CHICAGO - When International Paper Co.'s pension fund conducted an asset allocation study four years ago, pension executives decided to allocate 5% of total assets to alternative investments.
"There's no magic to that, it's just a round number," said Robert Hunkeler, vice president-investments, who oversees the $7 billion defined benefit plan.
And if Stamford, Conn.-based IP were to update its asset allocation study today, officials probably would stick with that 5% target to private equities, hedge funds and venture capital. "There's no art to it, other than we don't want more than that. You could say that's our upper tolerance," he said.
And that is the general state of the art of asset allocation: Institutional investors essentially stick a moistened finger in the air to determine their optimal allocation to these hot investment areas.
Now, Brinson Partners - which long has been at the forefront of asset allocation issues - has attempted to develop a systematic approach to investing in alternatives. If embraced by plan sponsors, the new tool could help boost - or at least intellectually justify - allocations to fast-growing non-traditional asset classes.
What makes sense
The upshot: an investor with a moderate level of risk should have a target of 20% to alternatives: 10% real estate, 5% private markets, 3% hedge funds and 2% natural resources. While many endowment funds and foundations have allocations that exceed that on an aggregate basis, few pension funds approach that level. Of the Top 200 defined benefit plans, the average alternative allocation was a shade less than 8%, as of Sept. 30.
"I guess there hasn't been a lot of thought given to what the appropriate allocation should be," said Jeff Bailey, manager of benefits finance at Minneapolis-based Target Corp., which has a $1 billion defined benefit plan, of which 20% is in alternatives. Instead, pension executives, in the short run, have focused on "how much (of alternatives) can we get our investment committees to agree to hold. The amount of rigorous analysis going on has been pretty limited."
Years of struggling
Institutional investors have struggled with how to set asset allocations for alternative asset classes, going back to the 1980s when they started hiking their allocations to real estate.
The problem with real estate is that appraisals - usually performed annually - made real estate valuations appear far less volatile than they actually were. When those understated standard deviations - the normal measure of volatility - were entered into the traditional means-variance optimizer used to determine the optimal asset mix, the error was magnified.
A similar problem exists with venture capital, which is revalued only when new financing or an initial public offering is made.
The problem is that optimizers magnify any estimation errors in the data. "You realize that if you called an optimizer `an error maximizer,' you wouldn't trust it," said John Ilkiw, director, global consulting practices for Russell Investment Group, Tacoma, Wash. Frank Russell officials optimize three or four of the broadest asset classes and then carve out subsets to specialized assets.
"Optimizers require robust numbers going in," said Barton Waring, managing director and head of the U.S. client advisory group, Barclays Global Investors, San Francisco, who praised the Brinson initiative. "Garbage in means garbage out."
In the early days, optimizers typically would tell investors to make seemingly bizarre allocations, such as investing solely in inflation-protected bonds and venture capital, "because those two asset classes had perfect complementary characteristics," said Bob L. Boldt, managing director at Pivotal Asset Management LLC, San Francisco, and former head of global equities for the California Public Employees' Retirement System, Sacramento.
As a result, most consultants constrain the optimizer by capping the maximum allocation that can be invested in specialized asset classes.
In one major departure, Wilshire Associates Inc., Santa Monica, Calif., four years ago developed a proxy for private equities that blended public market stock and bond indexes. Anne Casscells, chief investment officer of Stanford University's $8 billion endowment fund, Palo Alto, Calif., said her fund follows the Wilshire methodology. Fund officials estimate correlations, based on underlying assets, and enter "reasonable return estimates," she said.
Recently, the fund boosted its policy allocation to absolute-return strategies to 12% from 5%, based on expectations that these strategies will fare better than U.S. equities. In reality, the change is far more modest because the fund's actual absolute-return investments had reached 10% of assets. Ms. Casscells would not give further details on changes to the fund's asset mix.
But, in general, investors eschew a scientific approach implied by the use of an optimizer, and instead make allocations to alternatives based more on their comfort level, pension experts said.
Brinson officials are trying to remedy this, making a quantitative assessment of the riskiness of alternatives relative to traditional asset classes. In particular, they say, investors should be rewarded for the illiquidity of alternative asset classes - something the traditional optimizer does not normally take into account.
"We're trying to look at the true underlying risks" of alternatives, said Brian D. Singer, head of asset allocation and risk management at Chicago-based Brinson.
Instead of plugging their data into an optimizer, Brinson officials simulate risks and returns over the next 10 to 20 years, said Kevin Terhaar, Brinson's head of capital market analysis and strategy. Using a simulation also enables Brinson officials to calculate how investors should be compensated for liquidity risk.
In taking up this challenge, Brinson officials called on the resources of affiliated companies to do their research: private equity experts at Adams Street Partners, hedge fund pros at O'Connor and Global Asset Management, and their own real estate and timber professionals.
Using 14 factors, Brinson officials developed a forward-looking covariance matrix that considers the sensitivities of various kinds of alternatives to U.S. equity and fixed-income markets. The matrix shows, for example, that private equity has a 92% covariance with domestic stocks - although that measurement drops to 50% to 60% when leveraged buyouts are pulled out of the category.
To replicate the return and risk characteristics of a 60% stock and 40% fixed-income portfolio, the Brinson model produces a policy mix of 52% global equities and 28% global bonds, with the remaining 20% arrayed among a host of alternative investments, Mr. Singer said. And the risk and return characteristics of the total fund are more favorable than in the 60/40 portfolio: the expected return is boosted 50 basis points to 7.7%, while risk slips 10 basis points to 10.2%.
While Brinson and its parent, UBS AG, certainly have an ax to grind in promoting alternative investments, the recommendations do not appear unreasonably high. In fact, only the real estate allocation appears substantially higher than the 3.1% allocation among the Top 200 pension funds, as of Sept. 30.