Tuesday, January 04, 2005

Christmas reading - Kahneman and Tversky

I realize that the holiday season is over (boy do I realize it, since I am back in the classroom as of today), but I thought I would report on a couple of books I read over the break.

Choices, Values, and Frames
by Daniel Kahneman and Amos Tversky

This is a collection of papers on cognitive psychology as applied to economics and decision making. Kahneman and Tversky were pioneers in what is now known as behavioral economics. The papers are clearly written and offer a nice introduction to this field.

As you can tell from some of my previous posts mocking the idea of strongly efficient markets, I am not myself a big believer in the rationality or analytical power of individuals who make up markets. However, this does not mean I doubt the power of markets as aggregators of information, and on this subject I highly recommend Surowiecki's book The Wisdom of Crowds. Certainly, it is not excluded that a machine could function usefully even if many of its subcomponents are faulty :-)

I won't try to summarize everything in the book, which at minimum provides a nice compendium of cognitive quirks uncovered in surveys and laboratory observations. One that I found quite interesting is the prevalence of risk aversion, extending even to very small wagers. The typical person's utility function, as extracted from experimental observation, is not at all linear even for small amounts at risk - the pain of a loss exceeds the pleasure of a gain of equal size.

It seems reasonable to assume that individual utility functions are logarithmic - a gift of $1000 brings proportionally less pleasure to a billionaire than to a millionaire. But there is good evidence that this is not so, and indeed M. Rabin, in one of the papers, shows that the aversion to small losses cannot be accommodated by such a model of utility. It seems that our brains constantly set reference points ("zero points") relative to which we measure loss or gain.

If indeed our brains constantly (say on monthly to yearly timescales) reset their financial zero points, and are naturally risk averse, then volatility itself is a painful phenomena. One interesting idea of Rabin's is that this explains the (unnaturally?) high value of the equity risk premium.

In my opinion, investors have been trained in recent decades to believe that small losses are only temporary and that equity markets almost always go up over the long run. This makes them more resistant to loss aversion and leads to a smaller equity risk premium (and correspondingly high P/E ratios).


Anonymous said...

OK< OK, I read them as well :)

Anonymous said...

OK, OK, I'll read them as well :) These comments have a mind of their own. So, the machine works, though the parts, this part, may be faulty.


Anonymous said...


January 04, 2005

Equity Returns in the Future
By Brad DeLong

I tend to start from the equation:

r = E/P (rs-r)(E-D)/E rs(E*-E)/E

where r is the expected real rate of return, E/P is the earnings yield, rs is the rate of return the average company earns on its reinvested earnings (which Glenn Hubbard and company argue is significantly in excess of the market return becuase of various information and signaling problems), and E*-E is the gap between accounting earnings and the business's real Haig-Simons earnings (which Eric Brynnjolffson and company argue is significant because of all the trial-and-error investments in organizational form that are inaccurately counted as operating expenditures). This makes me think of the current earnings yield of 5% as a lower bound to expected equity returns, which I tend to see as 5.5-6%. And when I compare this to a real long Treasury bond return of 2-2.5%, and when I meditate on long-run inflation risk to which nominal bonds are uniquely vulnerable...

Well, the upshot is that the 3-4% annual expected equity premium return that I see still seems very large to me: to correspond to the preferences of a 62-year-old male expecting to spend his wealth in the next fifteen years or to the preferences of a money manager for whom reporting a big loss in the next four years is a career-limiting move, rather than to the risk preferences of the economy considered as a frictionless social welfare maximizing machine. And this means that there is still a powerful, powerful case for stocks for patient investors with a long horizon of a quarter century or more. If you can wait a quarter century, stocks do not look like a sure thing relative to Treasury bonds, but they do look like a 90% thing.


Anonymous said...

If there is a flw in the argument, I do not find it. And, yes, I find our magixcal thinking on Social Security saddening.


Anonymous said...

"[S]ocial security privatization is likely to bid up equity prices and depress their future returns. Imagine the following analogy: one day, foreign investors wake up and decide to increase their portfolio allocation to US equities. The result may be a buoyant stock market, but to what extent does this increase real value creation in our economy?"

I agree completely, there is every reason to believe that investing payroll tax revenues in the American stock market will buoy prices in the short term only. An important argument.



nonk9 said...

Rabin's comically overcomplicated "calibration theorem" in Econometrica (2000) and Rabin & Thaler's (2001) "lite" version are mathematically flawed. There are a half dozen papers on the web by economists sniffing around the impenetrably dense math trying to diagnose what the flaw is.

R&T are still pretending their math is sound, because once they admit that it isn't, their chance of winning the brass ring in "economic sciences (sic)" will be about zero.

A prediction: By the end of 2005, economists will agree that contrary to popular belief, Rabin did not prove mathematically what K&T proved empirically. Rabin claims to have mathematically shown that expected utility theory is dead and the new king is prospect theory. But the flaw he identified in EU cannot be fixed with loss aversion/prospect theory.

Blog Archive