Search interesting materials

Saturday, October 18, 2008

Mythbusting: Reserves edition

by Ajay Shah.  

Myth: When FIIs sell shares and capital leaves India, RBI's reserves go down

An FII sells shares, realises rupees, and then takes those rupees to the currency market. Under ordinary circumstances RBI's reserves are not affected. The currency market is a vast ocean and there are many buyers and sellers. The FII is just one seller of rupees on this market. Buyers and sellers make the price - RBI needn't be in the picture. E.g. an FII buying dollars might have his currency trade match against an exporter selling dollars.

Conversely, if Apple India buys \$500 million on the currency market in order to import a million iPhones, this is no different from an FII buying \$500 million because he sold shares. All buyers/sellers of rupees or dollars come together in the currency market. Reserves do not go down when Apple India buys \$500 million of foreign exchange in order to import iPhones, just as reserves do not go down when an FII buys \$500 million of foreign exchange by liquidating shares or bonds.

It is only in a fixed exchange rate system, with every currency transaction going through the central bank, that every decision by a private party to either import goods or take capital out of the country leads to a dimunition of foreign exchange reserves. The people who assume that sale of shares by FIIs yields a reduction in reserves are perhaps instinctively operating on the intellectual toolkit of a fixed exchange rate regime - something that India shed many decades ago.

RBI is a player on the currency market. When they sell dollars on this vast market, reserves go down. But there is no tight link between RBI's decision to sell dollars with the decision by an FII to take money out. RBI's trading on the currency market is designed to manipulate the price. Their motivation for trading is price based. As with any market manipulator, RBI buys when they want to drive up the price and sells when they want to drive down the price. A trade by RBI might match against anyone - e.g. RBI might buy dollars or sell dollars from importers or exporters. It's a vast market out there; it is impossible to pinpoint who in the market "sold dollars to RBI", and it is certainly not true that sales by FIIs somehow selectively pickup RBI as the counterparty. On a currency market which does \$10-\$30 billion a day, FII purchases/sales of about \$250 million a day are not particularly important in influencing the exchange rate and thus triggering off trading by RBI.

While I'm on this subject, I must mention a report that RBI regularly puts out named `sources of reserves accumulation'. When you think seriously about it, that report is conceptually confused. It would be best to discontinue the production and release of this report.

Myth: When reserves drop, RBI was selling

The level of reserves changes for three reasons: interest income on the securities held in the reserves portfolio, valuation changes, and transactions.

All of us tend to track reserves measured in USD. RBI holds other currencies in the reserves portfolio also. E.g. suppose they have Euros. When the Euro loses ground, it looks like the reserves (measured in dollars) declined.

Here's an example, where I use the `US Major Currencies Index' produced by the US Federal Reserve as a measure of fluctuations of the USD. It is an index of the USD against all the floating exchange rates of the world, weighted by their trade with the US. I abbreviate this as `USM'.

MonthUSMDelta ReservesRBI purchase
April 2008 70.914.99 4.32
May 2008 70.690.65 0.15
June 2008 70.96-2.54 -5.23
July 2008 71.73-6.42 -6.32
August 200875.06-9.80 1.21

In a month like April or July, RBI's purchases are similar to the change in reserves. But this is not always the case. August is a striking example. The market widely commented on the decline in reserves of \$9.80 billion. But with a long lag, when data for RBI purchases was released, we find that in that month, RBI was a net buyer of dollars. They bought \$1.21 billion. What happened?

In August, the USD gained strongly - the global flight to safety involved selling assets in other countries and buying assets in the US. As a consquence, the USD strengthened from an index level of 71.73 to 75.06 - a change of 4.6%. Suppose RBI had \$150 billion in non-USD assets and suppose these assets took a hit of 4.6% (i.e., that RBI's weights are the same as US trade weights). This yields a loss of \$6.9 billion.

Apart from valuation effects, some money gets added to reserves regularly, on account of the interest income on the fixed income securities held in the reserves portfolio. So even in a month when there was no trading by RBI on the market, and there were no currency fluctuations, the reserves would rise owing to interest income.

It might be useful to look at the weekly changes in reserves (which come out with a small delay) even though they're not the same as weekly data for RBI trading

The lack of transparency by RBI on their currency trading can have some interesting consequences:
  1. Suppose economic agents treat purchases by RBI as evidence that the INR will appreciate at future dates and vice versa.
  2. When you see RBI buying dollars, it's efficient to bring capital into the country, for you profit from the coming INR appreciation. Conversely, when you see RBI selling dollars, it's efficient to sell assets in India and take money out of the country, for you avoid the coming INR depreciation.
  3. The trouble is, you don't observe RBI's trading, except with a long lag, and all that is then given out is monthly aggregates.
  4. What's the closest proxy to what RBI might be doing? Delta reserves.
  5. So while Delta reserves is not an accurate measure of the extent to which the exchange rate is distorted by RBI, it's the only one you have.
  6. It might be efficient for you to use Delta reserves as a tool for forecasting the exchange rate, even though it's not a sound measure.
  7. Weekly data for delta reserves might influence capital flows.
The main argument here is that by using such a heuristic, people might often do the wrong thing because delta reserves does not quite predict the direction and magnitude of distortion of the exchange rate. E.g. in August, reserves fell quite a bit, and you might have thought that the rupee was overvalued, but actually RBI was pushing the rupee towards depreciation. But given RBI's opacity, this is the best that the private sector can do and it might be efficient to use this noisy data for gauging currency distortions.


  1. An excellent post esp, the point USM part clarifies...

    Ajay can you please talk a little bit more about USM.
    Is this the same thing, this article has mentioned?
    Trade Weighted USD Index


  2. last 5 yrs of credit orgy led to sahibs putting lots of money in india as capital inflows made indian current account look good, made rupee stronger and stock/real estae boomed. It was an illusion mungerilal was prouf off.
    Now the FII are rushing out as fast as they can. Expect rupee to fall to 60 against dollar (inflation differential between us & india is ~5%..5 yrs of cumulative difference and we have at least 25% overvaluation of rupee everything else remaining from 48 level at 2003..a decline to 60 seems ahead.
    The down side could be worse as US economy gets on the upswing and IT/BPO BUST and real estate bust in mungerilal cities are yet to arrive.

  3. on the RBI intervention, i think they intervene on a gross flow basis and not on a net number. So on a prticular day, say if RIL is bringing in $2bn through ECBs, RBI would probably intervene , although in that month there could have been net outflows through the FII route. sthg like that happened in Jan-08, due to the R-Power IPO, but the subsequent days saw fii outflows but RBI still had a net dollar buy figure

    some data on that

    % Fy 08 Fy 07 Fy 06
    Intervention as a % of Gross Capital Flows 10.4% 6.4% 3.1%
    Intervention as a % of Net Capital Flows 72.4% 58.6% 32.0%

  4. I always wondered what net buy by RBI would mean when reserves are going down. Does it mean that RBI is holding forex reserves in non US Currency and the decline in reserves is due to depreciation of the currencies being held? Isn't it possible to get data on RBI's holdings in respective currencies to come to such conclusions.


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.