## wpiea2019130

## Source details

**Canonical URL:** [wpiea2019130](https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019130.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2019/wpiea2019130.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2019/wpiea2019130.pdf.json)

---

### I. Motivation and focus
- Context:
  - Demographic and pension pressures, low labor market participation rates, and still high tax wedges in many EU28 countries motivate policy review (Attinasi et al, 2016).
  - Recovery from the Global Financial Crisis (GFC) saw unemployment decline, personal income tax rates decrease, and improved labor market flexibility in EU New Member States (NMS).
  - NMS: Bulgaria, Croatia, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia and Slovenia.
- Objective:
  - Study hours worked (per population aged 15-64) over 1995-17.
  - Assess role of taxes in accounting for differences in hours across NMS using:
    - Stylized facts on tax rates and hours.
    - A calibrated neoclassical growth model augmented with taxes (following Ohanian et al (2008)).
    - An econometric analysis estimating elasticity of hours to different types of taxes.
- Key summary finding:
  - Consumption and labor taxes significantly discourage labor supply and can explain close to 21 percent of the observed variation of hours across NMS.

### II. Stylized facts on hours and taxes
- Hours worked (definition and trends):
  - Hours = total hours worked in the economy divided by population aged 15-64.
  - Four periods (5-year moving averages): 1995-02 decline; 2003-08 highest growth; 2009-13 contraction post-GFC; since 2013 recovery to historically high levels.
  - Cross-country heterogeneity up to about 200 annual hours per person (Estonia highest, Slovakia lowest).
- Decomposition (1995 vs 2017 percentage-point changes, selected countries):
  - Bulgaria: Hours 13.6; Intensive margin -2.7; Extensive margin 16.3.
  - Latvia: Hours 11.3; Intensive margin -4.1; Extensive margin 15.4.
  - Lithuania: Hours 15.5; Intensive margin 6.2; Extensive margin 9.2.
  - Croatia: Hours 1.7; Intensive margin -3.5; Extensive margin 5.3.
  - Czech Republic: Hours -1.7; Intensive margin -6.1; Extensive margin 4.4.
  - Estonia: Hours 0.1; Intensive margin -12.2; Extensive margin 12.3.
  - Hungary: Hours 6.2; Intensive margin -12.3; Extensive margin 18.6.
  - Poland: Hours 2.2; Intensive margin -2.7; Extensive margin 4.9.
  - Romania: Hours -15.9; Intensive margin -2.7; Extensive margin -13.2.
  - Slovak Republic: Hours -2.3; Intensive margin -7.9; Extensive margin 5.6.
  - Slovenia: Hours 1.7; Intensive margin -4.4; Extensive margin 6.2.
- Grouping:
  - Group 1: countries with hours increased by at least 10 percentage points (Bulgaria, Latvia, Lithuania).
  - Group-level observations:
    - Hours per worker declined in all countries except Lithuania.
    - Employment rates increased in all countries except Romania.
    - Rise in employment rates partly due to higher female labor force participation.
- Tax measures (1995-17, national accounts; Croatia from 2001):
  - Consumption tax rate = total consumption tax revenue / total private consumption (VAT, excises, import duties).
  - Labor tax rate = total tax revenue on labor income / total labor income (personal income taxes and social security contributions).
  - Consumption tax rates: around 20 percent on average.
  - Labor tax rates: hover around 30 percent on average.
  - Consumption and labor taxes are not highly correlated.
  - Cross-country highlights:
    - Croatia and Hungary: highest consumption tax rates.
    - Czech Republic: highest labor taxation.
    - Effective tax rates: Hungary highest, Romania lowest.
- Stylized relationships:
  - Negative correlation between hours worked and taxes in levels and changes, though not robust to exclusion of some countries.
  - Cross-country correlations (Figure 5): Panel (a) Group 1: -0.09, Group 2: -0.26; Panel (b) total log changes: -0.25.

### III. Model setup, calibration, and main mechanisms
- Household and production setup:
  - Lifetime utility: sum_{t=0}^∞ β^t u(C_t, H̄ − H_t), with u(C_t, H_t) = α log(C_t) + (1−α) (H̄ − H_t)^{1−γ}−1 / (1−γ).
  - Budget constraint: [1+(1−τ_at) r_t] a_t + (1−τ_lt) w_t H_t = (1+τ_ct) C_t + a_{t+1}.
  - Production: Y_t = A_t K_t^θ H_t^{1−θ}; K_{t+1} = (1−δ) K_t + I_t.
- Tax wedge / effective tax:
  - MRS = (1−τ_lt)(1+τ_ct) F_2.
  - Define (1−τ_t^eff) ≡ (1−τ_lt)(1+τ_ct).
  - Effective tax rate main variable: after-tax wage w(1−τ_effective) with (1−τ_effective) = (1−τ_labor)(1+τ_consumption).
- Calibration and parameters:
  - γ = 1.
  - H̄ = 14*365 = 5110.
  - α = 0.37.
  - θ = 1/3.
  - Data sources: AMECO for hours, consumption, output; national accounts for taxes.
- Key model implication:
  - Higher taxation reduces net labor income and real purchasing power, inducing substitution of consumption for leisure (untaxed), reducing labor supply.

