Attraction grows for institutional investors

But many struggle with how to determine best asset allocation among all the classes

Institutional investors are shifting billions of dollars into alternative investment strategies, but figuring how much to allocate to areas such as hedge funds, private equity and real estate is a conundrum for most.

Some institutions such as TIAA-CREF, Morgan Stanley (MS) and Citigroup Alternative Investments, all based in New York, are wrestling with ways to make alternatives allocations. But observers believe nobody has come up with the perfect solution.

The problem: Left unconstrained, the traditional mean-variance optimizer — which institutions use to set their asset allocations — spits out high allocations to alternative investments, and none to traditional stock and bond investments.

"If you just follow the optimizer, you'd just have a portfolio combining hedge funds, private equities, timber and a little bit of commodities, for icing," said Stephen Nesbitt, chief executive officer at Cliffwater LLC, Marina del Rey, Calif., a consultant on alternative investments.

The optimizer, which stems from Nobel Laureates Harry Markowitz's and William Sharpe's work on portfolio theory, works well with traditional stocks, bonds and cash. But the tool fails when it comes to alternatives because historical data are scarce on these asset classes. As a result, the optimizer tends to understate risk and overstate diversification benefits.

"The reason is that risk is very hard to measure in these asset classes. It's really just a true bear of a problem," said Richard Michaud, president and chief investment officer of New Frontier Advisors LLC, a Boston-based research and investment advisory firm that specializes in asset allocation issues.

That's why most institutional investors constrain the optimizer. They might cap their allocation to private equities at, say, 10% of total assets, or hedge funds at 5%. While pragmatic, this practice has little intellectual rigor.

TIAA-CREF's core

P. Brett Hammond, senior managing director and chief investment strategist at TIAA-CREF, said an unconstrained portfolio might generate higher expected returns and a better Sharpe ratio than a conventional portfolio, but its asset mix could look very strange: under one scenario, the efficient frontier would allocate 31% real estate securities, 24% to venture capital, 20% to real estate, 10% to emerging market equity, 8% to private equity and 7% to commodities.

"But can you sell this to your committee?" Mr. Hammond said at a recent conference sponsored by Financial Research Associates LLC in Monterey, Calif.

Given the limited history of data on alternatives, plan sponsors don't have the same confidence in the expected returns, variances and co-variances generated by the optimizer, Mr. Hammond said. And it's even harder for smaller institutions, which might not have access to top-notch managers, he said.

"How confident are you in the stability of portfolios that have alternatives in them?" he asked the conference audience.

Mr. Hammond's research was based on work he had conducted with Martin Leibowitz, managing director at Morgan Stanley (MS), when Mr. Leibowitz was vice chairman and chief investment officer of TIAA-CREF. Both men have pursued the research separately.

Messrs. Hammond and Leibowitz both proposed creating a core alternatives allocation while using traditional stocks and bonds to round out the portfolio.

In a simplistic example, Mr. Hammond gave equal allocations to eight alternative asset classes totaling anywhere from 30% to 60% of total assets.

The resulting portfolio has a beta of 0.61 to 0.65 — virtually identical to the beta of a traditional pension fund portfolio — that is much higher than the beta of 0.44 generated by the all-alternatives portfolio. A consistent beta level might be much easier for an investment committee to swallow, Mr. Hammond said in an interview.

"Optimization may be fine, but you don't want to let it run hog wild," he added.

Citigroup's approach

Researchers at Citigroup Alternative Investments took a different approach. They said the traditional optimizer, which relies solely on rewards and risk to generate a series of the most efficient portfolios, makes a number of assumptions that don't apply to alternative investments.

For example, prices for many alternatives, such as private equity or real estate, cannot be measured continuously because they are not traded minute by minute. That makes it hard to compare them with publicly traded securities.

Another problem is the lack of liquidity for many alternative investments. Investors are often locked up for periods of seven to 10 years in real estate or private equity pools.

In addition, some event-driven hedge fund strategies, such as merger arbitrage, distressed debt and convertible arbitrage, bear a greater chance of a large loss than a large gain. That means the standard measure of volatility does not accurately portray their downside risk.

Instead, Citigroup researchers broke down returns for any asset class among four factors: economic fundamentals (beta), manager skill (alpha), liquidity and downside risk. Rather than setting allocations by a simple trade-off of risk and return, an institutional investor can construct his total portfolio based on a combination of the four sources of return. It's a more complicated exercise, but could result in a better asset mix, according to a paper by Citigroup researchers.

For example, Citigroup's asset allocation tool — which is made available free of charge to clients — breaks down private equity returns into an illiquidity premium, the manager's skill plus fundamental factors, explained Vineet Budhraja, senior research analyst.

Similarly, a long-short equity portfolio can be broken down between fundamental and skill factors, while a merger-arbitrage strategy can be split into fundamental, skill and downside risk factors, he said.

Skill, however, is less reliable than fundamental factors, as more managers discover the same market inefficiency or the manager changes strategies over time. "The chance that we're wrong on the skill-based side is much higher than on the fundamental side," Mr. Budhraja said.