One of the biggest stories of the coming decades will be whether pension funds, both public and private, will be able to earn an adequate return on their investments. If they do, great. If they don't, there could literally be trillions of dollars of unfunded liabilities, setting up waves of disappointed retirees who learn how fickle the words "guaranteed pension" can be.

Last month, I sat down with Robert Arnott, CEO of Research Affiliates and one of the brightest financial minds of recent history. I asked him his take on whether the return assumptions pension funds are using these days are reasonable. Here's his response, which includes some important financial wisdom. (Transcript follows.)

Morgan Housel: We spoke with the chief investment officer of a major pension fund recently who has a 7.5% assumed rate of return going forward, and he readily admitted that the fund will not meet that within the next five or 10 years, but he said over the next 20 or 30 years he thinks its feasible. When you look at private pensions and public pensions, what rate of return do you think they should be using for the next 10, 20 years?

Robert Arnott: Again, I'm going to be provocative. I think the right assumption is to work off the Treasury yield curve, because anything you earn beyond the Treasury yield curve, which of course tops out in the high twos right now; anything beyond that is earned by dint of either a risk premium or an alpha. And I think pensions shouldn't count on risk premium or alpha until it's been earned.

Now that, in turn, means that if you calculated based on the Treasury yield curve and you're expecting you can do better than that, that you should be content with less than 100% funding. So if you aim for calculating your funded ratio based on the Treasury yield curve, if you're 70 or 80% funded, that's OK. unfortunately, using 7 or 8% is the norm in the public pension fund community. They're not going to earn that unless they invest rather wildly out of mainstream. And if they do so, there's a reciprocal risk of sharp underperformance.

So with a reasonably conventional portfolio, let's say 60/40 stocks/bonds, comprises three-fourths of the portfolio and only 25% is invested outside of mainstream. If outside of mainstream gives you an opportunity to do 3 or 4% better, then you've just added 1% to your return expectations, and anything beyond a 4, maybe 5% return expectation, is heroically optimistic.