We analyze the extent and causes of the low Roma formal employment rate in Hungary between 1993 and 2007. It is also one of the few countries with reliable research data on Roma. Historical evidence on the well-being of Roma communities and their relationship to mainstream societies is relatively sparse.
The communist regimes hastened the dissolution of the Roma communities and set in motion a paternalistic assimilation process. At the same time, many of the ties within the Roma communities were destroyed. For the first and last time in the history of the HLFS, it also contained ethnic markers.
6 Using this definition, 8 percent of students in the sample have one or two parents identified as.
ROMA EMPLOYMENT IN HUNGARY
As a result, the employment gap refers to the Roma employment rate minus the national employment rate. The gap between the Roma employment rate and the national employment rate was 36-37 percentage points in 1994. The vast majority of jobs filled by Roma were unskilled jobs that were no longer productive in a market economy and were destroyed in the first years of transition.
Using data from parents in the 2007 Hungarian Life Course Survey (HLCS), we can take a closer look at the characteristics of the current labor market participation of Roma and non-Roma in a directly comparable way. Please note that the data is not representative of the population, both because of the age range (limited to persons between 30 and 59 years old, in the case of Table 2) and because of the over-representation of adults living with children. The majority of the non-working and non-job-seeking male population has a disability pension in both ethnic groups, but this is more pronounced among the Roma.
Similarly, most unemployed and non-job seeking women are at home with children in both ethnic groups, but more so among Roma.
DECOMPOSING THE EMPLOYMENT GAP
Part of the age difference between the population considered for the employment analysis is because Roma leave school earlier. Roma are significantly behind the majority ethnic group in terms of demographic transition (Hablicsek, 2002). The first term in the decomposition, nt'xt, measures the difference due to the different compositions of the two samples.
It measures the difference that is the result of the different composition of the Roma and non-Roma (or in 1994 and 2003 national) sample. The second term in the decomposition, nt' xrt, represents the part of the employment gap that is not due to differences in composition in terms of the variables on the right-hand side. Technically, the second term shows the part of the employment gap that results from the fact that the regression coefficients, including the constant, are different.
In all likelihood, quality-adjusted educational differences would account for an even larger part of the employment gap. Geographic differences explain a smaller part of the employment gap when other factors are taken into account. Compositional differences explain less of the employment gap between men than women due to the role of children.
The second robustness test involved restricting the Roma and the non-Roma subsamples to the general support of the right-hand side variables. It appears that, strictly speaking, the support is common in each survey year; all variables on the right are consistently nonmissing for both the Roma and the non-Roma subsamples. We then calculated the 5th and the 95th percentile of the propensity score separately in the Roma and the non-Roma subsamples.
Most of the reduction is related to a reduction in the composition term, which again is not surprising. However, the relative importance of the right-hand side variables within the composition term is unchanged.
WAGE DIFFERENTIALS
The Oaxaca-Blinder decompositions are based on linear regression models that use non-Roma coefficients for the composition terms (and Roma means for the coefficient terms). The standard errors in parentheses are based on heteroskedasticity-robust covariance matrix estimates of the coefficients. The composition term parameter estimates are all significant at the one percent level.
The age differences are slightly negative, reflecting the fact that the Roma are younger and that younger cohorts tend to earn less. Differences in the educational composition of the Roma versus the non-Roma workforce are again the most important elements of the composition concept; they account for a third to a half of the total difference. Differences in geography appear to play a somewhat more important role than in the case of the employment gap, but they are still much less important than education.
TRENDS IN THE ETHNIC DIFFERENCES IN EDUCATION
Figure 3, reprinted from Kertesi and Kézdi (2010), shows the proportion of the 2006 cohort of eighth graders in vocational schools and high schools over the survey years. The figure shows the proportion of cohorts in vocational or upper secondary school, even though they attended lower grades due to grade retention. Fraction of the cohort of eighth graders in 2006 in high school and vocational school in four years after graduation from eighth grade (in percent).
According to national figures, 71 percent of students are in high school in the year they should be in 12th grade, and another 21 percent are in vocational school. In the year when they should be in the 12th grade, 28 percent of Roma students are in high school and 33 percent in vocational schools. Overall, more than 85 percent of high school students are in the twelfth grade four years after the eighth grade.
These figures allow for a simple back-of-the-envelope calculation of expected vocational and high school completion. Assume that everyone in the twelfth grade will graduate from high school or vocational school, but only half of the grade-retainers will do so. Then, of those represented in the FDH sample, we can expect 66 percent overall to complete secondary school, but a rate of only 23 percent among Roma.
Taking all this together, we can expect the secondary school completion rate to be 63 percent nationally and 21 percent among Roma. Vocational school completion rates are expected to be 19 percent nationally and 24 percent among Roma. With the 11 percent Roma fraction in the group, we can expect the non-Roma secondary school completion rate to be close to 70 percent, the vocational school completion rate to be 18 percent, and the share of those with no to be 12 percent.
These figures indicate a significant increase in secondary and vocational school completion among the Roma in the most recent years. Yet more than half of the Roma population has less than an 8th grade education.
CONCLUSIONS
However, remember that not all members of the relevant birth cohort are represented by the HLCS, because not all of them reach eighth grade. Although more than 95 percent of a birth cohort completes eighth grade, the same figure among the Roma is around 90 percent (see footnote 6 above). The expected fraction of the birth cohort with neither secondary nor vocational degrees is 18 percent at the national level and 55 percent among the Roma (the corresponding figures for the 1980 birth cohort were 20 and 80 percent, respectively; . see Figure 2).
As a result, we can expect some improvement in the relative employment prospects of Roma in Hungary for the youngest groups, but the improvement is likely to remain modest for the overall Roma population. The increase in the employment gap is the result of the increased role of education in employment opportunities, which harms the Roma due to their low level of education. We have shown that the difference in education will probably still be large, as more than half of the Roma have fallen behind due to the increased participation of Roma in secondary and vocational schools.
As a result, we can expect some improvement in the relative employment opportunities of Roma in Hungary for the youngest cohorts, but this improvement is likely to remain modest for the entire Roma population. In the long term, the aim is to prevent the employment gap from reoccurring for future generations by reducing the education gap. An Analysis of the Dynamics of Self-Employment,” Journal of Labor Economics An Extension of the Blinder-Oaxaca Decomposition Technique to Logit and Probit Models” IZA Discussion Paper no.
Hablicsek László (2004), "Az idősödő népesség demográfiája Magyarországon." Munkaanyag, KSH Demográfiai Kutatóintézet. Kertesi Gábor (2005), "Roma foglalkoztatás az ezredfordulón." [The Employment of the Roma in End of the 20th Century.], Szociológiai Szemle, 16. Kertesi, Gábor, és Gábor Kézdi (2007), "Children of the Post-communist Transition: Age at the Time of the Parents' Job Veszteség és korai iskolaelhagyás” – a B.E.
Ladányi, János, and Iván Szelényi (2001), “The Social Construction of Roma Ethnicity in Bulgaria, Romania, and Hungary During Market Transition.” Review of Sociology, 7(2), pp. The estimate is limited to common support (by excluding all those below the 5th percentile of either the Roma or non-Roma subsample or above the 95th percentile of either the Roma or non-Roma subsample). The propensity score is estimated as a probit with Roma on the left and variables on the right of the decomposition regressions.
Detailed results of linear probability models for employment estimated on the subsample whose propensity score is within [5, 95].
BWP 2009/07
BWP 2009/08 2010
BWP 2010/02
BWP 2010/06
