Click here for Analysis Report Archives

Download .pdf file

November 2006

We have developed a quantitative model that explains movements in USD/INR since 2001. In building the model, we studied a range of variables to see which ones had an adequately high correlation with USD/INR movements over a reasonably long period. Our analysis showed that as long run drivers of USD/INR, oil imports as a percentage of total imports and FII inflows explain over 77% of the movements of the rupee at a monthly frequency. This is a remarkably good fit for an empirical exchange rate model.

Of course, there are other short run - or cyclical - variables that also affect the currency value, and the model has attempted to quantify the impact of these as well.

Oil and the rupee:

While many analysts have highlighted the importance of oil imports in India's balance of payments, it is quite remarkable to see that India's trade balance ex-oil has been positive for most of the period since 1999 (Figure 1). For an economy on a high growth path, particularly over the last few years, this speaks eloquently of our export competitiveness at current parameters.

Of course, since India imports more than 70% of its oil requirements, there is very little "give" available on the oil import front, which confirms the accepted wisdom that oil prices are a key driver of the value of the rupee.

In seeking to find an oil price related variable that could be used as a predictor, we first assessed the correlation between oil prices and the rupee, which, in recent months, has increased dramatically, with the three month rolling correlation running as high as 85% (Figure 2). However, this correlation was nowhere near as strong in previous years, when the rupee was much more controlled.

We found, however, that the correlation between [oil imports as a percentage of total imports (3M rolling sum)] and monthly USD/INR was above 65% over an acceptably long period (1998 to date) (Figure 3), which makes this a much better variable from a statistical viewpoint.

FII Inflows and the Rupee:

FII inflows in the Indian capital market have increased significantly in recent years, particularly from 2003 onwards. With the vast bulk of FII inflows focused on the equity markets, the BSE Sensex is very highly correlated with these flows (Figure 4). The most recent evidence of this was the outflow (of around USD 1.6 billion) in May 2006 as a result of the sudden increase in global risk aversion, which triggered a massive sell-off in Indian stocks and, not incidentally, the rupee (Figure 5),

Indeed, the rupee has become much more volatile as a result of FII flows and we found that the correlation between 12M FII inflows and the rupee is more than 85% from 2001 onwards.

Such a high correlation, again, over a reasonable length of time, makes this - the 12-month sum of FII inflows - an excellent statistical variable to determine long run movements in the currency.

The Model:

Using these two as independent variables, we built the model to forecast the rupee. Interestingly, the model showed that, even though oil imports are in the range of 33% of total imports (and over 100% of the trade deficit), whereas FII inflows are less than 25% of the trade deficit, FII infows are more significant as a predictor of the rupee. The beta coefficient for the FII variable (at -0.0083) is more than four times that of the oil variable (0.002). This translates to: if FII inflows increase by USD 1 billion (all other things being constant), the rupee will appreciate by more than 0.8%; whereas if there is a one percentage increase in the ratio of oil imports to total imports, the rupee will depreciate by around 0.2%.

The chart shows the fitted curve (Figure 6) superposed on the actual movements of USD/INR since 2001. In statistical terms, the model provides a reasonably good fit, since the deviations from the model are cyclical in nature, suggesting that the deviations are driven by short run (mean-reverting) forces, which dissipate completely over time (Fig 7).

We estimate that these deviations (long run residuals) have an average half life of 3 months. This means that on average, the deviations from our fitted long run values will be reduced by one half in three months.

Impact of short run deviations:

These short run deviations are explained by the action of other forces that influence the market, like interest rates or RBI intervention in the foreign exchange market. One noteworthy finding of our analysis is that these otherwise apparently significant variables are essentially short run in nature, and that their impact on the rupee is, in time, overpowered by the longer run forces - viz., FII inflows and oil prices.

According to the model, the long run value of USD/INR on October 31, should have been 45.25. In fact, its actual value was 45.00, indicating that it was overvalued by 0.55% Since then it has strengthened further and, at its peak of 44.35 (on November 10), had overshot its long run value by 2.03%.

The loudest short run force that has been driving this move away from the long run value is the U.S.-India interest rate differential.

Indian interest rates have been firming more so since the last monetary policy announcement (Oct 31). Again, strong economic growth and inflation knocking at the 5.5% upper target set by RBI suggest that Indian interest rates are likely to climb further. On the other hand, it seems clear that the U.S. Fed is not likely to raise rates any time soon - indeed, the Fed funds futures sees a 28% chance of the next move in U.S. interest rates being downward. Clearly, the interest rate differential has been moving in favor of the rupee, which explains why it has been overvalued compared to its long run estimate. [Figure 8 shows the three month rolling correlation between 2-year interest rate differential and USD/INR, which has recently (since September 2006) peaked at over 80%. Note that this correlation has been very volatile, which confirms that the interest rate differential is merely a short run indicator.]

On the other hand, another short run factor - RBI intervention - has been fighting this move. We note from reserves growth figures (Figure 9) that RBI has been intervening aggressively these past few weeks. The impact of this intervention was very clear on November 10, when the rupee fell from its high (of 44.35) to close at 44.70, which is now merely 1.2% from its long run value.

FORECAST:

To assess how the long run value of USD/INR could change into the future, we ran the model under different scenarios described by certain assumptions for oil prices and non-oil imports and FII inflows.

To project the value of oil imports, we assumed that oil volume growth remained at its 12-month average (of 14% a year) and made assumptions about oil prices [see below]. We also assumed that non-oil imports would also grow at its 12-month average, which, coincidentally, was also 14%.

