The index effect: Comparison of different measurement approaches

The purpose of this paper is to examine additions and deletions of the S&P500. Competing event studies will be performed to analyze the announcement effect with different benchmarks. The main components of the paper are a Market Model based event study and a GARCH(1,1)-M Model study. The GARCH-M Model assumes a relation between variance and mean of an asset, where the variance tends to follow a long-run variance, and is dependent upon the stock's preceding abnormal return and variance. The event study examines approximately 350 stocks over 240 trading days. I have to confirm a positive index effect for additions and vice versa for deletions. Price, abnormal return, volume and variance of added stocks rise on the day after the announcement. My analysis reveal a semi-strong reversal and therefore are consistent with the Price Pressure Hypothesis and partly with the Imperfect Substitutes and the Visibility Hypothesis. The hypothesis of a significant lambda factor - the impact of the variance on the mean - has to be rejected for the additions and deletions of the S&P 500 for all event studies conducted. The analysis find significant ARCH and GARCH terms for the variance equation. The obtained lambda factors are not normally distributed. In the additions set the use of the GARCH(1,5)-M model seems to extenuate the index effect.