I've conducted a regression analysis using an event study methodology to explore the Cumulative Abnormal Returns (CAR) linked to stock repurchases over one and two years. Interestingly, while year 1 showed statistical significance, year 2 did not. Now, I'm investigating the factors influencing CAR in both periods, using it as the dependent variable and multiple variables as independents in a cross-sectional regression.
However, in the year 2 analysis, I've found significant independent variables despite the CAR insignificance, prompting an econometric query: Is it valid to use initial insignificant CAR based on mean values while employing regression analysis on individual observations later? Does this methodological divergence potentially affect result validity, considering the initial CAR insignificance?
