Barth et al. (2004) indicate that data in the database that became available in 2001 are primarily from 1999. That is of the 107 responses reviewed, 13 were received in November 1998, 65 were reviewed in 1999, and 29 in 2000 (19 of which in either January or February).
12. It should be mentioned that all the descriptive statistics are after capping the extreme values of all financial variables. The influence of outliers can be severe in logistic regression. To deal with this problem we smooth all financial variables by replacing all observations above the 99th percentile with that value. This reduces the impact of outliers in the estimation of the parameters of the model while it allows retaining all observatories in sample.
13. A third specification was also estimated with all the bank and country specific variables as
well as country (dummy) variables. The rationale for the inclusion of the dummy variable was to control for any differences not considered through our three country-specific variables. We do not report the results in the paper, as none of the country variables was statistically significant