Clustering

Thirty Indian stocks are studied around 2 Mar 2026 (common event date). The average pairwise correlation of their market-model residuals is 0.10, and a naive test gives $t_{\text{BMP}}=3.0$. (a) Derive $\operatorname{Var}(\overline{AR})$ under equicorrelation. (b) Compute the Kolari–Pynnönen-adjusted statistic. (c) What simpler procedure avoids the problem entirely, and what does it give up?