The existence of a peak at 49 employees in the firm size distribution in France, followed by a permanent decrease in the number of firms has been the starting point of political discourses and academic studies on the cost of size-dependent regulations at 50-employee. These features of the distribution are visible when firm size is declared by employers in fiscal data but not when it is reconstructed from individual-level social security data. This working paper explores these differences both from statistical and institutional viewpoints. It provides evidence showing that a large proportion of employers manipulate the firm size they declare in their fiscal documents. This manipulation generates the particular shape of the size distribution in the fiscal data. We discuss the rationale for such behavior: the key point is that the under-declaration in fiscal data is not subject to substantial sanctions and it can allow firms not to comply with the labor law. Event studies and comparisons of firms below and above the 50-employee threshold suggest that this threshold may only have limited effects on firm performance or growth potential. Consequently the welfare costs of the regulations at 50-employee might be smaller than what was found by some of the studies that assume a perfect compliance with the law.