Économie et Statistique n° 357-358 Employment - Social Transfers - Pensions - Industry - Environmental Goods Evaluation
The employment effects of the 11 June 1996 law on the shorter working week
Many econometric studies have endeavoured to evaluate the prospective impact of shorter working week pol-icies and to explain the conditions for the success of these policies. However, little ex-post evaluation infor-mation is available. This second path is chosen to evaluate the employment effects of the 11 June 1996 law (the «Robien law») on the shorter working week. This is done by comparing available data on establish-ments that have implemented this process with other data sources (ACEMO, UNEDIC and DIANE). The characteristics differentiating establishments that have signed an offensive agreement (in order to hire new workers) under the 11 June 1996 law from the others vary depending on whether these establishments are compared with establishments that had not yet reduced their working week in September 2001 (first group of comparison) or with establishments that only reduced their working week after January 2000 (second group of comparison). In the first case, the important elements are size, previous growth in staff numbers and the cost of labour. In the second case, the differences have more to do with the businesses' economic and financial health. These differences reveal the existence of selection by the establishments and enterprises when they introduce the shorter working week. Given the significant differences between the establish-ments that reduced their working week under the 11 June 1996 law and those that had not yet reduced their working week in September 2001, only the estimate of the mechanism's employment effect calculated for the second group of comparison is retained. We hence show that, during the period of the mechanism's introduction, growth in staff numbers in establishments having reduced their working week under the 11 June 1996 law is significantly higher than the growth in the other establishments, even after taking the selection bias into account.