Friday, July 31, 2009

My Book is Obvious AND Wrong

I haven't gotten a lot of public feedback on my book, but the private feedback is rather circular. On one hand, there are those saying my findings are wrong. I'm 'saying the earth is flat', in one irate economist's view. My empirical findings are not rigorous, in that I'm using incorrect statistical tests and if I used a more powerful test I would not reject the hypothesis that risk explains returns as standard theory dictates. Indeed, many of the articles I reference in my support leave open this possibility. I admit it's possible. Anything's possible. It's just improbable. The current paradigm finds parochial risk factors for each asset class (a different one for currencies, cross sectional equities, bonds), without any intuition. Further, adding more complexity to any test lessens its statistical power, and so the 'failure to reject' would give one solace if the theory kinda worked. The theory sucks, it mainly has the wrong sign (higher risk assets in any asset class, in general, have lower than average returns!)

On the other side, I get comments like 'this is obvious', it's 'too simple', or 'everyone knows that'. I too think it is obvious and simple, but not everyone in finance knows that 1) risk and return, in general, are uncorrelated in any obvious way and 2) this is an equilibrium result if people have relative status orientation. Saying something new, true, important AND obvious&simple, is better than just being new, true, and important. But the people who recognize it is obvious and true tend to be new to the literature, and so aren't trained to not see what is directly before their eyes. Experts develop special lenses to observe reality, called a 'theory', and after a while can't see without it. They are like American Natives who supposedly could not see Columbus's ships as they appeared on the horizon because their brains were hardwired to not see what is impossible.

But, the newbies look at the data I present, which is mainly a survey of the literature and sample averages, and see it as obvious (see SSRN summary of evidence here). Those who have been in the field for years see it as untrue, those who don't have it as a primary concern, think it is obvious and rather trivial. I like the future of my idea on that score, because as they say, science advances one funeral after another.

But then I noted two Amazon reviewers found it did not have any implications for your average investor. Let me try again, because obviously anyone smart enough to buy and read my book is no fool, as this is all my fault.

In the book I go into 'finding alpha' strategies, based on the fact good strategies are highly parochial, most people fail at such endeavors, and people lie about what they are doing. But the 30,000 foot practical investing implication is rather obvious: buy 'low risk' stocks, relatively. This can be done many ways: while keeping total beta at 1.0, maximizing the Sharpe ratio, or maximizing the Information Ratio. All of these are improvements relative to the S&P500 passive benchmark at orders of magnitude greater than the lift of going from active to passive equity investing. As long as the CAPM does not work, and I argue it does not because most investors are benchmarking, this implies such obvious investing tactics that basically focus on the low volatility, and avoid the high volatility. See discussion below from my videos.

5 comments:

Symme7ry said...

Have you sent Tyler Cowen a copy of your book to review? I'd really like to see his take on it, and I hear he reads several books per day so I don't think you have to meet a very high bar for him to read it.

Anonymous said...

Maybe I'm a bit slow. But you write:
"[they say] My empirical findings are not rigorous, in that I'm using incorrect statistical tests and if I used a more powerful test I would not reject the hypothesis that risk explains returns as standard theory dictates."

Power is the ability to reject the null when the null is false. If you use a more powerful test, you can reject the null when a low-powered test failed to reject the null. But you already reject the null. So how does using a more powerful test enable you to NOT reject the null when your (presumably) less powerful test already rejected it?

Either I am missing something obvious or your critic is an idiot.

AHWest said...

I do think a typical retail investor could benefit from reading your book if they are unusually intelligent, and independent thinkers.

But I don't think most retail investors will be able to pull off stuff like long low beta, short high beta strategies. Most retail folks would be better off concentrating on avoiding common mistakes that kill them, that usually involve high fees for unattractive products.

Eric Falkenstein said...

they don't have to short! A long only, low beta, or minimum variance port generates a higher Sharpes and Information Ratio than the S&P500

J Mann said...

As a retail investor, I might look at a Vanguard fund that employed Eric's strategy, but actually picking stocks, and rebalancing every so often to keep on track, seems way too hard. Maybe I am not a sufficiently involved retail investory.

Cowen would be an interesting test. According to one of his econtalk interviews, he starts several books a day, but quits reading most of them before the halfway mark. (He has an idiosyncratic application of sunk cost and search cost theory as applied to reading).

Speaking of EconTalk, Eric, if you've got an hour of material and can interest Russ Roberts, I would love to hear you do a show. A combination of your views on capital reform generally + some book pimping would make for a good one.