67

In 2007 and 2008, experiments were carried out on Active Solidarity Income (RSA) in 34 French départements. Each of these départements defined one or more test zones or experimental zones, where the new benefits were implemented. In total, 51 test zones were chosen by the départements. So as to evaluate the effects of the RSA, we built and implemented a methodology enabling the definition of control zones in each experimenting département. These control territories were selected in such a way as to provide the most satisfying counterfactual possible for the evaluation of the test zones, taking into account the inherent limits of local statistics on return to employment behaviour. The first step, for a given département, was to establish a list of possible control zones whose socio-demographic characteristics were comparable to those of the test zone, notably as regards size (population, number of RMI, or Minimum Income Allowance, beneficiaries), the density of the population and the administrative situation ( canton, CLI, UTAS, CAS, etc.). The second step was to classify all of these possible zones, from the closest to the test zones as regards the possibility to entry into a job for the RMI beneficiaries as estimated using the CNAF’s (National Family Allowance Fund) administrative data. To do this, every semester there is a reconstruction of the rate of entry into a job for the Minimum Income beneficiaries on the communal level from June 2002 to June 2007 whilst neutralising the effects of constituting the beneficiaries according to characteristics observable in the administrative files. In the third step, the control zones were proposed to the experimenting départements, which were then able to make counterproposals backed up with arguments in case of disagreement. In practice, the choice of this iterative procedure enabled a good compromise between the principle of national unity in the selection process of control zones and putting local informational systems and knowledge of the field to the best possible use.