Supported by most theories, employment protection often is found to reduce economic growth. Almost all existing empirical studies, however, are based on data from continental Europe and Japan, where labor protection is generous. Using data for the United States, where labor protection is minimal, we find, by contrast, positive effects but only in knowledge-intensive industries. To reconcile these facts, we propose a simple theoretical model based on a hold-up problem arising from firm-specific investment. This makes some job security efficient and the relationship between job security and growth an inverted U-shaped, i.e., basic labor protection increases growth, but generous protection reduces it. Importantly, we show that a firm faces a time inconsistency problem so that its promise of job security is not credible. Thus, legal restrictions become valuable if they are well designed. Since job security is even less for financially distressed firms, interactions also arise between financial and labor laws, as powerful banks can demand more layoffs. Using U.S. state-industry data, we confirm these effects of bank competition and employment protection, as well as their interactions.