Taking bank recapitalisation to its logical conclusion


India’s decisive public sector bank (PSB) recapitalisation plan has been met with a big thumbs-up by markets. The PSU Bank equity index jumped 30% on the news and, almost three weeks later, has held on to all those gains. And why not? In one fell swoop, the government committed more capital than it has cumulatively done over the past decade. Furthermore, this front-loaded commitment brings a much-needed coherence to the resolution process, with banks hopefully becoming more willing to take the large haircuts that are needed to effectively resolve impaired assets. Both for the decisive recapitalisation and the growing invocation of the bankruptcy law, policymakers need to be strongly commended.

But to appreciate the broader macroeconomic implications, one has to dig deeper. Has enough “resolution capital” been provided to adequately solve the problem? Has enough “growth capital” been provided to boost PSB lending—which is the whole point of the exercise? But, to what extent can PSB lending sustainably pick up, given migration to private banks and non-bank sources of funds? Also, is this a fiscal free lunch, as some believe? And, most importantly, how do we prevent moral hazard, and ensure that a decade from now, we are not back in the same boat?

Has enough ‘resolution capital’ been provided?

Prime facie, the government’s bank recapitalisation package is very large, at Rs2.11 trillion, or 1.3% of the gross domestic product (GDP). To be sure, some of this is expected to be raised by banks from the market (Rs58,000 crore). But even excluding that, the government’s injection is about Rs1.72 trillion (recapitalisation bonds, cash from the budget, interest on the recapitalisation bonds), or about 1% of GDP—more than cumulatively allocated over the past decade.

Surely, this should be enough, right? Well, it depends crucially on the size of the haircut. The Reserve Bank of India (RBI) estimates total PSB non-performing assets (NPAs) to be between 13.2-14.2% of advances under different stress scenarios by March 2018. Add to this some restructured loans turning into NPAs, and the total PSB NPA pool would be Rs8.7-9.3 trillion.

Assuming a 50% haircut during the resolution process, and adding the incremental capital needed to adhere to Basel III by April 2018 (which is estimated to be Rs40,000-60,000 crore), total capital requirements would be around Rs4.8-5.2 trillion. Public sector banks have provisioned about Rs2.7 trillion. Add to that the Rs1.7 trillion of “certain” resources from the latest package, and the total resource envelope extends to Rs4.4 trillion—Rs60,000 crore less than needed. But the residual can plausibly be expected to be raised from markets once the government provides bulk of the capital.

But things change dramatically if one simply increases the haircut from 50% to 60%. Now the capital shortfall (after including the recent package) is Rs1.2-1.7 trillion, which banks will likely struggle to raise from markets. So the quantum of the haircut is absolutely critical. Small changes there will have big implications for the adequacy of the committed capital.

Has sufficient ‘growth capital’ been allocated?

Thus far the focus has largely been on “resolution capital”. The key, however, is whether the current package will jump-start lending and credit growth (and therefore GDP growth)—which presumably is the whole point of the exercise—with PSB credit growth slowing to just 1.8% in 2016-17.

We remain cautious about an early pick-up of credit growth. First, some part of the problem is clearly demand and not supply, and demand will be unaffected by the recapitalisation. That said, some of it is clearly a supply constraint, evidenced by the fact that private sector bank credit growth has held up, while PSB credit growth has collapsed. But if PSB credit growth accelerates, do banks have enough “growth capital” to underpin that?

Even before the recent deleveraging, public-sector bank credit growth as a proportion of nominal GDP growth had slowed sharply, as private banks and non-bank sources gained market share. However, between 2010 and 2013, this ratio settled at 1. Over the subsequent three years, however, as impaired asset recognition accelerated and capital increasingly became a constraint, that ratio progressively fell all the way to 0.2 in 2016-17.

As the supply constraint is released, we would expect the ratio to mean-revert, but not back to unity, given the permanency of some of this recently migrated credit to other sources. So, one should not expect a boom in PSB credit growth. For the sake of argument, say it slowly mean-reverts to 0.6 and 0.8 in two years. Under our baseline of 11.5% nominal GDP growth, this would mean relatively-modest 7% credit growth in 2018-19 and 9% in 2019-20. But even to grow at these levels, additional capital of Rs75,000 crore would be required. Arguably, banks could raise some of this in markets and through internal profits, but those are still imponderables.