### IV. Model results and counterfactuals
- Method:
  - Build equation (2) per country, normalize predicted and actual series to one in year 2000.
  - Compare model without taxes (effective taxes set to zero) and model with taxes.
- Model without taxes:
  - Group 1: actual hours grew 15 percent through sample; model without taxes predicts 2.5 percent increase.
  - Model without taxes fails to capture rise/fall around 2008 for both groups.
  - Group 2: model without taxes predicts a 4 percent overprediction between 2000 and 2015 compared with actual 2 percent rise.
- Model with taxes:
  - Group 1:
    - Model with taxes predicts 7 percent increase for 2000-08 (model without taxes predicts 2 percent); actual increase about 13 percent.
    - Between start of sample and 2013, model with taxes can help explain close to 50 percent of growth in hours worked.
    - Predictions do not track trend after 2013.
  - Group 2:
    - Model with taxes tracks evolution well, including GFC dynamics.
    - Entire sample: hours rose 2 percent; model with taxes predicts increase of 1 percent.
  - Pooled:
    - Incorporating taxes improves prediction for the average country.
    - A substantial break after 2013: actual hours present a significant rise which the model does not capture, partly because effective taxes increased after the GFC.
- Robustness:
  - Extending private consumption to include government spending on goods produced does not significantly change predictions.
- Appendix II counterfactuals (isolating taxes):
  - Group 1: labor-tax-only predicts higher hours than consumption-tax-only; labor taxes carry more weight in explaining rise in hours, consumption taxes also important after 2006.
  - Group 2: before GFC no large difference; after 2008 labor-tax-only predicts rise in hours while consumption-tax-only predicts decline — increases in consumption taxes post-GFC had important negative consequences for labor supply.

### V. Econometric specification and identification
- Baseline equation:
  - log(H_it) = β log(τ_it^eff) + δ′ X_it + φ log(C_it / Y_it) + c_i + μ_t + ε_it.
  - H_it: hours worked per working-age population.
  - τ_it^eff: effective tax (interaction of labor and consumption tax rates).
  - X_it: institutional covariates (openness, business regulation, labor market regulation, social benefits, remittances, migration).
  - c_i and μ_t: country and year fixed effects.
- Interpretation of β:
  - Percentage change in hours following a one percent increase in the tax rate (elasticity).
  - Also interpretable as fraction of government revenue lost due to decline in hours.
- Endogeneity and measurement:
  - Hours and taxes filtered with 5-year moving averages to remove short-run fluctuations.
  - Robustness: OECD tax measures used on subset (Appendix IV); results robust to tax data construction.
  - No instrumental variables used; results are correlational and should be interpreted with caution.
  - National accounts hours include informal and formal hours; larger informal sectors likely bias point estimates toward zero.
- Exploratory decomposition:
  - Replace effective taxes with consumption and labor tax rates; decompose labor tax into τ_it_PIT and τ_it_SSC and estimate separate elasticities.

### VI. Estimation results — magnitudes, margins, and gender
- Overall elasticity (Table 2):
  - Estimated β ranges between -0.390 and -0.543 across specifications (columns 1–8).
  - Interpretation: a one percent increase in effective tax rate leads to close to 0.5 percent decline in hours worked.
  - Implied deadweight loss: tax revenue increases only by 0.5 percent following a one percent increase in tax rates (other 0.5 percent is deadweight loss).
  - Significant controls:
    - log(social benefits/GDP): −0.087***.
    - log(business regulation): −0.129***.
    - log(openness): 0.121** (positive in some specifications).
  - R^2 up to 0.507 in full specification (column 8).
  - Taxes and consumption/output explain up to 21 percent of total variation of hours (column 1); including all covariates raises explained variation to 51 percent (column 8).
- By tax type and margin (Table 3):
  - log(tax cons):
    - log(hours): −0.138*** (0.036).
    - log(hours per worker): −0.159*** (0.020).
    - log(employment rate): 0.022 (not significant).
  - log(tax labor):
    - log(hours): −0.335*** (0.063).
    - log(hours per worker): −0.026 (not significant).
    - log(employment rate): −0.310*** (0.052).
  - Decomposed labor taxes:
    - log(tax PIT): −0.071* on log(hours) (0.036); −0.054*** on log(hours per worker).
    - log(tax SSC): −0.260*** on log(hours) (0.055); −0.219*** on log(employment rate) (0.045).
  - Summary:
    - Labor taxes have larger coefficients than consumption taxes.
    - Social security contributions show largest negative elasticity, followed by consumption taxes; personal income taxes play a smaller role.
    - Margin heterogeneity:
      - Consumption and personal income taxes affect hours per worker (intensive margin).
      - Social security contributions negatively associated with employment rates (extensive margin).
- Gender-specific results (Table 4):
  - Men:
    - log(tax labor): −0.312*** (0.055).
    - log(tax SSC): −0.217*** (0.047).
  - Women:
    - log(tax labor): −0.327*** (0.056).
    - log(tax SSC): −0.232*** (0.050).
  - Interpretation:
    - Labor taxes discourage both men’s and women’s employment via the extensive margin.
    - Women’s elasticity of employment rate with respect to labor taxes is around 5 percent higher than men’s (statistically significant at 5 percent).
    - Social security contributions affect women’s employment about 7 percent more than men’s (discussion in text).
- Robustness (Appendix IV, Table A1):
  - Using national accounts vs OECD tax measures and excluding Bulgaria, Croatia, Romania does not substantially change main results.
  - Selected coefficients (Table A1):
    - log(tax cons – National Accounts): Column (1) −0.236*** (0.034); Column (3) −0.154*** (0.020); Column (5) -0.084** (0.038).
    - log(tax on goods and services - OECD): Column (2) −0.336*** (0.046); Column (4) −0.260*** (0.041); Column (6) −0.077* (0.043).
    - log(tax labor – National Accounts): Column (1) −0.295*** (0.069); Column (3) −0.026 (0.072); Column (5) −0.273*** (0.078).
    - log(tax labor - OECD): Column (2) −0.189*** (0.064); Column (4) −0.077 (0.057); Column (6) −0.116* (0.059).
  - Observations: Columns (1),(3),(5): 75; Columns (2),(4),(6): 91.
  - R^2 range: 0.544 to 0.792 across reported columns.

