Thursday, December 14, 2006

What to do with $170 billion of reserves

Jaimini Bhagwati has a great article in today's Business Standard about the poor returns on India's foreign exchange reserves portfolio. In it, he says:

According to the Reserve Bank of India?s Annual Report for 2005-06, the nominal rupee (INR) denominated rates of return on India's foreign currency assets (FCA) for 2005-06 and 2004-05 were 3.9% and 3.1%, respectively. Foreign currency assets include foreign exchange reserves less gold holdings, special drawing rights and India's reserve position in the IMF. Inflation in India in the last one year has been about 5%. Therefore, the real rate of return on India's FCA for 2005-06 was around minus 1.1% (3.9-5).

The Reserve Bank of India (RBI) is currently responsible for managing the country's foreign exchange (FX) reserves. Since central banks hold gold as a last measure of protection against a balance of payments crisis, the discussion in this article is confined to FCA management. The real INR rate of return on India's FCA has been negligible to negative in the last two years. These low returns on India's FCA could be attributed to the RBI's cautious policies in which the guiding principles are to maintain mark-to-market value and liquidity by taking minimal credit and market risk. Effectively, the rate of return is a secondary concern and the RBI's options are accordingly limited to investing in short-dated triple-A rated government debt securities.

How should we measure the performance of the Indian FCA portfolio? For instance, should the numeraire currency be the dollar or INR? The currency in which the rate of return is measured should not matter in economic terms. It is the currency composition of the benchmark portfolio, against which the FCA portfolio's performance is measured, which needs to be calibrated carefully. Given the potential for disorderly US$ depreciation the benchmark portfolio may need to be tweaked frequently to take into account the exposure of the Indian current account to exchange rate risk.

A traditional yardstick to assess the sustainability of a sovereign's external borrowings is to compare the average cost of such borrowings with the GDP growth rate. By extension, India's FCA earnings should be comparable with the GDP growth rate. In the last one year, real growth has been 8% plus, making the GDP growth rate 9% [8 - (-1.1)] more than FCA earnings. Another construct, suggested by Dani Rodrik, is to compare FCA earnings with the cost of short-term external commercial borrowings (ECBs). At the margin, Indian FCA earnings would be about the same as the yield on 3-month US Treasury bills, currently about three-month LIBOR - 0.40%. The average cost of short-term ECBs for Indian firms is about 3-month LIBOR + 2.5%. That is, the opportunity cost of holding "excess" FCA would be at least about 3%.

Of the Asian countries which have accumulated significant volumes of FX reserves, Singapore has allocated the responsibility of managing reserves to maintain the stability of the Singapore dollar to the Monetary Authority of Singapore (MAS) and excess reserves have been assigned to the Government of Singapore Investment Corporation (GIC), which manages higher risk-return investment portfolios. More recently, South Korea has followed the same model and set up the Korean Investment Corporation.

Some questions about Indian FCA management need answers. Could the returns be increased significantly without undue risk to the country's ability to service its external debt and sustain current account deficits required for investment purposes? Is the allocation of Indian FCA management to the RBI conducive to a satisfactory rate of return? Further, if the mandate for FCA management is altered would that necessarily result in a higher rate of return?

Obviously, there are no definitive answers to these questions. In the context of adequacy of FX reserves, Guidotti-Greenspan have suggested that these should be at least equal to external debt maturing in the next one year. As of end March 2006, India's short-term external debt plus long-term debt maturing in a year was about $15 billion. On the same date, non-resident Indian (NRI) deposits amounted to approximately $35 billion and external commercial borrowings (ECBs) totalled about $26 billion. The market value of foreign institutional investors' (FII) portfolio investments in Indian equity markets is around $110 billion. That is, total external debt maturing in a year plus 50% of NRI deposits and ECBs and 25% of FII portfolio investments add up to about $70 billion.

How should the RBI maintain an adequate level of reserves for external debt servicing and other requirements while it manages excess FCA separately to maximise returns within acceptable credit and market risk limits? As of mid November 2006, Indian FX reserves amounted to about $170 billion. One option is for the RBI to manage a highly liquid, low-return portfolio of $70 billion as insurance against capital flight caused by unexpected market developments. The remaining $100 billion could also be managed by the RBI in separate portfolios with associated benchmarks which carry greater market risk and hence commensurately higher returns. This proposal for several portfolios with differing risk-return characteristics is not necessarily to take additional credit risk but diversified market risk. For instance, FCA managers could move up the yield curve and take some interest rate risk and invest in highly-rated municipal and asset-backed securities. They could also diversify out of fixed-income instruments into asset categories such as equity and real estate. Singapore and South Korea have set up separate investment corporations to manage their growing reserves because the skills required for managing investments in equities, asset-backed securities, and real estate are qualitatively different from those needed for managing portfolios consisting of short-maturity government-debt securities.

