Spatial and Temporal Aggregation in the Estimation of Labor Demand Functions

The consequences of aggregation, temporal or spatial, for the estimation of demand models are theoretically well-known, but have not been documented empirically with appropriate data before. In this paper we conduct a simple, but instructive, exercise to fill in this gap, using a large quarterly dataset at the establishment-level that is increasingly aggregated up to the 2-digit SIC industry and the yearly frequency. We only obtain sensible results with the quadratic adjustment cost model at the most aggregated levels. Indeed, the results for quadratic adjustment costs confirm that aggregation along both dimensions works to produce more reasonable estimates of the parameters of interest. The fixed adjustment cost model performs remarkably well with quarterly, but also with yearly, data. We argue that is may be one more consequence of the unusually high labor adjustment costs in the Portuguese labor market.