Making sense of RBI's capital buffers and dividends

In the ongoing Indian government bond market soap opera, one subplot relates to RBI’s dividend payment to the government. After three years of paying out all of its net income as dividend, for fiscal year 2016-17 (FY17), the RBI board retained INR 131.9B of its income for its own books.

Given current fiscal stress, this led to murmurs, including suggestions that RBI should now pay an ad-hoc dividend to the government. Any hasty roll back of this kind, without debate and consensus, would severely risk RBI’s credibility.

That said, we do need to debate the appropriate level of capital buffers on RBI’s balance sheet. There has been work done on this – by the Subrahmanyam Group (1997), Usha Thorat Group (2004) and the Malegam Committee (2013). More needs to be done.

Central Bank balance sheets and income statements can be tricky to grasp. In addition, this issue can excite touchy, ideological arguments about government credibility and RBI independence. This is not the easiest topic for an informed, dispassionate debate.

There are two opposing viewpoints. Conservatives contend that as of end FY17, RBI had retained earnings buffer of only 8% of its balance sheet, and that we actually need higher true buffer (the Subrahmanyam group suggested 12%). The opposing Liberal viewpoint, described in the Economic Survey of FY16, is that RBI holds wastefully high total capital reserves of 26% of its large balance sheet (amongst the highest in the world), denying the country scarce, investible funds. Paradoxically, both data points are correct, and both conservatives and liberals would be wrong to ignore the other.

As I hope to show in this note, given its sustainable income stream, existing reserves and “right-way” risk, RBI had more than ample capital buffer against reasonable risk as of end FY17. In fact, it is difficult to understand why RBI needed higher retained earnings for this fiscal.

Under extreme stress events, any buffer can seem inadequate. If we don’t want the RBI to ever approach the government for funds, RBI needs very large buffers. However, there are better ways of protecting RBI while reducing capital wastage.

Understanding RBI’s balance sheet & income

Figure 1 provides a snapshot of RBI’s balance sheet as of end FY17.

Without commercial intent, RBI has an enviable, regular income stream. It borrows cheap money from us, and deploys it in high quality foreign currency (FCY) and INR assets.

What are its sources of funds? Every INR currency note in circulation represents interest-free money borrowed by RBI. Also, commercial banks park their Cash Reserve Ratio (CRR) with RBI. Further, any surplus liquidity of the banking system and government is placed with RBI. As of end FY17, RBI had INR 24T from these three sources, of which INR 20T (83%) was interest free.

RBI broadly deploys these funds into two areas – INR 25T equivalent of FCY assets (including gold), and INR 8T of INR assets, largely Indian Government Bonds (IGBs). RBI’s total assets of INR 33T are comparable to that of SBI.

Given the above operating model, RBI earns a healthy, annuity net interest margin. Even in FY17, when interest-free currency notes were converted into interest bearing reverse repo post demonetization, RBI earned INR 619B of net interest income.  While we will consider risks, this annuity revenue stream is a source of overall risk comfort.

Understanding Revaluation Reserves

In figure 1, the difference between cash sources (INR 24T) and deployment (INR 33T) largely represents (i) FCY revaluation reserves of INR 5.3T (ii) INR bond revaluation reserves of INR 0.6T and (iii) retained earnings of INR 2.6T.

Of these, retained earnings (technically, Contingency Fund – CF, and Asset Development Fund - ADF) are the only ones that have been banked and have passed through the RBI income statement.

Revaluation reserves (i) and (ii) are unrealized, volatile, and arise (or deplete) when RBI’s assets are restated at current market prices.

FCY Revaluation Reserve

As INR depreciates, the INR equivalent of RBI’s FCY assets – deposits and securities - increases. The FCY revaluation reserve of INR 5.3T against FCY assets of INR 25T indicates that RBI purchased these assets at an average 21% lower INR cost. So if all the FCY were hypothetically in USD, against a USDINR of 64.50 as of end FY17, RBI’s average USD purchase cost would be 51.00.

What if USDINR dropped overnight to say 40.00, and wiped out all of RBI’s FX revaluation reserves and some more? Even if this extreme event were to take place, this feels like a right-way risk – RBI would be short of capital at a time when INR is extremely strong, and hopefully the country’s degrees of freedom are larger, than say if USDINR was to suddenly rise up to 80.00.

INR Bond Revaluation Reserve

The INR bond reserve of INR 0.6B, or 7.5% of INR assets of INR 8B, indicates that the RBI on average purchased INR bonds at higher yields. Assuming all of RBI’s holdings were in 10-year IGB, against the 6.50% yield as of end FY17, the holding yield of RBI would be 7.50%. Much of these reserves would have depleted now, given the increase in yields since end FY17 – highlighting their volatile nature.

