## _wp14153 - Section III. We find that, unlike in many other economies, high structural unemployment

## Source details

**Canonical URL:** [_wp14153 - Section III. We find that, unlike in many other economies, high structural unemployment](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14153.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14153.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14153.pdf.json)

---

### Key findings
- High structural unemployment coexists with relatively flexible labor markets, implying the main cause is not lack of labor market dynamism.
- Taxes on labor are high, raising the cost of hiring, particularly lower skilled workers.
- Skill mismatches appear to be a concern, and policies to address them are underused.
- Current levels of unemployment in 2013 are close to estimated NAIRUs for Latvia and Lithuania and somewhat below the NAIRU for Estonia.
- With output growing at potential, unemployment would not drop significantly without rising wage and inflationary pressures.

### Empirical framework
- Time-varying NAIRUs estimated using reduced-form Phillips curve equations and the Kalman filter, drawing on Laubach (2001), OECD approaches (Gianella et al., 2008; Guichard and Rusticelli, 2011), and IMF (2013).
- Domestic inflation measured by core consumer price inflation adjusted for changes in indirect taxes.
- Inflation expectations approximated by the past 4-quarter rolling average of the core CPI inflation.
- Import price inflation included to control for short-term supply shocks; a cost-push shock term is also present.
- The unemployment gap and the NAIRU dynamics:
  - NAIRU modeled as a random walk.
  - Unemployment gap assumed to follow an AR(1) process to ensure convergence of unemployment to its structural rate in the absence of shocks.
- Initial conditions and estimation choices:
  - Initial value of the NAIRU set equal to the unemployment rate in each country in 2005Q1.
  - Initial unemployment gap equals the unemployment rate in the initial sample period minus the specified initial NAIRU.
  - Signal-to-noise ratios for the variance of the NAIRU transition error explored in a range consistent with the literature.
  - Parameters and shock variances estimated with maximum likelihood using a nonlinear Kalman filter.
- Sample period: depending on data availability, starts in the early 2000s and ends in Q4 2013.

### Results: estimated NAIRUs and uncertainty
- General characterization: NAIRU in the Baltics is high, relatively stable, and close to the current level of unemployment.
- Lithuania and Latvia: time-varying point estimate ranges between 10¾ and 13¾ percent.
- Estonia: time-varying point estimate ranges between 10¾ and 15¼ percent.
- 95-percent confidence intervals around point estimates shrink over time; by the end of the estimation period they are +/- 1¼ percentage point (1½ percentage point for Lithuania).
- Interpretation: the still high unemployment in the Baltics appears to reflect equilibrium trends rather than purely cyclical conditions.

### Robustness checks
- Historical comparison: NAIRU estimates are consistent with historical experience of unemployment in the Baltics; the crisis-driven increase in unemployment was from a historical low.
- Real wage behavior: real wage growth is highly correlated with the inverse of the calculated unemployment gap, consistent with the expectation that real wage growth accelerates once unemployment falls below the NAIRU.
- Labor market flexibility: evidence of high downward flexibility of wages in the Baltics, consistent with relatively flexible labor market institutions.
- Additional formal tests: changes in the Beveridge curve and Okun’s law examined; overall checks indicate the NAIRU is high and relatively stable over time.

### Policy-relevant implications
- Because structural unemployment appears high and close to current unemployment levels, policies solely aimed at stimulating demand may not substantially reduce unemployment without inducing wage and inflation pressures.
- High taxes on labor increase hiring costs, especially for lower skilled workers — implying potential benefits from labor tax reform to reduce hiring costs.
- Skill mismatches are an identified concern, and existing policies to address them are underused — suggesting scope to strengthen active labor market policies and training/education alignment.

### NAIRU estimates and real wage dynamics
- Correlation between Baltic average real wage growth and deviation from NAIRU is 0.86 percent for the Baltics.
- Country correlations (2002q1–2013q2): Estonia 0.81, Latvia 0.82, Lithuania 0.87.
- The NAIRU estimation relied on inflation dynamics in the Phillips curve; a very similar result would be obtained using wage dynamics (NAWRU).
- Historical average and NAIRU unemployment (Percent):
  - Estonia: Historical average 9.7; NAIRU 12.4 (Historical average period: 1993Q1-2013Q3).
  - Latvia: Historical average 12.2; NAIRU 12.3 (Historical average period: 1996Q1-2013Q3).
  - Lithuania: Historical average 11.1; NAIRU 12.4 (Historical average period: 1998Q2-2013Q3).
- NAIRU estimates align with those from other international institutions (European Commission; OECD); on average, point estimates are very similar.

