What sense do stock prices make?
TABLE Rank Correlation between P/E and Net Profit Growth (at the level of twenty portfolios sorted by P/E) --------------------------------------------------- Forecast Forecast Horizon (years) made ------------------------------ in year 1 2 5 --------------------------------------------------- 1990 0.362 0.054 -0.062 1991 0.125 -0.107 0.493 1992 0.427 0.379 0.459 1993 0.182 0.513 0.641 1994 0.359 0.512 1995 0.821 0.668 1996 0.717 0.755 1997 0.680 ---------------------------------------------------Everyone who has looked at stock prices has experienced a certain dismay with their day by day, or minute by minute fluctuations. What sense do these fluctuations make? Many people feel there is something meaningless about these price changes; this leads to a discomfort with the importance of the stock market in the modern economy.
Economists have pondered this problem from the 1940s onwards. The first economists who subjected stock price data to statistical analysis were horrified to find that the prices did not have systematic patterns or trends. This is unlike all other economic time series -- the production of wheat, tourist arrivals into India, GDP, etc. are all economic data series where we see trends, seasonality, etc. But stock prices do not have such patterns. This could be viewed as an ``absence of order''.
It was around 1965 that the idea of market efficiency came up. This was the notion that stock prices move in response to forecasts made by speculators about the future prospects of firms. When news breaks, speculators compete in rapidly trading in exploiting new developments. If stock prices react to news and nothing but news, then future stock prices should be exactly as hard to forecast as future newspaper headlines. A price that can be forecasted is like a 500 rupee note lying on the ground; speculators will move quickly in grabbing it. Hence, patterns which yield forecasts will be hard to find. In this sense, the absence of patterns is what would denote an orderly, well functioning stock market.
In the years which have followed, thousands of studies of financial prices have been done, across different time-periods and different countries, including India. The broad answer that emerges from these is that prices have little forecast-ability. Strictly speaking, the amount of forecast-ability that is found is not zero, and numerous research papers are written on these `violations of market efficiency'. The daily returns on Nifty on two consecutive days has a correlation of 0.1, and not 0. However the forecast-ability is not large, and it can often not be converted into profits when all the costs of trading are taken into account. It would be easy to earn speculative profits in an inefficient market, so that's probably not a market that we see around us. From this perspective, economists feel that stock markets work fairly well.
This story, of financial prices fluctuating in response to news and being hard to forecast, is a powerful step forward in the understanding of finance. However, it's peculiarly unsatisfying insofar as it does not directly connect up with the `real economy'. It would be nice to see a link between information processing and forecasting by speculators, and future outcomes in terms of GDP, profits of companies, etc.
Recently, I set about exploring this question, using the P/E ratios observed on stock markets. Given two companies which have a net profit after tax of Rs.1 crore in the latest 12 months, why does the stock market award a valuation of Rs.20 crore to one company (a P/E of 20) and a valuation of Rs.100 crore to another (a P/E of 100)? Market P/Es are supposed to reflect forecasts of future net profit growth and a perception of future risk. The stock market is supposed to award a higher valuation to the company which is expected to produce better growth in profits the future.
Does this actually work? Do companies with higher market P/E ratios actually obtain higher profit growth in the future? There are a few hurdles in posing this question properly:
- At the simplest, surprises will always take place; the best of forecasting in 1993 could be overtaken by events in 1994 so that a firm which was expected to do badly in 1993 actually does very well in 1995. These surprises would afflict the best of forecasters, and the stock market's P/E (as a forecaster) would be no exception.
- In addition, differences in risk would play a significant role. Even if two firms have the identical growth prospects, the riskier firm would be awarded a lower P/E.
These problems are nicely addressed by forming portfolios. Instead of looking at one company at a time, we sort all companies by P/E and form 20 portfolios. If 1000 companies exist, the 50 companies with the highest P/E ratios are allocated into the first portfolio, the next 50 companies are allocated into the second portfolio, and so on. Within a portfolio, the differences in risk and the unexpected events would tend to cancel out.
Hence, we ask the question: do portfolios with high P/E ratios obtain higher (portfolio) net profit growth as compared with portfolios which have lower P/E ratios? The answers are summarised in the table.
The first line may be interpreted as follows. It involves using P/Es as of 31 March 1990. The rank correlation with profit growth into 1990-91 was 0.362, with profit growth to 1991-92 was 0.054 and with profit growth to 1994-95 was -0.06. Hence, the P/E ratios in March 1990 did fairly badly in anticipating future net profit growth. Similar poor results are found with the forecasting using P/E ratios as of 30 March 1991.
The picture seems to improve dramatically in recent years. The rank correlation using P/E ratios on 31 March 1997 and net profit growth in 1997-98 was 0.68. From 1993 onwards, all the values are highly significant statistically.
Why have these correlations changed? Two major explanations come to mind:
- The macro-economic policy surprises of the early 1990s generated an upheaval in terms of some firms doing well and some firms doing badly. After 1993, the pace of reforms has been much slower.
- From 1995 onwards, the structural improvements in securities markets (electronic trading, the clearing corporation, the depository) have reduced the costs of trading, and helped foster market efficiency.
This evidence tells us that market P/E ratios, and hence market prices, do reflect nontrivial information processing about the future. Recall that P/E ratios shot up after the early reforms, and remarkably enough, two consecutive years with nearly 100% net profit growth followed. Similarly, recall that P/E ratios dropped sharply in 1995, and net profit growth slowed down dramatically in the following years. There is more method in this madness than we commonly ascribe to it.
The vision of market efficiency is quite different from the simplistic supply/demand stories that are found in newspapers or among market practitioners. Every newspaper supplies the reader with the great insight that stock prices rose today because there was a lot of buying. It is a small leap from this remarkable insight to think that FII buying has a major impact upon stock prices. On the contrary: (a) FII transactions are miniscule compared with the size of Indian stock market, and (b) FIIs (as much as any Indian speculator) are concerned with buying stocks when they appear to be cheap (i.e. when future growth prospects seem to be bright when compared with the price) and selling stocks when they seem to be expensive (i.e. when future growth prospects seem to be poor when compared with the price). FII buying may be a symptom, but it can hardly be the cause of stock price movements.
Back up to Ajay Shah's media page