Abstract
This paper mainly examines fiscal decentralization, credit-loss recovery, and unemployment in Croatia. The degree of expenditure and revenue decentralization in Croatia appears limited relative to its peers. At about 16 percent of general government spending, subnational government spending in Croatia is modest compared to other southeastern European countries and to the EU-28 average, and particularly low compared to the most decentralized countries in the EU. Croatia's recovery since late 2014 has been moderate. Croatia's recession lasted six years and was thus the longest among the new EU member states. Croatia's structural and cyclical unemployment rates are very high, at about 11.5 percent and 5 percent respectively in 2015.
Regional Unemployment, the Minimum Wage, and Incentives to Work1
A. Labor Market Developments and Regional Unemployment
1. Croatia’s structural and cyclical unemployment rates are very high, at about 11.5 percent and 5 percent respectively in 2015. While the high level of cyclical unemployment reflects the still wide output gap, the high structural unemployment has been persistent, which reflects the lack of substantial structural reforms.
2. Unemployment disproportionately impacts lower-skilled labor. Workers with medium-skills2 constituted nearly 63 percent of all unemployed in 2014, followed by low-skilled workers—around 26 percent of total. Highly-skilled workers constituted only around 12 percent of the unemployed. Long-term unemployment has increased sharply during the recession (long-term unemployed are about 60 percent of the total unemployed) and stand much higher than in peers.
3. Overall employment has dropped by about 11 percent from its pre-crisis level in 2008.
The SME sector contributed most to the decline (5 percentage points), followed by the SOE sector (4.2 percentage points), and large privately owned enterprises (2.1 percentage points). However, SMEs remained the most important employer and continued to employ more than 60 percent of all workers, while the proportion of those employed by the public enterprises declined.
Evolution of Employment Losses during Recession, 2008-2014
(percent of 2008 employment level)
Citation: IMF Staff Country Reports 2016, 188; 10.5089/9781498321587.002.A003
Source: CNB and staff calculations.Evolution of Employment Losses during Recession, 2008-2014
(percent of 2008 employment level)
Citation: IMF Staff Country Reports 2016, 188; 10.5089/9781498321587.002.A003
Source: CNB and staff calculations.Evolution of Employment Losses during Recession, 2008-2014
(percent of 2008 employment level)
Citation: IMF Staff Country Reports 2016, 188; 10.5089/9781498321587.002.A003
Source: CNB and staff calculations.4. The geographic distribution of employment losses have been unequal. In relative terms, the east counties were hit hardest while the northern counties and the City of Zagreb fared better given their initial share of employment.
Initial share of total employment, 2008
(percent of total)
Citation: IMF Staff Country Reports 2016, 188; 10.5089/9781498321587.002.A003
Source: Croatian Bureau of Statistics and staff calculations.Initial share of total employment, 2008
(percent of total)
Citation: IMF Staff Country Reports 2016, 188; 10.5089/9781498321587.002.A003
Source: Croatian Bureau of Statistics and staff calculations.Initial share of total employment, 2008
(percent of total)
Citation: IMF Staff Country Reports 2016, 188; 10.5089/9781498321587.002.A003
Source: Croatian Bureau of Statistics and staff calculations.5. Regional disparity in unemployment rates is very high, with the unemployment rate in the eastern counties nearly four times as high as in Zagreb. Moreover, the difference in regional unemployment rates, especially among low- and medium-skilled workers, has been persistent for many years.
B. Factors Leading to the Large Differences in Regional Unemployment Structural characteristics
6. Croatian regions with lower economic development experience higher unemployment. The high-unemployment eastern and central regions have a GDP per capita that is about 40 percent lower than in the north and north-western regions. The differences in the unemployment rates across regions among the lower-skilled workers generally tracked closely the developments in the regional unemployment rates during 2000-2014.
| Region | Unemployment rate(%), 2014 | Unemployment rate among the low-skilled (%), 2014 | Value added per capita (Kuna) |
|---|---|---|---|
| East | 36.0 | 57.7 | 43,272 |
| Central | 29.0 | 52.4 | 50,443 |
| West | 24.6 | 40.7 | 54,500 |
| North | 20.0 | 35.9 | 50,978 |
| North-West | 15.4 | 28.6 | 82,239 |
| City of Zagreb | 11.2 | 24.4 | 116,786 |
| Average | 22.7 | 40.0 | 66,370 |
| Region | Unemployment rate(%), 2014 | Unemployment rate among the low-skilled (%), 2014 | Value added per capita (Kuna) |
|---|---|---|---|
| East | 36.0 | 57.7 | 43,272 |
| Central | 29.0 | 52.4 | 50,443 |
| West | 24.6 | 40.7 | 54,500 |
| North | 20.0 | 35.9 | 50,978 |
| North-West | 15.4 | 28.6 | 82,239 |
| City of Zagreb | 11.2 | 24.4 | 116,786 |
| Average | 22.7 | 40.0 | 66,370 |
7. Persistent differences in unemployment rates across regions can be attributed to several structural characteristics of regional economies: infrastructure development, economic distance to EU markets, and share of export-oriented manufacturing sector in regional economies. For example, the northern and north-western regions have strong industrial base and enjoy close economic proximity to the EU, whereas the central
and eastern regions are not well-connected to the EU markets. Moreover, the underlying production structure of the western coastal regions is heavily tilted toward services (especially tourism).