### Evidence from the Beveridge Curve
- Purpose: test whether coordinates in the Beveridge curve (vacancies vs unemployment) exhibit movements pointing to cyclical or structural shifts since the crisis.
- Interpretation:
  - Movements along the Beveridge curve → cyclical variations in unemployment.
  - Shifts of the Beveridge curve → increased inefficiencies of labor matching → increasing structural unemployment.
- Visual/initial evidence:
  - Latvia and Lithuania: Beveridge curve visually has not shifted since the crisis.
  - Estonia: appears a slight outward shift visually.
- Data and approach:
  - Used European Commission series of employers’ perceptions of labor shortages in manufacturing as proxy for vacancy rate developments.
  - Estimated country-specific Beveridge curves and a panel Beveridge curve with country-fixed effects.
  - Model regressed unemployment rate (U) on labor shortages (LS) and LS^2; included crisis dummy (value 1 since 2009Q1) and interactions to test slope changes and shifts.
  - Panel estimator: LSDVCE (Least Square Dummy Variable Corrected estimator) to address dynamic panel bias with large T and small N.
- Econometric findings:
  - Estimates do not reject hypothesis that Beveridge curves have not shifted since the crisis; Beveridge curve remained stable over time.
  - No statistically significant estimates of the parameters identifying slope changes or shifts (parameters for interactions and crisis dummy).
  - Slopes of Beveridge curves did not change since the crisis.
  - Movements in vacancies and unemployment since the crisis are consistent with short-term cyclical shocks; structural unemployment remained elevated but broadly stable.
- Table excerpt (selected coefficients):
  - Labor shortage coefficients: Pooling -0.171 ***; Estonia -0.18; Latvia -0.324 ***; Lithuania -0.161; LSDVCE -0.168 ***.
  - Labor shortage^2 coefficients: Pooling 0.0024 **; Estonia 0.0022; Latvia 0.0050 **; Lithuania 0.0021; LSDVCE 0.0023 **.
  - Unemployment rate t-1 coefficient ~0.617 to 0.761*** across specifications.
  - Observations and sample: Number of countries 3; Observations vary by specification.

### Evidence from the Okun Relationship
- Okun relationship links changes in unemployment to output dynamics; β (“Okun’s beta”) describes responsiveness of unemployment to output.
- Estimation details:
  - Specifications in levels and differences; levels form: (Y* - Y_t) related to (U* - U_t).
  - U* and Y* obtained using HP filter with λ=1600.
  - Quarterly data covering late 1990s through 2013Q3; OLS used; 20-quarter rolling regressions also performed.
- Key findings:
  - Okun’s beta for the Baltics is relatively high and relatively stable over time, with a slight downward trend.
  - Sample-period estimates (levels): average Okun’s beta = 0.42.
    - Estonia 0.41, Latvia 0.41, Lithuania 0.49.
  - Specification in differences yields broadly similar results.
  - Fit: high adjusted R2 values (examples: Estonia Adj. R2 0.71; Latvia 0.84; Lithuania 0.73 in levels table).
  - Rolling regressions: Okun’s beta has gradually trended down; trend set in before the crisis.
  - Time-trend regressions:
    - Time trend coefficient significant for Estonia and Latvia (negative), not for Lithuania.
    - Interaction of time trend with post-crisis dummy (2008Q1–2013Q3) not statistically significant for Estonia or Latvia; for Lithuania a small positive and significant time dummy coefficient implies a slight decrease in absolute value post-crisis.
- Comparative context (Ball et al. (2013) benchmarks):
  - Anglo-Saxon average Beta -0.42.
  - Examples: US -0.454*** (Adj. R2 0.82); Canada -0.432*** (Adj. R2 0.81); Sweden -0.524*** (Adj. R2 0.62).
- Interpretation:
  - The Baltics’ relatively high absolute value of Okun’s beta is consistent with relatively flexible labor market institutions.
  - No strong indication of a structural break in Okun’s relationship since the 2008/09 crisis; responsiveness of unemployment to output variations unchanged.

