Re-Examining the Technical Efficiency of Australian Banks: A Bootstrap DEA Approach

Abstract

This study re-examines both pure technical and scale efficiency of Australian banks using bootstrap data envelopment analysis (DEA). The aim is to improve the choice of variables of the core profit efficiency model which is commonly used in earlier Australian banking efficiency studies. After we introduce the "interest income" over "net interest income" variable in the core profit efficiency model, the proportion of fully pure technical efficient banks decreased to 23% which is significantly lower than 81% which was reported in a recent study. This research argues that the main issue that has contributed to this difference is that improving the choice of variables significantly increases the discriminatory power of efficiency estimates. Additionally, emphasising on statistical properties of efficiency estimates, this study employs bootstrap DEA to provide confidence intervals and bias corrected estimates of pure technical efficiency scores of the sample banks. The bootstrap results show the importance of incorporating sample variation and bias in estimating efficiency scores. Earlier Australian banking efficiency studies ignored such issues. The new findings from the sample banks could have important implications for the banking industry in Australia.