Evaluating the effectiveness of employment programs: can existing statistical methods be applied to the italian labour market and institutional context?

Abstract

In the first part of the paper we provide a brief summary of the statistical approach that has emerged as the paradigm for evaluating employment programs in the United States. We discuss the role that longitudinal data and methods play in this approach. In the second part of the paper we take on a more challenging task, motivated by the view that the appropriate statistical methods used for evaluation purposes should be tailored to the institutional context in which the program operates. We discuss three directions in which the U.S. evaluation paradigm should be developed if it has to be applied to a labour market context such as the one prevailing in Italy: (I) shifting the attention away from repeatable outcomes, such as earnings or repeated spells of unemployment, which characterize the U.S. approach and provide the basis for the use of longitudinal methods: because of the characteristics of the Italian labour market, very few meaningful repeatable outcomes can be found to evaluate the success of employment programs; (II) broadening the range of programs to be evaluated, with respect to the U.S. practice of focusing evaluations on training programs targeted to disadvantaged individuals; (III) broadening the objectives of the evaluation, from an emphasis on absolute effectiveness, to other types of evaluation questions, such as the effectiveness of service providers, and the issue of the adverse effects of interventions.