Cross-Sectional Variation in Stock Price Reaction to Bond Rating Changes: Evidence from India (Methodology-2)

The BSE 500 companies were then divided into two equal parts -large and small, each year on the basis of their characteristic value. Company below the median characteristic value was classified as small or low on characteristic otherwise it was classified as large or high on the characteristic1. The characteristic category of each case of bond rating change was taken as the category to which the case belonged for the year – end preceding the rating change. Liquidity had to be estimated as average trading volume. The liquidity of stocks was computed for each case of rating change. The companies on BSE 500 were arranged in the descending order of this ratio and divided into two equal parts – high and low based on their liquidity value for each relevant date. Company below the median liquidity value was classified as less liquid otherwise it was classified as highly liquid. The category of each case of rating change was taken as the category to which the case belonged on the relevant date of rating change. Thereafter, upgrade and downgrade portfolios based on each characteristic were analysed separately using the CAAR analysis.
In the second phase of research, the relationship between the stock returns and bond rating changes was analysed in light of factors like anticipation, magnitude of rating change, transition to, from or within speculative grade and impact of business cycle. The impact of these factors is also seen for the various firm characteristic based portfolios. Regression model given by Goh and Ederington (1999) was used for analysis.  It is calculated as the absolute value of difference between the numerical score of the rating after and before the rating change. BUSINESS_CYCLE is dummy variable for downturn of business cycle taking values either 1 or 0. In this study the period after September, 20085 was taken as the period where the business cycle took a downturn due to global economic crisis. The variable takes the value of 1 where the rating was changed after September, 2008. In all other cases where the rating underwent a change in or before September, 2008, the dummy variable is taken as 0. SPEC is a dummy variable taking the value of 1 for movement within speculative grade and 0 otherwise. FALLEN is also a dummy variable which assumes value 1 where the movement is to or from speculative grade and 0 otherwise.
This model was also applied to the various firm chacteristic based portfolios to examine whether this relationship changes due to the variables including anticipation, magnitude of rating change and impact of business cycle. The other two variables (FALLEN and SPEC) were excluded owing to the lack of sufficient cases in the portfolio data.
All the variables were tested for correlation with each other. Wherever, a significant correlation was found the variables were suitably transformed to avoid multicollinearity problems.

Tags: , , , ,