By conventional statistics, youth unemployment seems to be quite moderate in Korea: ‘only’ 9.6 percent of the ‘active’ youth labour force was unemployed compared to 21.4 percent in EU-27 in 2011. Germany, with a youth unemployment rate of 8.5 percent, is one of the very few European countries outperforming Korea. But the Korean case is in one respect unusual. From the perspective of intergenerational risk sharing Korea’s youth unemployment rate is 4.6 times higher than the unemployment rate of adults aged 45 to 54; in Germany, this figure is only 1.7. Further peculiarities come up if unemployment is measured by the number of youth not in employment, education or training (NEET) in percent of the total youth population. Korea’s NEET figures are at the top in OECD countries, especially for youth with tertiary education. This paper throws some light to explain this conundrum: It sketches, first, the main causes of youth unemployment and the general policy interventions; because a large part of the problem is structural, possible immediate measures to avoid long-term scar effects for the unemployed youth are briefly reviewed; differences between Europe and the United States show in particular the importance of automatic stabilizers like unemployment insurance in order to reduce the pressure on unfavourable risk sharing for youth in times of recession. The main part is devoted to possible lessons for Korea from Europe, in particular from Germany. Dual education and vocational training systems that emphasise middle level and market oriented skills are identified as institutional device both for fairer intergenerational risk sharing as well as for a smoother transition from school to work. In its outlook, the paper comes back to the puzzle of highly and academically inflated youth unemployment by referring to a possible hidden cause in Korea: A strong insurance motive might explain the overall striving for an academic degree inducing not only wasteful congestion at labour market entries but also unfair job allocation through credentialism.