Typical street scene in Santa Ana, El Salvador. (Photo: iStock)
IMF Survey: Is the 35-Hour Workweek Really Bad for France?
March 22, 2007
Lesure time: enjoying a drink in a street cafe in Paris, France. (Photo: Newscom/ndx)
Dear IMF Survey:
The paper by Marcello Estevão and Filippa Sà, summarized in the IMF Survey article, "Are the French Happy with the 35-Hour Workweek?," is a welcome addition to the empirical literature on work-sharing because it bolsters our meager empirical evidence of the effect of such laws and regulations on employment.
Estevão and Sà use panel data with large sample sizes to analyze the response of firms and the labor market to a law introduced in France in 1998 to reduce the standard workweek from 40 hours to 35 hours. The law did not initially apply to small firms, allowing the authors to use an approach known among economists as a "differences-in-differences" strategy.
In using this approach, the authors had to decide how exactly to align their three-year overlapping panels with the timing of the law change. The new law was announced in 1998, implemented for medium-sized and large firms in February 2000, and for small firms in January 2002. The data panels the authors had access to covered the periods 1996-1998, 1997-1999, 1998-2000 and 1999-2001 (and presumably 2000-2002), with the labor surveys carried out in March of each year.
At first glance, the 1999-2001 panel may appear to be the optimal one to use to follow workers from before to after the law was introduced. However, because the law is announced in 1998, it is possible that firms began to adjust early and that 1999 does not appropriately capture the "before" period. It is apparently this concern that led the authors to use the 1998-2000 panel instead. This is fine if one assumes that employers were in compliance with the law one month after it was introduced. The authors then chose the 1997-1999 panel as the second dataset spanning the law change. This choice puts 1999 in the "after" period even though the new law had not yet come into effect and thus seems less judicious.
To make sure the results are not biased toward finding no effect, the authors would be well advised to conduct a sensitivity analysis replacing the 1997-1999 panel with the 1999-2001 panel. Only then will we know for sure whether the French were in fact worse off overall after the introduction of the 35-hour workweek.
Sincerely,
Jennifer Hunt
Department of Economics
McGill University