How much do legal size-contingent thresholds impact the size distribution of French firms?

Several papers have shown that France has a lower share of firms with more than 50 employees than other developed counties (see for example Van Ark and Monnikhof, 1996). It has been argued that size-contingent thresholds in labor regulations at 10, 20 and 50 employees could explain part of this phenomenon. For instance, firms with more than 50 employees have harder firing rules and must regularly discuss working security conditions with representatives. Those constraints can discourage SMEs from growing over the threshold of 50 employees. For this reason, the French government has decided (law of economic modernization of august 4th 2008) to allow firms passing the thresholds to delay complying with the rules. In this paper, we extend the work of Cahuc et Kramarz (2004), who have shown that there are indeed threshold effects near 10, 20 and 50 employees, but probably of small magnitude. We look at different databases and different concepts to measure the number of employees. We show that threshold effects are of very different magnitudes depending on the source. Using the data with the strongest threshold effects, we then estimate an upper bound of the global effect of size thresholds on firm size distribution. We assume that without size-contingent labor market regulation there would be a smooth relation between size and growth probabilities. We estimate a counterfactual transition matrix by smoothing the conditional growth probabilities around the thresholds and use it to compute the long run effect on firm size distribution. We find a statistically significant but small effect. If the thresholds were smoothed, the share of firms between 0 and 9 employees would decline by 0.4 percentage point, while the share between 10 and 19 and the share between 20 and 250 would both increase by 0.2 percentage point.