Page 222 - ES 2020-21_Volume-1-2 [28-01-21]

Regulatory Forbearance: An Emergency Medicine, Not Staple Diet! 205

7.3 During the period of the global financial crisis (GFC), the policy worked well with
banks selecting genuinely distressed, but viable in the long-run, borrowers for restructuring.
Box 3 explains the careful panel regressions undertaken to control for various confounding
factors. The results show that, during the GFC, banks more likely to benefit from forbearance
do not differ in their selection of restructuring choices when compared to a bank with a
lower likelihood of utilizing forbearance. The propensity to restructure any given borrower,
including unviable ones is, however, significantly higher in the years after the crisis.
Evidently, once the banks got a signal about the continuation of forbearance despite the
economic recovery, several types of distortions crept in. As pointed out earlier, emergency
medicine indeed became a staple diet. For instance, figures 6 and 7 show that the proportions
of loans restructured increased significantly during this period. The share of restructured loans
increased from 0.74% in FY2008 to 6.94% in FY2015, as shown in figure 6. The increase in
the share of restructured loans among public sector banks was much higher, from 0.82% to
8.49%. However, the private sector banks also saw their share of restructured loans increase
from 0.64% to 2.87%. On the contrary, as shown in figure 8, the reported gross NPAs of banks
increased only modestly from 2.2% in FY2008 to 4.3% in FY2015. It appears that the banks
used the option of restructuring loans that were on the verge of defaulting without regard to
the viability of such loans, as shown subsequently in Section 8.27. During the forbearance
window, the proportion of firms in default increased by 51% after their loan(s) got restructured.
In the pre-forbearance era, there was only a marginal 6% increase in the likelihood of defaults
after restructuring. Forbearance thus helped banks to hide a lot of bad loans.

Box 3: Difference-in-Difference Framework to Show
Distortion in Banks’ Incentives
Mannil, Nishesh, and Tantri (2020) use a difference-in-difference methodology to test whether
and when forbearance induces lending distortions among banks. This strategy estimates the
lending activity by a bank in the counterfactual scenario of forbearance not being available.
The difference in the actual lending activity in the presence of forbearance and the one under
this counterfactual scenario is therefore a consequence of forbearance.

To this end, banks are classified into two separate groups - B1 and B2, such that the two types
of banks are similar in all respects except for their susceptibility to exploiting the forbearance
policies. Precisely, banks in B1 have a higher proportion of borrowers that are adversely hit
by the crisis. This naturally increases the likelihood of B1, relatively, exploiting the use of
regulatory forbearance measures. By a thorough comparison of attributes such as ownership,
capital, NPA, and age, Mannil, Nishesh, and Tantri (2020) show that the two categories of
banks thus formed are otherwise similar.
Banks in B1, o n average, would have a higher share of loans restructured during the crisis
on account of the higher shock faced by their borrowers. Therefore, a comparison of B1 and
B2 on aggregate outcomes would not be appropriate to understand the distortions forbearance
can induce. Subsequently, Mannil, Nishesh, and Tantri (2020) use a firm-level fixed effects
estimate that compares outcomes within a given firm and between the two types of banks.
If for the same firm, banks in B1 exhibit a higher restructuring activity during forbearance,
it implies that these banks, on average, are relatively less prudent in selecting cases for
restructuring and are likely to restructure even unviable projects. Because B1 and B2 are