Early indication of program performance: The case of a Swedish temporary employment program

This paper proposes an application of standard evaluation techniques to obtain an early indication of the efficiency of a social program. An explicit distinction is made between participation and treatment since non-participants in many cases can find treatment outside of the program. By estimating the amount of treatment participants would have received if there had been no program it is possible to derive the net treatment provided by the program as soon as the first participants have entered the program. Using propensity score matching to study substitutes in the recently implemented Swedish "Career Break" pilot program we obtain three main results: First, the selection into the program favours participants that would have received treatment even without the program. Second, the net treatment provided by the program is on average only about half of the expected program length. Third, a simulation shows how the net treatment could be increased by a simple change in the eligibility criteria.