Title: Copula-based models of productivity and efficiency
Authors: Artem Prokhorov - University of Sydney (Australia) [presenting]
Abstract: The aim is to discuss how to use copulas to measure firm efficiency and productivity. Copulas are mathematical concepts that model dependence between random variables. Traditional econometric analysis of efficiency and productivity based on parametric production functions (commonly referred to as stochastic frontier analysis) has ignored important aspects of production where dependence plays a key role. The dependence influences simultaneous decisions on how much to produce and what proportions of inputs to use. It influences how factors outside our control such as extreme weather or new legislation, affect how inefficient we are. Appropriate copulas capturing these dependencies allow for robust estimation and testing of production models and achieve remarkable improvements in productivity and efficiency. The focus will be on copula estimation for stochastic frontier analysis. The use of copulas to capture various dependencies that have not been accounted for before during production will be considered. One such important case study is endogenous choice of production inputs. Such endogeneity, if ignored, leads to biased estimates of productivity and return to scale and may understate inefficiency. Another is dependence overtime in a panel data setting, which allows for a more precise estimation of technical inefficiencies. Yet another is the case of modelling joint patterns in technical and allocative inefficiencies of firms.