### VII. Policy-oriented insights and conclusions
- Quantitative importance:
  - Taxes matter substantially for labor supply in NMS: estimated elasticity of hours to overall taxes close to 0.5.
  - Consumption and labor taxes together can explain close to 21 percent of observed variation in hours across NMS (in some specifications).
  - In some countries, tax changes account for close to half of the increase in hours between 1995-13.
- Tax-type differentiated policy implications:
  - Reductions in social security contributions may increase employment rates (extensive margin), with potentially larger benefits for women.
  - Reductions in consumption or personal income taxes may increase hours per worker (intensive margin).
  - Consumption taxes matter for labor supply and are relevant to long-term hours growth in NMS with effective tax rates below 40 percent.
- Distributional and gender considerations:
  - Social security contributions reduce employment rates, with larger negative effects on women — policy design should consider intra-household incentives, childcare, and tax treatment of married couples.
- Fiscal and reform cautions:
  - Deadweight losses from taxation are large relative to other labor-market determinants.
  - Reforms should be country-specific; aging populations and pension/health spending pressures imply fiscal implications must be carefully calibrated.
- Suggested further research:
  - Break down social security contributions by employer and employee shares (not available in national accounts).
  - Differentiate cash vs in-kind social benefits and poverty-targeted social spending vs unemployment insurance.
  - Expand micro-determinants of labor market participation (unemployment insurance, childcare), especially for women.
  - Further investigate episodes where the model fails to predict trend changes in hours (post-2013 puzzling event).

*Source: wpiea2019130 — REFERENCES and Appendices I–IV.*

### REFERENCES ___________________________________________________________27

### REFERENCES ___________________________________________________________27

### I. MOTIVATION
- Context:
  - Demographic and pension pressures, low labor market participation rates, and still high tax wedges in many EU28 countries motivate policy review (Attinasi et al, 2016).
  - Recovery from the Global Financial Crisis (GFC) saw unemployment decline, personal income tax rates decrease, and improved labor market flexibility in EU New Member States (NMS).
  - NMS refer to Bulgaria, Croatia, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia and Slovenia.
- Focus:
  - The paper studies hours worked (per working age population, defined as population aged 15-64 years old).
  - Objective: assess the role of taxes in accounting for differences in hours worked across NMS over the 1995-17 period using:
    - Stylized facts on tax rates and hours.
    - A calibrated neoclassical growth model augmented with taxes following Ohanian et al (2008).
    - An econometric analysis estimating elasticity of hours to different types of taxes.
- Key summary finding:
  - Consumption and labor taxes significantly discourage labor supply and can explain close to 21 percent of the observed variation of hours across NMS.

### II. STYLIZED FACTS
- Hours worked (definition and trends):
  - Hours = total hours worked in the economy divided by population aged 15-64.
  - Panel (a) Figure 2: average hours across NMS (5-year moving averages) show four periods:
    - 1995-02: hours declined.
    - 2003-08: highest growth period.
    - 2009-13 (post-GFC): contraction lasting five years.
    - Since 2013: recovery to historically high levels.
  - Heterogeneity across countries: differences up to about 200 annual hours per person between highest (Estonia) and lowest (Slovakia).
- Cross-country and income relationships:
  - NMS on average work more hours than EU15 (Eurofound 2008); lower productivity may be compensated by more hours.
  - For NMS, no clear negative correlation between hours and GDP per capita for the 1995-17 average: Estonia and the Czech Republic (high income) worked more hours than other NMS.
- Decomposition of hours (identity):
  - Hours worked = Hours per worker (intensive margin) * Employment rate (extensive margin).
  - Table 1 reports changes in hours and margins between 1995 and 2017 (in percentage points). Selected values:
    - Bulgaria: Hours 13.6; Intensive margin -2.7; Extensive margin 16.3.
    - Latvia: Hours 11.3; Intensive margin -4.1; Extensive margin 15.4.
    - Lithuania: Hours 15.5; Intensive margin 6.2; Extensive margin 9.2.
    - Croatia: Hours 1.7; Intensive margin -3.5; Extensive margin 5.3.
    - Czech Republic: Hours -1.7; Intensive margin -6.1; Extensive margin 4.4.
    - Estonia: Hours 0.1; Intensive margin -12.2; Extensive margin 12.3.
    - Hungary: Hours 6.2; Intensive margin -12.3; Extensive margin 18.6.
    - Poland: Hours 2.2; Intensive margin -2.7; Extensive margin 4.9.
    - Romania: Hours -15.9; Intensive margin -2.7; Extensive margin -13.2.
    - Slovak Republic: Hours -2.3; Intensive margin -7.9; Extensive margin 5.6.
    - Slovenia: Hours 1.7; Intensive margin -4.4; Extensive margin 6.2.
  - Grouping:
    - Group 1: countries where hours increased by at least 10 pp. in the whole period (Bulgaria, Latvia, Lithuania).
  - Observed trends:
    - Hours per worker declined in all countries except Lithuania.
    - Employment rates increased in all countries except Romania.
    - Rise in employment rates partly due to higher female labor force participation.

### TAXES IN NMS
- Tax measures constructed:
  - Consumption tax rate = total consumption tax revenue / total private consumption (includes VAT, excises, import duties).
  - Labor tax rate = total tax revenue on labor income / total labor income (incorporates personal income taxes and social security contributions).
  - Data source and period: national accounts, 1995-17 (Croatia from 2001).
  - These measures represent average tax burden households face in each country over the time span; novel application for NMS in this study.
- Stylized relationship:
  - Negative correlation documented between hours worked and taxes, both in levels and changes.
  - Heterogeneity in tax levels across NMS corresponds with heterogeneity in hours.

