Your Author's Mistake: Rushing to a Conclusion

Ready to Believe The Financial Times alerted me to a German study demonstrating that "smokers--a group of people who lack self-discipline--earn lower returns than nonsmokers."

When I read that note, I nodded fervently. As Jason Zweig summarizes, many researchers have concluded that Vulcans are better investors than Andorians. (The latter being a touchy breed.) The very model of a major investment general, Warren Buffett, is famous for remaining calm while the investment markets seethe. His demeanor permits him to succeed, while this guy fails. Every seasoned investor appreciates the value of patience.

Inspired by The Financial Times' item, I decided to write a column about how investing rewards ants, not grasshoppers. As with smoking, making a rash investment trade is damaging. The logical connection between the two acts is obvious. In each case, the subject does what feels good today but brings regret tomorrow.

The problem is, the German study doesn't show what The Financial Times purports. Overall, the study's smokers achieved slightly higher returns than did its nonsmokers. As their portfolio risks were also lower, they unambiguously enjoyed better results. I endeavor to be creative, when possible, but chastising smokers for failing when they had in fact succeeded was taking ingenuity a step too far.

Help Helped "Smoking Hot Portfolios? Self-Control and Investor Decisions" (Charline Uhr and Andreas Hackethal, Goethe University Frankfurt; Steffen Meyer, University of Southern Denmark) evaluates the results of 19,000 German investors over the seven-year period from 2012 through 2018. Because those investor accounts were held in an online brokerage that issued credit/debit cards, the authors were able to determine who smoked by tracking their subjects' cigarette purchases. (So much for privacy!)

Although the study ostensibly concerns smoking, its chief finding is that investors who received professional help fared better than those who went solo. This, surely, is a sentence that financial advisors will highlight--but it should be noted that the authors' definition of professional assistance includes both: 1) meeting with an advisor and 2) owning mutual funds rather than buying stocks directly. The paper does not indicate which of those two paths was more important.

What it does indicate, though, is that those who obtained help were better diversified, traded somewhat less frequently, and enjoyed higher returns. That those characteristics also described smokers' portfolios, in aggregate, isn't because smoking improved investment decisions. It is, instead, because smokers were more likely to seek assistance. Their self-directed portfolios performed even worse than did nonsmokers', but proportionately, there were fewer of them.

The authors reach two main conclusions:

1) When left to their own devices, smokers are worse investors than nonsmokers who are left to their own devices, because smokers lack self-control.

2) Fortunately, smokers tend to be aware of their limitations, and thus are likelier to employ investment professionals.

It's Complicated Those findings require some caveats.

Of the study's 19,000 total participants, about 5,000 were identified as smokers, with the latter group subsequently being split into two camps: 1) those who received help, and 2) those who did not. Although the paper's main conclusions are statistically significant, I hesitate to generalize them. They are based, after all, on a single, modestly sized test group, from one brokerage firm, in one country, during a portion of one decade.

In addition, there are meaningful differences between the characteristics of the pool of smokers and that of the nonsmokers. On average, those in the smoker group were younger, earned a lower income, and were likelier to be employed. They were also likelier to be male, although in either case men were so dominant that the study can't really be said to apply to women. All those distinctions were significant, at the 1% level.

Which means that to evaluate smoking's effect, the authors needed to adjust for the differences. For example, if younger investors tend to assume more risk than retirees, then the analysis must adjust for that fact. For such problems were multiple-regression techniques developed. Nevertheless, controlling for those factors, many of which are interrelated, is a messy task. The computer program loudly provides precise calculations, then quietly whispers the error terms.

(Table 7, "The Effect of Low Self-Control on Demand for Advice," provides a sense of the quantitative adventure. The paper finds that receiving investment advice is correlated with age, wealth, and smoking, at the 1% significance level, and with being female, at the 5% level. That regression's R-squared--which measures how well the model fits the data--is 0.01%. In contrast, the R-squared for the notoriously idiosyncratic Fairholme Fund FAIRX against the S&P 500 is 0.14%.)

This discussion assumes that the authors pinpointed the relevant factors. Perhaps the smokers differ in a critical but unidentified fashion from the nonsmokers, and that difference creates the results. It could be, for example, that smokers tend to hold public-sector positions, and that those who work in government don't much understand the private markets, making them clumsy investors. Fortunately, they recognize their weakness and are therefore more likely to seek professional help. In such a case, smoking would be a side effect, not a cause.

I wouldn't bet heavily on that particular conjecture being true, but it's certainly possible that the paper's conclusions could be altered by additional analysis. By necessity, many potentially meaningful factors could not be evaluated.

A Cautionary Tale The Financial Times' comment, while incomplete, was accurate in its own way. Nor do I have any quarrel with the paper itself (which, it should be noted, is in draft form; it is not yet finished). It is clear. Nevertheless, because my first encounters with this topic appeared to confirm what I thought to be true, I rushed to judgment.

Fortunately, I could not write this column without reading the paper, which corrected my error. However, the incident serves as an important reminder. Today's investment studies tend to be abstruse, as the simple subjects were covered long ago. It is tempting when encountering such complicated works to simplify the analysis, by seizing upon what we think that we already know. That is not how to learn.

John Rekenthaler has been researching the fund industry since 1988. He is now a columnist for Morningstar.com and a member of Morningstar's investment research department. John is quick to point out that while Morningstar typically agrees with the views of the Rekenthaler Report, his views are his own.