Productivity Drivers of Efficiency in Banking: Importance of Model Specifications

We use nonparametric method with various specifications to estimate efficiency of banks and then use truncated regression with double-bootstrap (Simar and Wilson, 2007) to analyze how various bank-specific factors explain the differences in the estimated levels of inefficiency across the banks in Ukraine, with a particular focus on foreign ownership. We show that some results are very robust, yet others depend on the chosen Data Envelopment Analysis specification for the efficiency model. We also find that the efficiency of banks with foreign ownership is not distinguishable from the efficiency of the locally-owned banks if the degree of ownership control is not accounted for. On the other hand, when we account for degree of control, we find evidence that banks that were 100% foreign-owned were significantly more efficient than the locally owned banks, while the partially foreign-owned banks were significantly less efficient than the locally owned banks (on average and ceteris paribus), thus providing empirical support for the agency and corporate governance theories in economics.