What if yields shot up to 8.50%, and created a hole in the revaluation reserve? Again, this feels like a right way risk. As an INR liquidity provider and market stabilizer, RBI would generally buy IGB when yields are high, and sell them when yields are low – witness RBI selling through OMO in the current fiscal. RBI has captive low-cost liabilities to fund the bonds. Further, when IGB is held to maturity, all bond revaluation reserve would fall away – indicating we could look through INR bond revaluation reserve fluctuations.

Reconciling Conservative & Liberal viewpoints on buffers

The conservatives are right in that revaluation reserves are unbanked, and should not be paid out as dividend to the government.

However, the extent of available revaluation reserves should be an input towards determining the level of retained earnings. After all, the biggest risk that the RBI faces is from possible adverse valuation of its assets. If RBI has ample revaluation reserves, it needs fewer buffers of retained earnings.

In this context, it is interesting to see how the Subrahmanyam group (1997) arrived at the 12% number for retained earnings – the magical number that conservatives quote.

The group suggested:

(i)            5% for losses from FX and INR intervention and operations, and depreciation of INR/ FCY securities

(ii)          5% for FCY and gold revaluation shocks – over and above the 5% revaluation reserve that existed in 1997

(iii)         An additional 2% for systemic risks & developmental activities – essentially a buffer on a buffer

The risks (i) and (ii) highlighted by the Subrahmanyam committee are clearly mitigated by extant FCY and INR revaluation reserves, besides its revenue stream.

Note that the group actually considered the then existing 5% FCY revaluation reserves while seeking the additional 5% for FCY revaluation risk – so in effect, they called for 17% total reserves including revaluation reserves. However, somewhere, this morphed into a standalone 12% retained earnings goal.

The Usha Thorat group (2004), likewise, called for a total buffer of 18% of balance sheet, across retained earnings and revaluation reserves. However, the RBI continued to work towards 12% standalone retained earnings.

After reviewing the work done by the prior two groups, the Malegam Committee (2013) recommended retained earnings should continue to be enhanced, but the exact amount to be retained each year was left to management to decide.

Subsequently, for three years in a row under Dr. Rajan, RBI transferred all of its net income to the government. Given RBI had high level of revaluation buffers in each of those years there was little need to enhance retained earnings.

In contrast, the FY17 RBI decision to increase retained earnings (CF) by 40bps from 7.6% to 8.0%, ignoring high extant 18% of revaluation reserves looks questionable.

Some normal risks, and some abnormal ones

Does the work done by the previous groups, years ago, adequately look at all possible risks? After all, some of the percentages suggested there look suspiciously like back of the envelope round numbers.

Using the Basel framework, let’s consider some “normal” commercial bank type risks, and then address some truly stress events.

Given RBI carries very low credit risks, and has ample capital reserves, it scores very highly on traditional Basel credit and leverage capital buffers.

RBI officials often point to an increased cost of funds from banking liquidity surpluses as a risk. Given reverse repo is only against IGB held by the RBI, it is very unlikely higher reverse repo amounts will lead to a negative net interest margin. The demonetization exercise was a good illustration – it increased RBI’s cost of funds, but the RBI still produced a healthy net interest margin.

What if we faced some really unthinkable stress events? After all, RBI is far beyond a commercial bank, and its job is to worry for all of us. What if the USA froze our USD assets overnight? What if a currency that we held suddenly dropped to zero value on account of an unthinkable geopolitical event?

Stretched to these extremes, even 26% of total reserves would be insufficient to protect the RBI’s balance sheet. However, should the RBI hold buffers to cater for such extreme events?

If we wish the RBI to be financially independent of the government even through times like these, there will be no end to the level of buffers needed. The cost of buffers is that the country is denied scare, clean, investible earnings. RBI reserves are lent back to the government, but they transform the country’s equity into debt, and that impacts our capacity to invest. We need better ways of protecting RBI’s independence, than through costly super-inflated buffers.

Summary

While conservatives are right to insist that no dividend should be paid out of unbanked revaluation reserves, changes to retained earnings reserves should not ignore the extent of revaluation reserves available. RBI’s decision to marginally enhance retained earnings in FY17, when 18% of revaluation reserves were available, does not look right.

We should move away from a standalone retained earnings goal, to an overall minimum level of total capital, including revaluation reserves.

If our total level of buffers was significantly above this threshold, we could consider reducing the level of retained earnings. On the flip side, if total reserves dropped below the threshold, the government would need to pump in capital – no questions asked.

For this to happen, the government needs to put in place a credible framework that protects RBI’s financial independence. This cannot be a one-way street for the government to take out money from RBI (and eye its substantial INR 2.6T of retained earnings), and demur when the direction of capital flow needs to be the other way around.