Bear in mind that viewing it as what a reasonable person should expect is my phrasing of how that sort of method should be used, not theirs. They simply call them forecasts.

But, with respect, I think that the objections come not from believing that their forecasts are useless, but from the fact that nobody wants to believe that low returns are in the offing. In 1998 they predicted ten year S&P returns of about -1%/year, and the actual result was about zero. Given that the average US investor was expecting 20%/year, that's a pretty good estimate.

It's not as if the method makes you a permabear either: In June 2009 they predicted 12.3%/year in the next seven years, and the actual result was 12.9%/year.

The problem many people have is that when the predicted returns are low, people just can't accept that, and assume that there has to be something wrong..."the structure of the market has changed" or some such nonsense. Denial, it ain't just a river in Egypt.

Nope, things are just expensive these days. Real returns from the future were pulled into the present, both by valuation multiples and debt bingeing, so we don't get those returns a second time.

With any stock price forecast, the result you get is critically depending on the luck of the draw at the end date. The price might be crashing precisely 7 years later, or soaring. That doesn't change the fact that it's pretty darned easy to estimate roughly what the return will be from now and ending on any future date that the market is neither unusually high nor unusually low. Starting now, the answer is clearly "somewhere between pretty negative and merely well below average".


So the defense of GMO forecasts is that they were spectacularly accurate in 1998 and for a small window after the greatest financial crisis since the Great Depression. Wonderful. Now can we discuss the other 20 years?

Nobody disputes that valuations matter or that expensive markets tend to produce lower future returns. The dispute is whether GMO particular definition of “normal” has been remotely useful for predicting those returns.

The most fascinating part - when GMO got 1998 right, it was evidence of a brilliant model. When it got the subsequent multi decade badly wrong, it was apparently just the unfortunate “luck of the draw” at the seven-year endpoint. What a remarkably convenient arrangement!

As for dismissing structural changes in corporate America as “nonsense,” perhaps someone should tell Microsoft, Apple, Google, and other globally dominant businesses that their economics must eventually conform to the profit margins of 1970s industrial America. Apparently, capitalism is allowed to innovate in everything except GMO spreadsheets.

They should do what reasonable people do - updating one's assumptions when the evidence changes.

The real question is not whether valuations could revert to historical averages. Of course they could. It is whether assuming that they must revert within an arbitrarily chosen 7 year window is a useful investment framework.

You say it's “pretty darned easy” to estimate future returns. After so many years of systematically underestimating U.S. equities, perhaps the evidence suggests it's not quite as easy as you think.

Now comes my really really favorite part of this argument - that investors who ignored GMO and compounded their wealth were supposedly in denial, while those who repeatedly predicted returns that never materialized were the rational ones!

Denial, as you say, ain't just a river in Egypt. Apparently, it also runs through the valuation department.