Asset Allocation Weekly (February 28, 2014)

by Asset Allocation Committee

At the founding of our asset allocation process in the second quarter of 2000, the prevailing method of asset allocation was strategic, which is usually defined as an allocation based on a time horizon of at least 7-10 years.  However, in practice, strategic programs assumed that adjustments to the model were usually unnecessary, and only required when truly secular changes occurred in markets, society, government or geopolitics.  For the period of the 1980s through the 1990s, when equities were in a secular bull market, unchanging models generally performed well, although it could be argued that the risk-adjusted returns they provided were rather lackluster.

In the late 1990s, when Mark Keller, the current CIO and CEO of Confluence Investment Management, was appointed to head the Investment Strategy Committee at A.G. Edwards, there was a concern that static models were exposing investors to excessive risk.  For example, by December 1999, the Shiller Cyclically Adjusted P/E had reached 44.2x earnings, an all-time high.  Financial advisors at the firm wanted Mark to devise a program that would take market and economic conditions into account in making asset allocation decisions.

At the same time, there was hesitancy among the members of the committee to create a purely tactical allocation program.  These programs are usually rules-based by design, using past trends to establish positions.  The promise of tactical systems is that they can “capture the upside or avoid the downside.”  And, if the past perfectly matches the future, they will perform as advertised.  However, the Investment Strategy Committee was concerned that unusual events that had not occurred over the period for which the tactical program was created would likely not be incorporated into the “model” and thus lead to adverse results.

And so, the Investment Strategy Committee created a hybrid between these two polar opposites, a Cyclical Asset Allocation Program.  The process incorporated a variety of viewpoints, including economic, political, geopolitical, and financial factors that evaluated risk, yield and total return of 10 (now 12) different asset classes on a rolling three-year time horizon.  Each member of the committee submitted forecasts for these asset classes and presented these forecasts to the entire committee in a series of meetings.  After all the forecasts were submitted, the committee would establish a consensus that, in the end, best incorporated the committee’s expectations for the markets over the next 12 quarters.  The asset allocation process used today here at Confluence is unchanged from that original process.

This process was designed to address the weaknesses of both the strategic and tactical allocation models.  Strategic models usually are first established via a fundamental process—the manager looks at the world and designs a program that will work over the long term, assuming that underlying trends in markets and the economy will roughly remain the same.  Tactical models, on the other hand, tend to be “twitchy,” moving positions around fairly often to try to capture short-term gains.  Tactical models have tended to evolve into algorithms that have been heavily back-tested; although the back-tests offer some comfort, as we noted above, they will only work as expected if the future resembles the past rather closely.

Our method, instead, looks at the world through a variety of viewpoints; each member of the committee has a unique background and methodology in creating their forecasts.  The consensus drawn is not a mere average of forecasts; instead, it works to establish a viewpoint that reflects the basic expectations of how the next three years will unfold.  It does not purport to be perfect but the program does create a process where an ever changing world can be evaluated with a consistent approach and rebalanced at least every three months.  We believe our cyclical asset allocation program offers a unique method of addressing uncertainty that is flexible enough to make changes, but not so rules-based as to miss major changes in the world order.

View the PDF