Once upon a time, if your finance professor asked you what the long-term return on equities was, the answer was nine percent.

I wonder what the answer would be today.

We are in the midst of a great rethinking,the research puzzle | See my earlier effort, “rethink everything.” with many key concepts and assumptions on the table for examination, so let’s pick one up and look at it.

The asset class returns normally used in standard allocation models have changed during the last quarter century, and that migration has affected the investment choices that have been made throughout the markets. According to Ibbotson’sMorningstar | Disclosure: Ibbotson is owned by Morningstar, with which I have a business relationship. yearbook, probably the most popular source for asset class return information, in 1982 the historical return on “large company stocks” was just over nine percent. Within fifteen years, it had cleared eleven percent, due to sustained strong returns during the Nineties (when the average return for some trailing five-year periods approached thirty percent). Investor surveys taken around the turn of the millennium showed that average expectations of future returns had been ratcheted up even higher.

Despite the fact that it would seem statistically unlikely for the market to continue camping out at one end of the distribution (and despite the fact that historically markets valued as highly as those of the time tend to yield subsequent subpar returns), the only thing that had changed in most models used by financial advisers for asset allocation was that the expected return had gone up to match the higher historical average return. Even the severe whacking earlier this decade didn’t dissuade the tendency to look backwards rather than forwards to come up with inputs for the models. When in doubt, it was always easiest to point to the chart on the wall showing what would happen over the long term if you stayed the course.

A prospective look at the returns likely to be available to investors back then would have at least cast doubt on the likelihood that such certainty would make for good investment policy, a point forcefully made at the time by Rob Arnott, Jeremy Grantham, and others.  But urged on by the prevailing view, investors took on risk because that was the thing they were supposed to do, in accordance with the colored pieces of pie on the plan with their name at the top of the page.Sometimes, the allocation is more forced than urged upon the investor: Last week, my daughter’s employer switched providers for her defined contribution plan, and by default her existing assets were transferred into a “life cycle” fund. It has almost ninety percent in risky assets that are highly correlated to each other, high fees, and fourteen — count ’em — fourteen different sub-advised portfolios in U.S. equities, providing something trumpeted as “manager diversification.” It’s down forty percent for the year.

Now their hands are being held, they are being told to hang in there, and pamphlets to read are sent their way. I have seen several examples, and they are remarkably consistent in their advice, no matter the firm that produced them: “Stay calm, remember the long term, no one ever got rich panicking, and, by the way, look at how frequently these rare big moves down are followed by gigantic snapbacks.” All much better advice now that the hit that was thought unlikely has been taken.It’s too bad that the quality of such communications varies so wildly.  Take the example of one asset manager whose Fall 2008 magazine for its clients provided this historical perspective: From the end of 1970 to the end of 2007, U.S. stocks returned eleven percent, U.S. bonds nine percent (!), and “a diversified portfolio,” helped by some other assets and some unusual historical events, averaged 10.7%. The title of the article promoted using “a bond cushion for turbulent times,” and it indicated that volatility could be muted without giving up hardly anything in return.  It was nothing more than a tour de force combination of data mining and deceptive marketing — and no help whatsoever to the investor.

And so, what number is going to be put into the model now?   Will it be the newest historical average, fresh from the barber chair?  If the buy-and-hold bubble is the last in a series to burst, and sellers greet every rally, it may be that even those diminished expectations won’t be met anytime soon.  But the odds of achieving good long-term returns are changing, and not for the worse as the behavioral reflex would have it, but for the better, as the pummeling creates value over time.

Individuals, corporate pension plans, endowments, and others who had liabilities or plans based upon unrealistic expectations now have projected return forecasts that are more defensible than they were just a few months ago.  Unfortunately, they probably also have asset bases that are depleted to the extent that their original goals still cannot be met, albeit for a different reason.

For those charged with making asset allocation recommendations, there are many issues that need to be addressed, e.g., the tendency for the correlation of risky assets in supposedly different asset classes to increase at just the wrong time.  For each of the analytical building blocks, there are interesting questions to pose and debates to have, and the result should be not an answer but a process.  When it comes to projecting expected return, the approach should focus on the distribution of likely outcomes, and the choice of any single-point estimate needs to be more than plucking a number off of a grid because “it’s the best we have.”  The range of investor possibilities and opportunities is ever-changing, and asset allocation should be informed by an analysis of historical experience but not bound to it.

Start by asking yourself:  “What’s the probability of attaining that famous nine percent now?”