![Sample our Area Studies journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5922/TOC-Subject-Sample-2021-Area-Studies.jpg"})
Abstract
This article analyses the evolution of private returns to tertiary education during the period of transition from a socialist to a market economy using the personal income tax data of all Slovenian workers employed between 1994 and 2008. We document a rich interplay between supply and demand in the labour markets of high school and university graduates. We show that, in spite of significant increases in the labour supply, the demand for university graduates outweighed this and increased the rates of return in the early period of transition (1994–2001), while in the later period (2001–08) the opposite was the case. We also provide evidence of considerable heterogeneity in rates of return between genders, levels and fields of study, with particularly large (low) returns to the fields that were suppressed (favoured) during socialism. These initial differences in returns have, however, gradually declined.
Notes
1. See De Arriba Bueno (Citation2010) for description and analysis of the objectives of transition.
2. The literature on returns to different fields of study is also relatively scarce for established market economies. Authors find large and consistent differences in the rates of return for four-year college graduates (Arcidiacono Citation2004), MA programmes (Yoram Citation1973) and Canadian graduates (Finnie and Frenette Citation2003, Stark Citation2007). Non-American studies of rates of return to education by fields of study are similarly limited. The exceptions are Livanos and Pouliakas (Citation2008) studying the Greek labour market, Blundell et al. (Citation2000), O'Leary and Sloane (Citation2005) and Walker and Zhu (Citation2011) studying returns to higher education in Britain, Kelly et al. (Citation2010) studying returns in Ireland and Buonanno and Pozzoli (Citation2009) studying the Italian labour market.
3. To avoid the ability bias if using an ordinary least squares estimator without some measure of ability, researchers have used for example (i) some proxy of ability (Griliches Citation1977, Nordin Citation2008), (ii) an instrumental variables estimator (Angrist and Kruger Citation1991, Citation1992, Card Citation1995, Harmon and Walker Citation1995, Citation2000, Denny and Harmon Citation2000), or (iii) data on siblings or twins (Ashenfelter and Krueger Citation1994). For a more comprehensive review see Harmon et al. (Citation2000) and Meghir and Rivkin (Citation2011).
4. A person is unemployed if she or he is registered at the local employment office. This definition is less strict than the standard ILO definition. Hence Slovenian unemployment rates based on registry exceeded survey unemployment rates by as much as 6 percentage points.
5. The exchange rate in 2007 was EUR 1 = USD 1.37.
6. Note that the grading scale is changed every year so that the points achieved in each cohort are Gaussian distributed.
7. We also used non-normalised points but the results were qualitatively similar.
8. The total population of Slovenia is around two million; we exclude employees with less than four years of high school completed and sole proprietors from our analysis.
9. We estimate separately returns to levels of education and returns to fields of education. In the first case j is equal to a level of education (two-year UG, four-year UG, MSc/MA or PhD). In the second case j is equal to a degree in a specific field (e.g. four-year UG degree in education). The omitted category is always high school.
10. Other studies find the returns to schooling range between 2% and 13% (e.g. Harmon et al. Citation2000, Boarini and Strauss Citation2007, Citation2010). For example, Harmon et al. (Citation2000) estimate the returns for males in 1995 to be 8.9%, 7.8% and 13.0% for Slovenia, US and Great Britain. However, direct comparison of these returns with our results may be problematic owing to the differences in assumptions and estimation techniques.
11. The results for other levels of tertiary education are omitted for the sake of brevity. They are, however, consistent with the results presented.
12. The difference in regression coefficients for two-year UG is not statistically significant. (see Table A3; Appendix available from authors on request).