Replication and reconciling of the CICE evaluations on employment (2013-2014)

We reproduce the main estimates of the effects of the CICE (Tax Credit for Competitiveness and Employment) on employment presented by the TEPP and LIEPP teams for France Stratégie based on the methodological indications described in their first research reports (september 2016 and march 2017). After the reconstitution of the samples, globally validated by comparisons in descriptive statistics, we reproduce their different empirical strategies. The effects on employment reproduced are consistent with the results presented by TEPP: positive, significant and of the same order of magnitude for the firms most exposed to the CICE (fourth quartile of exposure to the CICE). On the other hand, the effects on employment by socio-professional category do not reproduce the results of TEPP. For LIEPP, the replication of the methodology leads to results similar to those presented in their report on aggregate employment in 2013 and 2014 (no significant and positive effect), and as well as on the breakdown by socio-professional category. We then compare the specifications of the two teams. The differences in results persist on a common sample, and are robust to the linear or non-linear specification of the treatment. However, we observe a particular sensitivity of the results to the presence or absence of past productivity of firms, in level, among the explanatory variables of employment growth. We present two reconciled specifications that lead to consistent results, shedding light on origins of the divergence. We do not conclude on employment effects of the CICE, as the work of the two teams is still ongoing.