Essentially, the two models developed for ROA and ROE behave very similarly in the sense that both the models have same set of variables which are statistically significant (and the same set of variables which are statistically not different from zero). The direction of causality for all these variables whether statistically significant or not significant, is also same. Thus, it is reasonable to say that for the given data, the ROA and ROE models behave in the same way.The empirical model developed above shows that the coefficient of market share is statistically significant for both the ROE and ROA model.
It is also evident from the model that the coefficient for market share is positive. This implies that market share has a positive bearing on profitability. Another observation here is that the effect of concentration is statistically not significant on profitability of the banks. All these findings are in line with the inferences and deductions of other researchers like Smirlock (1985) and Goldberg & Rai (1996).
The paper deals with developing an empirical relationship between the profitability of the banks and their market share and market concentration. For this purpose, the paper uses the data of three largest banks in four different countries across five years. The banks included in the study have both positive and negative ROA and ROE so that diversity in terms of their performance can be achieved through the data. To capture changes in the market structure across time and geographies, different countries and different years are taken. This when combined together with different banks in each geography and year, gives a dataset diverse enough to capture variations in the market and conduct as well as structural efficiencies of the banks.
The analysis of the data using ordinary least square regressions shows that the behaviour of the profitability is consistent across both the indicators used to measure profitability. The analysis uses two separate OLS regressions to develop causality between profitability measures and the set of independent variables. The variables of interest here are market share and market concentration. The other variables namely the GDP, CPI and the deposits are used as control variables. The control variables do not directly impact the dependent variables, but they do control for changes or reactions in them indirectly.