Essays on Credit Ratings
Abstract: This thesis consists of four self-contained articles, all of which contribute to the empirical research on credit ratings. Broadly speaking, the first two papers highlight two less ordinary “uses” of credit ratings, in the context of (1) measuring financial constraints and (2) bond market segmentation. The last two papers focus on a relatively new concept in the credit ratings’ literature, specifically the time variation in rating standards, by looking at the sovereign credit ratings, and at how the time variation in rating standards has been measured and interpreted in prior literature.By using an extensive set of tests and corporate investment decisions indicating various degrees of financial constraints, the first paper provides first-hand evidence of the ability of credit ratings to capture corporate responses consistent with various levels of financial constraints. We find that they outperform six common financial constraints proxies in each of our tests and shocks.In the second paper, I use an exogenous change in the index inclusion criteria for a large, widely followed bond index, to show that bonds also exhibit an “index” effect following either addition or deletion to the bond index. Furthermore, this exogenous change leads to different types of index effects, allowing me to compare, for the first time, the relative effect of positive (addition) and negative (deletion) changes in a bond index composition. In this comparison, I find that index addition triggers a market reaction twice as large as index deletion, suggesting that investor awareness may be an important channel for bond pricing.The third paper is the first to provide evidence of rating conservatism for sovereign (i.e. country) credit ratings. I find that a sovereign that received an AA rating in 1993 would only get an A+ twenty years later in spite of maintaining the same economic environment, and this change in conservatism is not justified by an increase in the default rates or by a riskier global environment.In the final paper, I revisit the method commonly used for measuring time variation (i.e. conservatism) of rating standards over time, and I find that the decrease in the year coefficients (i.e. the measure for conservatism according to prior research) is largely a result of common secular trends in the financial variables included in the model. While my findings do not disprove the existence of conservatism, they suggest that its true magnitude may be significantly lower, and that other empirical methods might be more suitable for its assessment.
CLICK HERE TO DOWNLOAD THE WHOLE DISSERTATION. (in PDF format)