Semi-parametric analysis with double bootstrap: an econometric procedure for productivity analysis

A two stage semi-parametric technique is a popular methodology in productivity analysis. Traditional two stage methodology, in which non-parametric Data Envelopment Analysis and some sort of regression as a parametric part of methodology are used, may produce unreliable results. This unreliability is due to upward bias in the technical efficiency (TE), serial correlation in the error term, correlation of the error term and environmental variables and lastly due to bounded nature of the TE scores. Though Simar and Wilson (2007) proposed a Double Bootstrap Procedure as a remedy towards these problems, but even then few studies adopt this methodology for two stage semi-parametric analysis. Primary reason for not adopting this methodology is non availability of such econometric software that can perform this analysis in an automated way. A SAS Program is being presented here that can perform this analysis automatically and in a considerably lesser time as compared to the time taken, for such types of analyses, reported in the literature.