Monday, February 13, 2006

Buy high P/E stocks in India

I noticed an fascinating article Growth beats value on the Bombay Stock Exchange by Satneet K. Sabharwal and Timothy Falcon Crack, December 2005 [link to it on SSRN]. It looks at data for a sample of stocks on the BSE from 1990 till 2004. There are few other papers in the field of Indian equity + asset pricing theory, so it is really interesting to read a well implemented and pioneering paper.

I am surely a non-specialist, for I don't read papers in asset pricing theory. But I felt uncomfortable about two things about this one.

Difficulties with sampling: They use panel data (the 203 firms with atleast 173 readings of returns over 175 months). But there was a huge phenomenon in India of birth and death of firms over this period. Their results pertain to: the few old firms who survived. This amounts to a nonrandom selection of low-risk firms, since the high-risk firms are likely to have died. I don't know how that changes things, but this worries me. I suspect you get an upward bias by looking at a few low-risk firms that managed to survive. Also, cross-sectionally, across these firms, there may be a low variation of risk thus adversely affecting the experimental design.

Liquidity premium: I think that over 1994-2001, there was a huge change in liquidity owing to the reforms of the stock market. This improvement in liquidity gave a liquidity premium. The gain in liquidity was concentrated amongst large stocks. So it looks like big stocks got strong returns - but what might have been going on was a liquidity premium story. At first, there was a spike in liquidity of the biggest stocks (Nifty). Later, that has been percolating into other stocks. E.g. the gap between the liquidity of Nifty and Nifty junior has declined in recent years (accompanied by powerful returns to Nifty Junior). But all this is about one-time changes in response to a modified market design. It isn't about a timeless relationship between risk and return.

Can liquidity premia explain the large differences in return across the portfolios formed in a paper like this? I believe they can. As an example, the diagram of excess returns on A group stocks when badla was banned (in the JFE article by Eleswarapu & Berkman) shows huge effects. And the improvements in liquidity that came through the reforms were even bigger than the widening of the bid-offer spread that came from the ban on badla.

The dataset is now building up to do these kinds of papers - the CMIE firm level data is mature from 1989-90 (i.e. year ended 31 March 1990) onwards, and the CMIE returns data starts from 1/1/1990. So I hope there will be more work of this nature - we are all ears.

While I'm on this subject, data sizes in India are becoming quite cool. My data for daily Nifty returns goes from 3/7/1990 till 7/2/2006 and is 3,680 points of a fairly homogeneous methodology. The daily returns time-series for the BSE Sensex goes from 3/4/1979 till 7/2/2006 and is 5,888 points (!). Unfortunately, they messed up along the way (shifted to float-weighted index), so they've lost out both on consistent methodology across a long series and on having a sensible methodology.


  1. can we really isolate the effect of liquidity premia in this context? (as in to what extent is the liquidity premia driving this)

    also, iitb doesnt subscribe to ssrn. ergo i coudnt read the paper. :(

  2. The paper can be downloaded without having a subscription to SSRN (e.g. I don't have one).

    There is a huge literature on measuring liquidity premia. It is a big goal of economists, to be able to write theoretical models and have empirical estimates which drive (a) measurement of liquidity through some metrics and (b) mapping these liquidity metrics to liquidity premia.


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.