### What explains the high level of structural unemployment in the Baltics? — Overview
- Section objective: examine traditional and non-traditional explanations for high and persistent structural unemployment in the Baltics.
- Approach: (A) Traditional factors—labor market characteristics; (B) Non-traditional factors.

### A. Traditional Factors — Labor Market Characteristics
- Micro-flexibility indicators:
  - Minimum wages (as share of mean wage): below or close to OECD average.
  - Unemployment benefits: much less generous than OECD average.
  - Employment protection: on par with OECD average.
- Tax wedge:
  - Labor tax wedges are high in the Baltics, largely because of high social security contribution rates.
  - Tax wedge measured as difference between labor costs to employer and net take-home pay of employee.
  - Economic theory: high labor tax rates depress labor supply and employment, expand shadow economy; can reduce worker effort and raise labor costs if shifted to employers.
- Econometric evidence on tax wedge and structural unemployment:
  - Cross-country studies find tax wedge coefficients in range 0.17 to 0.36.
  - Implication: a reduction in the tax wedge by 10 percentage points would lead to a reduction in structural unemployment by 2 to 4 percentage points, all else equal.
  - Event study of nine episodes of large declines in structural unemployment (≥ 2.5 percentage points) correlates these declines with reductions in tax wedge for unskilled workers.
  - Correlation results: simple OLS regression shows coefficient 0.4 and R^2 = 0.3 → reductions in tax wedge explain about 30 percent of variation in structural unemployment in the sample.
- Examples of historically large declines in structural unemployment (Country | Period | Reduction in tax wedge | Reduction in structural unemployment):
  - Bulgaria 2001-08: 8.5 | 8.3
  - Finland 1996-2008: 6.7 | 5.4
  - Germany 2005-13: 2.3 | 3.6
  - Ireland 1996-2001: 10 | 6.1
  - Italy 1996-2007: 5.5 | 2.5
  - Poland 2002-08: 3.4 | 5.9
  - Slovakia 2001-08: 5.2 | 3.7
  - Spain 1996-2007: -1.3 | 4.5
- Conclusion on traditional factors:
  - High tax wedges, combined with benefit changes, can generate unemployment and inactivity traps by reducing financial gain from employment, especially for lower-wage earners.
  - Baltics face higher unemployment and inactivity traps than CE4, Anglo-Saxon, or other emerging market OECD countries.

### B. Non-Traditional Factors — Skill Mismatches and ALMPs
- Skill mismatches:
  - Baltics score close to OECD averages on PISA tests; variation across Baltics: Lithuania lags, Estonia outperforms OECD average.
  - Tertiary enrollment rates in line with OECD averages; male tertiary enrollment is low.
  - All three Baltics score worse than average on skill field mismatch.
  - Mismatches exist in educational attainment required within professions and in training for the right profession/sector.
  - Regional divides: e.g., in Lithuania, rural unemployment (much above urban) consists mostly of people with below-tertiary education.
- Active Labor Market Policies (ALMPs):
  - ALMP spending: Baltics 0.5 percent of GDP vs Nordics over 2 percent of GDP.
  - ALMP training spending: Baltics 0.1 percent of GDP — less than half that in the Nordics.
  - In Baltics, passive labor market support (out-of-work income support; in Latvia and Lithuania also early retirement) accounts for a large share of total ALMP spending.
  - ALMP spending did not increase dramatically since 2008/09 (exceptions: some increases from a small base, e.g., Estonia).
  - Low ALMP spending reflects very low participation in these programs rather than high efficiency.
  - Longer-term averages of unemployment (especially youth) show a negative correlation with ALMP spending, indicating potential of ALMPs to address skill mismatches.
- PISA and tertiary indicators (selected figures):
  - Math scores: Estonia 521; Latvia 491; Lithuania 479; Baltic average 497; OECD average 494.
  - Low performers in Math (Percent): Estonia 10.5; Latvia 19.9; Lithuania 26.0; Baltic ave. 18.8; OECD ave. 23.0.
  - High performers in Math (Percent): Estonia 14.6; Latvia 8.0; Lithuania 8.1; Baltic ave. 10.2; OECD ave. 12.6.
  - ALMP spending and tertiary enrollment figures reported in source tables (Baltic ALMP spending 0.5 percent of GDP; ALMP training spending 0.1 percent of GDP).

