This paper studies the value of past coworkers as a source of information about the labour market. Using Swedish register data and the unexpected death of past coworkers as a natural experiment, I find that losing a professional connection lowers annual wage earnings by 0.9 per cent after two years and employment by 0.5 percentage points. The findings are consistent with a job search model where connections affect the arrival rate of job offers for the unemployed. Unemployed workers lower their reservation wage after losing a connection, but not enough to fully undo the effect of a lower arrival rate on employment. I then use Generalised Random Forests to test different theories about the determinants of the value of a connection. Significant negative earnings effects are concentrated in the bottom quartile of the distribution of predicted effects. The most affected individuals lose connections who in turn had few other connections, suggesting that competition for information leads to congestion in job search networks. Homophily in general varies less across predicted effect quantiles, though some forms of homophily between deceased and affected individuals do predict more negative effects. In general, the predicted effect varies more across deaths than within deaths, again suggesting that the degree of competition for information from the deceased individual is the primary determinant of the value of a connection.

Sébastien Willis is a labour economist, currently working as a postdoctoral researcher at the department of economics, Uppsala University, and research coordinator of the Uppsala Immigration Lab. He conducts empirical research on immigrant labour market integration and on the use of social networks in job search. He received his PhD in economics from Universitat Pompeu Fabra in 2021.