Golfers Are Better Investors ... Right?

Data Mining On my desk what may be the oddest finance paper ever written.

Heard on the Green: Sociability, Golf Courses, and the Performance of Institutional Investors, an unpublished paper by two Singaporean professors, claims to show that professional investors benefit from playing amateur golf, as the knowledge gained from their social network improves their investment performance. That might be true, I suppose. However, the authors lack relevant data, making their article a strange journey indeed.

For starters, the authors do not know which investment managers play golf. Thus, they cannot compare the performance of managers who play golf against the performance of those who do not play golf. They cannot do so directly, and they cannot do so indirectly. In no way, shape, or form can the authors create two subgroups, one of golf-playing managers and one not, and demonstrate how the first group beats the second.

Mere mortals would be daunted by this data vacuum. But these authors are made of sterner stuff. Lacking knowledge about managers' golf habits, they take a side path by measuring how far managers live from golf courses. The assumption is that the closer that managers live to golf courses, the likelier they are to play golf, and thus the likelier they are from their golf outings to learn valuable information that makes them better at their jobs. The subgroups therefore consist of 1) managers who through geography are deemed to be more likely to play regular golf, and 2) managers who through geography are deemed to be less likely to play regular golf.

That is one Moby Dick of an assumption.

That's only the beginning. Because--you guessed it!--the authors also do not know where investment managers live. Who this side of NSA has a database of home addresses? What the authors actually know is the location of the managers' employers, collected from SEC-required 15F filings. So, their assumption is that managers who work near golf courses are deemed to be more likely to play golf regularly.

But not just any courses. The paper extends even further out on a limb by assuming that managers learn the most from their golf games when they play at the most-prestigious courses, as measured by the amount of their greens fees--the higher the fee, the more prestigious. What's more, these top courses are the ones that are relatively open to the general public by having "reciprocal guest policies," rather than highly exclusive and closed courses.

(That managers who play golf tend to frequent prestigious courses seems obvious. However, that they would benefit more from semipublic gossip at relatively open courses than they would from potential insider tips given to them at the most-exclusive clubs seems anything but obvious.)

Let's review: Playing amateur golf helps professional investors, as measured by the distance from an investment manager's workplace (as recorded in 15F filings) to various golf courses, hand-selected by the authors, that are regarded as prestigious and that have relatively open guest policies.

The "distance-to-golf" measure, as the authors call it, is created by calculating the distance in miles from the workplace to each of these selected golf courses that exists within a state, averaging those several distances to get a single figure, and then standardizing that figure to account for the different sizes of different states "by dividing by the value of state median distance-to-golf."

Convinced Yet? Aside from the various assumptions embedded into choosing only certain golf courses as being meaningful, assuming that the distance from those cherry-picked courses to the manager's workplace is an important factor in the manager's golf habits (and learning), and assuming that an average of those various distances is more important than, say, the distance to the closest single golf course, there is that awkward matter of using the workplace address as a proxy for the manager's home location. By that reasoning, every manager who works at Fidelity shares the same address.

Of course, the problem is much worse than that, because effectively all Boston-based investment-management companies share the same location, given that their offices are but a hop and a skip apart. Managers who work at Fidelity, Putnam, Wellington, MFS, Hancock, and Eaton Vance share a single data point. As do all midtown New York managers, all downtown New York managers, and all Greenwich, CN, managers.

To their credit, the authors do acknowledge the problem with assigning portfolio managers locations according to 13F filings. They claim to address that issue by studying how managers perform when a major golf tournament is held nearby. The authors find that "investor performance spikes in the years in which major championships [i.e., the U.S. Open or PGA Championship] are held within the state."

That is interesting, I guess, but I don't see how that relates to either the price of tea in China or whether playing golf regularly improves a portfolio manager's results. For one thing, as is customary with this study, the authors have no direct data. They don't know whether the presence of a U.S. Open causes managers to play more golf, or to meet while watching the Open, or engage in other activities that somehow goose their investment returns. For another, the observations are a mere 44 data points--fewer than the data points for the spurious Super Bowl Indicator.

Finally, several of the data points are effectively corrupt because of the problems inherent in the notion of "sharing the same state." For example, the 2013 PGA Championship held at Oak Hill in Rochester, NY, is treated as a home-state event for New York City-based investors, although it was several hundred miles away in the other corner of the state, while the 2005 Baltrusol event just across the border in Springfield, NJ, was not.

To cap things off, the authors claim a connection between weather and investment performance. They find that when precipitation in the managers' regions is above average for a three-month period, that performance for the manager pool is weaker, at a statistically significant level, than when precipitation is below average. This, naturally, occurs because of the amount of golf that is played. "We find that their outperformance occurs during times of low precipitation around golf courses, evaporating when bad weather keeps golfers off the greens."

That was when I figured the paper must be a spoof. But I am assured it is not. The authors are indeed employed at Nanyang Technological University's business school, they have Ph.D.s, and they are putting their quantitative tools to work.

They might have been better off playing golf.

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.