### Box 1. Iceland: Active Labor Market Policies During the Crisis
- Context:
  - Crisis period: 2008/09 banking and economic crisis in Iceland.
  - Problem addressed: sharply rising unemployment following the crisis.
  - Initial emphasis: expanding registration for unemployment benefits and educating the public about available options.
- Measures Implemented:
  - Job retraining.
  - Subsidized hiring for trial periods.
  - Study programs.
  - Subsidized hiring (general).
  - Volunteer work placements.
  - Opening secondary education to anyone under age 25.
  - Programs emphasizing work-related education.
  - Greater cooperation between social partners and the education system.
- Outcomes and Effectiveness:
  - Scope and evolution: The wide scope of ALMPs and their gradually changing role over the course of the crisis helped to increase the number of participants in the programs.
  - Program-specific success: Programs providing on-the-job training/apprenticeships or employment in specific projects seem to have increased chances of participants “de-listing” from the unemployment rolls.
  - Youth outcomes: Available information suggests that about half of unemployed youth found jobs after participating in the programs.
  - Nature of unemployment addressed: ALMPs in this episode addressed mostly cyclical unemployment problems.
- Lessons and Implications:
  - ALMPs can be successful even when primarily addressing cyclical unemployment.
  - Program design that includes on-the-job training and project-based employment appears especially effective at facilitating exit from unemployment.
  - Gradual expansion and diversification of ALMPs can increase participation and improve outcomes.
  - Coordination between education providers and social partners enhances the relevance of training and education initiatives.

*Source: _wp14153 - Section III. We find that, unlike in many other economies, high structural unemployment — https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14153.pdf*

### Section III. We find that, unlike in many other economies, high structural unemployment

### _wp14153 - Section III. We find that, unlike in many other economies, high structural unemployment

### Key findings
- High structural unemployment coexists with relatively flexible labor markets, implying the main cause is not lack of labor market dynamism.
- Taxes on labor are high, raising the cost of hiring, particularly lower skilled workers.
- Skill mismatches appear to be a concern, and policies to address them are underused.
- Current levels of unemployment in 2013 are close to estimated NAIRUs for Latvia and Lithuania and somewhat below the NAIRU for Estonia.
- With output growing at potential, unemployment would not drop significantly without rising wage and inflationary pressures.

### Empirical framework
- Time-varying NAIRUs estimated using reduced-form Phillips curve equations and the Kalman filter, drawing on Laubach (2001), OECD approaches (Gianella et al., 2008; Guichard and Rusticelli, 2011), and IMF (2013).
- Domestic inflation measured by core consumer price inflation adjusted for changes in indirect taxes.
- Inflation expectations approximated by the past 4-quarter rolling average of the core CPI inflation.
- Import price inflation included to control for short-term supply shocks; a cost-push shock term is also present.
- The unemployment gap and the NAIRU dynamics:
  - NAIRU modeled as a random walk.
  - Unemployment gap assumed to follow an AR(1) process to ensure convergence of unemployment to its structural rate in the absence of shocks.
- Initial conditions and estimation choices:
  - Initial value of the NAIRU set equal to the unemployment rate in each country in 2005Q1.
  - Initial unemployment gap equals the unemployment rate in the initial sample period minus the specified initial NAIRU.
  - Signal-to-noise ratios for the variance of the NAIRU transition error explored in a range consistent with the literature (see Laubach, 2001).
  - Parameters and shock variances estimated with maximum likelihood using a nonlinear Kalman filter.
- Sample period: depending on data availability, starts in the early 2000s and ends in Q4 2013.

### Results: estimated NAIRUs and uncertainty
- General characterization: NAIRU in the Baltics is high, relatively stable, and close to the current level of unemployment.
- Lithuania and Latvia: time-varying point estimate ranges between 10¾ and 13¾ percent.
- Estonia: time-varying point estimate ranges between 10¾ and 15¼ percent.
- 95-percent confidence intervals around point estimates shrink over time; by the end of the estimation period they are +/- 1¼ percentage point (1½ percentage point for Lithuania).
- Interpretation: the still high unemployment in the Baltics appears to reflect equilibrium trends rather than purely cyclical conditions.

