What works in Zurich: Dividend yield

After analyzing the PER, let's move on today to his little brother, the dividend yield, or more commonly "Yield" in English. I have to admit that putting this ratio under the microscope within the Swiss market makes me feel like a spoiled brat who is being offered the latest Playstation. I have read and written so much on this subject since the launch of this site in 2010 that I feel like nothing can surprise me anymore.

I have already told you several times about the mythical example of Bank of America (NYSE: BAC) during the subprime crisis. For those who missed an episode, here is the summary. The bank American stock was showing a yield of over 10% in June 2008. The price had lost almost half of its value since the highs reached almost two years earlier. The investor who bought BAC at that time, believing he was getting a good deal, found himself nine months later with a stock that was paying almost no dividend and whose value had been divided by a further 3.5 since the purchase.

This is what is called a yield trap or "Yield Trap" in English. We had seen the equivalent phenomenon for the PER ("Value Trap") during our analysis on the French market.

Yield Trap - BAC example

BAC is far from being an exception. Eastman Kodak and General Motors were among in 2002 among the "Dogs of the Dow" (Dow Jones stocks that pay the highest dividends). Their story ended even worse as they both went bankrupt a few years later.

All this to tell you that focusing only on the high yields in dividends, it can be a very bad idea. In any case, that's what I've been writing to you for almost 15 years. However, something is tickling me deep in my gut: what if I'm wrong, at least for Switzerland? After all, the results for the PER on the Swiss market are surprisingly good, so why not the Yield? Since the dividend is nothing more than a share of the profit paid to the shareholder, what works for one must also work for the other, right?

Dividend yield

Dividend yield is even easier to calculate than P/E. You simply divide the annual dividend paid by the stock price.

If it is paid quarterly, as in the USA for example, the last dividend is multiplied by four. We can also use the variant of the last twelve rolling months. However, I prefer the one that consists of annualizing the last dividend paid. This gives a more accurate picture of what the investor will receive during the coming months.

For the backtest, stocks were ranked from lowest to highest return and then divided into quintiles. The process was repeated every twelve months since 2004 and the performance of each quintile was analyzed.  

Global market

The stocks that display the best dividend yields tend to outperform the market over the period analyzed (11.71 vs. 8.47). Conversely, those that are the least generous to their shareholders underperform quite significantly compared to the market (3.56 vs. 8.47). This result is interesting and quite close to what we saw for the PER in Switzerland. The only notable difference is in the 4th quintile, which is slightly below the 3rd. The latter is not that far from the 5th quintile (10.63 vs. 11.71).

The result is therefore a little less clean than with the profits, since this time we do not find perfect progression across the quintiles.

The last quintile shows a nice annual return, with 11.71%. This is very good for such a simple strategy. But then... Have I been wrong all these years? Not so fast. Let's continue our analysis first.

Comparison with peers

The backtest of the PER with peers, within industries, had given results to be taken with a pinch of salt. The lowest PERs compared to peers certainly tended to outperform, but the progression across quintiles was less regular. We find for the dividend yield a result that goes in the same direction, but with an additional particularity: the best quintile is the third. In other words, companies that pay moderate dividends compared to their competitors obtain the best performance on the stock market.

As with the PER, we must remain somewhat circumspect with regard to this result due to the relatively small size of the Swiss market, and a fortiori of its industries. However, let us keep this image in a corner of our mind. We will talk about it again a little later.

Big and Mid-Caps

The dividend yield backtest for large and mid-caps gives us even stranger results than with the PER. Here too we have a sample size problem, since each quintile only includes about ten companies.

Among large and medium-sized companies, those that pay moderate dividends (2nd and especially 3rd quintile) have performed best on the stock market. This reminds us of the comparison within industries.

Small, Micro and Nano-Caps

Logically, the situation should be clearer for smaller capitalizations. This is in any case what we saw with the PER.

However, we find the same phenomenon as with larger companies and with the comparison within sectors: the third quintile passes ahead of everyone. Certainly, the 5th quintile displays a nice result, with 12,71%, but moderate dividends do even better, with 13,29%.

This time, the sample size excuse no longer holds water, since each quintile contains three to four times more companies than in the analysis with the Big and Mid Caps.

So, 3rd or 5th quintile?

We saw that when we looked at the entire market, the highest yields gave the best stock market performance. But the moderate yields, with the third quintile, were right behind. And then, every time we started to go down a level, within industries, among large and mid-caps or among the smallest companies, the moderate dividends worked better.

This gives the impression that there is another variable at play. Since the beginning of the dividendes.ch adventure, when I write that we should favor moderate dividends, I always add a condition: growth. Companies that pay juicy dividends can hardly afford to increase them, unlike those that remain reasonable in relation to the payment to shareholders.

If we add the dividend growth parameter to the yield parameter, we get a different picture:

Note that in both backtests above, I compared which approaches worked best for each strategy: Highest Increasing Returns in the Swiss global market, with respect to increasing moderate returns within industries. Note that if we had taken moderate and increasing dividends in the global market, we would also have beaten the high increasing yields (11.77% vs 10.66%).

Dividend growth criteria used:

• average annual growth over 5 years > 2%
• Growth of the last 12 months compared to the previous 12 months > 6%
• Growth in the last half-year compared to the same half-year a year ago > 6%
• Growth of the last year compared to the previous year > 0%

We deduce that:

• Moderate and growing dividends are more profitable than high yields alone
• Moderate and growing dividends are more profitable than high growing yields
• High increasing yields are less profitable than high yields alone (this can be explained by the fact that continuing to increase a dividend that is already very high can disadvantage investments in the company and its debt)
• Moderate and increasing dividends are less risky than high increasing yields (lower maximum loss, beta and market correlation)
• Reward/risk ratios (Sharpe and Alpha) are better for moderate and growing dividends than for high growing yields

Conclusion

In Switzerland, we have seen that a very simple strategy, targeting high dividend yields, gave good results. However, moderate dividends, on the global market, come in just behind. The latter are even more effective, when we focus on smaller samples.

By pushing the analysis a little further, we saw that another important variable had to be taken into account: dividend growth. When we do, the results become significantly different: moderate and growing dividends are more profitable and less risky than high yields.

With 15,72% per year on average since 2004, moderate and increasing dividends are for the moment the best strategy that we have tested on the Swiss market (and we are only at the beginning of our approach!). This is really a very good result for a relatively simple approach.

It is even possible to further optimize this result by ensuring that dividends are well covered by profits (growth in profits over the last 12 months compared to the previous 12 months > 3% and average annual growth in profits over the last 5 years > 0%). With this, we increase the average annual profitability to 19,42%!

There is, however, a rather big drawback: given the relatively small size of the Swiss market and the restrictive criteria used, the portfolio contains on average only four stocks for moderately growing dividends and even less than 3 stocks for moderately growing dividends covered by earnings growth. This means that this approach is difficult to use in real life. It is necessary either to use it in parallel with other strategies or to expand the sample size by taking other markets into account.

By the way, what about France? We will see this in our next article.