A0339
Title: Endogeneity, temporal variation, and heterogeneous inefficiencies in panel stochastic frontier models
Authors: Abbas Hakeem - University of Adelaide (Australia) [presenting]
Firmin Doko Tchatoka - The University of Adelaide (Australia)
Abstract: Endogeneity in stochastic frontier models arises when production inputs are simultaneously chosen with output, leading to the correlation between inputs and either statistical noise, inefficiency, or both. Existing models that address endogeneity, particularly those allowing for input correlations with inefficiency, often use copulas to model these correlations. However, the use of copulas can be computationally intensive, especially in panel data settings, and may introduce misspecification errors. A panel stochastic frontier model that does not require the use of copulas is proposed. The proposed model accommodates heteroskedasticity in both noise and inefficiency while allowing inefficiency to follow an autoregressive process. Estimation is performed using marginal maximum simulated likelihood (MMSL). Conditional mean, conditional mode, and restricted conditional mode estimators are also developed to estimate inefficiency. Through Monte Carlo simulations, it is demonstrated that the estimators perform well under varying scenarios.
