Column on SAA: Simply Awful Allocations?

From aiCIO Magazine's April Issue: Many institutions are looking for a new approach and are adopting risk parity and other de-risking strategies, but many of these approaches reflect faulty applications of the theories.

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

When my information changes, I change my mind. What do you do?” There is some uncertainty as to whether this quote can be attributed to John Maynard Keynes or Paul Samuelson, but investors should heed this advice. Sadly, one of the bad applications of modern portfolio theory is to do just the opposite—namely, anchor the portfolio for five to 10 years in a Strategic Asset Allocation (SAA).

This practice has led to serious funding problems. Many institutions are looking for a new approach and are adopting risk parity or de-risking strategies for their SAAs, but many of these approaches, by ignoring the wisdom of Keynes or Samuelson, run the risk of making the SAA a “Simply Awful Allocation.” The root of the trouble is the bad application of the theories of another set of brilliant economists (Harry Markowitz, Bill Sharpe, James Tobin), who developed the tools of modern day portfolio management. Markowitz’s optimization approach is the basis of possibly every SAA and the expected return forecasts used as inputs in the SAA development process are derived from the Capital Asset Pricing Model (CAPM). Under the theoretical CAPM, the long term (single period) expected equity risk premium for stocks is positive. I recall polling major investment banks and managers in 1998 when I was helping develop the SAA for the World Bank Pension Plan and without exception, every one of them assumed a real equity risk premium of 3 to 3.5%, giving us an expected return for bonds of 5 to 6% and an expected return for equities of 8 to 9%. In a recent discussion with a CIO, he informed me that he had gone back over the last 10 years and evaluated the forecasts of his SAA consultants. What he discovered was that in short, many of us, yours truly included, did not recognize how poor we were at forecasting long-term returns. 

Second, in nearly all institutional portfolios, the individuals with ultimate fiduciary responsibility for the fund—the boards—delegate decisions either to internal staff or to an outsourced entity. In economic theory, this is called “the principal-agent problem.” Given the possible concern of the boards that staff, consultants, or external managers are lucky rather than skillful, there is a need to restrict the freedom given to agents and to benchmark performance to some passive benchmark. Tracking error became the dominant risk measure and risk budgeting became the buzzword of choice. Again, many of us who managed institutional funds were restricted in how much we could deviate from the SAAs by rebalancing ranges that were often no more than +/-3% around strategic targets. In effect, these tight ranges implicitly assumed extreme confidence in long-term asset allocation forecasts and the positive expected returns we had projected, but dragged portfolios lower in down markets. 

Given the volatility in markets, boards members and CIOs would be well advised to take a different approach from that used in the past. Some people have instituted programs such as risk parity, but this has its own problems—namely static leverage. The smarter route would be to have a more dynamic approach to asset allocation incorporating the various market factors that influence the future performance of assets. In effect, this would be the Samuelson-Keynes approach. One often hears that this is market timing, and “market timing is bad!” However, all investing, even a static SAA, is market timing, so the choice is only between naïve market timing and a more intelligent approach incorporating market information (Dynamic Asset Allocation).

Alternatively, if boards are not comfortable with this approach, then CIOs need to be empowered and allowed to deviate from the SAAs. There is nothing worse than forcing a CIO to rebalance to a portfolio with a 50%-60% allocation to equities when equities are likely to suffer a negative year. This means that funds should relax the ranges within which a portfolio is allowed to move and empower internal staff with adequate resources to make these decisions with the rebalancing bands as implemented by San Bernardino County.  

A third approach would be to hire asset managers focused on managing the drawdown of the portfolio and not minimizing tracking error. In other words, managers should not be selected for their Sharpe or Information Ratio, but rather on their ability to protect client capital. 

Some innovative funds, like the CERN Pension Fund, have done away with the traditional SAA completely because they feel that this leads to mediocrity and the ultimate benchmark for a pension fund is the liability. (One can make an equivalent case for endowments and foundations as shown recently by the University of Chicago Endowment.) As Rahm Emanuel pointed out, a crisis is a terrible thing to waste, and hopefully it provides a learning moment for all us to move away from the Simply Awful Allocations.  


Dr. Arun Muralidhar is Chairman of Mcube Investment Technologies LLC (, and CIO of AlphaEngine Global Investment Solutions (AEGIS). Arun is the author of A SMART Approach to Portfolio Management: An Innovative Paradigm for Managing Risk (Royal Fern Publishing LLC, 2011) and three other books. He has worked as a plan sponsor, asset manager, and supplier of investment and risk management technologies.