Q2. What data do the authors use to tackle this issue?
To tackle that issue, the authors use country-level data on banking sector profitability and risk from the World Bank Financial Structure dataset (see Beck, Demirgüç-Kunt, and Levine, 2010).
Q3. What is the main challenge of using loan rejection rates to define risk taking?
A second challenge is that using loan rejection rates to define risk taking may be prone to a selection bias as applicant firms may be a systematically truncated sub-sample of all firms.
Q4. Why is the skewed industrial composition of localities dominated by banks?
This could be because the industrial composition in localities dominated by banks domiciled in countries with higher barriers to entry is skewed towards sectors that for technological reasons do not need much external finance.
Q5. What is the effect of the reduced monitoring ability on credit availability?
Given the organizational frictions associated with lending a la Stein (2002), this reduced monitoring ability could have a disproportional effect on credit availability.
Q6. What are the variables that are excluded from the rest of the exercises?
In terms of the exclusion restriction, the variables competition, subsidized, and corruption are included in this demand model, but excluded from the rest of the exercises.
Q7. What is the probability of a firm having a loan rejected?
In the latter case, the authors estimate that when self-selection is accounted for (column (6) of Panel B), an informationally opaque firm has a 9% lower probability of having its loan application rejected if it is dealing with banks at the 75th percentile instead of with banks at the 25th percentile of the sample home-country regulatory stringency.