### MODELING AND EMPIRICAL APPROACH
- Model:
  - Calibrated neoclassical growth model augmented with taxes to account for time-trends in hours following Ohanian et al (2008).
  - Mechanism: higher taxation reduces households’ net labor income and real purchasing power, inducing substitution of consumption for leisure (untaxed), thereby reducing labor supply.
  - Comparison: model with taxes versus model without taxes.
    - Model without taxes performs poorly in tracking evolution of actual hours in most countries.
    - Inclusion of taxes significantly improves tracking, especially for countries where hours increased most.
    - For some countries, changes in tax rates can account for close to half of the increase in hours between 1995-13.
    - Identified puzzling period: 2013-17 where the model fails to predict trend changes in hours.
- Econometric analysis:
  - Objectives:
    - Complement model findings by including other determinants and institutions: trade openness, business and labor market regulation, social benefits, remittances, migration.
    - Estimate elasticity of hours (and margins) to different tax types and assess gender differences.
  - Main empirical findings:
    - Conditional on other factors, taxes are an important determinant of hours.
    - Point estimates suggest a high elasticity of hours to taxes, close to 0.5; robust to inclusion of other factors.
    - By tax type:
      - Social security contributions deter hours the most.
      - Consumption taxes deter hours next most.
      - Personal income taxes have a smaller negative effect.
    - By margin:
      - Consumption and personal income taxes affect hours per worker (intensive margin), but not employment rates.
      - Social security contributions are negatively associated with employment rates (extensive margin), but do not seem to affect hours per worker.
    - By gender:
      - Women’s employment rate is more sensitive to changes in tax policies than men’s.
      - Elasticity of employment rate to social security contributions is 7 percent larger for women vis-à-vis men.

### COMPARISON WITH LITERATURE
- Relation to Prescott (2004), Ohanian et al (2008), McDaniel (2011):
  - Prior studies attribute divergent trends in hours across countries largely to changes in taxation; this paper extends analysis to NMS.
- Other explanations in literature:
  - Alesina et al (2005): unionization and bargaining.
  - Maoz (2010): stronger preference for leisure.
  - Rogerson (2007): government investment in non-market activities offsets tax disincentives.
  - Causa (2009): marginal taxes disincentivize women working longer hours.
  - Recent micro and disaggregated analyses: Bick et al (2018a), Chakraborty et al. (2015), Bick and Fuchs-Schundeln (2018).

### POLICY-ORIENTED INSIGHTS
- Taxes are quantitatively important in explaining cross-country differences and time variation in hours worked across NMS.
- Heterogeneous effects by tax type and margin imply:
  - Reductions in social security contributions may increase employment rates (extensive margin), with potentially larger benefits for women.
  - Reductions in consumption or personal income taxes may increase hours per worker (intensive margin).
- Quantitative accounting:
  - Consumption and labor taxes together can explain close to 21 percent of observed variation in hours across NMS.
  - In some countries, tax changes account for close to half of the increase in hours between 1995-13.

*Source: wpiea2019130 - REFERENCES ___________________________________________________________27*

### Appendix I, we provide a detailed description of the construction of taxes.

### Appendix I, we provide a detailed description of the construction of taxes.

### Effective tax rate: definition and construction
- The effective tax rate combines consumption and labor tax rates to represent the total tax burden on individuals.
- If w is the wage, the after-tax wage is defined as: w(1−τ_effective) with (1−τ_effective) = (1−τ_labor)(1+τ_consumption).
- The effective tax rate embodies both changes in consumption and labor taxation and is the main variable of interest.
- Footnote origin: “The origins of this definition come from the first order condition of the model, explained in the next section.”

### Stylized facts: levels, cross-country dispersion, and groups
- Consumption tax rates: around 20 percent on average.
- Labor tax rates: hover around 30 percent on average.
- Consumption and labor taxes are not highly correlated.
- Cross-country highlights:
  - Croatia and Hungary have the highest consumption tax rates.
  - Czech Republic exhibits the highest labor taxation.
  - Effective tax rates: Hungary has the highest, Romania the lowest.
- Country grouping for hours dynamics (1995–2017):
  - Group 1: countries where hours increased by at least 10 percentage points in 1995-17.
  - Group 2: all other countries with a small increase or decline in hours.
- Hours and taxes by group (summary of Figure 4):
  - Hours worked increased substantially for Group 1; remained stable in Group 2.
  - The Global Financial Crisis (GFC) produced a sharp decline in hours after 2008 in both groups.
  - Countries in Group 1 had lower effective tax rates (below 40 percent) throughout the period.
  - A negative correlation between hours and taxes is observed for most of the sample.
  - For Group 1, the large increase in hours between year 2000 and the GFC happened when taxes declined.
- Cross-country correlations (Figure 5):
  - Panel (a) correlations: Group 1: -0.09, Group 2: -0.26.
  - Panel (b) correlation (total log changes): -0.25.
  - Relationship is negative but not robust to exclusion of some countries.

### The model: setup, calibration, and mechanism
- Representative household utility:
  - Lifetime utility: sum_{t=0}^∞ β^t u(C_t, H̄ − H_t) with 0 < β < 1.
  - Period utility: u(C_t, H_t) = α log(C_t) + (1−α) (H̄ − H_t)^{1−γ}−1 / (1−γ).
  - Parameters: 0 ≤ α ≤ 1 (taste for consumption vs leisure); γ ≥ 0 (elasticity of substitution between consumption and leisure).
- Budget constraint (including taxes):
  - [1+(1−τ_at) r_t] a_t + (1−τ_lt) w_t H_t = (1+τ_ct) C_t + a_{t+1}.
  - Taxes: labor income tax τ_lt, capital income tax τ_at, consumption tax τ_ct.
- Production:
  - Cobb-Douglas: Y_t = A_t K_t^θ H_t^{1−θ}, 0 < θ < 1.
  - Capital law of motion: K_{t+1} = (1−δ) K_t + I_t, 0 < δ < 1.
- First-order condition and tax wedge:
  - MRS = tax-adjusted marginal product of labor: U_2 / U_1 = (1−τ_lt)(1+τ_ct) F_2.
  - Define (1−τ_t^eff) ≡ (1−τ_lt)(1+τ_ct) as the tax wedge (one minus the effective tax rate).
- With γ set to 1 and algebraic rearrangement, hours satisfy:
  - H_t = (1−θ)^{α/(1−α)} (1−θ)^{α/(1−α)} + (C_t / Y_t)
    1/(1−τ_t^eff) H̄
  - (Displayed in source as equation (2); the presentation emphasizes that hours depend on the effective tax and the consumption share.)
- Calibration and data:
  - γ = 1 (elasticity of substitution between leisure and consumption).
  - H̄ = 14*365 = 5110.
  - α = 0.37.
  - θ = 1/3.
  - Sources for hours, consumption, and output: AMECO database.
  - Average taxes used (national accounts); model has complete data for all 11 NMS for 1995-17 (tax data missing for Croatia for 1995-01).