### Robustness checks
- Historical comparison: NAIRU estimates are consistent with historical experience of unemployment in the Baltics; the crisis-driven increase in unemployment was from a historical low.
- Real wage behavior: real wage growth is highly correlated with the inverse of the calculated unemployment gap, consistent with the expectation that real wage growth accelerates once unemployment falls below the NAIRU. This corroborates NAIRU estimates derived from inflation dynamics.
- Labor market flexibility: evidence of high downward flexibility of wages in the Baltics, consistent with relatively flexible labor market institutions.
- Additional formal tests: the study examines changes in the Beveridge curve (vacancies-unemployment relationship) and Okun’s law (output-unemployment relationship) as part of robustness analysis; overall checks indicate the NAIRU is high and relatively stable over time.

### Policy-relevant implications (as presented)
- Because structural unemployment appears high and close to current unemployment levels, policies solely aimed at stimulating demand may not substantially reduce unemployment without inducing wage and inflation pressures.
- High taxes on labor increase hiring costs, especially for lower skilled workers — implying potential benefits from labor tax reform to reduce hiring costs.
- Skill mismatches are an identified concern, and existing policies to address them are underused — suggesting scope to strengthen active labor market policies and training/education alignment.

*Source: _wp14153 - Section III. We find that, unlike in many other economies, high structural unemployment — https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14153.pdf*

### section III).

### _wp14153 - section III)

### NAIRU estimates and real wage dynamics
- The correlation between Baltic average real wage growth and deviation from NAIRU is 0.86 percent for the Baltics.
- Country correlations (2002q1–2013q2): Estonia 0.81, Latvia 0.82, Lithuania 0.87.
- The NAIRU estimation relied on inflation dynamics in the Phillips curve; a very similar result would be obtained using wage dynamics (NAWRU).
- Historical average and NAIRU unemployment (Percent)
  - Estonia: Historical average 9.7; NAIRU 12.4 (Historical average period: 1993Q1-2013Q3).
  - Latvia: Historical average 12.2; NAIRU 12.3 (Historical average period: 1996Q1-2013Q3).
  - Lithuania: Historical average 11.1; NAIRU 12.4 (Historical average period: 1998Q2-2013Q3).
- NAIRU estimates align with those from other international institutions (European Commission; OECD); on average, point estimates are very similar.

### Evidence from the Beveridge Curve
- Purpose: test whether coordinates in the Beveridge curve (vacancies vs unemployment) exhibit movements pointing to cyclical or structural shifts since the crisis.
- Interpretation:
  - Movements along the Beveridge curve → cyclical variations in unemployment.
  - Shifts of the Beveridge curve → increased inefficiencies of labor matching → increasing structural unemployment.
- Visual/initial evidence:
  - Latvia and Lithuania: Beveridge curve visually has not shifted since the crisis.
  - Estonia: appears a slight outward shift visually.
- Data and approach:
  - Used European Commission series of employers’ perceptions of labor shortages in manufacturing as proxy for vacancy rate developments.
  - Estimated country-specific Beveridge curves and a panel Beveridge curve with country-fixed effects.
  - Model regressed unemployment rate (U) on labor shortages (LS) and LS^2; included crisis dummy (value 1 since 2009Q1) and interactions to test slope changes and shifts.
  - Panel estimator: LSDVCE (Least Square Dummy Variable Corrected estimator) to address dynamic panel bias with large T and small N.
- Econometric findings:
  - Estimates do not reject hypothesis that Beveridge curves have not shifted since the crisis; Beveridge curve remained stable over time.
  - No statistically significant estimates of the parameters identifying slope changes or shifts (parameters for interactions and crisis dummy).
  - Slopes of Beveridge curves did not change since the crisis.
  - Movements in vacancies and unemployment since the crisis are consistent with short-term cyclical shocks; structural unemployment remained elevated but broadly stable.
- Contrast: Bonthuis et al. (2013) found outward shifts in the Euro area except Germany; Baltics’ results differ.

