How to explain policy outcomes that seemingly contradict the logic of the institutional environment they are embedded in? The residual welfare states of Estonia and Latvia boast the most generous unemployment insurance schemes in Central and Eastern Europe, whereas two of the most generous post-Communist welfare states, Poland and Hungary have the most residual unemployment insurance schemes among post-Communist EU member states. The article is a contribution to historical institutionalist explanations of institutional change by showing the process of how in a relatively short period, political decisions made at a critical juncture in history can lock in distinctive policy pathways in the long run. It is also a contribution to the literature on Central and Eastern European welfare state development by opening the black box of the "muddling through" argument and identifying the causes of seemingly contradictory outcomes.
The analysis identifies the cause for this outcome in the first years of transitions when decision makers in the two groups of countries choosing different policies to alleviate the cost of transition and pacify resistance to economic reforms, and in the subsequent feedback mechanisms. Hungary and Poland opted for mass exit from the labour market into early and disability retirement for redundant workers as the main tool of alleviating the deep social costs of the transition. In Estonia and Latvia in the same period, however, pensioners were deliberately pauperised and pension spending was effectively cut. These initial conditions presented governments with very different opportunities for subsequent social policy development in the two group of countries: in Poland and Hungary social and labour market policy development has been constrained by the costly fiscal and political burden of these initial pensioner booms ever since, whereas the Baltic states could follow policies to support their flexible labour markets with setting up systems along the "flexicurity" principle of short duration and relatively high replacement rates.
The analysis is based on a new, yet unpublished policy database covering policy change in 12 social, labour market and economic policy fields in 37 democracies, on a time-frame of 40 years, containing more than 13.000 policy reforms. Thus, this unique database makes the detailed comparison of policy outcomes across countries and policy fields possible, and allows for the exploration of how policy change in one domain is contingent upon change in others.