### Model results: role of taxes in predicting hours
- Approach:
  - Build equation (2) for each country and compute average predicted trend of H_t by Group.
  - Normalize predicted and actual series to one in the year 2000.
  - Compare model without taxes (set effective taxes to zero) and model with taxes.
- Model without taxes (Figure 6):
  - For Group 1: actual hours grew by 15 percent through the whole sample; model without taxes predicts a 2.5 percent increase.
  - The model without taxes fails to capture rise and fall of hours around 2008 for both groups.
  - For Group 2: model without taxes predicts a positive trend (4 percent overprediction between 2000 and 2015 compared with actual 2 percent rise); does not capture decline between 1997 and 2000.
- Model with taxes (Figure 7):
  - Group 1:
    - Model with taxes predicts an increase in hours of 7 percent for the 2000-08 period, while model without taxes predicts a 2 percent rise.
    - Actual hours increased about 13 percent in this period.
    - Between beginning of sample and 2013, model with taxes alone can help explain close to 50 percent of the growth in hours worked.
    - Predictions do not track trend after 2013.
  - Group 2:
    - Model with taxes tracks evolution of hours well, especially rise and fall around the GFC.
    - Entire sample: hours rose by 2 percent; model with taxes predicts an increase of 1 percent.
    - Excluding taxes leads to an overprediction in the rise in hours of 4 percent.
    - In pre-2008 rise, models with and without taxes predict the same increase, suggesting taxes did not play role in those years for Group 2.
  - Pooled (Figure 8):
    - Incorporating taxes improves prediction of hours for the average country.
    - A substantial break appears after 2013: actual hours present a significant rise which the model does not capture, partly due to the increase in effective taxes after the GFC.
    - Suggests other factors behind post-2013 rise in hours.
- Robustness on government spending:
  - Extending private consumption to include government spending on goods produced predictions not significantly different.

### Econometric analysis: specification, intuition, and identification
- Baseline econometric specification (derived from structural determinants, eq. 2):
  - log(H_it) = β log(τ_it^eff) + δ′ X_it + φ log(C_it / Y_it) + c_i + μ_t + ε_it.
  - H_it: hours worked over working age population.
  - τ_it^eff: effective tax in country i, year t.
  - C_it / Y_it: consumption-output ratio (controls for intertemporal decisions).
  - X_it: vector of institutional covariates (openness, business regulation, labor market regulation, social benefits, remittances, migration).
  - c_i and μ_t: country and year fixed effects.
- Interpretation of β:
  - Measures percentage change in hours following a one percent increase in the tax rate (elasticity of hours to tax rates).
  - Also interpretable as fraction of government revenue lost due to decline in hours: a one percent increase in tax rate (e.g., from 40.0 percent to 40.4 percent) leads to an increase in government revenue of 1+β percent.
- Addressing endogeneity and measurement:
  - Hours and taxes filtered with 5-year moving averages to remove short-run fluctuations.
  - Robustness using OECD tax measure on subset (Appendix IV) shows elasticity not dependent on tax data construction.
  - No instrumental variables used; results should be interpreted with caution.
  - National accounts hours include informal and formal hours; larger informal sectors likely bias point estimates toward zero.
- Exploratory decomposition:
  - Semi-aggregate analysis decomposes hours into hours per worker and employment rate.
  - Replace effective taxes with consumption and labor tax rates:
    - log(y_it) = β1 log(τ_it_cons) + β2 log(τ_it_labor) + δ′ X_it + φ log(C_it / Y_it) + c_i + μ_t + ε_it.
  - Further decompose labor tax into personal income tax on labor (τ_it_PIT) and social security contributions (τ_it_SSC) and estimate:
    - log(y_it) = β1 log(τ_it_cons) + β2 log(τ_it_PIT) + β3 log(τ_it_SSC) + controls + fixed effects.

