Comment

Abstract

Previous article FreeCommentChristopher A. PissaridesChristopher A. PissaridesLondon School of Economics Search for more articles by this author PDFPDF PLUSFull Text Add to favoritesDownload CitationTrack CitationsPermissionsReprints Share onFacebookTwitterLinked InRedditEmailQR Code SectionsMoreMuch effort in modern labor economics is devoted to finding “natural experiments,” which can be peculiarities in legislation that give rise to different treatments of otherwise similar subjects. When the interest is labor market policy, in Europe usually the unit of analysis is the nation. Countries are small and policies are national, but there are still significant policy differences across countries. Cross‐country analyses, usually with panels of data going back to the 1970s, have shed important light on the impact of labor market policies on the economy. But they have been plagued by the fact that despite the convergence that is taking place in the context of the European Union, there are still large differences in many other dimensions across the union. Identifying the impact of policy from that of other national characteristics has been difficult.Chemin and Wasmer’s paper identifies a peculiarity in the coverage of legislation in France across its regions that enables a more reliable analysis of the impact of policy, on the assumption that other regional differences within France are less important than, say, differences between France and Germany. An area of France, Alsace‐Moselle, was under German jurisdiction between 1870 and 1918, and many laws that were brought in at that time are still in force. Subsequent changes in French legislation sometimes applied differently in Alsace‐Moselle than in the rest of France, and so one can do a “difference‐in‐difference” comparison of the response to the legislation change in Alsace‐Moselle with the response in the rest of France. Given that the reason that policy coverage is different in Alsace‐Moselle than in the rest of France is exogenous, this would be a valid comparison.Of course, a simple comparison without any correction requires that Alsace‐Moselle be identical in all other respects to the rest of France. This, however, is not likely to be the case. For example, they are the only regions in France that share a border with Germany, and this alone introduces differences since proximity to the German border increases their exposure to German shocks. A thorough study of the impact of the differences in the laws of Alsace‐Moselle and the rest of France will need to control for the other differences between the two regions and the rest of the country. The authors do correct for permanent differences between the regions and the rest of France and find some differences in the policy impacts that have interesting implications for more general policy impacts.The first difference in policy coverage is income support. Alsace‐Moselle had in operation a system of support for low incomes since its German days (presumably associated with the Bismarck welfare support programs), but the rest of France introduced it only in 1989. The support amounts to bringing up low incomes to some level below the minimum wage, so it applies to unemployed workers. The question is whether this support discourages job search and increases unemployment durations.The authors compare the responses of long‐term unemployment to this change in 1989 in Alsace‐Moselle and the rest of France. They find a larger rise in long‐term unemployment in the rest of France than in Alsace‐Moselle across the policy change, and they attribute the difference to the impact of policy. They conclude that the disincentive effects of the policy are strong.There is a large amount of evidence that unemployment compensation reduces search incentives and increases unemployment. Reasonable panel estimates with cross‐country data put the impact at about 1.1 percentage points of unemployment for a 10‐percentage‐point difference in unemployment compensation. However, there is virtually no evidence about income support programs associated with low income. Is this policy change equivalent—in terms of coverage, administration, and so forth—to unemployment compensation or is there something more to it? A comparison with what we know about the impact of unemployment compensation would have been useful. The main impact studied in the paper is the one that obtains after unemployment compensation is exhausted. This is a valid restriction only if at lower durations unemployment compensation brings incomes above the level guaranteed by the new policy, a point not investigated in the paper.A second welfare policy reform in France not affecting Alsace‐Moselle was the introduction in 1978 of more generous sick leave for employees with more than 3 years' tenure. Alsace‐Moselle already had a more generous sick leave system in operation for all employees, so the new policy did not change coverage in it. Difference‐in‐difference comparisons between Alsace‐Moselle and the rest of France across the policy change show a substantial increase in absenteeism among French workers with more than 3 years' tenure but no change for those with lower tenures. Although the computed coefficients are not used in the paper to calculate days lost by the introduction of the new policy, they seem rather high. They are of about the same order of magnitude as the impact of the first policy change on the incidence of long‐term unemployment.Another impact of the sick leave policy change is on labor turnover. Again using the same methodology, the authors find that the policy reduced labor turnover in the rest of France when compared with Alsace‐Moselle. The entitlement to more generous sick leave kicked in after 3 years' tenure in the same job, so a quitting employee loses all her accumulated entitlement. Making the entitlement transferable would alleviate this problem, which could have other negative consequences for the performance of the labor market.Finally, the authors explore the implications of a third “anomaly” in the comparison between Alsace‐Moselle and the rest of France. Traditionally, Alsace‐Moselle had two extra days of holiday a year. As 2 days amount to 16 hours of work, following the introduction of the shorter workweek in France in 2000, employers in Alsace‐Moselle could satisfy the law by reducing annual hours by 16 hours less than employers in the rest of France. This privilege, however, lasted only for 2 years, as the courts ruled otherwise in 2002. The authors investigate the impact of the smaller reduction in annual hours in Alsace‐Moselle on unemployment for these 2 years by comparing unemployment before and after the policy change in 2000 in the two regions.Here the authors find a rather large impact of the reform but one that is not plausible. They find that unemployment in Alsace‐Moselle increased by more than in the rest of France, and if this is interpreted as the impact of the reduction in hours of work, it leads to the conclusion that the reduction in hours reduced unemployment by a large amount. The authors reject this interpretation and attribute the bigger “recession” in Alsace‐Moselle in 2000–2002 to a deeper recession in Germany than in France affecting the region. This part of the research, however, is not convincing, and the impact of the reduction in hours of work on unemployment remains an open question.It seems to me that the difference in the reduction of hours—about 16 hours per year for 2 years—is too small to have an impact on employment in a country that has strict employment protection legislation. Especially during a recession, employers are likely to wait before changing employment when the law changes the maximum hours of work. Given the small difference in hours between Alsace‐Moselle and the rest of France, in both magnitude and duration, one should not expect to see an immediate impact on employment that could be picked up in a difference‐in‐difference regression. This concurs with the authors’ preferred interpretation, that the policy change had no impact on unemployment, but it is not compelling to generalize from the failure to pick up any effect here to the conclusion that the change in hours of work had no impact nationwide. The experiment here seems to be too small and too short lived to enable a substantial impact on employment, even if one took place nationwide.If one looks at employment and hours trends in France generally, there is clearly something unusual about them that needs explanation. Figure 1 compares data for France and Italy since 1970. The trend in their employment rates is about the same, but weekly hours of work in the working‐age population (not just working population) fell by much more in France than in Italy. In the comparison of hours per employee (not shown), among the main European economies, France and Germany experienced similar large reductions that exceed reductions elsewhere, but French employment fell by more than in the other countries. It seems that France has succeeded in reducing weekly hours, although not necessarily by legislation, since the dip after 2000 in figure 1 is small compared with the overall trend before then. But it failed to attract more women into the labor force or reduce unemployment by as much as the other European countries of similar size have done. The overall trends in hours of work and in employment do not justify optimism that a reduction in weekly hours over and above the reduction in other countries will be accompanied by a larger increase in employment than in those other countries.Fig. 1. Aggregate hours of work and employment in France and Italy, 1970–2005View Large ImageDownload PowerPoint Previous article DetailsFiguresReferencesCited by Volume 5, Number 12009 Article DOIhttps://doi.org/10.1086/596014 Views: 97Total views on this site © 2009 by the National Bureau of Economic Research. All rights reserved.PDF download Crossref reports no articles citing this article.