1 Introduction
In the implicit contract literature firms act as insurers for their employees: risk-averse
workers get insurance against income fluctuations in that wages are not linked to changes
in productivity (see Rosen, 1985, for a survey).
Principal-agent models stress instead the role of incentives in determining the optimal
compensation scheme, starting from the consideration that the interests of firm and workers
rarely coincide. Workers enjoy leisure, and this may come at the expense of the firm’s
performance. If workers’ actions in the workplace were perfectly observable it would be easy
to devise ways of inducing them to produce the amount of effort agreed upon in a contract.
But actions are only partially observable, or observability may require costly monitoring.
A flat compensation scheme, while offering perfect insurance, removes any incentive for
the worker to exert effort. One way to create proper incentives is to link compensation
to the firm’s performance. However, such risk-sharing has a cost, as it requires paying
workers a premium that increases with their risk aversion. Thus, providing incentives
curtails insurance.
The trade-off between incentives and insurance in the firm-worker relation has received
a great deal of attention in the theoretical literature and is at the heart of modern contract
theory. Much progress has been made in studying the design of incentive contracts under
a variety of theoretical situations (see Gibbons, 1998, for a recent survey). Substantial
progress has also been made in confronting some of the implications of the theory with the
data, with two main strands: one attempting to measure whether incentives actually im-
prove firm performance, the second aimed at testing the structure of the insurance-incentive
model. Prendergast (1999) offers a thorough account of the empirical achievements.
Most of the empirical literature to date has been concerned with executives’ compensa-
tion (see, for example, Jensen and Murphy, 1990; Margiotta and Miller, 2000). However, if
one is interested in assessing the insurance role of the firm in the labor market, it is unlikely
that CEOs provide a proper benchmark. First, CEOs and executives are in general a tiny
fraction of the labor force. Second, they are a highly self-selected group of low risk averse
(or perhaps even risk loving) workers who probably have only marginal interest in wage in-