### Estimation results: magnitudes, margins, and gender
- Table 2 (log(hours worked) on log(effective tax)):
  - Estimated β ranges between -0.390 and -0.543 across specifications (columns 1–8).
  - Interpretation: if the average NMS increases its effective tax rate by one percent, then hours worked would decline by close to half percent.
  - Implied deadweight loss: tax revenue would increase only by 0.5 percent following a one percent increase in tax rates (the other 0.5 percent is deadweight loss).
  - Controls with statistically significant associations:
    - log(social benefits/GDP): −0.087*** (negative association).
    - log(business regulation): −0.129*** (negative association).
    - log(openness): 0.121** (positive association in some specifications).
  - R^2 ranges: up to 0.507 in full specification (column 8).
  - Taxes and consumption/output explain up to 21 percent of total variation of hours (column 1); including all covariates increases explained variation to 51 percent (column 8).
- Table 3 (by tax type; dependent variables: log(hours), log(hours per worker), log(employment rate)):
  - log(tax cons):
    - Column (1) log(hours): −0.138*** (0.036).
    - Column (3) log(hours per worker): −0.159*** (0.020).
    - Column (5) log(employment rate): 0.022 (not significant).
  - log(tax labor):
    - Column (1) log(hours): −0.335*** (0.063).
    - Column (3) log(hours per worker): −0.026 (not significant).
    - Column (5) log(employment rate): −0.310*** (0.052).
  - Decomposed labor taxes:
    - log(tax PIT): −0.071* on log(hours) (0.036); −0.054*** on log(hours per worker).
    - log(tax SSC): −0.260*** on log(hours) (0.055); −0.219*** on log(employment rate) (0.045).
  - Summary:
    - Both consumption and labor taxes are important determinants of hours worked.
    - Labor taxes have larger coefficients than consumption taxes.
    - Social security contributions show the largest elasticity (largest negative effect), followed by consumption taxes; personal income taxes on labor play a smaller role.
    - Heterogeneous margin effects:
      - Consumption taxes and personal income taxes associated with lower hours per worker (intensive margin).
      - Social security contributions associated with lower employment rates (extensive margin).
- Table 4 (gender and type of tax; dependent variable: log(employment rate)):
  - Men:
    - log(tax labor): −0.312*** (0.055).
    - log(tax SSC): −0.217*** (0.047).
  - Women:
    - log(tax labor): −0.327*** (0.056).
    - log(tax SSC): −0.232*** (0.050).
  - Interpretation:
    - Labor taxes discourage both men’s and women’s employment via the extensive margin.
    - Women’s elasticity of employment rate with respect to labor taxes is slightly higher (around 5 percent) than men’s (statistically significant at the 5 percent level).
    - Social security contributions affect women’s employment decision/status about 7 percent more than men’s (discussion in text: higher discouragement for women likely consequence of intrahousehold decisions and second-earner effects).

### Key policy implications and conclusions
- Taxes matter substantially for labor supply in NMS:
  - Model and econometric evidence support that taxation plays a significant role in explaining hours worked across countries.
  - Estimated elasticity of hours to overall taxes close to 0.5.
  - Differences in tax burden can explain up to 21 percent of variation in hours worked across NMS (in some specifications).
- Deadweight losses from taxation are large relative to other labor-market determinants.
- Consumption taxes matter, not only labor taxes:
  - Consumption taxes are associated with lower hours per worker.
  - Long-term growth in hours observed in NMS with effective tax rates below 40 percent.
  - Policy design should consider overall tax burden (combined consumption and labor taxes) when assessing incentives.
- Distributional and gender considerations:
  - Social security contributions reduce employment rates, with larger negative effects on women.
  - Reforms should consider gender dimension, intra-household incentives, childcare policies, and tax deductions affecting married couples.
- Nuanced reform advice:
  - Avoid one-size-fits-all reforms: each NMS has idiosyncratic challenges.
  - With aging populations and spending pressures (pensions, health), fiscal implications of tax reforms must be carefully calibrated.
- Suggested avenues for further research:
  - Break down social security contributions by employer and employee shares (not available in national accounts).
  - Evaluate differences between cash and in-kind social benefits; differentiate poverty-targeted social spending and unemployment insurance.
  - Expand to micro-determinants of labor market participation (e.g., unemployment insurance, childcare opportunities), especially for women.
  - Further investigate episodes where the model fails to predict trend changes in hours (referred to as “puzzling events” in related literature).

*Source: Appendix I, detailed description of the construction of taxes, wpiea2019130.*

