has received little attention in the empirical as well as in the theoretical literature. However,
it is plausible that the extent to which shocks are passed on to wages depends on their
degree of persistence.1 For instance, it may be that only transitory shocks to output are
absorbed by the firm, while permanent shocks are shared, at least partially, with workers.
Insofar as both types of shock are present, ignoring the distinction may bias the results
towards insurance or incentives, depending on the relative importance of the transitory and
permanent components. Fourth, our methodology allows for wage shocks that are unrelated
to firm performance. Thus, we can compute how much of the observed earnings variability
can be traced to workers sharing firms’ risk and how much to idiosyncratic shocks to wages;
while the former can potentially be insured within the firm, the latter are unlikely to be
insurable. Finally, we propose a novel identification strategy that can also be applied to
analogous problems arising in different areas of research.
The rest of the paper proceeds as follows. In Section 2 we review the insights of the
standard principal-agent model. In Section 3 we characterize our empirical approach to the
problem, considering a stochastic specification for firm performance and workers’ earnings.
We show that, in the spirit of the principal-agent model, if worker compensation is related
to firm performance, a set of orthogonality conditions obtains that can be used to answer a
number of empirically relevant questions. In particular, one can examine whether shocks to
firms’ performance are passed on to wages, and to what extent this is affected by whether the
shock is transitory or permanent. Section 4 discusses how identification of the parameters
of interest can be achieved. Section 5 describes the matched firm-worker data set used
in the empirical analysis, and Section 6 presents the estimation of the stochastic model
of firm performance and workers’ earnings. The main empirical results are presented in
Section 7, where we focus on the estimates of the incentive-insurance trade-off for the total
sample and then examine the implications of the agency model for the sensitivity of wages to
performance. Section 8 discusses the results and contrasts our findings with the implications
of perfectly competitive models with and without frictions. Section 9 concludes and traces
1One paper we are aware of that addresses this point theoretically is Govindaraj and Ramakhrishnan
(2000). They show that the sensitivity of the agent’s payment to firm performance increases with the
persistence of the latter (see their Proposition 2).