THE DETERMINANTS OF SOVEREIGN CREDIT RATINGS: A NEW EMPIRICAL APPROACH

Abstract

The purpose of this paper is to extend the literature on the determinants of sovereign credit ratings by, firstly, showing empirically that there is a probable bias in the results obtained by previous studies and, secondly, to combine a General-to-Specific (GETS) model selection strategy with Dynamic Panel Data estimation techniques to resolve these biases. The results are encouraging: the preferred model is not only statistically adequate, but also economically significant and consistent with the categories of variables deemed to be important by the agencies who construct sovereign credit ratings.