Labor supply and reservation wage
8. The implicit tax on returning to work from unemployment—the unemployment trap3—seems to be too high in many counties. This makes work a financially less attractive option relative to receiving welfare benefits. For example, for a family of four with one wage earner (67 percent of the average wage), a non-working spouse, and two children, the implicit marginal effective tax rate on returning to work varies between 85 and more than 100 percent.
Unemployment Trap—Implicit Tax on Returning to Employment, 2014
(percent of gross earnings)
Citation: IMF Staff Country Reports 2016, 188; 10.5089/9781498321587.002.A003
Note: Staff calculations based on OECD methodology for one earner household consisting of a married couple with 2 children younger than 14 earning 67 percent of average county gross wage.Pension contributions (20 % of gross wage); Guaranteed Minimum Benefit (GMB)–1600 Kuna; Unemployment benefits are estimated as an average monthly payment during the first year of unemployment; electricy subsidy–200 Kuna.Unemployment Trap—Implicit Tax on Returning to Employment, 2014
(percent of gross earnings)
Citation: IMF Staff Country Reports 2016, 188; 10.5089/9781498321587.002.A003
Note: Staff calculations based on OECD methodology for one earner household consisting of a married couple with 2 children younger than 14 earning 67 percent of average county gross wage.Pension contributions (20 % of gross wage); Guaranteed Minimum Benefit (GMB)–1600 Kuna; Unemployment benefits are estimated as an average monthly payment during the first year of unemployment; electricy subsidy–200 Kuna.Unemployment Trap—Implicit Tax on Returning to Employment, 2014
(percent of gross earnings)
Citation: IMF Staff Country Reports 2016, 188; 10.5089/9781498321587.002.A003
Note: Staff calculations based on OECD methodology for one earner household consisting of a married couple with 2 children younger than 14 earning 67 percent of average county gross wage.Pension contributions (20 % of gross wage); Guaranteed Minimum Benefit (GMB)–1600 Kuna; Unemployment benefits are estimated as an average monthly payment during the first year of unemployment; electricy subsidy–200 Kuna.9. Staff estimates that a higher marginal effective tax rate leads to an increase in the unemployment rate in the labor force with less than tertiary education. For example, a one percent increase in the unemployment trap is associated with an increase in unemployment of more than 2 percentage points. The results are robust to the inclusion of controls for GDP per capita, productivity, and minimum wage.
Regression Analysis: Poverty trap and Lower Skilled Unemployment 1,2
p-values in parentheses (*** p<0.01, ** p<0.05, * p<0.1)
The analysis uses OLS estimation for regressions (1) and (3), and robust regression estimation for regression (2). Regression (3) excludes four outliers (the City of Zagreb, Medimurje, Krapina-Zagorje, and Varaždin).
Regression Analysis: Poverty trap and Lower Skilled Unemployment 1,2
| Dependent variable: unemployment rate in the labor force with less than tertiary education. | |||
|---|---|---|---|
| (1) | (2) | (3) | |
| Poverty trap—the implicit tax on retuning to work for unemployed persons, (%) | 2.75** (0.012) | 2.90*** (0.007) | 2.08*** (0.009) |
| Minimum-to-average wage ratio (%) | −3.22** (0.011) | −3.07** (0.012) | −1.22 (0.216) |
| Value added Per Capita (thousand Kuna) | −0.320** (0.030) | −0.41** (0.018) | −0.30** (0.019) |
| Constant | −72.20 (0.239) | −88.62 (0.138) | −95.05** (0.023) |
| Sample: 21 counties, 2014 | 21 | 20 | 17 |
| R2 | 0.64 | 0.63 | 0.85 |
| Adjusted R2 | 0.57 | 0.56 | 0.82 |
| Degrees of freedom | 17 | 16 | 13 |
| RMSE | 6.270 | 6.010 | 3.677 |
p-values in parentheses (*** p<0.01, ** p<0.05, * p<0.1)
The analysis uses OLS estimation for regressions (1) and (3), and robust regression estimation for regression (2). Regression (3) excludes four outliers (the City of Zagreb, Medimurje, Krapina-Zagorje, and Varaždin).
Regression Analysis: Poverty trap and Lower Skilled Unemployment 1,2
| Dependent variable: unemployment rate in the labor force with less than tertiary education. | |||
|---|---|---|---|
| (1) | (2) | (3) | |
| Poverty trap—the implicit tax on retuning to work for unemployed persons, (%) | 2.75** (0.012) | 2.90*** (0.007) | 2.08*** (0.009) |
| Minimum-to-average wage ratio (%) | −3.22** (0.011) | −3.07** (0.012) | −1.22 (0.216) |
| Value added Per Capita (thousand Kuna) | −0.320** (0.030) | −0.41** (0.018) | −0.30** (0.019) |
| Constant | −72.20 (0.239) | −88.62 (0.138) | −95.05** (0.023) |
| Sample: 21 counties, 2014 | 21 | 20 | 17 |
| R2 | 0.64 | 0.63 | 0.85 |
| Adjusted R2 | 0.57 | 0.56 | 0.82 |
| Degrees of freedom | 17 | 16 | 13 |
| RMSE | 6.270 | 6.010 | 3.677 |
p-values in parentheses (*** p<0.01, ** p<0.05, * p<0.1)
The analysis uses OLS estimation for regressions (1) and (3), and robust regression estimation for regression (2). Regression (3) excludes four outliers (the City of Zagreb, Medimurje, Krapina-Zagorje, and Varaždin).