- Table excerpt (Beveridge Curve Estimates for the Baltic Countries; Dependent variable: Unemployment rate at each quarter (LFS data))
  - Labor shortage coefficients (selected): Pooling -0.171 ***; Estonia -0.18; Latvia -0.324 ***; Lithuania -0.161; LSDVCE -0.168 ***.
  - Labor shortage^2 coefficients (selected): Pooling 0.0024 **; Estonia 0.0022; Latvia 0.0050 **; Lithuania 0.0021; LSDVCE 0.0023 **.
  - Crisis dummy and interactions: not consistently significant across specifications.
  - Unemployment rate t-1 coefficient ~0.617 to 0.761*** across specifications.
  - Observations and sample: Number of countries 3; Observations vary by specification (e.g., 1385, 338, 471, 138, 138 shown).

### Evidence from the Okun Relationship
- Okun relationship links changes in unemployment to output dynamics; β (“Okun’s beta”) describes responsiveness of unemployment to output.
- Estimation details:
  - Specifications in levels and differences; levels form: (Y* - Y_t) related to (U* - U_t).
  - U* and Y* obtained using HP filter with λ=1600.
  - Quarterly data covering late 1990s through 2013Q3; OLS used; 20-quarter rolling regressions also performed.
- Key findings:
  - Okun’s beta for the Baltics is relatively high and relatively stable over time, with a slight downward trend.
  - Sample-period estimates (levels): average Okun’s beta = 0.42.
    - Estonia 0.41, Latvia 0.41, Lithuania 0.49.
  - Specification in differences yields broadly similar results.
  - Fit: high adjusted R2 values (examples: Estonia Adj. R2 0.71; Latvia 0.84; Lithuania 0.73 in levels table).
  - Rolling regressions: Okun’s beta has gradually trended down; trend set in before the crisis.
  - Time-trend regressions:
    - Time trend coefficient significant for Estonia and Latvia (negative), not for Lithuania.
    - Interaction of time trend with post-crisis dummy (2008Q1–2013Q3) not statistically significant for Estonia or Latvia; for Lithuania a small positive and significant time dummy coefficient implies a slight decrease in absolute value post-crisis.
- Comparative context (Ball et al. (2013) benchmarks):
  - Anglo-Saxon average Beta -0.42.
  - Examples: US -0.454*** (Adj. R2 0.82); Canada -0.432*** (Adj. R2 0.81); Sweden -0.524*** (Adj. R2 0.62).
- Interpretation:
  - The Baltics’ relatively high absolute value of Okun’s beta is consistent with relatively flexible labor market institutions.
  - No strong indication of a structural break in Okun’s relationship since the 2008/09 crisis; responsiveness of unemployment to output variations unchanged.

### What explains the high level of structural unemployment in the Baltics? — Overview
- Section objective: examine traditional and non-traditional explanations for high and persistent structural unemployment in the Baltics.
- Approach: (A) Traditional factors—labor market characteristics; (B) Non-traditional factors.

### A. Traditional Factors — Labor Market Characteristics
- Micro-flexibility indicators:
  - Minimum wages (as share of mean wage): below or close to OECD average.
  - Unemployment benefits: much less generous than OECD average.
  - Employment protection: on par with OECD average.
- Tax wedge:
  - Labor tax wedges are high in the Baltics, largely because of high social security contribution rates.
  - Tax wedge measured as difference between labor costs to employer and net take-home pay of employee.
  - Economic theory: high labor tax rates depress labor supply and employment, expand shadow economy; can reduce worker effort and raise labor costs if shifted to employers.
- Econometric evidence on tax wedge and structural unemployment:
  - Cross-country studies find tax wedge coefficients in range 0.17 to 0.36.
  - Implication: a reduction in the tax wedge by 10 percentage points would lead to a reduction in structural unemployment by 2 to 4 percentage points, all else equal.
  - Event study of nine episodes of large declines in structural unemployment (≥ 2.5 percentage points) correlates these declines with reductions in tax wedge for unskilled workers.
  - Correlation results: simple OLS regression shows coefficient 0.4 and R^2 = 0.3 → reductions in tax wedge explain about 30 percent of variation in structural unemployment in the sample.
- Examples of historically large declines in structural unemployment (Country | Period | Reduction in tax wedge | Reduction in structural unemployment):
  - Bulgaria 2001-08: 8.5 | 8.3
  - Finland 1996-2008: 6.7 | 5.4
  - Germany 2005-13: 2.3 | 3.6
  - Ireland 1996-2001: 10 | 6.1
  - Italy 1996-2007: 5.5 | 2.5
  - Poland 2002-08: 3.4 | 5.9
  - Slovakia 2001-08: 5.2 | 3.7
  - Spain 1996-2007: -1.3 | 4.5
- Conclusion on traditional factors:
  - High tax wedges, combined with benefit changes, can generate unemployment and inactivity traps by reducing financial gain from employment, especially for lower-wage earners.
  - Baltics face higher unemployment and inactivity traps than CE4, Anglo-Saxon, or other emerging market OECD countries.

