Paul C. Lipson speaks in irreverent tones about Modern Portfolio Theory and related practices commonplace at big pension fund sponsors.
"We don't go through that mean/variance nonsense," said the chief investment officer of the nearly $9 billion Federal Reserve Employee Benefits System, Newark, N.J., talking about its $5.8 billion defined benefit fund. (The system also has a $2.8 billion defined contribution fund.)
"We don't have a strategic asset plan," he added.
Is he wacky? No. The way the system oversees its defined benefit fund, he acknowledges, is unconventional.
"We delegate to our managers the very important asset allocation decisions," he said. "Each manager is selected based on asset allocation capability and equity and fixed-income capabilities."
"We select managers who use different asset allocations methodologies," he said. "That gives us a kind of diversity no other pension plan has - an asset allocation diversity." The fund has eight managers.
"All we do is set broad parameters" for the managers. "We give them parameters - 40% to 80% equities" and the rest fixed income. "You can be far more nimble (using this approach) than pension funds that have to go through their boards to change their allocations."
"Virtually all pension funds ...use mean/variance analysis" - a tool used to optimize a portfolio's expected return and risk - "in developing their strategic allocations," he said. "What it is, in effect, is like the general fighting the last war. Mean/variance is laden with historical returns. So (sponsors) wind up outsizing positions in assets that performed well in the past, but not necessarily in the future."
"The mean/variance of (Harry) Markowitz's 1952 paper always has the dead hand of history," Mr. Lipson added.
This model for overseeing the pension fund has roots going back to 1934, when the Fed set up the plan. The issue of asset allocation raised a potential conflict of interests, he said. How could the Fed make such decisions for the fund when it has a powerful influence on the economy? "We give money managers authority pension sponsors would normally exercise themselves," said Mr. Lipson.
He said the system set its equity parameters "to have a portion not in a risky asset. ...It's not a precise analysis." The fund's benchmark is 60% Standard & Poor's 500 stock index and 40% Lehman Aggregate bond index.
As a composite, the fund is now 58% equities. That "is not a meaningful number," he said. "We don't influence that allocation at all. But if you ask why a manager went from (say) 80% to 70%, then as a fiduciary I have to know. I talk with them."
Mr. Lipson doesn't eschew sophisticated techniques for evaluating managers. Tools he uses include a Wilshire multifactor model and a price-to-book value model based on research by academics Eugene F. Fama and Kenneth French.
The fund has no private equity or non-dollar-denominated securities. International investing has high transaction costs and only a small fraction of the risk reduction academics claim, he said. "But we have to keep an open mind," he said. "Every two years or so we review (international investing)."
The system hasn't had to make a contribution to the pension plan since 1986. Its annual report, available to the public, details of its performance history and its managers. In 2001, the fund returned -3.2%, bettering the -3.7% return of its 60/40 benchmark. "We have done well," he said, "without being faddish."