Asymmetric Effects of ISCED4 Graduates on Total and Youth Unemployment in Romania: Evidence from a NARDL Framework

The aim of this article is to analyse how changes in the number of post-secondary non-tertiary education (ISCED4) graduates influence the total unemployment rate and the youth unemployment rate in Romania over the period 1991-2023. Using a nonlinear autoregressive distributed lag (NARDL) framework, we decompose lnISCED4 into positive and negative partial sums to test for potential short-run and long-run asymmetries while controlling for EU accession and the global financial crisis. The NARDL results for total unemployment suggest a long-run relationship and significant short- and long-run asymmetry. However, with controls included, the long-run coefficients on lnISCED4+ and lnISCED4– are imprecise, implying that asymmetry mainly reflects short-run adjustment. A benchmark linear ARDL model provides stronger evidence of cointegration and points to a modest negative long-run association between lnISCED4 and total unemployment. For youth unemployment, the NARDL estimates show robust short-run asymmetry. Increases in ISCED4 graduates are followed by short-run increases in youth unemployment, whereas decreases are also associated with subsequent short-run increases, while long-run education-related coefficients remain statistically insignificant. Robustness checks using labour force normalised ISCED4 intensity measures indicate that long-run asymmetry and levels relationships are sensitive to scaling and a shorter 1996-2023 window. These findings underline that expansions in intermediate vocational supply may generate temporary youth labour market pressures when absorption and matching mechanisms are weak and may support policy-makers in designing complementary measures to improve school-to-work transitions.