### B. Non-Traditional Factors — Skill Mismatches and ALMPs
- Skill mismatches:
  - Baltics score close to OECD averages on PISA tests; variation across Baltics: Lithuania lags, Estonia outperforms OECD average.
  - Tertiary enrollment rates in line with OECD averages; male tertiary enrollment is low.
  - All three Baltics score worse than average on skill field mismatch.
  - Mismatches exist in educational attainment required within professions and in training for the right profession/sector.
  - Regional divides: e.g., in Lithuania, rural unemployment (much above urban) consists mostly of people with below-tertiary education.
- Active Labor Market Policies (ALMPs):
  - ALMP spending: Baltics 0.5 percent of GDP vs Nordics over 2 percent of GDP.
  - ALMP training spending: Baltics 0.1 percent of GDP — less than half that in the Nordics.
  - In Baltics, passive labor market support (out-of-work income support; in Latvia and Lithuania also early retirement) accounts for a large share of total ALMP spending.
  - ALMP spending did not increase dramatically since 2008/09 (exceptions: some increases from a small base, e.g., Estonia).
  - Low ALMP spending reflects very low participation in these programs rather than high efficiency.
  - Longer-term averages of unemployment (especially youth) show a negative correlation with ALMP spending, indicating potential of ALMPs to address skill mismatches.
- PISA and tertiary details (selected figures):
  - Math scores: Estonia 521; Latvia 491; Lithuania 479; Baltic average 497; OECD average 494.
  - Low performers in Math (Percent): Estonia 10.5; Latvia 19.9; Lithuania 26.0; Baltic ave. 18.8; OECD ave. 23.0.
  - High performers in Math (Percent): Estonia 14.6; Latvia 8.0; Lithuania 8.1; Baltic ave. 10.2; OECD ave. 12.6.
  - Tertiary enrollment rates (Percent) — total 2010: Estonia, Latvia, Lithuania, Baltic average, Nordics, Anglo-Saxon, Korea, OECD (ave.) shown in source.
  - Unemployment rate by educational attainment (Lithuania, 2012): urban vs rural and by attainment categories shown in source (not reproduced as single numbers beyond statement that rural unemployment is much above urban and concentrated among below-tertiary education).

*Source: IMF staff calculations and analysis contained in section III of the source document.*

### Box 1. Iceland: Active Labor Market Policies During the Crisis

### Box 1. Iceland: Active Labor Market Policies During the Crisis

### Context
- Crisis period: 2008/09 banking and economic crisis in Iceland.
- Problem addressed: sharply rising unemployment following the crisis.
- Initial emphasis: expanding registration for unemployment benefits and educating the public about available options.

### Measures Implemented
- Job retraining.
- Subsidized hiring for trial periods.
- Study programs.
- Subsidized hiring (general).
- Volunteer work placements.
- Opening secondary education to anyone under age 25.
- Programs emphasizing work-related education.
- Greater cooperation between social partners and the education system.

### Outcomes and Effectiveness
- Scope and evolution: The wide scope of ALMPs and their gradually changing role over the course of the crisis helped to increase the number of participants in the programs.
- Program-specific success: Programs providing on-the-job training/apprenticeships or employment in specific projects seem to have increased chances of participants “de-listing” from the unemployment rolls.
- Youth outcomes: Available information suggests that about half of unemployed youth found jobs after participating in the programs.
- Nature of unemployment addressed: ALMPs in this episode addressed mostly cyclical unemployment problems.

### Lessons and Implications
- ALMPs can be successful even when primarily addressing cyclical unemployment.
- Program design that includes on-the-job training and project-based employment appears especially effective at facilitating exit from unemployment.
- Gradual expansion and diversification of ALMPs can increase participation and improve outcomes.
- Coordination between education providers and social partners enhances the relevance of training and education initiatives.

*Source: IMF (2011)*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14153.pdf_