Referral Hiring and Labor Markets: a Computational Study

Abstract

Many empirical studies emphasize the role of social networks in job search. The social network implicated in this process is known to be characterized by diversified properties, including communities, homophily or ties having various strength levels. Nevertheless, previous models of the labor markets fail to capture the complexity of social networks, as each specific network requires the development of specific algorithms. In this paper, we rather rely on an independent generic network generator for creating detailed networks describing friendships, colleagues, communities and various degrees of connectivity. We build a simple model of the labor market in which individuals find positions solely through their acquaintances, and update their network when being hired. This original experimental setting facilitates the analysis of various characteristics of networks in the labor market, including various sizes, the number of friendship links or the impact of communities. Experiments confirm the ”strength of weak ties” phenomenon. However, the initial characteristics of the network like the existence of communities are shown to be destroyed by the implausible mechanisms integrated into this simplistic model; this suggests that the impact of plausible networks on models’ dynamics may only be studied when the mechanisms of this economic model are plausible as well - in other words, ”a model is only as descriptive as its most implausible component”. Addition of probabilities on ties allow to study close friends configuration and weak, or Facebook friends, configurations. Experiment show that, whereas friend are always more useful than colleagues, weak friends are especially useful when the market is stable and with low unemployment and vacancy rate.