CalPERS' board is expected to decide in September whether to remove alternative investments from its tracking-error calculation used to control risk.
Deciding to do so would make sense, consultants say. When it comes to alternative investments, tracking error doesn't work because these asset classes lack an investible benchmark.
By comparison, it's "fairly easy to find benchmarks" for public equities, said John Delaney, Philadelphia-based portfolio manager for Willis Towers Watson PLC.
Using tracking error for alternative investments means that investors are constraining "real life strategies" to a certain level of correlation or volatility relative to a proxy benchmark, he said.
Many asset owners use a benchmark reflecting the opportunity cost, or the return over a public markets index for alternative investments. For example, CalPERS' private equity benchmark is Custom FTSE All World All Cap Equity, plus 150 basis points.
For private assets, tracking error limits can limit investment in assets that "don't look like the benchmark," reducing the value of investing in alternative investments, Mr. Delaney said.
The issue came up at the June 14 investment committee of the $469.8 billion California Public Employees' Retirement System, Sacramento. Staff would like to switch to what they call an "actionable tracking error" approach.
According to proposed investment policy changes shared with the investment committee, CalPERS would remove a requirement that the asset allocation for its entire portfolio be managed within a target forecast annual tracking error compared to its benchmark of 0.75%. The asset allocation limit of 0.75% implies that during any one-year period, there will be less than a 5% probability that the active asset allocation return will fall less than 1.2%, the existing investment policy said.
The asset allocation requirement in the existing investment policy captures the asset class deviations in CalPERS' total fund relative to its asset allocation targets, said CalPERS spokeswoman Megan White in an email.
"Any underweights or overweights in any asset class relative to the policy benchmark create tracking error because our portfolio weighting differs from the policy weighting," she said.
Arnold B. Phillips, interim deputy CIO, said in an email that the separate, annual asset allocation tracking-error limit of 0.75% wouldn't be needed because the actionable tracking-error calculation for the total portfolio would take that into account.
"The asset allocation impact on tracking error comes from the deviations between the total fund portfolio and the policy benchmarks for the asset classes defined in the strategic asset allocation," Mr. Phillips said. "The asset allocation contribution is just one piece of the calculation of both total, and actionable tracking error."
If adopted, the total fund would instead have an active risk target "consistent with forecast tracking error" of up to 1% relative to its benchmark. The proposed methodology would remove alternative investments from its calculation as well as eliminate annual tracking-error limits.