### REFERENCES

### REFERENCES

### Bibliographic sources
- Alesina, A., E. Glaeser, and B. Sacerdote (2005). “Work and Leisure in the U.S. and Europe: Why So Different?”, NBER Macroeconomics Annual, Vol. 20 Issue 1.
- Atoyan, R., L. Christiansen, A. Dizioli, C. Ebeke, N. Ilahi, A. Ilyina, G. Mehrez, H. Qu, F. Raei, A. Rhee, and D. Zakharova (2016). “Emigration and Its Economic Impact on Eastern Europe”. IMF Staff Discussion Note. SDN/16/07
- Attinasi, MG, F. Berardini, R. De Stefani and S. Osterloh (2016). “The Effects of Labor Income Taxes on Labor Market Performance: An Empirical Analysis”. Working Paper.
- Bakija, J. (2011). “Tax Rates and Labor Supply in OECD Countries”. MIMEO.
- Betcherman, G. and C. Pagés (2007). “Estimating the Impact of Labor Taxes on Employment and the Balances of the Social Insurance Funds in Turkey”. World Bank.
- Bick A. and N. Fuchs-Schündeln (2018). “Taxation and Labor Supply of Married Couples across Countries: A Macroeconomic Analysis”, Review of Economic Studies, 85(3), July 2018, 1543-1576.
- Bick A., B. Brüggemann, N. Fuchs-Schündeln, and H. Paule-Paludkiewicz (2018a). “Long-term Changes in Married Couples' Labor Supply and Taxes: Evidence from the U.S. and Europe since the 1980s”, NBER Working Paper 248995, National Bureau of Economic Research.
- Bick A., N. Fuchs-Schündeln and D. Lagakos (2018b) “How Do Hours Worked Vary with Income? Cross-Country Evidence and Implications”. American Economic Review, 108(1), January 170-199.
- Blundell, R., Graber, M. and M. Mogstad. (2015) "Labor income dynamics and the insurance from taxes, transfers, and the family." Journal of Public Economics 127 pp. 58-73.
- Booth A. L., P. Katic (2010) “Estimating the Wage Elasticity of Labour Supply to a Firm: What Evidence Is There for Monopsony?” IZA Discussion Papers. DP No. 5167.
- Boppart, T. and P. Krusell (2016). “Labor supply in the past, present, and future: a balanced-growth perspective”. NBER Working Paper No. 22215
- Chakraborty, I., H. A. Holter, and S. Stepanchuk (2015): “Marriage Stability, Taxation and Aggregate Labor Supply in the US vs. Europe,” Journal of Monetary Economics, 72, 1–20.
- Causa, O. (2009). “The Policy Determinants of Hours Worked Across OECD Countries”, OECD Journal: Economic Studies, Volume 2009.
- Chetty, R., Guren, A., Manoli D. and A. Weber (2011). "Are Micro and Macro Labor Supply Elasticities Consistent? A Review of Evidence on the Intensive and Extensive Margins." American Economic Review, 101 (3): 471-75.
- Chetty, R, Guren A, Manoli D, and A. Weber (2013). “Does Indivisible Labor Explain the Difference between Micro and Macro Elasticities? A Meta-Analysis of Extensive Margin Elasticities”. NBER Macroeconomics Annual, 27(1), 1-56.
- de Boer, H., Egbert J. and M. Mastrogiacomo (2018). “Tax-Benefit Reforms and Structural Models for Labour Supply.” Structural Reforms, pp 239-264
- Eurofound (2008) “Comparative Analysis of Working Time in the European Union”. Report.
- Evers M., De Mooij R. and D. Van Vuuren (2008) “The wage elasticity of labor supply: a synthesis of empirical estimates”. De Economist 156, No. 1
- Giannis, M. (2014) "The Effect of Relative Wage on Hours Worked". Honors Projects. Paper 127.
- Gollin, D. (2002) “Getting Income Shares Right.” Journal of Political Economy 110(2) April 2002: 458-474
- Grigoli, F., Koczan Z. and P. Topalova (2018) “Drivers of Labor Force Participation in Advanced Economies: Macro and Micro Evidence.” IMF Working Paper series 18/150.
- Guvenen, F., Karahan, F. Ozkan, S. and J. Song. (2015) “What do data on millions of US workers reveal about life-cycle earnings risk?” No. w20913. National Bureau of Economic Research.
- Hoffmann, E. B. and D. Malacrino. (2019) "Employment time and the cyclicality of earnings growth." Journal of Public Economics 169 pp.160-171.
- Jones, C.I. (2016). "The Facts of Economic Growth", Handbook of Macroeconomics, 2016, Vol. 2A, pp. 3-69
- Keane, M. (2010). “Labor Supply and Taxes: A Survey” Working paper n°160. UTS School of Finance and Economics.
- Keane, M. and R. Rogerson. (2012). "Micro and Macro Labor Supply Elasticities: A Reassessment of Conventional Wisdom", Journal of Economic Literature, 50 (2): 464-76.
- Krause, E. and I. Sawhill (2018). “What Do We Know and Don’t Know about Declining Labor Force Participation: A Review”. The Brookings Institution, May 2018.
- Maoz, Y.D. (2010). “Labor Hours in the United States and Europe: the role of Different Leisure Preferences.” Macroeconomic Dynamics, 14 (2010), 231–241
- McDaniel, C. (2007). “Average Tax Rates on Consumption, Investment, Labor and Capital in the OECD 1950-2003”. Arizona State University.
- McDaniel C. (2011) “Forces Shaping Hours in the OECD”. American Economic Journal: Macroeconomics, Volume 3, Number 4
- Peterman, W.B. (2015). “Reconciling Micro and Macro Estimates of the Frisch Labor Supply Elasticity”, Economic Inquiry, 2016, 54 (1), 100-120
- Prescott, E. C. (2004). “Why Do Americans Work So Much More Than Europeans?”, FRB Minneapolis Quarterly Review, July issue, 2-13.
- OECD (2018), Revenue Statistics 2018, OECD Publishing, Paris, https://doi.org/10.1787/rev_stats-2018-en.
- Ohanian, L., A. Raffo, and R. Rogerson (2008) “Long-term Changes in Labor Supply and Taxes: Evidence from OECD Countries, 1956–2004”. Journal of Monetary Economics, 2008, 55, 1353–1362.
- Rogerson, R. (2007). “Taxation and Market Work: Is Scandinavia an Outlier?”, National Bureau of Economic Research, Working Paper 12890.
- Rogerson, R. and J. Wallenius (2009). “Micro and Macro Elasticities in a Life Cycle Model with Taxes”, Journal of Economic Theory, 144 (2009), 2277-2292.

### Appendix I. Construction of tax rates — methodology and formulas
- Data:
  - Source: AMECO database for the period 1995-2017 for 11 NMS.
  - Variables and definitions: GDP, TPI (Taxes on production and imports), Sub (Subsidies), W (Compensation of employees), OSMI (hh) Gross operating surplus & mixed income, C (Private final consumption expenditure), I (private) Gross fixed capital formation – (gov) gross fixed capital formation, HHT (hh) Current taxes on income & wealth, SS (gov) Social contributions received.
- Construction of average household income tax rate:
  - τ_income = HHT / [GDP − (TPI − Sub)]
- Labor share calculation (let α be capital share, (1−α) labor share):
  - Accounting identity: W + (1−α) OSMI = (1−α)(GDP − (TPI − Sub))
  - Solving for labor share:
    - (1−α) = W / [GDP − (TPI − Sub) − OSMI]
- Total labor taxes paid by households:
  - HHTl = τ_income (1−α)(GDP − (TPI − Sub))
- Average labor tax rate (including social contributions SS):
  - τ_l = [SS + HHTl] / [(1−α)(GDP − (TPI − Sub))]
- Decomposition of τ_l:
  - τ_l = τ_SSC + τ_PIT
  - τ_SSC = SS / [(1−α)(GDP − (TPI − Sub))]
  - τ_PIT = HHTl / [(1−α)(GDP − (TPI − Sub))]

- Construction of consumption tax:
  - Taxes on production and imports, TPI, are taken to include VAT, excise taxation, import duties and property taxes; property taxes are allocated to consumption taxation.
  - Let λ be the share of taxation relying only on consumption goods while (1−λ) relies on both consumption and investment.
  - Total tax revenue levied on consumption:
    - TPIc = (λ + (1−λ) (C / (C + I))) (TPI − Sub)
  - The authors assume λ = 1/3 (following Prescott (2004)).
  - Average tax on consumption:
    - τ_c = TPIc / [C − TPIc]

