Improved methods for identifying the operational determinants of a bank’s capital ratio

Abstract

Published research using econometric models to identify the determinants of bank capital ratios has produced inconsistent results. This is partly due to the failure of model formulations to distinguish between operational and managerial effects. This paper explains how the use of bank leverage can separate these effects, how to prevent financial ratios from undermining model interpretability, and how to identify and avoid ratio-related biases. The application of these improvements is tested on a panel of East Asian retail bank data from 2004 to 2014, covering China, Hong Kong, Indonesia, India, Japan, Singapore, Malaysia, the Philippines, and Thailand. Conclusions are drawn from quasi-experimental designs comparing two-way GMM models before and after implementation of the improvements. The results show that the improved models identify the operational determinants of capital ratios and avoid simultaneity and omitted variable bias and ratio-induced opacity in the results. The use of complementary data segments helps to interpret the results and identify paradigmatic cases for a better understanding of the relationship between regulatory capital and risk.