26 while loan level impact in credit quantity is undone
by firm-level adjustments, the same is not true for other credit terms and conditions. As banks with real estate exposure become more willing to extend credit, there is greater competition fora given firm’s overall debt capacity. The competition results in borrowing
firms receiving softer, more favourable credit terms.
18
Moreover, columns (4) through (6) of Table IX show that firms borrowing from banks with greater real estate exposure
do not experience any differential change
in propensity to default, sales or number of employees. Hence, results suggest that despite of large effects at the bank- firm level, the crowding-out completely mitigates these effects for firm real outcomes in good times for firms with established banking relationships.
4. Conclusions Our results are interesting for macroprudential policy and theory. The results suggest that credit supply booms lead to modest positive real effects but to substantial higher bank risk- taking, including softer lending standards and new loans that are significantly more likely
to default during the downturn, thus credit expansions add fragility to the financial system. These results are in contrast with the crisis, with a sharp reversal in credit. Our framework also uses credit register data that are available inmost countries around the world (see e.g.
Djankov, McLiesh and Shleifer (2007)). Our framework is thus practical to implement and should help central banks to gain abetter understanding of the overall strength of the bank lending channel in the economy. This is even more important nowadays given the new macroprudential supervision powers for the Federal Reserve
and the European Central Bank, and also given the new bank capital and liquidity regulation (Basel III).
18
Note that because firm fixed effects do not significantly change the estimated coefficient of real estate exposure, the bias-corrected estimated coefficient is not different statistically.
In addition, all these results are robust to the other tests we have done for the loan-level analysis (Table III and IV. In addition, we do
not find significant effects, even for the smaller firms (see Table III of the Online Appendix. In the Online Appendix, we also instrument the credit granted by ex-ante real estate exposure in explaining real effects with similar results.