### Appendix II. Extension of the model: isolating consumption and labor taxes — counterfactuals and findings
- Experiment setup:
  - Construct effective tax in equation (2) while setting either labor taxes to zero or consumption taxes to zero to isolate each tax's role in trend predictions of hours.
  - Compare three series: (i) both taxes (benchmark), (ii) labor tax only, (iii) consumption tax only.
- Findings by group:
  - Group 1:
    - Labor tax only series predicts a higher level of hours than the consumption tax only series.
    - Interpretation: labor taxes carry more weight than consumption taxes in explaining the rise in hours in Group 1.
    - Consumption taxes are also important, particularly after 2006.
  - Group 2:
    - Before the GFC, no large difference between consumption-only and labor-only series; both taxes carry similar weight.
    - After 2008:
      - Labor-tax-only predicts a rise in hours.
      - Consumption-tax-only predicts a decline in hours.
    - Interpretation: increases in consumption taxes post-GFC had important negative consequences for labor supply in these countries; rising consumption taxes are needed to drag the combined series downward.

### Appendix III. Description and source of variables
- Hours worked: Total hours divided by population aged 15-64 years old — AMECO
- Hours per worker: Total hours divided by total employment — AMECO
- Employment rate: Total employment divided by population aged 15-64 years old — AMECO
- Effective tax: Interaction between labor and consumption tax rates. See Appendix I for construction — AMECO
- Openness: Total exports plus total imports divided by GDP — World Bank
- Social benefits: Amount of social benefits divided by GDP. Transfers in cash or in kind to protect specific segments of the population against certain social risks — WEO
- Business regulation: Higher number indicates more business regulation; based on an index including: (i) administrative requirements, (ii) bureaucracy costs, (iii) starting a business, (iv) extra payments / bribes / favoritism, (v) licensing restrictions and (vi) cost of tax compliance — Fraser Institute
- Labor mkt. regulation: Higher number indicates more regulation in the labor market; index includes: (i) hiring regulations and minimum wage (ii) hiring and firing regulations (iii) centralized collective bargaining (iv) hours regulations (v) mandated cost of worker dismissal (vi) conscription — Fraser Institute
- Remittances: Personal remittances as a percentage of GDP — World Bank
- Migration: Total emigrants minus immigrants divided by population aged 15-64 years old — Atoyan et al (2016)
- Consumption/Output: (Private consumption + Government spending)/GDP — AMECO
- Employment rate by gender: Total (wo)men employed divided by total (wo)men aged 15-64 — Eurostat and AMECO

### Appendix IV. Robustness analysis — alternative tax measures and main quantitative results
- Alternative data:
  - OECD tax measures used as an alternative to national accounts construction; OECD measures: taxes on goods and services, personal income tax and social security contributions as a share of GDP.
  - OECD database does not include Bulgaria, Croatia and Romania.
- Replication strategy and comparisons:
  - First replicate Table 4 results excluding Bulgaria, Croatia and Romania to have a benchmark.
  - Then perform same econometric specification using OECD tax measures and compare to benchmark.
- Key empirical comparisons and findings:
  - Excluding Bulgaria, Croatia and Romania (columns 1, 3 and 5 of Table A1):
    - Consumption taxes: slightly higher elasticity with respect to hours worked (column 1) relative to original Table 4.
    - Hours per worker: similar estimated coefficient maintained.
    - Employment rate: now negative and statistically significant effect of consumption taxes (was not significant before).
    - Elasticity of hours to labor taxes: slightly smaller than in Table 4.
    - Conclusion: dropping three countries does not substantially change main results.
  - Using OECD measures (columns 2, 4 and 6 of Table A1) versus national accounts:
    - OECD consumption tax coefficient is slightly larger than national accounts consumption tax coefficient.
    - OECD labor tax coefficient is smaller in magnitude than national accounts labor tax.
    - Both OECD estimates are statistically significant at the 1 percent level.
    - Column 4: consumption taxes have a significant impact on hours per worker; labor taxes do not show that effect.
    - Column 6: labor taxes play a larger role than consumption taxes through the employment rate; magnitude of labor taxes is somewhat smaller when employing OECD data.
  - Overall conclusion:
    - The econometric results in section IV are not particularly sensitive to the source of tax data (national accounts vs OECD) or to the sample of countries (including or excluding Bulgaria, Croatia, Romania).

### Table A1. An Alternative Measure of Taxes — selected regression results (as reported)
- Dependent variable columns: log(hours) (1,2), log(hours per worker) (3,4), log(employment rate) (5,6)
- Coefficients and statistics (standard errors in parentheses):
  - log(tax cons – National Accounts):
    - Column (1): −0.236***  (0.034)
    - Column (3): −0.154***  (0.020)
    - Column (5): -0.084**   (0.038)
  - log(tax on goods and services - OECD):
    - Column (2): −0.336***  (0.046)
    - Column (4): −0.260***  (0.041)
    - Column (6): −0.077*    (0.043)
  - Log(tax labor – National Accounts):
    - Column (1): −0.295***  (0.069)
    - Column (3): −0.026     (0.072)
    - Column (5): −0.273***  (0.078)
  - log(tax labor - OECD):
    - Column (2): −0.189***  (0.064)
    - Column (4): −0.077     (0.057)
    - Column (6): −0.116*    (0.059)
- Controls: Yes in all columns
- Country FE: Yes in all columns
- Year FE: Yes in all columns
- Observations:
  - Columns (1),(3),(5): 75
  - Columns (2),(4),(6): 91
- R^2:
  - Column (1): 0.792
  - Column (2): 0.720
  - Column (3): 0.573
  - Column (4): 0.544
  - Column (5): 0.591
  - Column (6): 0.625
- Note: Cluster robust standard errors in parentheses. * p<0.1; ** p<0.05; *** p<0.01

*Content derived from wpiea2019130 - REFERENCES, including Appendices I–IV, as presented in the source PDF.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019130.pdf_