Measuring performance in the Portuguese banking industry with a Fourier regression model

Abstract

This article analyses the determinants of banks' profitability in the Portuguese banking sector during the period 1990 to 2005. The study extends the established literature on modelling the banks'k performance by applying a Fourier approximation in order to detect for possible nonlinearities between the profitability variables and the explanatory variables. In so doing, we verify that the introduction of the Fourier coefficients in the analysis quite improved the quality of the adjustments, the need to accept the existence of nonlinear relationships among the variables involved in the study thus becoming evident. The results of this article suggest that the best performing banks in the Portuguese banking sector are those which have endeavoured to improve their capital and labour productivity, those which have maintained a high dimension and, finally, those which have been able to reinforce their capital structure.

Notes

¹For more details on this methodology see Becker et al. (2002).

²These results, as well as those for all the other models that use ROE and the size variables Sizea and Sizepb, are available on request from the authors.

³Note that F* does not correspond to the critical value associated to the hypothesis of nullity of all the Fourier coefficients but instead to the critical value associated to the test of nullity of the pairs sin/cos of the Fourier coefficients (a_ji  = b_ji  = 0).