Reliability and Agreement of Credit Ratings in the Mexican Fixed Income Market

Abstract

Credit ratings play an important role in the fixed income market as the entire regulatory framework of this market segment is based on them and a significant part of what investors can and cannot do is dictated by ratings. Also, a number of ratings-based metrics are employed globally to estimate capital reserves, liquidity buffers, and solvency standards for many institutional investors such as insurance companies and pension funds. A critical assumption at the root of this regulatory architecture is that the credit-rating scales of the three leading agencies (Moody’s, Fitch, and Standard & Poor’s) are completely equivalent.In this study we focus on the Mexican fixed income market. We find that the ratings of all three rating agencies exhibit a very high degree of inter-rater reliability. This means that in terms of ranking a group of bonds based on creditworthiness the three rating agencies would produce very similar results.On the other hand, using a non-parametric statistic, the Wilcoxon matched-pairs test, we conclude that there are significant discrepancies among the ratings of the three agencies. This is consistent with a low level of inter-rater agreement detected. These findings challenge the suitability of credit ratings as a useful metric for regulatory purposes as they create the possibility of arbitrage.