Standard models of dynamic labor demand rely on the presence of adjustment costs to explain the observed smoothness in employment patterns, although the costs are often difficult to quantify. The experience rating feature of the U. S. Unemployment Insurance (UI) system provides a measurable linear cost of adjustment. Using a unique data set with administrative data on over 8000 firms, I estimate the effect of a UI-induced linear adjustment cost on seasonal labor demand in retail trade. I find strong support for the large role of adjustment costs in reducing the employment response of firms to seasonal fluctuations in demand.

This paper considers the revealed preference implications of a fairly general optimization hypothesis under any finite number of constraints. It extends Afriat's seminal nonparametric analysis. The usefulness of the results are illustrated in two examples.