The hidden dangers of the traditional asset allocation model – Part 1

In the last few decades, institutional investors have migrated en mass from vanilla domestic equity/fixed-income portfolios into the multi-asset class “endowment” style pioneered by David Swensen at Yale.  Swensen preached the diversification benefit and additional risk premium that can be captured when long-term investors tap into alternative and illiquid assets.  Yale’s performance and thought leadership led many to follow.  As a result, the contrarian strategy has slowly become the norm.

Unfortunately, results for those that have adopted this approach have been mixed.  Most recently during the financial crisis, illiquid assets cruelly handicapped portfolios, while favorable returns in esoteric asset classes mean-reverted in an abrupt and highly correlated fashion.  This left some investors re-thinking the logic behind their approach.

A core belief at Novus is that the complexity of grappling with an endowment portfolio has become an acute challenge.  Most institutions – armed with a skeleton staff and a consultant’s rating system – have adopted asset-allocation models (or their en vogue evolution, the “risk allocation” model) to cope with data complication.  These models generalize all of the clutter across intricate investment structures by bucketing fund allocations into groupings such as “Domestic Equity” or “Inflation-Sensitive”, allowing allocators to measure their aggregate exposure in bite-sized strategic allocations.

There is an inherent danger in this simplification process, and it is especially problematic with active investments.  As a simple example, an endowment’s “European Equity” bucket can easily fit an allocation to the Vanguard FTSE Europe ETF (VGK), but doing so for a European long/short equity hedge fund may prove problematic.  If you peel the onion behind the fund’s actual risk exposures, you may find the fund is investing in geographies outside of the Eurozone.  In fact, in 2013, many European equity hedge funds Novus tracks tilted heavily towards North American exposure, as the opportunity set was far riper.  This leads to unmeasured risk creeping into the asset allocation process.

‍

Apply this example across an entire portfolio of myriad active investments and you begin to see the dangers in any simplified asset-allocation model.  Your actual exposure to your generalized bucket does not reflect your true risk exposures as the graph above shows.  The only true method of understanding risk and asset class exposure is by parsing all of the underlying managers’ risk exposure/position-level data at its most granular level, and aggregating it to the sub-portfolio and portfolio levels. Analyzing such big data requires flexible interfaces and database infrastructures that can string together normalized time-series of category exposure on command with a readily digestible UI.

Charlie Munger, fabled investor and partner of Warren Buffet once gave a speech to the Foundation Financial Officers Group.  He implored investors to avoid the layerings of fees and the unnecessary complexity behind modern portfolio construction, in place for concentrated investing in high quality individual stocks.

Our perspective at Novus is that endowment-style portfolios demand tooling and true data aggregation that can provide an accurate picture of real exposure and risk.  Without it, they are left vulnerable to the latent risks inherent in their oversimplified models.  Once they do so, they will be positioned to capture the illiquidity and diversification benefit of their endowment-style allocation model.  Otherwise they better simplify things and listen to Uncle Charlie.