Lawrence Summers, speaking at an LK Jha memorial lecture in Mumbai on March 24, 2006, remarked that India could expect to earn about 6% in real terms on its FCA if these were invested in global capital markets. On balance, it appears that it should be possible to earn an additional 5%, compared to current FCA earnings by investing excess Indian FCA through a dedicated investment platform set up by the RBI/Ministry of Finance. An additional 5% return on $100 billion is $5 billion, which is about 0.6% of GDP. This number would be lower if the investment guidelines for the higher risk-return reserves portfolios were to be restrictive but it is still likely to be a significant figure. All things considered, it is time for the RBI and the Ministry of Finance to set up a separate investment firm or platform with appropriate performance benchmarks and incentives for the staff.

In my understanding, the quasi-fiscal costs of holding reserves in this fashion come in two parts. First, there is the well known opportunity cost of being invested in a low return asset (US government bonds or USD denominated bank accounts). As a country, India is paying higher rates for the equity/debt capital coming into the country, but earning low rates for the USD assets held by RBI - doesn't sound like a very good deal. More narrowly, if you focus on the consolidated government of India, there has been a gap between the high price at which GOI borrows on the bond market versus the low returns obtained on the reserves portfolio - this is a fiscal cost to GOI. This issue has been known in the literature for a while.

The second component, in my opinion, is more important: this is the losses that come about in the reserves portfolio when the currency adjusts. Suppose RBI tries to block an INR appreciation by purchasing USD, at a time when the exchange rate is Rs.46/$. This involves buying $100 for Rs.4600. Suppose you believe that you are in a situation where the central banks can only slow down the inevitable market process but not block it. So, if an appreciation was coming, it would come anyway - RBI is unable to prevent the inevitable, only delay it. In this case, after some period of time, you end up at Rs.43/$. Now the RBI portfolio is valued at Rs.4300. In other words, a loss has taken place from Rs.4600 to Rs.4300.

Some of this difficulty goes away by keeping score in USD. If RBI holds 25% of the reserves portfolio in EUR, and keeps score in USD, then the returns on the reserves portfolio as measured in USD looks great when the USD depreciates. I think this is illusory. I think it's safest to think in INR. In the Indian case, there are two manifestations of the fiscal costs of reserves. First, the `market stabilisation scheme' (MSS) was a very nice thing in converting some of the hidden cost of reserves into a line item in the budget. I think it's wrong to waste money on holding reserves, but if you must do so, I think it's better to be transparent about it. In addition, if you look at the time-series of the dividend paid by RBI to GOI over the years, expressed as percent of GDP, there's been a sharp drop. This drop has come about largely because of the growth of reserves.

Jaimini asks: Can we do better than RBI's management of the reserves portfolio? I'm sure we can. Jaimini is right in saying that the improved performance could amount to as much as 1% of GDP, which is a huge number.

I think it's equally important to ask: Do we need reserves beyond $70 billion? Does it make sense for any bureaucrat to hold a portfolio of 20% of GDP? I think it is preferable to not hold such assets in the public sector in the first place; it is better for the citizens of the country to hold globally diversified portfolios rather than placing these in the hands of a government agency, regardless of whether it's an RBI-style agency which earns low returns or an ADIA-style agency which does better. As an example, consider this Economic Times article titled Indian Temasek Challenger to be Launched (an inaccurate title). I think Indian politics does not mix well with such an entity. We are better off without it; we are better off with a government which just sticks to public goods and stays out of the management of the stock of assets of the citizens of the nation.

In 2003 and 2004, when India was building up reserves at a huge pace, Ila Patnaik wrote an excellent group of articles in Business Standard warning about these kinds of issues. In particular, see The USD quagmire, Dining with the devil and Feeding an elephant. This sequence of articles ended in March 2004 when India got off that tightly-pegged exchange rate. On the subject of `how much reserves is adequate', see her India's policy stance on reserves and the currency.

The only saving grace about the Indian story is that things are better here when compared with East Asia. The reserves managers in East Asia must feel terrible, holding gigantic USD assets that they dare not sell, while week after week they endure the agony of watching the USD drop.

