Why Labor Markets May Underprovide Benefits
Sulka: "Providing Benefits to Uninformed Workers" CRC Discussion Paper No. 566
Pecuniary workplace benefits (such as health insurance and pension plans) make up a large and growing share of workers’ total compensation. In the United States, for example, they account for around 31% of what the average employee earns. Yet many workers have a poor understanding of what their employer actually provides. Moreover, since that understanding improves with tenure, workers appear to learn about benefits through experience rather than upfront.
In “Providing Benefits to Uninformed Workers,” Tomasz Sulka (HU-Berlin, Project B05) studies what happens when firms design compensation packages knowing that workers cannot fully evaluate the benefits component before accepting a job offer. In contrast to wages, benefits are modeled as something workers observe only after joining a firm, and the paper asks how this shapes the compensation packages firms choose to offer in equilibrium.
The paper identifies two main findings. First, the unobservability of benefits distorts firms’ incentives toward providing inefficiently low benefits. For a given wage offer, raising benefits increases a firm’s costs without improving its ability to attract workers, since workers do not observe benefits when deciding whether to accept an offer. Cutting benefits, on the other hand, reduces costs while leaving hiring unaffected. Low benefits lead to higher turnover once workers discover what they are receiving - but this may not be sufficient to discipline firms into providing high benefits. As a result, an outcome where all firms provide high benefits is more difficult to sustain in equilibrium than one where all firms provide low benefits, even though the former is more efficient and more profitable for firms.
Second, when the workforce includes both workers who correctly account for the relationship between wages and benefits, and workers who treat the two components as independent, firms can profitably offer distinct compensation packages even when their costs and workers’ preferences are identical. These equilibria resemble the prediction of the classical theory of compensating differentials -- the idea that jobs with worse non-wage attributes must pay higher wages to attract workers -- but the underlying logic and efficiency properties differ sharply. Crucially, the wage premium offered alongside low-benefit packages need not fully compensate workers for the difference in benefits.
Together, these findings offer a unified explanation for several empirical patterns that are difficult to rationalize under standard full-information models: why benefit provision appears inefficiently low, why benefits exhibit large dispersion across firms relative to wages, why lower benefits are associated with higher turnover, and why empirical estimates of compensating differentials are often hard to interpret. The model also yields a policy implication: measures aimed at increasing benefit provision are likely to be more effective if they target firms’ incentives directly (for instance through tax advantages) rather than relying on workers’ ability to evaluate benefits and sort into better compensation packages during job search.
Link (pdf): Providing Benefits to Uninformed Workers
