Costly Transmission of Information. A comparison of signals and networks

Economists and sociologists disagree over markets’ potential to substitute for networks of private connections. In this paper we focus on the transmission of information in labor markets and study a model where workers differ in their (ex ante unobservable) productivity. Before entering the market, individuals can pay a cost and either establish a link to a currently employed worker ("network"), or take an exam that may certify them as high productivity ("signal"). High productivity individuals have a higher probability of connecting to a high productivity worker in the network case, and of achieving certification in the signaling case. The main difference between the two mechanisms is that networking, as opposed to signaling, reveals information only locally and thus ties the worker to one specific firm. The lower return from networking also makes it less informative. The final result is that under networking firms make positive profits (because of the informational rents), wages to high productivity workers are depressed (because information is less precise and because the rents are shared), and wages to low productivity workers are higher (because information is less precise). Our model then comes down in qualified support of the economists’ view: at equal cost, networking is not a very effective mechanism because it reveals the information only locally, creates positional rents and reduces the difference in rewards between high and low productivity workers. ∗Department of Economics, Columbia University, GREQAM, NBER and CEPR. [email protected] †Earth Institute, Columbia University, [email protected]