LDI: Does It Make the CAPM a CRAPM?

From aiCIO's November Issue: A column scrutinizing the impact of the Capital Asset Pricing Model (CAPM) on asset management. 

To see this article in digital magazine format, click here.

The portly G.K. Chesterton once remarked to the slim George Bernard Shaw that if folks saw Shaw they would think there was a famine in the United Kingdom—to which the quick-witted Shaw replied, “And if they saw you, they would know why!”

The Capital Asset Pricing Model (CAPM) did a similar injustice to effective asset management as Chesterton did to the food supply of the UK—it sucked up all the oxygen. Many naïve proponents of liability-driven investing (LDI) extend CAPM to develop LDI approaches and try to ground their products in finance theory. But if everyone invests based on LDI, CAPM would be wrong and extensions of wrong models are rarely right. The global investment industry is based on CAPM/Mean-Variance Optimization (MVO). Is this the cause of the declines in solvency and potentially of future crises?

MVO and CAPM make two assumptions that should cause any reader of this magazine, who are meant to be represented in CAPM, to sit up and pause. First, every investor seeks to earn the highest absolute return per unit of volatility. Second, they make all investment decisions themselves (i.e., they function as “principals”). While heavily divorced from our reality, along with other simple assumptions, this allowed theorists to develop the CAPM, which is elegant and easily learned by even my undergraduate students. For example, the expected return of a risky asset is easily discerned by its beta and the expected return of a “market portfolio” and everyone allocates between a risk-free asset (cash) and the risky market portfolio—no mention of liabilities here. Yet CAPM is silent about the return of bonds; bonds account for a reasonable share of total financial markets and portfolios, and are a core asset in any LDI approach. The massive decline of rates led to meaningful solvency declines globally, yet CAPM has little to say on the expected return of bonds, factors impacting it, and hedging liabilities. LDI proponents are seizing this opportunity to market products as opposed to fixing the original problem—the flawed CAPM theory behind the portfolio allocation decisions.

Every investor, whether institutional or retail, sets aside money to service some future consumption or expenditure (termed the “liability”), as shown by the late Nobel winner Franco Modigliani. Investment decisions need to be made on this pool of savings. These decisions need to be made relative to the liability, however defined, and not on an absolute risk-adjusted return basis as assumed by CAPM. While common in pensions, Mark Schmid, the CIO of the University of Chicago endowment, demonstrates how even an endowment can apply LDI concepts, moving away from the CAPM-anchored “endowment model.” In short, investment portfolios must be highly positively correlated to liabilities to prevent future solvency declines. Many of us who used the CAPM approach overallocated to equity/hedge funds (negatively correlated to liabilities!) as opposed to the liability hedge, and paid the price from 2000 to 2012. Given current underfunding and equity allocations in portfolios, a smart investor needs to invest in strategies that earn a return greater than the liability (to grow solvency), and correlate positively to liabilities, and negatively to equities (to lower solvency risk and provide a tail risk hedge).

The second area of relative decision making is that very few investors manage their own investments. Boards (“principals” in CAPM) typically hire CIOs and consultants who then hire external managers (“agents”) to implement the investment decisions. CAPM misses the nuances that follow. The principal is unsure whether the “highly paid” agent is lucky or skillful. High relative risk conveys low confidence that the agent is skillful. To preserve the fee annuity, agents manage portfolios to correlate closely to the benchmark (low relative risk), however flawed the benchmark might be in replicating the liability or in minimizing drawdowns (e.g., in 2008), thereby hurting investors.

A new and different model—let’s call it the Relative Asset Pricing Model (RAPM), which academics need to develop—would price all assets relative to the value they provide to hedging the liability and not just a market portfolio, and this pricing would be affected by how much relative (or correlation) risk one can bear relative to the liability. Every investor would allocate between the liability hedge, cash, and risky assets. Using RAPM, the decline in interest rates from 2000 can be attributed to changes in regulator-imposed constraints on relative risk to liabilities in Europe and the US (because of PPA), leading to a greater demand for nominal bonds/liability hedges. When the regulators in Europe relaxed this constraint in June 2012, rates backed up as RAPM would predict. CAPM is just a very specialized/constrained case of a more realistic RAPM and assumes no liability and no delegation. Noting the flaws of CAPM, innovative funds are dispensing with a strategic asset allocation that is produced from some CAPM-based models, benchmarking themselves to the liability, and ensuring that they are empowered by the board to make investment decisions as opposed to being encouraged to delegate to third parties.

Rather than try to flog LDI products, our focus should be on helping investors think about using RAPM to rethink asset pricing, portfolio structuring, rebalancing, and manager compensation. Otherwise, agents will earn good fees, but client portfolios could end up in a pile of CRAPM.

Arun Muralidhar is Chairman of Mcube Investment Technologies LLC (www.mcubeit.com), and CIO of AlphaEngine Global Investment Solutions (AEGIS).