The Dark Side of Asset Allocation, Part 1
That’s a rather provocative title don’t you think? How can I possibly say that a great good, like asset allocation, has a dark side? In this series I hope to demonstrate to you how asset allocation is actually one of the hidden sources of fragility in the global financial system. Yet, its pernicious effects are largely unexamined because of the concept’s universal acceptance as one of the best innovations in all of finance, and hence, in all of business.
Taking On Unnecessary Risk
In this first part I want to discuss the fact that asset allocation leads investors to blindly take on risks that they otherwise would not have. This is because thinking about risk at a portfolio level is very different activity than thinking about risk at the individual security level. When thinking about risk at the portfolio level, as Modern Portfolio Theory instructs us to do, we abstract and even ignore the individual risky characteristics of a security.
Said another way, when we ignore the absolute level of risk in a security and convert it to relative risk courtesy of correlation, covariance, and beta – the tools of Modern Portfolio Theory – then we obscure the underlying absolute level of risks. Case in point: the mortgage backed securities of the Great Recession. If individual mortgages, like jumbo subprime, had been evaluated individually these securities would, in all likelihood, have never been underwritten, issued, and purchased.
Yet, when we view a single asset as only one little tiny piece of a portfolio, not only do we obscure and ignore the absolute level of risk, we may even seek out such characteristics because of the way in which these securities complement others within a “diversified” portfolio from a mathematical point of view. In fact, this near cataclysm of the MBS meltdown can almost entirely be blamed on asset allocation being accepted as the dominant form of risk management. Ouch!
To be absolutely clear about my message, I am saying that asset allocation, because it uses mathematical measures of "risk" - things like, standard deviation, covariance, beta, and so forth - actually misses the underlying risk of an individual security. So, ironically, the idiosyncratic risk is put into the portfolio, unexamined in any way at all.
Doubtless there are positive outcomes contained herein, too. For example, the ability to diversify risk via Modern Portfolio Theory may allow capital to be allocated to opportunities that otherwise would go uninvested. Yes, fully acknowledged. But notice the hidden assumption in this point. Namely, that an opportunity has individual merits, but that its risks make it difficult to gain buy order conviction. In other words, this thinking acknowledges a view of the security on an individual basis. Which is my point: thinking in terms of asset allocation, thinking only of the overall beta portfolio contribution of an asset, leads to outsized risk-taking.
Insurance companies, whose entire portfolio of policies are underwritten risks, think about risk in absolute terms, and not just in relative terms. In fact, when they begin to lose sight of the fact that certain risks should never be underwritten, this is exactly when they run into trouble and when the combined ratio soars over 1.0.
Deworsification
Within asset management asset allocation leads to deworsification as investors dilute their alpha-opportunity set with beta-worshiping over diversification. So, another pernicious effect of asset allocation is that the work done by active managers to identify and underwrite the purchase of individual securities is undermined when the portfolio management team keeps adding in unnecessary assets to diversify away “risk.”
When I was a portfolio manager of the Davis Appreciation and Income Fund we typically had 30-35 assets in the portfolio. Something that our third-party evaluators like Morningstar called “concentrated” – their polite way of saying that we were risky. Yet our retort was always the same.
We spent months and years evaluating the risk of these assets at the individual business level and the individual security level. In fact, we considered dozens of individual elements of risk when evaluating businesses. Among the factors was the quality of management, the ability of management to perform in recessionary times, the externality risk, the risk if the company could not finance its plans, and on and on.
Simply put, we did not purchase assets that we believed were risky. Implicit in this concern on the part of our third-party evaluators about a “concentrated” portfolio is that asset allocation is the only true way of removing risk from a portfolio. Bollocks! In fact, as this article should make clear, asset allocation actually leads to unnecessary risks being underwritten.
In conclusion, when you evaluate risk as an investor never lose site of the fact that individual securities carry absolute levels of risk, and not just risk at the relative level. My next article in this series will demonstrate how asset allocation impoverishes huge swathes of the world.
Key Account Manager/Certified RTT Practitioner
6 年I am sorry, perhaps in the Alchemy of financial institutions, asset allocation becomes overly complex, but the basics of eg even a stock and bond portfolio has reduced downside risk. The problem is trying to OVERLY fix a problem in the wrong way. Which I believe is what you have spoken about
Customer Operations Manager at Royal Mail
6 年Spot on. I can't agree more.
Director - Real Assets, Portfolio Manager and Partner at BSW Wealth Partners, a Public Benefit Corporation
6 年Jason enjoyed reading your article. I would substitute the word asset allocation with indexing as a more accurate target for your critique. Asset allocation in my opinion and from my understanding are the broad strokes one can take to steer the tanker of risk return in your portfolio. Whether you do that with concentrated slices or diversified indexes of asset classes it doesn’t matter. You end up getting to the same point which is some desired expected level of risk and return that is appropriate for the constitution and goals of the investor. Indexing is a path to accepting idiosyncratic risk but knowing statistically that it gets diversified away in a large portfolio. I’m not sure we are there yet but the thing to be worried about would be the systemic risk of too much of the market indexing.
IBA, ACCA student, Candidate CFA Level II, Engro Fertilizers Ltd, Ex-GlaxoSmithKline, Ex- