Price to Earnings ratio
As I mentioned earlier probably the most used metric to determine a value of a company is Earnings. This is why the most heavily used tool to determine if a stock or an index is overpriced is the Price to Earnings ratio, or some kind of modified version of this. The reason behind this is quite obvious. The Price to Earnings (PE) is the inverse of returns, so the higher the PE the lower the returns, that we can expect. A PE of 10 means 1/10 in returns or 10%, a PE of 20 means an 1/20 in returns or 5%.
This the most widely used valuation metric, so it is interesting to see if it is actually that useful.
I analyzed the data from 2000 to 2015 and also a dataset going back to 1881.
This firs graph illustrates the change in the Price to Earnings ratio starting from 2000. This time period is quite limited, however it contains the full deflation of the Dot-com bubble, the bull market until 2008, and the deflation of that boom, and the growth of the market since.
The data starts at the height of the Dot-com bubble. The valuations are quite high, near 30 compared to the historical average of 15. The price of the index starts to drop, however because of the recession, earnings drop as well, and that is why in 2002 the PE ratio is even higher than it was in 2000. Prices came down a lot, but earnings just collapsed.
Starting from 2002 prices and earnings started to rise. Earnings raised faster than prices, so the PE ratio came down. The next peak in PE ration was in 2009 at the absolute lows of the market.
From this plot it is clear that the Price/Earnings ratio can increase dramatically during a recession when Earnings collapse.
This observation is strange and contradicts conventional wisdom, that low PE ratio means that we can expect high returns in the coming years.
Viewing the scatter plot it is hard to find any correlation between the PE ratio and the returns in the next 1 to 5 years, and the week correlation that do exist actually signals, that the higher the PE ration the bigger the returns. This is because of the recession years when Earnings collapse more than stock prices, and the Price/Earnings ratio rises.
Another observation is that the correlation is stronget if the investment period is longer.
Going back, and using all the data starting from 1881 the picture changes dramatically. This graph shows more than a century of data. Based on this data the historical average Price to Earnings ratio is 15, although this average is higher in the past couple of decades.
The correlation in this case is weak, just as for the shorter period, however in the opposite direction. This is strange because based on conventional knowledge I expected a strong return, when the PE is low stocks are underprized, and when the PE is high stocks are overpriced.
Based on these correlation plots it is not the best idea to use the PE ratio as a measure of a value of the stock market. This may be different for single stocks though the highest gains on stock are made when a company that has problems makes a turnaround.
Probably it is a good idea to split the data into bull markets and bear markets, and analyze them separately in that case the correlation would be stronger, still this means that the PE ratio cannot be used in a simple equation to determine if the market is underpriced or in a bubble.
If we take a longer investment horizon, for example 5 year, the picture does not improve.
It is important to notice, that because the PE is heavily used maybe it is already included in the selection of the stocks that make up the index, or it is priced in optimally that is why does not offer any information about the future price of the index.
Future analysis of this metric will come to extract as much information from this metric as possible.