It's been the best deal imaginable for the US: buy cheap Chinese goods through a distorted exchange rate, pay for them using IOUs, and then have those IOUs depreciate in value so they don't even have to be paid back in full. The Chinese get hurt twice: first when selling cheap goods, and second when suffering losses on the USD portfolio. The same double-whammy hurt India also, but on a much smaller scale.


  1. Eshwar Prasad and Raghuram Rajan came up with a proposal last year where the basic idea was to "securitize" the capital inflows that an economy like India faces..
    It basically entails the central bank licensing fund managers who can issue shares in INR to Indian investors (like you and me) and purchase USD from RBI and use it to invest in a global portfolio of stocks, bonds and other assets..The main benefits they point out are two fold..One, investors get the benefit of global diversification and two, the RBI being able to control the timing and quantum of outflows/ running down of its reserves over the minimum required limit...
    (Link to the proposal can be found here:

    What do you think of such a proposal especially when compared to the Temasek/Singapore GIC model??It does have a few messy things like licensing fund managers (which may also involve setting quotas for each of them, preventing free competition etc.), but isn't it a lot more feasible politically while at the same time allowing RBI to follow a gradual/cautious approach which it is prone to do anyway??

  2. Some responses on the Prasad and Rajan proposal. A useful way of
    describing the proposal, or of selling it, would be to say something as follows. Under the proposal, we'd be scooping up capital inflows when a surge comes in, and smearing them over a predictable range of dates in the future. Through this, the effect of the surge upon the economy would be smoothed over a range of predictable dates.

    I don't think like this. I think that if a surge comes in the currency should quickly adjust! But it's a nice thing to say to an old-style central bank.

    Exchange-traded funds would fit this implementation well, giving global diversification, eliminating the complexity of picking an active manager, and giving low costs.

    I think it's important to have multiple competing private fund managers. The main reason for this is to cater to the diverse array of preferences of domestic households. It is useful to harness the entire domestic financial sector for the education problem. Every agent in the country should be selling global diversification to households.

    There will almost certainly be persistent discounts/premia, which
    will be hard to arbitrage away because of the capital controls. The
    closed end fund discount is an impediment to the viability of closed end funds in the best of times. In this proposal, we're dealing with the closed end fund discount in the worst of times: An environment where capital controls inhibit arbitrage.

    When the INR/USD is a manipulated rate, and it's overvalued, most
    economic agents expect it to appreciate. At such a time, few will want to hold any currency but the USD. Thus, at the time when the RBI most wants to push foreign capital out of the country, the proposal will be the least effective. The discount/premium on the closed-end fund will fluctuate over time reflecting the currency views of private agents.

    I know, they are saying that this will work best without a distorted exchange rate. But in the limit, if the country had a clean floating exchange rate, they would have convertibility, and this proposal would be unimportant. Their proposal is intended to help in a country which hasn't quite figured out it's act.

    I think that if I'm a domestic household in the midst of a capital
    surge, and with a pegged exchange rate I expect local currency appreciation, I will not want to own offshore fixed income assets. But
    when we think of the numerator and denominator of the Sharpe's ratio
    of foreign equity portfolios, a few percent of currency movements do
    not swamp the decision. The gains from diversification could be large
    enough to overcome currency expectations. So it might be possible to sell offshore equity funds but not offshore bond funds.

    I think any such effort should be accompanied by modifying regs for
    local mutual funds, local pension funds and local insurance companies,
    so that they are able to benefit from global diversification. Financial system stability will be enhanced when these institutional investors are able to reduce their portfolio vol, that they wouldn't have all their eggs in one basket.

  3. I think both options - GIC vs closed end fund securitization model - will be open for political manipulation, although GIC model is more likely to be. We can see that in UTI pension-fund fiasco. Instead of getting higher returns, it’s not a stretch to think of taxpayers probably ending up bailing out RBI to keep up minimum reserves with higher interest payments to foreign lenders.

    It's best to keep our benign government out of securities markets at home and abroad - that is keep excess reserves to a minimum.


Please note: Comments are moderated. Only civilised conversation is permitted on this blog. Criticism is perfectly okay; uncivilised language is not. We delete any comment which is spam, has personal attacks against anyone, or uses foul language. We delete any comment which does not contribute to the intellectual discussion about the blog article in question.

LaTeX mathematics works. This means that if you want to say $10 you have to say \$10.