Conclusion
We have shown that, mostly for persons 35 years old and under, in both France and the United States,
movements in the real minimum wage are associated with significant employment effects, typically in the direction predicted by competitive labor market theory. In France, as the real SMIC increased over the period from 1981 to 1989, a certain share of young French workers had real wages that fell between the increasing consecutive real minimum wages. For workers in this situation, subsequent employment probabilities fell significantly. However, participation in employment promotion programs apparently shielded the youngest of these workers from some of the effects of the increasing real SMIC, and when this eligibility ended, the probability of subsequent nonemployment shot up dramatically.
In the United States, a comparable effect from the real minimum wage moving in the opposite direction occurred, as many workers had market opportunity wage rates that were passed by the declining real minimum wage over the period from 1981 to 1987. American workers whose current real wage rate would have been below the real minimum wage in earlier periods were much less likely to have been employed in those earlier periods.
By comparing effects of minimum wage movements on workers employed at the minimum with those employed marginally above it, we identify the direct effects of the minimum wage, as distinct from heterogeneity across the wage distribution in labor force attachment and response to macroeconomic shocks. We suppose that workers in these two groups have identical labor supply behavior. We find that those employed between two real minima have much lower subsequent employment probabilities in France and much lower prior employment probabilities in the U.S. Across the population as whole, however, our results also suggest that youth in both countries and women in the U.S. are most affected by movements in the real minimum wage.
Even if the conditional elasticities in question are large, the at-risk groups (workers between two minimum wages) are relatively small, specifically, 1% of adult men and 2% of adult women in France, 6.5% of adult men and 11.6% of adult women in the U.S. Thus, overall unconditional elasticities tend to be much lower than the elasticities conditional on being between
the two minima. If the relevant policy question concerns the impact of the minimum wage on those individuals most likely to be affected by it (i.e. those currently paid at the minimum wage), our results suggest that there are much larger negative employment effects on this group, especially as compared to the group in the wage distribution marginally above the minimum, than other research has found.
Our results, which are based on direct data evidence from households, contrast sharply with the results of Card and Krueger (1994, 1998), which are based on direct data evidence from establishments. Recently, Kramarz and Philippon (1998) have analyzed the French data for 1990 to 1997, focusing carefully on the effects of targeted payroll tax subsidies on the total labor cost of minimum wage and low-wage workers. Their results, for a period of analysis that contains intervals in which the total labor cost of minimum wage workers rises and falls, are essentially the same as the ones we find here.
A priori, there is no reason to prefer household analyses over those based on establishments;
however, the strong similarity between the French and American results in our paper present a stronger challenge to the Card and Krueger results than would be the case for an analysis of either country taken by itself.
