Even though human rights clauses are increasingly included in preferential trade agreements (PTAs), not all PTAs contain social law. What accounts for this variation? We still know relatively little about what explains the inclusion of human rights in trade agreements. Seeking to fill this gap, this paper proposes a theoretical argument that stresses the selection process. The argument is that stakeholders - such as multinational firms, voters, NGOs and governments - respect and comply with clauses, which are not in their favor, in order to receive the benefits of other provisions included in the PTA. Consequently, the selection of human rights is interdependent with other design features of the agreement. In particular, it is expected that deep agreements are more likely to include social law. A time-series-dataset on the design of PTAs signed between 1957 and 2009 allows testing the arguments. Descriptive evidence and probit regressions support the theoretical expectations.
