مقاله سال ۲۰۱۷

This paper investigates the link between social networks and welfare in the context of a frictional labour market with occupational mismatch. Empirical studies show that 30-60% of new hires find jobs via referrals.1 At the same time, there is evidence that up to 47% of workers in some occupations are mismatched (Robst, 2007). Thus it is natural to ask whether the network channel of job search contributes to higher occupational mismatch. From a theoretical perspective, Bentolila et al. (2010) and Horvath (2014b) show that social networks with weak homophily may generate more mismatch compared to the formal channel of search. Weak homophily here means that workers have many social contacts in occupations other than their own. However, more mismatch is not equivalent to lower welfare, especially in the presence of volatile output. On the contrary, it may be optimal for workers to diversify their networks across occupations in order to reduce the risk of unemployment even if this strategy is associated with more mismatch. This study fills the gap in the analysis of network implications for social welfare and investigates the optimal level of network diversification in a setting with stochastic output.

The ingredients of the model are as follows. There are two worker types and two occupations. The type of worker is given by the initial training in one of the two occupations. Every worker can be unemployed, employed in the primary occupation or mismatched, which is associated with a wage penalty. Output fluctuations are described by a time-homogeneous transition matrix and are correlated between the two occupations. In the benchmark case there are binary fluctuations in each occupation, e.g. a period of expansion with high output and many vacancies and a period of recession with low output and fewer vacancies. Every worker has a fixed total number of social contacts, which is the network size. The level of network homophily is characterized by the proportion of contacts with other workers of the same type. Thus a higher level of homophily implies a less diversified network and vice versa.

In this setting the primary contribution of the paper is a detailed characterization of the optimal network diversification level from the individual and social perspective. Given that workers are risk-neutral and take aggregate indicators as given, they choose full homophily of the social network under some realistic conditions. This is different when the problem is considered from the perspective of social welfare. The social planner internalizes externalities that individual network choices impose on other labour market participants (other workers and firms) and takes into account changes in aggregate variables, such as the equilibrium unemployment rate and vacancies. So for a large range of parameter values, there exists an interior homophily level which is maximizing workers’ expected present value of income. Hence this paper supports policies targetting stronger occupational diversification (i.e. interdisciplinary projects and educational programms) as a means of reducing network homophily.