Page 242 - ES 2020-21_Volume-1-2 [28-01-21]

Regulatory Forbearance: An Emergency Medicine, Not Staple Diet! 225

The inadequate clean-up of bank balance sheets
7.33 In reality, the AQR exercise significantly under-estimated the full extent of NPAs as well
as the resultant capital infusion that was required to ensure that the bank balance sheets indeed
become healthy. In terms of additional (gross) NPAs, public sector banks added about INR
5.65 lakh crores from FY2016 to the end of FY2019. To put this amount in perspective, the
5
additional NPAs translated to about 7.9% of the total tax revenue over this period.
7.34 To be sure, the AQR did lead to some clean-up of the toxicity in the bank balance sheets.
Figure 22 plots the accumulated proportion of restructured loans reported by the banks (FY2009
to FY2015) against their NPA divergence during the AQR regime (FY2016 to FY2019). NPA
divergence is the difference in banks’ reported NPA numbers and the NPA assessed by the
RBI, taken as a proportion of total loans. A positive correlation indicates that the AQR did
identify some bad loans lent through restructuring activities. However, the effectiveness of the
AQR exercise cannot be assessed from such a simple positive association, especially given the
statistically low correlation (0.45). Also, notice that most banks are found to lie below the 45°
line. This means that, in most cases, the identified NPAs were smaller in comparison to the loans
restructured by the bank.

7.35 Kashyap, Mahapatro, and Tantri (2020) argue that the AQR was mostly restricted
to targeting bad lending through restructuring, rather than identifying subtle ever-greening
activities. Notice that loan restructuring warrants a disclosure whereas fresh lending does not.
Therefore, rather than restructuring, banks could have easily lent a new loan to an existing
borrower on the verge of default. To further camouflage their incentives, they could have
disguised the payment in the form of fresh lending to a network of related parties of the actual
firm in distress. Figures 23 and 24 plot the accumulated proportion of lending (FY2009 to
FY2015) in the form of these two subtle ever-greening possibilities against the gross NPA
divergence disclosed by banks. Figure 23 plots the proportion of direct lending to borrowers
with interest coverage less than one while figure 24 plots lending activity for such borrowers
through the channel of related parties. The extremely low correlations (which is, in fact,
marginally negative in one case) between divergence and ever-greening measures signify that
the AQR exercise failed to recognize subtle ever-greening and thus may have been unable to curb
distortionary lending. Almost all banks in the two figures lie below the 45° line which further
indicates the nominal extent of ever-greening unearthed during the exercise. Interestingly, Yes
Bank, which was recently rescued by the RBI, stands as an outlier in both these graphs. While
the bank’s divergence was just around 5%, its ever-greening indicators were as high as 38%-
52%. The sharp rise in the bank’s reported NPAs (0.75% in FY2016 to 16.8% in FY2020)
seems unsurprising from this analysis.

5 The only mention of AQR is in the Financial Stability Report of June 2016, which mentions “The gross
non-performing advances (GNPAs) ratio increased sharply to 7.6 per cent from 5.1 per cent between
September 2015 and March 2016, largely reflecting reclassification of restructured standard advances
as non-performing due to asset quality review (AQR).” Clearly, the report gives no details about the
assumptions involved or the procedure followed in the exercise.