Abheek Barua: Interest arbitrage: How effective?

Traditional exchange rate theory is too trade-centric to describe today's emerging markets

Bond yields have climbed up over the last three months and the rupee has appreciated considerably against the dollar. Is there a connection?

A common perception is that there is indeed a link""high domestic bond yields (one can interpret these as interest rates) are directly responsible for the rupee appreciation.

A rise in domestic interest rates widens the difference between local and foreign (say, the US) rates of return and invites more capital inflows. The rupee appreciates as a result.

Prima facie, this has simple intuitive appeal. The problem, however, is that the recent history on interest rate differences and rupee movements doesn't validate it at all.

I use monthly data for the last three years on the differences between the yields on one-year Indian and US bonds, on the one hand, and the rupee-dollar exchange rate, on the other, and find that the rupee has actually depreciated in periods when the yield difference (Indian yields minus US yields) has risen and appreciated when the difference in yields has fallen.

Thus, the movement we've seen in the last couple of months is actually a bit of an aberration, given the past trends.

This apparently perverse phenomenon of the last couple of years is hardly surprising if one gets foreign institutional (FII) equity inflows into the picture.

However, it is an interesting example of how cause and effect relationships actually pan out in the local currency and bond markets. It also brings into focus some of the less conventional drivers of domestic liquidity and exchange rates, which deserve closer and more systematic attention. But first, let me explain how it actually works.

Quite a bit of the variation in India's foreign exchange reserves over the last few years can be attributed quite directly to variations in dollar flows into the Indian equity markets.

In this period, foreign exchange assets were also the critical driver of the monetary growth. (On November 19, roughly 119 per cent of base or reserve money was accounted for by foreign exchange assets.)

In short, variations in FII inflows have been, to a significant degree, been responsible for changes in monetary or liquidity conditions.

Now, imagine what happens if there is a surge in FII inflows. Reserve money goes up, as does liquidity in the local money markets. This pushes down domestic yields and the yield differential between domestic and US bonds.

Greater foreign demand for domestic equities also means stronger demand for rupees. The rupee appreciates as a result. Thus, falling yield differentials and an appreciating rupee are perfectly compatible.

This sort of cause and effect relationship is not restricted to FII inflows alone. Other large inflows like private remittances, which are themselves insensitive to interest rate changes, would have the same effect as FII inflows, pushing down domestic yields and pushing up the rupee simultaneously.

Surprisingly, the observed currency and interest movements seem to preserve one of the most simple and elegant principles of economics, uncovered interest parity (UIP). UIP claims, as indeed much of economic theory does, that there is no free lunch.

To be a little more specific, it argues that cross-border arbitrage opportunities in the bond market cannot exist for long. Thus, the return in dollar terms (or rupee terms, as long as we use the same currency to measure both) on Indian and US bonds should equalise.

Thus, going by the UIP principle, if Indian bonds offer higher yields than US bonds, the potential profit you could make by borrowing in the US market and lending in the Indian money markets would be neutralised by a depreciation of the rupee.

This is exactly what my data show""whenever the margin between our domestic yields and yields in the US has increased, the "potential" arbitrage gain has been eaten up by currency depreciation. The converse has also held.

There are a couple of inferences to be drawn, in my opinion, from this bit of somewhat informal empiricism. For one, the impact of interest-arbitrage driven flows on the rupee's movements appears to be negligible at least in the short term.

This perhaps reflects the fact that the capital account is not fully convertible and there are significant barriers to moving capital in and out of the domestic money markets even if there is an arbitrage opportunity to be exploited.

Some of my colleagues at our treasury tell me that the arbitrage process is far more complex and movements in non-dollar currencies and associated interest rates need to be factored in.

Whatever the underlying reason, it invalidates the simple and direct correspondence between the difference between US and domestic rates, and the rupee-dollar exchange rate, which is often assumed to hold.

From a trader's or a currency forecaster's perspective, it offers a simple rule to judge the direction of bond yield movements. The rule suggests that if the rupee has appreciated for a period and bond yields haven't moved much, one of two things is likely to happen.

Either domestic yields are likely to go down or US yields are likely to move up. Conversely, if the rupee has depreciated, be prepared for a rise in Indian yields.

More generally, the fact that a relationship of this kind holds with a fair degree of consistency emphasises the importance of things like equity flows in currency or interest rate determination.

Conventional economic theory and the tools that come with it tend to ignore these elements and stand the risk of being somewhat irrelevant in today's context.

For one, traditional exchange rate theory (which gives us tools like the real effective exchange rate) is way too trade-centric to describe today's capital account-driven world of emerging markets.

The focus in this literature is on whether the currency is "competitive" enough for the right mix of export and import growth for the economy. Any deviation from this optimal level is assumed to trigger a process of correction to re-establish optimality.

This correction works essentially through the trade balance. If domestic inflation is higher than in competing economies, exports taper off and imports rise. This curbs the demand for local currency and induces depreciation.

This is fine as a theoretical construct but a more relevant analysis of currency and interest rates in the present market situation has to look a little more closely at critical capital account drivers.

To forecast the currency, for instance, it is now more important to get a fix on the relative attractiveness of the domestic stock market vis-à-vis other markets than to assess whether the currency is overvalued or undervalued in terms of relative purchasing power.

Thus, analytical devices like price-earnings multiples and the earnings growth of Indian companies relative to competing markets have become more important than inflation differentials or the trade balance.

Given the close connection between exchange and interest rates, if you can predict the exchange rate correctly, you are likely to get your interest rate views right.

It is also important to understand what drives some of the other "non-trade" components of the current account like private remittances. These are, today, as important as the merchandise trade balance in determining exchange rate movements.

Yet, while there are reams analysing variations in merchandise trade, a thing like private remittances remains a black box. Perhaps, the RBI, which has access to primary data on some of these components, could educate us a bit on the nature and drivers of these flows.

(The author is Chief Economist, ABN-AMRO Bank. The views are personal.)

He can be reached at

abheek.barua@in.abnamro.com)