Employment effects of tax cuts in a transition country: evidence from Serbia

Abstract

During the years preceding the current crisis Serbia had relatively high GDP growth rates. It seems, however, that these growth rates did not have much impact on the employment growth rate, which has been rather low and even negative since 2000. Given that there is quite a high tax wedge in the country, we analyse the impact of labour tax cuts through a reduction in social security contributions on the employment rate for workers of different skill types. Results show that tax shifting is higher, that is, wages will increase more as a result of reduced social contributions for high-skilled than for low-skilled workers. From a policy perspective our results indicate that it could be more effective to focus on selective tax reductions instead of applying across-the-board tax cuts.

Acknowledgements

I am deeply grateful to Mihail Arandarenko for valuable comments and continuous support. Sincere thanks to Radmila Dragutinovic-Mitrovic for comments while working with the data and estimating the model. I am also thankful to Lara Lebedinski, Suncica Vujic and Marija Djordjevic for helpful suggestions. I finally thank the Statistical Office of the Republic of Serbia and National Bank of Serbia for providing data.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

^1. The tax wedge measures the difference between labour costs and take-home pay of workers. It expresses the costs of social security contributions by employers and employees and the personal income tax of employees as a share of total labour costs. (Figure A1 in the Appendix, available from the author on request).

^2. This is a peculiar feature of social security contribution systems in the Western Balkan region. The most drastic example is Macedonia, in which the mandatory base is set as high as 65% of the average wage (Mojsoska-Blazevski 2011)

^3. The instruments are different lags of higher degree of the same variable.

^4. The system GMM estimator improves the precision of estimates of regression parameters and reduces finite sample bias.

^5. Since 2000 all governments have been reluctant to privatise socially owned firms, fearing job losses could cause social discontent.

^6. According to Hamermesh (1993, p. 248) the median length of the adjustment lag is equal to ln(1/2)/ln(β₁).

^7. All reported elasticities are the absolute value.