We provide a matching model where identical workers are embedded in ex-ante identical social networks. Job arrival rate is endogenous and wages are bargained. We study the evolution of networks over time and characterize the equilibrium distribution of unemployment rates across networks. We emphasize that wage dispersion arises endogenously as the consequence of the dynamics of networks, firms' strategies and wage bargaining. Moreover, contrary to a generally accepted idea, social networks do not necessary induce stickiness in unemployment dynamics. Our endogenous matching technology shows that the effects of networks on the dynamics mostly hinge on search externalities. Our endogenous framework allows us to quantify these effects.