What works in Zurich: the Price-Earnings Ratio

Today I am inaugurating a new series of articles devoted to the micro and macroeconomic factors that influence stock prices. I have already spoken several times on this site, as well as in my book, about many of them:

• valuation ratios: P/E, P/B, P/S, P/FCF, EBITDA/EV, FCF/EV, Yield, Value Composite, etc.
• qualitative ratios: ROE, ROA, ROIC, profitability, Piotroski score, etc.
• (il)liquidity of assets
• company size
• momentum
• unemployment rate
• volatility
• etc.

The title of this series of articles is directly inspired by the reference work "What Works on Wall Street" by James O'Shaughnessy. The author evaluates the impact of various indicators on stock performance. Does what works on Wall Street work the same here? This is what we are going to see, starting with the best-known stock market ratio.

The Price-Earnings Ratio

For those who have just arrived from another planet or those already affected by Alzheimer's, the PER is equivalent to the share price divided by the earnings per share. At first glance, it seems quite simple, but when you get down to it, it quickly becomes open to interpretation. Profit, yes, but over what period? That of the last half-year, that of the last twelve months or that of the last full fiscal year? Profits can also be quite easily manipulated by all sorts of accounting tricks. If you don't pay attention to the little footnotes, you can quickly be misled about the goods, because of the extraordinary elements.

For my backtests, I will rely on the earnings per share reported for the most recent half-year (annualized), excluding extraordinary items (this allows us to erase what is not recurring in the company). I chose the most recent half-year because it gives slightly more convincing results than the last twelve months, even if the difference is not significant.

Stocks were ranked from highest (most expensive) to lowest (cheapest) PE ratios and then divided into quintiles. The process has been repeated every twelve months since 2004 and the performance of each quintile analyzed. JI also tried to rebalance twice a year, thus aligning myself with the half-yearly results. The result was, however, slightly lower, although the difference was not significant. 

Global market

Good news! Value investing is not yet dead in Switzerland. Even with a basic ratio known as the PER, we obtain very interesting results. The 1st quintile, which includes the most expensive stocks (highest PER) has significantly underperformed compared to the other quintiles. The profitability, with 3.42% is more than twice lower than the entire Swiss market (8.47%).

From the 3rd quintile, profitability exceeds that of the Swiss market. This indicates that we can already beat the market just by avoiding the most expensive stocks.

The last quintile shows a very nice annual profitability of 11.36%. This is frankly not bad at all for such a simple strategy. Especially since it owes nothing to chance. There is a close (inverse) relationship that exists between the PER of a stock and its performance. The higher we go in the quintiles (the cheaper the stock), the more the profitability increases.

In theory, the easy money policies followed by central banks since 2008 (including the BNS with its negative rates) should have caused capital to flow into growth stocks, to the detriment of value stocks. However, despite this, this strategy worked very well throughout the backtest period.

A low P/E means that the market expects little future earnings growth, while a high P/E means that investors expect strong earnings growth. However, we see here that stocks that no one expects anything from do much better than those that are highly sought after. This may seem paradoxical. However, it is quite logical when viewed through a "value" prism. When we don't expect much, we settle for what we have, which is the current value. Anything beyond that is a pleasant surprise, which will immediately translate positively into the stock price. Conversely, when we expect a lot, the chances of being disappointed are significant. Even a very good result can be a bad surprise if it is lower than expected. Growth stocks (with high P/Es) are thus regularly punished by the market, even when they publish juicy profits, just because these are lower than analysts' estimates.

Comparison with peers

It is often recommended to compare ratios such as the PE ratio with similar companies in the same industry or sector. Various factors specific to the company's activity, such as capital structure, profitability and profitability, are in fact influenced by the field of activity and in turn influence the PE ratio. If we perform a backtest by distributing companies into quintiles according to their PE ratio relative to other companies in the same industry, we actually obtain an even better result for the 5th quintile.

However, the progression across the quintiles is less clear and regular. The first quintile shows almost no difference compared to the second. It is not even that far from the fourth. This can be explained by the relatively narrow Swiss market. By targeting industries, we end up with even smaller samples that can partly distort the results.

Big and Mid-Caps

What about if we focus on large and mid-cap companies? Oddly enough, we find something quite similar to that on the overall market, which compared the PER within industries. The peak is even more marked for the 3rd quintile. It even exceeds that of the lowest PERs. This seems a bit of a mess at first glance, but the reason is quite simple. As was the case with industries, the number of big and mid caps is relatively low in Switzerland. Each quintile therefore only has about ten companies. It is enough for a few of them to have posted exceptional performance, or conversely disastrous, to confuse the issues.

For Swiss Big and Mid Caps, stocks with lower PEs outperform those with higher PEs.

Small, Micro and Nano-Caps

The result is much clearer with the smallest capitalizations, which are much more numerous. Here we find a regular progression through the quintiles, which give more robustness to the backtest result. The slope is even more pronounced, with a 1st quintile much less efficient and conversely a 5th quantile which displays a very nice annual performance of 12.61%. Here too, it is remarkable for such a simple strategy.

We note that it is not the size effect of companies (their capitalization) which explains this phenomenon, because the universe of small Swiss capitalizations has performed "only" by 8.37% during this period.

Conclusion

One would have thought, over time, that such a basic approach would no longer work. An efficient market is supposed to integrate this kind of thing. However, we see that using a ratio as simple as the PER allows us to beat the market quite clearly.

The advent of the Internet and the growing sophistication of finance over the last twenty years has not changed anything. It is a safe bet that AI will not change anything either. It must be said that there is no need for highly developed intelligence, whether human or artificial, to follow this strategy.

What about in France? We will see this in our next article.