Scenarios

• 1A) Oil prices and FII inflows remain steady at present levels (of $57 per barrel) and USD 9.5 billion (for 12 months), respectively: Since non-oil imports are greater than the oil imports and, based on our assumptions [see above] grow at the same rate, the ratio of oil imports to total imports will fall gradually, resulting in minor appreciation in the long run value over the period.

• 1B) Oil prices remain steady at present levels and FII inflows rise by USD 500 mn each month: Since FII inflows are a stronger influence than oil imports as a ratio of total imports, the rupee appreciates strongly in this scenario from 44.80 by the year end to 42.90 by Sep-07.

• 2A) Oil prices fall by $1 per barrel each month and FII inflows remain at the same level: The rupee appreciates modestly from 45.15 by year end to 44.60 by Sep-07.

• 2B) Oil prices rise by $1 per barrel each month and FII inflows remain at the same level: The rupee remains more or less steady around 45.00.

• 3) Oil prices fall by $ 1 a barrel each month and FII inflows rise by USD 500 mn each month: Unsurprisingly, the model suggests that the rupee will appreciate significantly to 42.60 levels.

• 4) Oil prices rise by $1 a barrel each month and FII inflows fall by USD 500 mn each month: The rupee will fall significantly to 47.30 levels.

Reconciling the model with our views:

Our view is that the base case for next year is likely to be closest to scenario 1B, above - viz., that FII inflows continue to rise (to about $ 15-16 bn in the 12 months to September 2007) and oil prices stay more or less at current levels, although volatile.

This would suggest that the long run value for Dec 06 would be 44.80, for Mar 07 would be 44.00, for Jun 07 would be 43.40, and for Sep 07 would be 42.90.

However, as we have already seen, there are short run forces that could create possibly significant deviations from these forecasts - empirical models always need to be modulated by macroeconomic views.

As Figure 7 shows, the cyclical variations from the long run value have, in the past, lasted for as long as 9 months (average 6 months), and have reached as far as 3.5% (average 1.5%) away from the long run value. This means that theoretically the value at any point in time - say, Mar 07 - could diverge by this much; this would give a range of 42.45 to 45.55, which is so wide as to be meaningless. However, these numbers do determine limits beyond which the rupee will not move, and, hence, can provide key strong trading signals, if, indeed, any of these extreme levels are met.

Trying to assess the impact of short run forces in the future, we note that we have recently seen two of them in action - the increasing interest rate differential pushing the rupee upwards by 2% from its long run value and RBI interventing bringing it back towards its long run value by 0.7%.

From the price action on November 10, it seems that RBI will continue to exert its influence to prevent the rupee from strengthening; this also plays into its monetary policy need to sustain liquidity in the system, provided, of course, that inflation remains under the 5.5% mark. However, given the continuing boom in the economy, it would seem that sooner or later, RBI will have to relent on interest rates - our bet is that this will happen at the next credit policy announcement in January, and that the market will recognize this somewhat before - say, in mid to late-December.

Thus, we expect that the short run force 1 (RBI) will probably be successful in pushing the rupee back towards - and possibly below - its long run value for December (of 44.80), after which short run force 2 (interest rate differentials) will bring the rupee back into strength, driving it towards its long run value for March (of 44.00).

In fact, this continuing tension suggests that we could see a sudden breakout in rupee volatility soon. Indeed, other research we have done studying the ratio of maximum to average volatility also suggests this. Given the broadly bullish trend for the rupee, this suggests that it may be a good time to buy dollar calls - options are best bought when you believe the market will move in your favor.

Another short run variable that could influence the rupee could be a change in perception on U.S. interest rates. However, there has been so much uncertainty about the future direction of U.S. interest rates, and, interestingly, increasing talk about the general unreliability of economic data. In any event, given that U.S. interest rates only act on the rupee at a secondary level (as part of the interest rate differential), we believe that this is unlikely to have any substantive impact. This means that any action on the rupee prior to Fed meetings could provide likely trading signals, since, in our belief, this variable should have little or no impact on the rupee.

Again, political instability either domestically or globally, could create some short run variations, but, it is difficult to see - and, hence, plan for - such imponderables.

Turning back to the long run variables: while we have selected Scenario 1B as the most likely, we do need to consider the possibility of changes in FII inflows later in 2007 and/or an increase in volatility in oil prices. Given the need that international investors have for double digit nominal returns and the fact that, with 2% inflation (one of the sterling results of globalization) and 3+% maximum growth, developed countries markets will be unable to deliver this, it seems certain that the underlying attractiveness of India (and other "risky" assets) can only increase - at worst, it will stay at current levels. The price of risk is falling, however, and it is possible - indeed, likely - that there will be another bolus of risk aversion to hit global markets at any time.

For instance, the sudden differences emerging between the ECB and the U.S. Fed, if they become more entrenched, could turn markets nervous - let us recall that one of the widely accepted proximate causes for Black Monday in 1987 was increasing differences in opinion between the U.S. Fed and the Bundesbank. Again, an increase in geopolitical tension (driven, say, by the recent atrocities in Gaza, or attempts in Iraq to push the new U.S. Congess to accelerate the departure of U.S. troops), could create a bout of risk aversion and, perhaps, a rise in oil prices. Either of these could cause sharp variations from the long run value of the rupee.

However, RBI remains the main arbiter of the market, and we assume they will act - in either direction, given their mantra of managing the volatility not the value of the rupee - if the market moves too far out of line. Quantifying this trigger by their action on November 10 (when the market had moved 2% from its long run value) and assuming wider ranges as we go farther into time, we project the following ranges for USD/INR:

Dec 06 44.60 to 45.00
Mar 07 43.60 to 44.40
Jun 07 42.80 to 44.00
Sep 07 42.10 to 43.70