Consistency conditions for bank efficiency analysis in Ghana: A comparison of parametric and non-parametric techniques

This paper extends the concept of methodological crosschecking by examining whether bank efficiencies computed by the two frontier techniques, stochastic frontier analysis (SFA) and data envelopment analysis (DEA), are consistent. The study used a panel of 220 unbalanced observations from 27 Ghanaian banks between 2007 and 2016 to estimate cost and technical efficiencies and check for consistency using five criteria: efficiency distribution, ranking, ability to identify best or worst banks, stability of efficiencies, and relationship with accounting ratios. The results suggest that there is no consistency in the way parametric and non-parametric techniques rank or identify the best or worst banks. Also, there exists a weak relationship between the efficiency scores generated by both SFA and DEA and the non-frontier accounting ratios of Ghanaian banks. This suggests that the latter may contain some exogenous variables that make them weak measures of efficiency and should be used with caution, especially for bank supervision. However, the SFA approach yielded efficiency scores that were comparatively more stable over time. Therefore, the study concludes that the SFA approach is more practical and thus more appealing for regulatory purposes in Ghana due to the relatively consistent efficiency scores under the SFA approach compared to those under the DEA.