Endogeneity in panel stochastic frontier models: an application to the Japanese cotton spinning industry

Abstract

We present a panel stochastic frontier model that handles the endogeneity problem. This model can treat the endogeneity of both frontier and inefficiency variables. We apply our method to examine the technical efficiency of Japanese cotton spinning industry. Our results indicate that market concentration is endogenous, and when its endogeneity is properly handled, it has a larger negative impact on the technical efficiency of cotton spinning plants. We find that the exogenous model substantially overestimates efficiency in concentrated markets.

Investigating space‐time patterns of regional industrial resilience through a micro‐level approach: An application to the Italian wine industry

“…Another relevant issue that has not been addressed by the existing spatial frontier frameworks is the potential endogeneity of the inputs, associated with reverse causality or omitted variable bias. 2 Despite its impact on the consistency of the estimators (Amsler, Prokhorov, & Schmidt, 2016), this important concern has been addressed only recently in the SFA literature (Karakaplan & Kutlu, 2017a, 2017bKutlu, 2010;Tran & Tsionas, 2013), while traditional SFA models tend to ignore this potential bias.…”

“…• A panel SFA specification that allows to overcome the potential endogeneity of inputs (Karakaplan & Kutlu, 2017b).…”

“…where (lat i , long i ) is the coordinates vector of the ith point in space and β i (lat i , long i ) is a realization of the continuous function β (lat, long) at point i. In this study, we combine the GWR approach with the endogenous SFA specification proposed by Karakaplan and Kutlu (2017b): this method was developed as an extension of the original model for cross-sectional data (Karakaplan & Kutlu, 2017a) to solve the endogeneity problem of inputs in an SFA panel setting through a single-stage approach. The model is expressed as follows:…”

“…The aim of this paper is to contribute to advance the current empirical literature on economic resilience by proposing a novel parametric model that allows to study local responses to disturbances and to detect the presence of specific space-time patterns of firms' performance in the data. The proposed specification combines two frameworks-the geographically weighted regression (GWR) and the endogenous panel stochastic frontier model in the style of Karakaplan and Kutlu (2017b)-to locally estimate the temporal patterns of technical efficiency in specific firm populations, overcoming the main limitations of the existing methodologies. The empirical approach presented in this paper is particularly beneficial to evaluate the resilience dynamics of firm populations affected by relevant shocks and disturbances: during these periods, when the local responses of firms are likely to be highly asymmetric, assuming the presence of a single global trend of performances and ignoring spatial factors is particularly unrealistic.…”

Estimating market power using a composed error model

“… As explained by Karakaplan and Kutlu (,b, ), estimations with this model are done in a single stage. So, to avoid confusion, we do not use the name “first‐stage statistics” and instead, call them prediction equations. …”