Motivation: The importance of the architecture of relationships has been particularly recognized in labor markets, since friends and acquaintances are sources of employment information. The number of connections an agent has, as well as the way they are structured, may affect the information about vacancies that she receives, becoming a key determinant on her labor outcomes (Calvó-Armengol and Jackson, 2004, 2007; Granovetter, 2018).
Methods: We model explicitly the role of local density of links isolating the effect of clustering, so that this feature is not confounded with other network features. More precisely, we analyze the role of cycles, as distinct from the number of direct and indirect neighbors, which are kept constant.
Results: We show that two agents with different clustering will have a different probability distribution of offers coming from neighbors. Information flows do not only depend on the number of connections, but on the geometry created by these links. Thus, two networks where the players have the same number of links but with a different index of clustering may induce different employment levels and wage distributions. Cycles reduce the individual probability of employment. Short cycles also reduce expected wages, since the probability of being unemployed is higher. However, the expected wage of individuals employed is higher in a short cycle.
Discussion: The direct contribution of this paper is to shed light on the effect of cycles on labor market performance. Our results point to a negative relationship between short cycles and employment and expected earnings, and a positive effect on the average wages of those employed through contacts. This means that individuals immersed in loose-knit communities may enjoy information advantages even when the kind of information they receive is the same as in a highly cohesive network.