The Puzzle Between Banking Competition and Profitability can be Solved: International Evidence from Bank-Level Data

Nevertheless, the Claessens and Laeven (2004) study did not include Singapore, South Korea, Taiwan, or Sweden.

Claessens and Klingebiel (2001) analyze the costs and benefits of different degrees of financial sector competition and permissible activities, as well as related implications for regulation and supervision. According to country experiences and theory, they suggest that competitiveness does not necessarily require many financial institutions, as a concentrated system can be competitive if contestable (i.e., open to competition). The degree of contestable entry in the financial sector, together with competition from other forms of financial intermediation, has been an important determinant of the performance and efficiency of financial sectors. Claessens and Laeven (2003) relate a competition measure to industrial growth for 29 banking systems. They find that the effects of competition on access to financing (and growth) depend on the financial system’s level of development.

Bond et al. (2001) indicate that the system dynamic GMM method dominates the difference dynamic GMM proposed by Arellano and Bond (1991), because it deals with the problems of weak instruments and measurement error bias.

In order to make sure that the GMM model is superior to the other method, Wu-Hausman statistics (for details, see Wu 1973 and Hausman 1978) are adopted to examine whether there exists the problem of endogeneity among all explanatory variables. The test result shows that the null hypothesis of exdogeneity for variables is rejected, violating the assumption of the traditional OLS approach and causing the estimation bias. Thus, adding appropriate instrumental variables is essential. The estimation results based on the extended model find that the coefficients of CR4 and HHI become significant compared with the static model with insignificant results. Therefore, it demonstrates the advantage of dynamic GMM.

We also consider NIM (Net Interest Margin) as a proxy of profitability. The coefficients of the competition index remain unchanged, such as the coefficients (t values) for CR4, HHI, and PRH being respectively 0.081 (1.152), 0.009 (1.283), and −0.237 (−1.199). Aside from this similarity, other variables are not very consistent. The reason behind this inconsistency probably means that NIM is only one part of profitability. NIM is a measure of the difference between the interest income generated by banks or other financial institutions and the amount of interest paid out to their lenders (for example, deposits), relative to the amount of their assets. Thus, NIM measures traditional banking activities only (loans and deposits). However, NIM neither counts banks’ ability at generating non-interest revenue, nor reflects management skill, such as non-interest expense. Thus, some researchers prefer to either use return on assets (ROAE) or return on equities (ROE) in their analyses, such as Beck et al. (2005), Boubakri et al. (2005), and Goddard et al. (2004), etc. Both variables ROAE and ROE could be separated (through the Du Pont Equation) into the net return rate and turnover rate of assets (equity), which measure the profitability and the ability of management at the same time.

New Zealand’s banking system appears very profitable and has a low concentration—this is because virtually all banks there are controlled by Australian banks.

Nevertheless, the main purpose of this paper is not to emphasize the causal relationship between banking competition and profitability, but rather to point out that the linear relationship is unable to clearly point out banking competition influences banking profitability.

All dynamic panel regressions in Tables 5, 6, 7 and 8 pass the specifications tests. Specifically, none reject the Sargan test and the regressions do not exhibit significant serial correlation.

However, they are available upon request.