Mass Customization and Tactical Asset Allocation

Six employees of BNP Paribas Asset Management have posted a paper on SSRN, “Mass Customization of Asset Allocation.”  The abstract reads, in part:

The digital transformation is creating a need for mass customization of tactical asset allocation [TAA].  Asset managers publish tactical asset allocation qualitative views regularly.  However, the construction of robust portfolios from such views at large scale is often over-simplified.  We propose a robust framework that enables the industrialization of highly customized tactical asset allocation portfolios from a single set of investment views.  Our framework links the conviction of investment views to the consumption of risk and derives the expected returns accordingly.

Note the inclusion of “tactical” in the description that’s not found in the paper’s title.  Thus, the recommendations deal with the interim adjustments to asset class weightings within a strategic asset allocation framework.

Such adjustments could be applied to hundreds or thousands or “even millions” of accounts overseen by a firm on behalf of institutional and/or individual clients.  Changes are usually the province of an investment committee or asset allocation committee or a portfolio manager (or team) responsible for a product line.

Typically there is an implementation framework that captures two aspects of the decision makers’ beliefs about the prospects for each asset class — direction and conviction — in some combination of words, colors, numbers, and symbols.  An appendix in the paper provides a number of examples from well-known asset management firms to demonstrate how their beliefs are communicated to others.

Those beliefs then need to be translated into action across the array of portfolios being managed.  That’s where the challenge of so-called “mass customization” comes into play.

According to the authors, the views of the decision makers “should be implemented on an industrial scale and with no delay in all portfolios and mandates, even in those with the highest levels of customization.  Creating a viable industrial TAA process is thus part of the asset manager’s fiduciary duty towards all its clients.”

Framework

Those interested in the specifics of the recommended framework can dive into the details provided in the paper.  Here’s how the steps involved are described:

1. Construct a unique unconstrained active portfolio using an active risk budgeting approach that fully reflects the [investment committee’s] views at a given level of tracking error.

2. Calculate the implied active returns that render this unconstrained active portfolio optimal under [robust portfolio optimization] by reversing the optimization problem (reverse RPO).

3. Run the RPO with the implied active returns and the risk model as inputs while imposing portfolio constraints and defining the benchmark, the universe of financial instruments allowed and the tracking error tolerated.

There are a variety of exhibits which illustrate how the framework is applied.  Notably, the authors believe that the transparency of the approach allows those involved “to explain easily the impact of constraints on tactical tilts and to gauge the extent to which their constrained portfolios reflect the [investment committee’s] views.”

Considerations

Here are some items to consider regarding tactical asset allocation practices:

As mentioned in the paper, mean-variance optimization models have important shortcomings, yet they remain the foundation of most asset allocation work throughout the industry.

Implicit in the paper’s recommendation is the belief that tactical asset allocation adds value.  That may or may not be the case.  Simple rebalancing might be a better solution.

Those combinations of “words, colors, numbers, and symbols” used by firms to describe the desired direction and conviction of tactical movements are less than optimal.  They should be replaced by more granular numerical scales.

It is an open question as to whether those recommendations are best derived quantitatively or qualitatively or by using some advanced form of man-machine interaction.

The dimension of time wasn’t addressed in the framework.  Perhaps the varying time horizons of clients are captured in the risk assessment and budgeting processes, but perhaps not (or not sufficiently).

Another point of interest comes from a sentence which was quoted earlier:  “Creating a viable industrial TAA process is thus part of the asset manager’s fiduciary duty towards all its clients.”  In a business of scale, the need for “industrial” activities is very real, but there are trade-offs.  Some might argue that industrialization is largely for the benefit of providers rather than clients (resulting in improved margins and allowing for even more scale).

And, while firms need to do what they say they do — and to act in their clients’ best interests — having an “industrial TAA process” is not a necessary element of fiduciary duty.

The paper is a good example of the dance between theory and practice, as well as the push and pull in the industry between customization and industrialization.  There are a wide range of circumstances among the individuals and families served by investment advisors.  When considering solutions regarding tactical asset allocation (or any other investment function), implementation should be practical while being theoretically sound — and efficient yet personal.