A Hierarchical Panel Data Model for the Estimation of Stochastic Metafrontiers: Computational Issues and an Empirical Application

Abstract

In the metafrontier literature, firms are put into groups, generally defined by technology or geography. Each group has its own technological frontier, and the metafrontier is the upper bound of these group frontiers. The aim of this literature is to measure a firm’s inefficiency, and to decompose it into its inefficiency relative to its group’s frontier and the inefficiency of its group’s frontier relative to the metafrontier. A previous paper (Amsler et al., Empirical Economics 60:353–363, 2021) proposes a hierarchical stochastic frontier model to accomplish this, where the hierarchy is firms in groups in the overall set of groups. This chapter gives an empirical implementation of this model, with emphasis on computational issues.