For so many investors, regret and misunderstanding have painfully followed the portfolio wreckage caused by COVID-19. "I didn't know I had that much exposure" and "We knew our exposure, but we couldn't justify the cost of hedging" are expressions of both.
None of this is new. Over the past 20 years, shock after market shock, investors have felt the pain of surprising outcomes and difficulty explaining why a risk decision was made. Why, in modern markets, does this continue to happen? Why do investors still not understand the risks they have? Why is the value of risk management still unclear?
Glyn Holton said it well in his 2004 paper, "Defining Risk." We can only operationalize perceived risks, so rather than asking if a risk metric captures all risks, we should ask if it is useful. Either investors do not measure the risks they perceive, or their risk metrics aren't useful.
Since the global financial crisis, many investors have embraced risk measurement tools. But the language that we use to articulate risk decisions to boards, fiduciaries and other key stakeholders is not useful.
The most commonly used allocation decision tool, mean-variance optimization, requires investors to trade off mean returns against risk (namely, standard deviation). But investors typically do not express their preferences in that way. They express them in terms of the discrete outcomes they want, like growth above a target, avoiding drawdowns beyond some level and meeting particular payout commitments. Thus, the relationship between mean-variance portfolios and actual desired outcomes is indirect.
If investors are unable to articulate a mean-standard deviation trade-off or if they do not understand how portfolio weights translate into abstract notions of risk, the decision metrics are not useful. They can regret having chosen any mean-variance "optimal" portfolio.
Still, even as investors have embraced more sophisticated metrics, none directly link risk choices to desired outcomes in a way that all investment stakeholders can understand. It is not surprising that even the best risk-management program is second guessed before and after experiencing a shock or that valuable risk-based solutions are perceived as being too expensive.
As a trustee, former CIO and risk management practitioner, I decided it was time to create new metrics that could measure the benefits of risk management in the context of an investor's own needs. In particular, I wanted to link risk choices to desired outcomes and enable investors to easily understand the value of those choices. Thus, with academic colleagues, we created "Portfolio Pi" and "Eta" as metrics that can assess the value proposition of risk and many other interrelated investment program choices.
Portfolio Pi is a simple, weighted average of the probabilities that an investor will achieve desired investment outcomes, including risk. It takes into account the probability that each desired investment outcome will be achieved and the relative importance the investor assigns to each.
An investor can use Pi to evaluate any quantifiable change in an investment program. Pi can help to answer, for example, "Could less risk improve our average chance of meeting our desired outcomes?"
Eta measures whether increases in Pi are "worth it." Eta is the economic value that the investor stands to gain by making changes that lead to a higher Pi value. Investors communicate Eta as a percentage return or dollar amount. Eta can help to answer, for example, "If the proposed changes increase Pi, what is the equivalent increase in the real end value of our current portfolio?"
Pi can be recalculated for any combination of changes to areas such as asset class weights; risk-factor sensitivities; new managers, products or securities; alpha assumptions; tax strategies; and expense ratios and other investment fees.
We illustrate Pi and Eta with a simple, stylized example of an investor that has the following desired outcomes for its $200 million portfolio over a five-year investment horizon:
- Grow the real value of the portfolio by 3%.
- Limit the portfolio drawdown to -10%.
- Pay out 3% of the portfolio annually, along with fees and costs.
The investor plans to meet the third outcome each year no matter what. Thus, we measure probabilities that outcomes one and two will be met after payouts, fees and costs, and taking other context into consideration. We assume the investor values each outcome equally (50% weight applied to each): Pi = 59% (chance of achieving outcome 1)*50% + 73% (chance of achieving outcome 2)*50%= 66%.
With a fresh look at risk capacity, the investor evaluates the impact on Pi of reducing risk via a lower equity beta and manager changes: Now Pi equals 64.7% (chance of achieving outcome 1)*50% + 83% (chance of achieving outcome 2)*50%=73.9%.
Pi increases as the proposed changes increase the probabilities that each outcome is met.
What is this increase worth? Eta, the return that the investor would add to the portfolio in year five to achieve a Pi of 73.9%, is 8.2% (1.6% annualized).