Among important reasons that lead to a banking crisis, the inherent opacity of banks is one of the most prominent. Several influential research papers have shown that it is extremely hard to assess the true fundamentals of a bank by looking only at the bank balance sheet. This is because unlike a regular business, banks can easily transform the nature of an asset or a liability. In this article, based on some of my own research and those of others, I suggest that banks should be asked to disclose the economic situation of their large borrowers.
It is no secret that banks the world over ever-green loans. In simple terms, ever-greening is a process where an in-spirit delinquent loan is shown as current by showing a fictitious repayment and issue of a new loan. Let me illustrate with a hypothetical example. Assume that a borrower A borrowed Rs 100 crore from a bank B. Also assume that A’s business suffers heavy losses, and hence, he/she cannot repay the loan. In many such cases, the bank B, instead of recovering whatever is possible and writing off the part not recoverable, gives a new loan, probably of higher value, and borrower B repays his/her original loan using the proceeds of this new loan. Of course, this is a simplified version but captures the spirit of ever-greening.
It is easy to see that, on paper, the bank appears healthy as the loan is not recorded as an NPA. To add insult to injury, the bank shows higher credit growth because of the fictitious loan. However, when exposed to severe economic crisis where the bank cannot continue renewing dubious loans or a thorough audit, the practice gets exposed. Unfortunately, markets have repeatedly failed to recognise this practice in advance and discipline the banks by bringing down their share prices. Markets, in most cases, act as enablers of ever-greening by rewarding banks that indulge in ever-greening with rich valuations. However, when the deeds get exposed, the stock prices collapse, depositors run for cover, regulators step in, and the whole cycle repeats leading to another round of ever-greening. The above description fits in well with what happened in India after the 2008 global financial crisis.
In a recent paper titled “Do measure of bank audit quality really measure audit quality?” (Abraham and Tantri (2019)), Jefferson Abraham of the London Business School and I, argue that an audit of bank books is unlikely to detect ever-greening practices. We suggest a way of keeping a tab on banks. The idea is to burrow a layer below the bank books and examine the health of the borrowers. Using the natural experiment provided by the asset quality review, we estimate a model that predicts the likely bank NPAs based on the average financial health of the borrowers.
The underlying idea is very simple: It is unlikely that a bank could have low NPAs when many of its large borrowers are under distress. Consider a bank that reports close to 0% NPAs when a large chunk of its borrowers do not have enough cash flows to meet their loan obligations. It is likely that the low NPA levels are engineered by ever-greening. Between 2011 and 2014, we had high credit growth and low NPAs even when the underlying economic growth was slow. It appears that most of the high credit growth was driven by ever-greening. Our measure flags such cases in advance.
As per our measure, the most important variables that matter include (i) interest coverage ratio; (ii) medium term profitability; (iii) quality of auditors; (iv) volume of related party transactions; and (v) credit ratings of the borrowers. Using machine learning algorithms and more granular data, it is possible to estimate a more sophisticated model. However, data availability is a big challenge for researchers like us.
The findings have two important policy implications. First, banks should be asked to report changes in key financial variables of their large borrowers. This will enable investors and depositors to take action before it is too late. An adverse stock price reaction could act as a credible deterrent. In the absence of this information, unfortunately, markets fail to do the surveillance job. Second, using a supervised machine learning program, it is possible to estimate the actual relationship between borrower fundamentals and bank fundamentals. The regulator should regularly calculate the expected NPA from the algorithm and ask the banks to explain significant deviations. If this measure is implemented, the regulators need not resort to draconian measures such as the AQR (asset quality review). This stress test information need not be made public.
It is possible that, in line with the Lucas critique, the banks and firms will collude and manage the numbers together. While I cannot completely rule out such a possibility, I believe that collusion is practically difficult as there are a number of third-party sources, including the GST returns and audits available to track firms.
It is also important to invest more efforts in designing early warning measures of ever-greening. In my paper titled “Identifying Zombie Lending: Evidence Using Loan Level Data”, I attempt to show that a loan that is renewed quickly just before the due date by the same officers who lent the initial loan is likely to be an ever-greened loan. If given data and resources, researchers can design better methods. I hope that the right lessons are learnt from this crisis and regulators push for more transparency in bank balance sheets.
The author teaches at the Indian School of Business (ISB)