1. Leads-and-lags graph of ROA for 446 firms with high (above-median) versus 447 firms low (below-median) InfoCost2000. Graph shows annual point estimates and 95% CIs from regression of ROA on high InfoCost2000 dummy * year dummies, with DMO covariates, constant term, and firm and year fixed effects. Coefficient for 1997 is pegged to zero;C Panel;Outcome is ROA. Treatment group is based on InfoCost2000
2. Similar to Panels A and C but graphs show annual DiD coefficients for three "treated" groups (1) 171 treated firms (Audit Committee IV = 1) and low InfoCost2000; (2) 298 control firms (Audit Committee IV = 0) with high (above median) InfoCost2000, and (3) 148 treated firms with high InfoCost2000), relative to pseudo-control group of 276 control firms with low InfoCost2000. Regressions include DMO covariates, constant term, and firm and year fixed effects, Coefficient for 1997 is pegged to zero. F-test for differences between;E Panel;Outcome is ROA. Treatment Groups is based on both Audit Committee IV and InfoCost2000,2005
3. Obituary