Looking at the Nifty P/E

So as Warren Buffett says, "Be fearful when others are greedy, and greedy when others are fearful." A quantitative way of interpreting this is - invest when market valuations are low.

The most followed measure of market valuation is the Price/Earnings, or the P/E ratio. So I decided to analyze Nifty P/E data from Jan 1999 (from when data is available on the NSE website) to Jun 2016 to see how it has moved over time and what it has meant for forward returns. I have looked at trailing P/E since i) I do not believe analyst's forward estimates, ii) data is available readily only for trailing P/E.

Here is how the P/E has moved during this period:

The median P/E for this period is 18.75.

The max was 28.47, just before the dot-com crash and nearly reached again just before the 2008 financial crisis.

The min was 10.68 in Oct 2008, after the Lehman bankruptcy, closely matched in 2003 when the market bottomed out after the dot-com crash.

The 25th percentile is at 16.01 (which means 25% of the time, the Nifty P/E ratio was below 16.01) and the 75th percentile is at 21.14. This puts the P/E of 22.75 as of Jun 30, 2016 comfortably in the top quartile of P/E for the period (in fact, it is in the top 10% of P/Es for this period), which is not very comforting.

The next step is to understand quantitatively what this has meant in the past for forward returns. First things first though, is trailing P/E actually a predictor for future returns?

I ran correlation of 1, 3 and 5-year annualized forward returns with P/E and got correlation coefficients of -0.73, -0.73 and -0.75, respectively. So, yes, at least at those time scales, there does seem to be a significant negative correlation between trailing P/E and forward returns. I believe in this case at least that correlation does mean causation.

I then looked at 1, 3 and 5-year forward returns bucketed by starting P/E quartile. Below is what I see:

Some trends seem pretty clear to me:

• On average, investing at a low P/E (bottom quartile) has provided significantly higher returns over a 1-5 year timeframe than investing at a high P/E (top quartile). This average difference of returns is 54%, 29% and 19% (annualized) for 1, 3 and 5-year timeframes respectively which is very material.
• Staying invested over longer periods of time has meant that investors have made modest amounts of money even when starting P/E has been high. This would be expected since Indian markets overall have risen over time.
• Variance (or uncertainty) of returns is much higher for shorter time periods vs. longer periods. This is expected as markets can be much more volatile in the short term but tend to mean-revert over the longer term. This is why we see 1-year returns varying from less than -40% to greater than 100%. However, over a 5-year timeframe, the annualized returns vary only between 0% to 40% (with very few outliers on both sides).
• Not only that, the short-term variance is also higher for each P/E quartile bucket. This means that there is no guarantee that investing at a low P/E will give you stellar returns in the short term, or investing at a high P/E is a sure recipe for disaster.
• For e.g., in the 1-year returns chart, we see a significant number of cases where returns were negative (less than -20% though) even when starting P/E was in the bottom quartile, and we see some cases of returns in the 40-60% bucket even when starting P/E was in the top quartile.
• However, over a longer time period, the predictive power of starting P/E becomes higher as returns are bucketed more consistently.
• Another interesting trend I see above in all the 3 charts is that the forward returns profile when starting P/E is in bottom 2 quartiles varies much more than when starting P/E is in the top 2 quartiles. What this seems to suggest is that market euphorias die down much faster than the time it takes markets to come out of the doldrums. This is consistent with the way people tend to behave in the markets - every one jumps in when markets are going up, creating a frenzy which then dies down soon after. Then, when the markets fall (and are much safer to invest in!), people don't want to have anything to do with the stock markets.

Caveats to the above analysis:

• I am well aware that the time period selected for analysis can have significant bearing on the outcomes of the analysis. However, this is the data I have access to.  
• I have not considered dividends in the analysis to keep it simpler. If anything, I believe the conclusions will only be reinforced by considering dividends too. 
• I'm aware that Nifty composition changes pretty regularly. I don't know how to normalize my analysis for that, nor do I have the motivation to. I'm analyzing this from the perspective of investing in a Nifty ETF/mutual fund.
• Some people will say that since interest rates have gone lower over time, the acceptable P/E needs to be adjusted higher for that. Maybe (it would require analysis which I'll keep for a future post), but the correlation of returns to P/E even without considering that is pretty high so I don't want to unnecessarily complicate it.
• Past returns are no guarantee of future returns!

What this means for me:

• I was a bozo, investing at the peak in 2007 and exiting close to the bottom in 2008!
• Markets require significant amount of patience. While it is not easy to invest your money and then see the number swing around like a yo-yo, that is what is exactly required since most of us tend to invest at the top and exit at the bottom (ask me!).
• There seems to be significant correlation between starting P/E and future returns for Nifty. While I sat out the rally in the markets in 2009 and also 2011/12, now doesn't seem to dive in, given current valuations. I'll wait and keep my money in debt mutual funds till valuations come down (either earnings pick up or markets fall).

Please do provide your views/counter-views on this!