Adding this growth capital to the remaining resolution capital, the recent government commitment of Rs1.7 trillion—as large and decisive as it is—constitutes but 40-60% of what will eventually be needed, across different scenarios.

This is by no means to undermine the existing package. Policymakers need much credit for that. Instead, the recent capital injection should be thought of more as the “end of the beginning” rather than the “beginning of the end”. Significantly more resources may well be needed to take this process to its logical conclusion.

What are the medium-term fiscal implications?

There is a perception in some quarters that, fiscally, this is a free lunch. But there’s never a free lunch. To the extent that debt levels go up, the present discounted value of future tax liabilities rises in tandem, reflecting but an inter-generational transfer. By issuing bonds, the government is simply smoothing the fiscal cost inter-temporally. How then should we measure the fiscal cost? Think of how much tighter future deficits would need to be to bring debt levels back to what would have existed absent the recapitalisation bonds.

Consolidated public debt is currently around 68% of GDP and, under our baseline growth, interest, and fiscal deficit assumptions, public debt/GDP will gradually decline to 63% by 2023. If the government were to provide sufficient resolution and growth capital (eventually about 2% of GDP, in our view), debt/GDP would be about 64.5% in 2023. To reduce this to baseline levels (63%), the consolidated annual fiscal deficit would need to be 0.3% lower for the next five years. This should be thought of as the true fiscal cost. Furthermore, if the government were to accept the Fiscal Responsibility and Budget Management (FRBM) review committee’s recommendation to bring debt/GDP to 60% in 2023, the combined deficit would need to be lower by about 0.8% of GDP for the next five years, which appears very ambitious.

The key, however, is growth. Under the more comprehensive plan of providing both resolution and growth capital, it’s more likely that GDP growth will pick up. If nominal GDP growth picks up by just 0.5% every year for the next five years, the bank recapitalisation plan essentially pays for itself. No additional fiscal tightening is needed, and debt/GDP would converge to 63% by 2023. The key, therefore, is to provide sufficient capital for credit and nominal GDP growth to lift, but controlling for the moral hazard risks discussed below.

How should the recapitalisation bonds be designed?

The near-term objective, however, will be to place the recapitalisation bonds as non-disruptively as possible. Already, benchmark bond yields have hardened by 40 basis points in two months as global yields have risen, oil prices are firming, and some fiscal slippage has been priced in. The risk is when the Rs1.35 trillion of bonds are placed with banks, there would be a portfolio- substitution effect: the incremental demand for other bonds could fall. To minimize this, the recapitalisation bonds should be designed to be as imperfectly substitutable to other bonds as possible, on all dimensions: liquidity, statutory liquidity ratio (SLR) status, and tradability.

If the bonds were non-SLR and non-RBI-repo-able, they are unlikely to cannibalize demand for government or state bonds, but could still crowd out corporate bonds. Policymakers may therefore want to apply some temporary trading restrictions, as Indonesia did in its recapitalisation programme of 1998, initially allowing only 10% of the bonds to be traded, and progressively liberalizing that fraction.

Of bank recapitalisation, reform and riders

As demonstrated above, the key is to go the whole hog: to provide enough resolution and growth capital to lift GDP, such that the plan pays for itself. A large fiscal injection but not enough to induce stronger growth, will leave us neither here nor there.

But precisely because so much public money needs to be committed, it’s imperative that the recapitalisation is accompanied by reform, to prevent moral hazard. The first-best solution would be to reduce the government’s share below 50%. If that’s not feasible, then (i) there must be a meritocratic element to the allocation of capital, in conjunction with (ii)implementing the PJ Nayak Committee recommendations, to improve governance structures, and credit and risk management practices. Without this, policymakers are orchestrating a transfer of resources from households to firms, which is likely to keep recurring in the future.

Conversely, it’s equally important that the recapitalisation bonds don’t come with any riders about who to lend to. More “directed credit” will simply result in more impaired assets and vulnerabilities down the line. Instead, the allocation of capital needs to be adequate, and credit practices improved, such that allocative efficiency is achieved through the market mechanism.