Perfomance anxiety

Under Article IV of the IMF’s Articles of Agreement, a team of Fund staff visits each member country once a year, collects economic and financial information, and holds extensive discussions with officials on policy matters.  This is then published in its website.  The latest Article IV report for Bangladesh came out in September, stating that Failure to effectively address the problems in the banking system, including high non-performing loans pose a medium likelihood risk to the economy, with a medium-to-high impact in the near term if it hit — High and increasing non-performing loans and low capital adequacy would hamper the banking sector’s ability to finance business investment, add fiscal burden, and hamper growth.

Let’s unpack this.  In doing so, we are going to look at official data.  Yes, there is considerable scepticism about the veracity of official figures.  But official data is all there is to go on, and nihilism of trust nothing but one’s gut instincts is not analysis. As it happens, even official data tell a potentially scary story.

Non-performing loans are those loans where the borrower has failed to make interest or principal payments for an extended period of time, typically 90 days.  Obviously, the higher these loans are relative to the total amount of loan a bank (or the banking system collectively) has made, the worse things are.  This chart, from the Article IV report, shows that about 10% of total loans made by Bangladeshi banks aren’t performing.

The first question to ask is, is 10% NPL high?  One way to answer that is to look at what the ratio in other similar countries. The next chart, based on research by Junkyu Lee and Peter Rosenkranz of the ADB, shows that it is much higher than Southeast Asian countries, though maybe not so compared with India (where it has been rising fast) and Pakistan (where things appear to be improving).

Things are, however, more complicated than the headline ratio suggests.  Firstly, the problem appears to be far worse in the state-owned commercial banks — Sonali, Janata, Pubali, and Agrani — where 30 per cent of loans weren’t performing at the end of 2018.  A related problem is that these state banks are dangerously undercapitalised — that is, they don’t have enough capital to cover their risks.  This is shown by the next chart — from the Article IV — of ‘capital adequacy ratio’, a measure of how well placed a bank is to cover its risk: according to internationally accepted Basel III, this ratio should be at least 8%.

More importantly, the problem is potentially worse because rescheduling and restructuring are dramatically on the rise.  Stress assets — that is, the sum of non-performing as well as restructured and rescheduled loans — went from less than 15% of total loans in 2014 to over 20% in 2018.  Further, the rescheduling and restructuring are happening not just in the state owned banks, but also private commercial banks — see the next chart, again from the IMF report.

The high level of stressed assets limits banks’ ability provide fresh lending for new projects, dampening economic activity and income.  And today’s rescheduled or restructured could well be tomorrow’s non-performing loan.

So, what’s causing the banks’ troubles?  And if the stressed assets did become non-performing, how bad could the hit on the economy be?  The ADB economists’ paper can suggest some answer on both counts.

Studying 165 commercial banks in 17 emerging market countries in Asia over two decades, Lee and Rosenkranz find that the most important reason for the rise in non-performing loans are usually macroeconomic — that is, banks find their borrowers are unable to repay loans when the economy is in a recession or is faced with adverse shocks such as sudden spike in inflation.  But according to official data, Bangladesh is not in macroeconomic difficulty.  To quote the Fund — Despite solid economic performance, the NPL ratio remains high, particularly for SOCBs, and the amount of restructured and rescheduled loans continues to increase.

If the answer is not macroeconomic, is it microeconomic? That is, could it be that the banks have been lending to poor borrowers?  For example, undercapitalised banks tend to find themselves lending to more risky borrowers in the hope earning higher yields.  The ADB economists do find that the low-capitalised banks tend to have higher NPLs — consistent with the Bangladeshi situation.

Digging yet deeper, maybe the state banks were just poorly managed — not a difficult thing to imagine in the context of, say, Janata Bank, is it?  This could manifest in two ways: to save costs, banks cut their monitoring function, and inadvertently makes things worse for themselves; or they engage in excess lending — in the Bangladeshi context, this second possibility doesn’t sound far fetched at all.

As for what might happen if the stress assets become NPL, the authors’ result imply that if the NPL ratio rose from 10% to 20% over three years, GDP growth would slow by about 1 percentage point, all else being equal.

That is, even by official data, there is a clear problem with the banks.

So, what should be done?  The good news is that there are technocratic solutions to these problems — tightening the criteria for rescheduling / restructuring of loans, strengthening the banks’ corporate governance, enforcing bankruptcy laws and such like.  Coupled with selected recapitalisation of banks, these measures can de-risk the banking sector without causing a hit to the economy.

The bad news is that these technocratic solutions require technocrats — regulators and judiciary — to work independently, transparently, and with credibility.  None of which is present in today’s Bangladesh.