A Production Frontier Model With Flexible Temporal Variation In Technical Efficiency

Abstract

Abstract Stochastic frontier models originated with Aigner, Lovell, and Schmidt (1977) and Meeusen and van den Broeck (1977). These models were intended for cross-sectional applications, and they rested on strong assumptions about the errors. Statistical noise was assumed to be normally distributed, while technical inefficiency was assumed to be distributed according to a specific one-sided distribution, such as exponential or half-normal. Furthermore, statistical noise and technical inefficiency were assumed to be independent of each other and of the explanatory variables (inputs). This line of research has been continued, with an emphasis on alternative distributional as­ sumptions for technical inefficiency. For examples, see Stevenson (1980) and Greene (1990). Surveys are given by Schmidt (1986), Lovell and Schmidt (1988), and Bauer (1990), and by Greene in Section 2.4 of this book. It is now well understood that the basic parameters (regression coefficients) of the model can be estimated under weaker assumptions, but the separation of technical inefficiency from statistical noise [as in Jondrow, Lovell, Materov, and Schmidt (1982)] intrinsically requires strong distributional assumptions.