We evaluate the effects of a central French reform that made the conditions for firm-level union recognition more democratic after 2008. The law gave equal chances to all unions to be recognized for bargaining, putting an end to the quasi-monopoly given to five historical unions until then. The law also introduced votes and minimal electoral requirements for union recognition. These new regulations only became effective at the first firms’ work councils elections following the promulgation of the law in August 2008. Those elections occur within each firm according to a pre-defined frequency– usually every four years–, so that election dates around August 2008 only depend on former election dates, and can be considered as quasi-random with respect to the new law, at least in firms that are old enough. The identification thus relies on a regression discontinuity design in which the running variable is the firms’ work councils election date: we compare in 2011 firms that had those elections just before or just after the law was passed. We find that the democratic rules introduced in 2008 increased social capital. Namely, both employers’ and workers’ satisfaction and trust towards unions are increased. Union coverage and membership are also increased. There is also limited evidence that the reform induced more labor disputes and less volontary quits, consistent with Hirschman’s Exit, Voice and Loyalty theory. Joint with Philippe Askenazy.