ABSTRACT
The objective of Chapter 9 is to describe the concepts of five techniques that can be utilized to complement understanding of efficiency after analysis, using DEA. This procedure is known as the two-stage analysis. The five techniques are Bootstrapping, CausalImpact, Tobit Regression and Ordinal Least Squares (OLS), Malmquist Index and Artificial Neural Networks (ANN). Brief descriptions of each technique are provided. The authors strongly suggest that these techniques be utilized in conjunction with DEA. Bootstrapping can be used to correct efficiency scores in view of random errors inherent in the data utilized, whereas CausalImpact, developed by Google researchers, can be utilized to attribute causality to effects observed in efficiency over time. Tobit Regression and Ordinal Least Squares and Artificial Neural Networks can be utilized to search for variables that explain the efficient behavior observed. The Malmquist Index can be utilized to evaluate progress or regression in efficiency and technology.
