Wednesday, August 04, 2010

Understanding the ADR premium under market segmentation by Matthieu Stigler, Ajay Shah and Ila Patnaik.

Many practitioners in India believe that net FII flows matter greatly to stock market returns. But this belief is not borne out when you carefully look at the evidence. The VAR in the above paper says this one more time: in the context of the specification used in this paper (aimed at uncovering insight into a different issue, the ADR Premium) you don't see an impact of a rise in net FII flows upon stock prices.

On an interesting story about the ADR Premium for one stock (as opposed to macroeconomic thinking), see: Spike in the ADR Premium on Tata Motors by Anmol Sethy.

You might like to see: the stock of papers from the NIPFP Macro/Finance Group.

1. About your conclusion of FII inflows, I am but 100% on this with you.

I did my own, research, did some correlation analysis with FII fund inflow and NIFTY growth. Apart from some positive correlation, which might be attributed to pure chance(didnt do any t-test) I see seldom predictability.

Another marker is, what layman eagerly wait for is: "follow-the-DOW" principle. Long Term correlation is negligible at best(positive though)

Soham

2. Is the relationship on daily changes lagged 4 days (I think it is this) or, 4 day cumulative changes lagged 4 days?

I'm wondering about the latter because the lag could be variable and perhaps a larger sampling period for returns would capture that variation, smooth out the effects? Or, maybe not, I'm not sure....

A few days back, Prof Jayanth Varma had posted comments on a RBI paper: RBI on India and the Global Financial Crisis. Interesting/confusing to note that the report found a causal relationship from FII and Nifty (opposite direction), apparently using similar modeling tools.

3. Vivek,

Reading up Dr Varma's report, I think that research holds better water, because I used stationary correlation in my work, and markets are anything but stationary, so spurious conclusion can always creep in. I think even a test for stationarity on FII inflows, might show a slight upward trend.

Soham

4. Soham,

This paper claims stationarity for all series.

Its possible that the RBI report has issues because of using a different sampling period. Dr Varma points out that this information is not provided in the RBI report.

Along the same lines, it seems that using cumulative changes (as I was saying in my previous comment) is a bad idea with Granger causality since it can indicate spurious causality.

5. The stationarity issue is a critical one. If stationarity is not present, standard VAR procedures are wrong.

In getting to stationarity, the daily net FII series needs to be carefully handled in two respects. The raw series has increasing mean and increasing variance. The first point is to rescale it by the overall market cap (which is also increasing mean and increasing variance) so as to generate a series of net fii as percent of overall market cap.

The second point is to exclude some of the early period where FII flows were just growing explosively. If that period is not removed from the analysis, then also there are problems with stationarity.

Once these two things are done -- as in our ADR paper -- the net FII series is indeed stationary.

Oh, and you have to be careful in that the data reported by SEBI on date t are not about the net FII on date t-1.

The ADR paper is interesting in that we get a glimpse into the impact of net FII upon Nifty while controlling for a few other things. We have worked on this pure question also (the impact of net fii upon Nifty) and by and large, when the econometrics is done carefully, you get a one-way impact from Nifty to net fii and not the other way around. I know that this result tends to surprise practical men, but if careful economic analysis did not sometimes surprise practical men, then it would not be worth doing, would it? :-)

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

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