#25 Models
MPT: Many Problematic Techniques
Investing is fundamentally about generating cash flows. Whether it’s for retirement or your child’s college tuition, the goal is a common one: A stream of inflation-adjusted cash flows, starting many years in the future, with some fixed or uncertain maturity.
Modern portfolio theory (MPT) has failed. In its place has increasingly stepped goals-based investing (GBI). But even GBI is rooted in the complex, risky, error-prone MPT models, using assets unsuited for our goals—hence, “Many Problematic Techniques.”
A recent survey found 40% of American adults are financially illiterate, unable to answer how much to save, how to invest, and how to decumulate. Ignoring this concern—and if, over the objections of the behavioralists, we accept that MPT is the appropriate model for asset pricing and asset allocation—we compound our errors.
The typical investor has to choose among stocks, bonds, commodities, and alternatives. None have the cash flow of the goals described above to allow a risk-free, robust, low-cost, and cash-flow-matched immunization. So we use mean-variance optimizers, even though many MPT models ignore the use of funds. This immediately contravenes the basic point of GBI—to reach a goal—because MPT is focused on wealth maximization, not a future cash flow.
Ignoring this concern, research has shown that our ability to forecast the optimization inputs is dismal. Ignoring this concern, the variables themselves are dynamic—yet MPT models use static inputs. Ignoring this concern, even if we perfectly forecast all variables, the time needed for these variables to converge to their true value is 40 years. We are thus trading on noise. Ignoring all these concerns, MPT assumes the risk-free asset is a bond (either nominal or inflation-adjusted), but a typical bond would not have a cash-flow profile similar to any of the goals listed above. So, a traditional bond portfolio is not a risk-free asset, but rather a risky one.
There is a simple solution: Develop new bonds that mimic the cash flows of our goals. Each bond would pay a real coupon, starting on the date needed (retirement, college entrance) and for the period of the goal. The accumulation and decumulation decision is folded into one asset. No complex optimizations and error-prone models needed; all an investor would have to know is multiplication to calculate the number of bonds required to achieve their respective goals. Until we create these instruments, a warning: as Albert Einstein noted, “If you can’t explain it to a six-year-old, you don’t understand it.”
Arun Muralidhar is the chairman of Mcube Investment Technologies and AlphaEngine Global Investment Solutions.