Scientists at the European Organization for Nuclear Research (CERN) in Geneva study the basic constituents of matter—the fundamental particles of everything we see (and don't see) around us. By understanding the constituent parts, they can try and unlock the secrets of the forces of nature.

The people responsible for ensuring that these scientific minds—the greatest of our generation—are adequately remunerated in retirement have taken a similar approach.

The first job undertaken by Theodore Economou, CEO of the organization's CHF 3.7 billion pension fund, was to break the scheme down to its barest essentials when he joined a little over three years ago. In 2009, the fund had a portfolio divided 60% in risk assets and 40% in bonds. It was 60.5% funded on an IAS19 basis, down from over 70% a year earlier, according to its accounts.

It must be remembered that the financial crisis was in full flow at this time and investors around the globe were suffering similar shortfalls. The challenge for Economou was to ensure that CERN's fund got out of that position and did not end up there again, whatever the economic climate.

In 2009, the fund had, like the majority of pension funds continue to have, a return-based approach. Economou realized that had to change. "It's an absolute-return approach to the entire fund—with a key term being 'liability-aware,'" he told aiCIO in 2010. This term means the fund is aware of its liabilities, but is not tied to them.

By the end of 2011, the fund had moved to holding 28% in "diversifying assets," which included real estate, commodities, and other absolute return securities. Bonds and equities made up just over 30% of the portfolio each, with the remainder held in cash.

Before any assets are allocated, however, a rigorous risk management framework must be followed. Economou explained: "We look at risk management as two processes. One: the overall risk management process, showing us the acceptable risk constraints. This tells us what the size of the sandbox we can play in is." The measurement of this "sandbox" is through a system called Conditional Value-at-Risk ("not the perfect measure but you need to start somewhere"), which shows trustees the guidelines within which investments will sit.

The next step is to "figure out how much fun" the pension fund can have in the sandbox. This is where Gregoire Haenni, the CIO, comes in. Haenni is a risk-modelling specialist, having spent his extended education studying it. He joined the fund just after Economou, in 2010.

The system used by the CERN pension fund would not look out of place in the scientists' lab. The team identifies the characteristics of fund managers and creates a type of DNA for each one. The manager is then plotted on a pair of axes illustrating investment approaches. The system looks at 20 dimensions of correlation that are distilled down to three, for visual purposes, and those in unacceptable parts of the chart are discarded.

Then the investment can start. "We don't pretend we can call the market," Economou said in 2010. "That's borderline delusional. People spend hundreds of millions trying to do this." As a pension fund, Economou says, they cannot argue with or try to compete with fund managers who are dead-set on one thing. "What we can do is identify different market regimes, different areas of risk that are excessive, and we can hedge."

The fund does not take large bets on asset classes or market movements, which means it may miss large swings to the upside. "The real test is when markets are up 20% and we're below that," Economou says. "We have told the board that we look at it as how much have we left on the table on the upside to protect the downside. I think they'll be with us in this scenario, due to the usual mantra: education, education, education." —EHP