“To the maximum extent possible we want to be transparent so our beneficiaries and others can see into the methods, assumptions and processes we’re using to develop the asset allocation,” investments chief Joe Dear says. “There is no enduring proprietary secret that would be protected by trying to do it all behind closed doors, and a lot to be gained by engaging people as much as possible.” In March the investment committee kicked off the analysis with a review of the role of asset classes, concluding that CalPERS’ current asset class structure was heavily influenced by GDP growth and masked underlying risk exposures. The May workshop, and subsequent three-day offsite in July, engaged the board in a discussion about the underlying capital market assumptions, in an attempt to better understand the drivers of investments, portfolio behaviour and risks.
“The purpose of the May workshop was to explore the issue and engage the board in a more comprehensive way. In the past the board was just given the consensus assumptions and not allowed or asked to weigh in with a view,” Dear says. The outcome of that process was an indicated set of return assumptions considerably lower than the current 7.75 per cent, at 7.29 per cent, a median of the work done by staff and consultants. As a result of these first two steps, about a third of the way to setting new strategic targets, CalPERS is now running a parallel process with portfolio models built on the standard tools of mean variance portfolio optimisation, but also on an alternative factor model identified in March. “One’s the fail-safe, the classic method. The factor model is to try and do something to improve upon the asset allocation toolset,” Dear says.