Thursday, July 10, 2008

A study of Rupee in relation to FII flows - 17th June 2008

Chart 1

Chart 1 captures the broad trend of institutional investment popularly termed as “hot money” and the domestic exchange rate a little over a 4 year period (Obtained by plotting the monthly average rupee rates vis-à-vis the corresponding Net FII flows in equity and debt.)

A closer look reveals an interesting relationship in that 67% of the times over the past 53 months, whenever FII net flows turned negative to the tune of Rs 3000 crs or more in a single month; it has had the effect of hitting sentiment for the domestic currency and has spurred a default dollar rally.

There seems to be a steady pattern even in this relationship:
- the first trigger is usually caused by FII outflows which takes the market by surprise and leads to heightened volatility leading to a sharp spurt in the dollar in a very short time span.
- This is followed by a continuation of the trend for a few months (anywhere between 3 and 6 months) wherein if FII flows stage a comeback to positive territory and remain positive for about 3-4 months in a stretch, then the dollar momentum wears out and gives way to rupee rise.

However in all of the above cases, the extent of rupee weakness in the face of FII outflows is governed by the principle trend of the overall price action. For instance, when price staged a break in the multi month uptrend of the $ in March 2007, an outflow of Rs 17000 crs in Jan 2008 coincided with the rupee hitting a high of 39.37(which is an exception to the above “normal” rupee response).

STATISTICAL (Regression) ANALYSIS - 4 instances of FII outflows and a resultant Weak Rupee

1) APRIL 2004 – AUGUST 2004
Chart 2
Zooming in on the first setback that the rupee faced in May 2004, chart 2 clearly shows, the inverse relationship (though to a lesser extent) between the independent variable (FII flows) and the dependant variable (rupee).


A mapping of per day’s FII flows and the corresponding rupee rate via Scatter plot in Chart 3 reinforces the underlying negative slope as obtained by the regression equation :
y = -0.0004x + 45.474

However the linear relationship is not very significant as the R square which spells the coefficient of determination is only .016. The R-Squared illustrates how well the Linear Regression Trend line approximates real data points.

Chart 3
An R-Squared of 1.0 indicates a perfect fit. Statistically the equation implicitly points to certain unexplained extraneous variable(s).This would typically relate to the jitters following the surprise outcome in the General Elections, trade deficit worries in the face of a dip in inflows and to a large extent sentiment, amongst others.

Chart 4 The final step in the above analysis was to plot varying values of the independent variable, namely FII flows and run a simulation based on the fitted regression equation and compare it against the actual rupee movement which was registered in the months, post August 2004.As can be seen in Chart 4, the variation has been quite substantial, as the fitted projection mostly overestimated the $

2) SEPTEMBER 2005 TO DECEMBER 2005

Chart 5

The next significant spell of rupee weakness is analyzed below using the same methodology employed earlier. The observations are as follows:


- The rupee witnessed a severe 6% round of weakness following multi month lack luster consolidation.






Chart 6


- Here, however the slope of the linear regression line turned mildly positive as seen in Chart 6.

-Though the initial trigger was provided by FIIs turning net sellers, it quickly reversed back into positive territory but rupee continued to remain weak
- Here again there is a huge unexplained component and surprisingly oil doesn’t feature in the list as it in fact declined 19% in the same period.

Chart 7

-This could be explained by the NDF related play which led to a trigger of a number of leverage structures and the sheer speed of rupee weakening sparked panic buying in dollars.
- However despite R square at .01, the actual rate that prevailed post December 2005 nearly coincided with the fitted projected curve after a lag of 3 months.

3)FEBRUARY 2006 to SEPTEMBER


Chart 8







Analyzing one of rupee longest streaks of weakness witnessed in 2006 validates yet again the impact that FII flows has had in moving the rupee into weaker territory.









Chart 9


Chart 10


One significant observation emerging from the regression analysis is that whenever the slope of the linear regression turns negative (as was the case even in April – August 2004); the simulated fitted trend line tends to consistently overestimate the dollar in the future for varying levels of FII flows.

4) APRIL 2008 till DATE

Chart 11


The new fiscal was greeted by the dollar virtually reversing all its previous year’s losses. Given the striking negative sloping linear trend between the explained and explanatory variable, it can be inferred from the above fitted regression equation that since the FII trend has not yet bounced into positive territory, chances are that a further slack in capital flows could prove positive for the dollar







Chart 12
Based on the above fitted curve, a projected scenario analysis at various possible levels of FII net flows would be as follows:

Chart 13
However since the negative slope has historically resulted in an overestimation of $ rates, there is every likelihood that the projected rates could possess an element of deviation. From the above projected, on the upside 43.71 would be a formidable barrier and likewise 40.68 on the downside in the event of rupee appreciation

No comments: