Figure 5 referenced).
- Back-of-envelope: changes in employment protection regulations moving a country from the 80th percentile to the 20th percentile of the speed of adjustment distribution are associated with an increase in medium-term labor productivity growth (Caballero et al. 2013 framework referenced).

*Sources: Authors’ calculations based on data from the International Labour Organization, World Bank, World Development Indicators, and World Bank, Doing Business Indicators. Appendix references and figures/tables cited as in source text.*

*Italic: Source — wpiea2020019-print-pdf, Section 2. Selected LAC Economies.*

### 0.5 percentage points (pp) per annum (Table 5, last column). Similarly, changing the minimum

### wpiea2020019-print-pdf - 0.5 percentage points (pp) per annum (Table 5, last column). Similarly, changing the minimum

### Main findings
- Changing the minimum wage from the 80th to the 20th percentile of the distribution when government effectiveness is high can increase growth by approximately 0.75 pp per annum.
- A change equivalent to 0.5 percentage points (pp) per annum is reported (Table 5, last column).
- Labor market regulations (employment protection legislation, EPL, and high minimum wages relative to labor productivity) lower labor productivity growth, with larger effects in labor-intensive sectors.
- Informality:
  - Dampens effects of macroeconomic shocks on employment (provides a buffer for low-skilled workers in downturns) but makes employment adjustment more sluggish.
  - Is linked to strict labor market regulations (certain dimensions of stricter EPL increase informality).
  - High minimum wages relative to labor productivity increase informality, lower the speed of adjustment to shocks, and hamper growth.

### Methods and empirical strategy
- Difference-in-difference (DID) approach inspired by Rajan and Zingales (1998) to exploit heterogeneous sectoral labor intensity:
  - Estimated equation (4): g_ijt = α_jt + β * (reg_i * α_jt-1) + γ X_ijt + θ_i + ε_ijt (notation as in source).
  - ijt g is sector j’s labor productivity growth in country i at time t.
  - α_jt is the labor share in value added of sector j at time t (median labor share across countries).
  - reg_i is either the proxy for employment protection legislations or the minimum wage over labor productivity ratio.
  - X_ijt controls include sector’s capital share, sector’s initial share in total value added, and country-year fixed effects.
- Goal: minimize endogeneity between growth and labor market regulations.
- Sectoral Common Correlated Effects Error-Correction Model estimated for agriculture, industry, services to obtain speed of adjustment coefficients.
- Caballero et al. (2013) framework used to link microeconomic flexibility (speed of adjustment) to long-term growth differences via a simple AK growth model; parameter values used by CCEM to compute Table 5.

### Quantitative results and robustness
- Table 8 highlights:
  - Sector's share in country's total value added, t-1: coefficients range (columns 1–6) from -0.156*** to -0.174*** and -0.153*** (exact values listed in table).
    - Exact entries: -0.156*** (0.0174), -0.175*** (0.0180), -0.176*** (0.0183), -0.174*** (0.0180), -0.154*** (0.0178), -0.153*** (0.0174).
  - Sectoral labor share, t-1: -0.132*** (0.0327), -0.0466 (0.0519), -0.0366 (0.0546) across columns.
  - Country's job security * Sectoral labor share, t-1: -0.0884** (0.0423).
  - Country's rel. minimum wage * Sectoral labor share, t-1: -0.267** (0.122).
  - Country's job security * Sectoral capital share, t-1: -0.0249 (0.119).
  - Country's rel. minimum wage * Sectoral capital share, t-1: -0.00348 (0.377).
  - Constant terms: 0.0455*** (0.00159), 0.0628*** (0.00459), 0.0629*** (0.00462), 0.0624*** (0.00459), 0.0568*** (0.00359), 0.0564*** (0.00350).
  - Country-Year Fixed Effects: YES (in all columns).
  - Sectoral Capital Share control: NO (columns 1–3), YES (columns 4–6).
  - Observations: 26,552; 26,539; 25,977; 26,539; 25,856; 26,418 (by column).
  - R-squared: 0.196; 0.197; 0.199; 0.197; 0.199; 0.197 (by column).
- Figure 5 (Speed of Adjustment and Growth):
  - Coefficient calculated from regression like Table 5; controlling for sectoral fixed effects; speed of adjustment is statistically significant at the 95 percent confidence level.
- Appendix C — sectoral error-correction model:
  - Industry has the highest average estimated speed of adjustment coefficient; services the lowest.
  - Informality reduces the speed of adjustment parameter for services and agriculture but not for industry.
- Estimation details:
  - In Caballero et al. (2013) two-step estimation, φ is estimated then used to construct the employment gap with Murphy-Topel standard errors in the second stage to account for estimation error.
  - Instruments: Δv_ijt − Δv_.jt^0 instrumented with Δw_ijt−1 − Δw_.jt^0.

### Policy implications and trade-offs
- Two key trade-offs when designing labor market regulations:
  1. Between employment protection and labor market dynamism & job quality:
     - Stringent EPL may limit employment losses in downturns but can increase informal employment (typically lower paying) and lead to sluggish recoveries.
     - Possible partial resolution: complement less-stringent EPL with a modern safety net (robust unemployment insurance and other benefits).
  2. Between labor market legislation and effectiveness of macro-stabilization tools:
     - EPL can undermine the employment benefits of macro-stabilization tools such as exchange rate flexibility.
     - Policy design could consider contingency mechanisms (examples provided):
       - Severance payments or minimum wage adjustments contingent on the state of the business cycle.
- Structural reforms that increase labor productivity are recommended as the best way to address the informality determinant related to high minimum wages relative to labor productivity (as suggested in IMF (2019b)).

### Data sources and variable definitions (selected)
- Real GDP growth: Change in log of real GDP (in %), 178 countries, 1990-2017, IMF WEO.
- Total employment: Share of employment to population 15+ (in %), 140 countries, 1990-2017, World Bank WDI.
- Informal employment (ILO): Share of informal employment on total non-agricultural employment (in %), 119 countries, 2016, ILO (2018).
- Informal employment (IDB): Share of total active workers that do not contribute to social security (in %), 17 LAC countries, 1990-2017, IDB SIMS.
- Informal employment (WB): Share of self-employed workers (in %), 142 countries, 1991-2017, World Bank WDI.
- Minimum wage ratio: Ratio between the national minimum wage and GDP per worker (labor productivity), 188 countries, 2014-2018, Doing Business indicators.
- Redundancy costs: Dismissal costs in weeks of salary, 189 countries, 2014-2018, Doing Business indicators.
- Government effectiveness: Dummy = 1 if government effectiveness estimate in 1996 above global median in that year, 214 countries, 1996, Kaufmann, Kraay, and Mastruzzi (2010).
- Exchange rate flexibility: Dummy = 1 if IRR index values 3 and 4 (flexible), zero if 1 and 2 (non-flexible), 194 countries, 1990-2016, Ilzetzki, Reinhart and Rogoff (2019).
- Job security indicator: Index following Botero et al. (2004), normalized 0–1, 186 countries, average for 2014-2018, authors’ calculations based on Doing Business indicators.
- Employment, output, and wages by sector: 173 countries, 1963-2017, INDSTAT and STAN databases.

*Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020019-print-pdf.pdf*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020019-print-pdf.pdf_