Revealed Preference Differences Among Credit Rating Agencies

Access to files

•   FULL-TEXT.PDF (pdf)

Abstract

Revealed Preference Differences Among Credit Rating AgenciesThe thesis studies the factors which underpin the allocation of credit ratings by the two major credit rating agencies (CRAs) namely Moody’s and S&P. CRAs make regular headlines, and their rating’s judgements are closely followed and debated by the financial community. Indeed, criticism of these agencies emerged, both in this community and the popular press, following the 2007-2008 financial crisis. This thesis examines several aspects of the allocation of credit ratings by the major agencies, particularly in relation to (i) their revealed “loss function” preference structure, (ii) the determinants underpinning the allocation of credit ratings and (iii) the reasons determining the circumstances when the two agencies appear to differ in their opinions, and we witness a split credit rating allocation. The first essay empirically estimates the loss function preferences of two agencies by analyzing instances of split credit ratings assigned to corporate issuers. Our dataset utilises a time series of nineteen years (1991-2009) of historical credit ratings data from corporate issuers. The methodology consists of estimating rating judgment differences by deducting the rating implied probability of default from the estimated market implied probability of default. Then, utilising judgment differences, we adapt the GMM estimation following Elliott et al. (2005), to extract the loss function preferences of the two agencies. The estimated preferences show a higher degree of asymmetry in the case of Moody’s, and we find strong evidence of conservatism (relative to the market) in industry sectors other than financials and utilities. S&P exhibits loss function asymmetry in both the utility and financial sectors, whereas in other sectors we find strong evidence of symmetric preferences relative to those of the market. The second essay compares the impact of financial, governance and other variables (in an attempt to capture various subjective elements) in determining issuer credit ratings between the two major CRAs. Utilising a sample of 5192 firm-year observations from S&P400, S&P500 and S&P600 index constituent issuer firms, we employ an ordered probit model on a panel dataset spanning 1995 through 2009. The empirical results suggest that the agencies indeed differ on the level of importance they attach to each variable. We conclude that financial information remains the most significant factor in the attribution of credit ratings for both the agencies. We find no significant improvement in the predictive power of credit rating when we incorporate governance related variables. Our other factors show strong evidence of continuing stringent standards, reputational concerns, and differences in standards during economic crises by the two rating agencies. The third essay investigates the factors determining the allocation of different (split) credit ratings to the same firm by the two agencies. We use financial, governance and other factors in an attempt to capture various subjective elements to explain split credit ratings. The study uses a two-stage bivariate probit estimation method. We use a sample of 5238 firm-year observations from S&P 500, S&P 400, and S&P 600 index constituent firms. Our results indicate that a firm having greater size, favourable coverage and higher profitability are less likely to have a split. However, smaller firms with unfavourable coverage and lower profitability appear to be rated lower by Moody’s in comparison to S&P. Our findings suggest that the stage of the business cycle plays no significant role in deciding splits, but rating shopping and the introduction of regulation FD increase the likelihood of splits arising.

Keyword(s)

Credit Rating Agencies ; Credit Ratings ; Loss Function

Bibliographic metadata

Institutional metadata

Record metadata