The Minimum Wage
10. The national minimum wage is high relative to peer countries and for Croatia’s less developed counties. When compared to peers, Croatia’s minimum wage of around €400 a month is higher than that of peers both in absolute terms and relative to income per capita. The high earnings disparity among Croatia’s counties translates into large relative differences when the national minimum wage is measured in percent of the average wage in each county.
For example, in Istria, which is one of the most developed and has the lowest unemployment, the minimum wage is only about 39 percent of the average wage. In contrast, the minimum wage exceeds 45 percent of the average wage in the eastern counties, where the unemployment rate is around 35 percent and almost 60 percent among low-skilled persons for which the minimum wage would be more binding.
11. The regression results suggest that the effect of the higher minimum wage on lower-skilled unemployment is mixed. The estimates of regressions 1 and 2 imply an association between higher minimum wages and lower unemployment. This is probably due to the high disincentives to work (unemployment trap), which leads to a high reservation wage as explained above. Higher wages would be needed under these circumstances to provide sufficient incentives to seek employment. However, the demand for lower-skilled labor would be constrained by higher wages, especially in counties with lower productivity. When the three northern counties with large export-oriented manufacturing sector and high productivity are excluded (regression 3), the results become statistically insignificant. As regards to the effects of the minimum wage on unemployment disparity among Croatia’s counties, the econometric results suggest that a higher ratio of the minimum-to-the average wage is associated with an increase in low-skilled unemployment relative to Zagreb. For example, an increase in the minimum-to-average ratio by 10 percent is associated with an average 1.2 percentage point rise in the unemployment rate among low-skilled persons in a region with 25 percent low-skilled unemployment rate.4
Regression Analysis: Minimum Wage and Low Skilled Unemployment
Regression Analysis: Minimum Wage and Low Skilled Unemployment
| Dependent variable: Δ log of unemployment rate | OLS | s.e. | |
|---|---|---|---|
| Δ log of minimum-to-average wage ratio | 0.46 | 0.26 | * |
| Δ log of unemployment rate in Zagreb | 0.44 | 0.09 | *** |
| Constant term | 0.02 | 0.01 | ** |
| R2 Sample: 20 counties, 2008-2014 | 0.19 |
Regression Analysis: Minimum Wage and Low Skilled Unemployment
| Dependent variable: Δ log of unemployment rate | OLS | s.e. | |
|---|---|---|---|
| Δ log of minimum-to-average wage ratio | 0.46 | 0.26 | * |
| Δ log of unemployment rate in Zagreb | 0.44 | 0.09 | *** |
| Constant term | 0.02 | 0.01 | ** |
| R2 Sample: 20 counties, 2008-2014 | 0.19 |
12. The results above highlight the need to rationalize social benefits in order to reduce the “unemployment trap” and reservation wage and to advance structural reforms to support employment creation. Furthermore, the relatively high unit labor cost in Croatia points to the need to maintain competitiveness by ensuring that wages do not move out of line with productivity gains.
Reference
Carone G., H. Immervoll, D. Paturot and Salomäki A. (2004): “Indicators of Unemployment and Low-Wage Traps: Marginal Effective Tax Rates on Employment Incomes”, OECD Social, Employment and Migration Working Papers.
Eurostat (2015), Reference Metadata in Euro SDMX Metadata Structure (ESMS).
International Monetary Fund, 2010, Slovak Republic: Staff Report for the 2010 Article IV Consultation, International Monetary Fund, Washington DC.
OECD. (2016), Taxing Wages 2016, OECD Publishing, Paris.
Stubbs, Paul; Sinsia Zrinscak, 2015, “ESPN Thematic Report on minimum income schemes, Croatia”, European Commission.
Prepared by Murad Omoev.
Skill levels are defined according to educational attainment: (i) low-skilled persons are those with less than lower secondary education; (ii) medium-skilled persons are those with upper secondary and post-secondary non-tertiary education, and (iii) high skilled persons are those with tertiary education.
The unemployment trap–or the implicit tax on returning to work for unemployed persons—measures the part of the additional gross wage that is taxed away in the form of increased taxes and withdrawn benefits such as unemployment benefits, social assistance, housing benefits when a person returns to work from unemployment.
The results are stronger if the outliers for northern counties of Međimurje, Varaždin, and Krapina-Zagorje are excluded from the sample. These results suggest that an increase in the minimum-to-average ratio by 10 percent is associated with a 1.65 percentage point rise in the unemployment rate among low skilled workers in a region with 25 percent low-skilled unemployment rate.
