As a further robustness check, we allowed for the most flexible model which includes hotel-specific lagged dependent variables and hotel-specific error variance that may be different before and after the event. In this model, we estimate 274 coefficients in addition to 42 variances of the error term. The model allows for event-specific and market-specific characteristics, yet it is able to obtain remarkably precise estimates of the main coefficient of interest. For all three measures of performance, the estimate of β is negative and strongly statistically significant, while the estimates of µ are weakly significant. These results are consistent with the results reported in Table 6.