Measuring retail productivity: proposition of a new method for decision making

Abstract

This paper proposes an elegant way of estimating retail productivity through a mathematical model inspired by FGLS multivariate linear regression, using the error equation obtained as a hidden variable estimator to be used as inferred efficiency. This methodology was applied to a dataset obtained from a leading ready-to-eat cereal (RTEC) manufacturer and demonstrated substantial results. Unlike conventional ordinary least squares (OLS) regressions, the proposed approach offers benchmarking information relative to the best-performing retailers, as opposed to the typical relative-to-average outcomes of regressions. Furthermore, in contrast to traditional data envelopment analysis (DEA), this methodology relies on the entire dataset to establish the efficiency index, thereby addressing limitations imposed by DEA and allowing for executive discretion and flexibility in the analysis.