Aggregation bias in stochastic frontier models employing region-level data: An empirical analysis with Korean manufacturing data

A stochastic frontier study often employs aggregate data to analyse the productivity and technical efficiency of regions. In this article, a stochastic frontier model is run on plant-level data and region-level aggregate data. Comparisons of estimated coefficients and characteristics of regional production based on estimation outcomes suggest that an empirical model employing regional-level data can provide misleading results concerning the production function faced by a representative plant in a region.