The first part of the thesis addresses a programme evaluation problem: the estimation of the short- and long-term effects of the Swedish active labour market programmes on participants’ subsequent labour market outcomes, in particular individuale mployment probability and collection of unemployment benefits overtime.
Exploiting a unique and comprehensive new Swedish dataset with extensive information on more than 110,000 individuals followed for five years, semiparametric propensity score matching techniques are adapted to the Swedish institutional context to investigate the overall effectiveness of the ‘Swedish model’ of labour market policy in the context of the high unemployment atypically experienced by Sweden in the 1990s. The performance of the Swedish system is thus considered in its entirety, combining all the programmes into one and focusing on the interactions between the unemployment benefit system and the programme system. Subsequently, the relative performance of the six main types of programmes available to unemployed adults in the 1990s is analysed, both relative to one another and vis-à-vis more intense job search in open unemployment. The differential performance of labour market training, workplace introduction, work experience placement, relief work, trainee replacement and employment subsidies is investigated using a multiple-treatment extension of the propensity score matching method.
The second part of the thesis deals with the evaluation problem in the returns to education framework. Different non-experimental estimation methods to recover the effect of education on earnings – ordinary least squares, instrumental variables, control function and matching – are reviewed and contrasted in the context of alternative microeconometric models – single- and multiple-treatment, homogeneous-and heterogeneous-returns models. The methods are subsequently applied to high quality data – the British 1958 NCDS birth cohort – to estimate private returns to schooling and to illustrate the sensitivity of the different estimators to model specification and data availability.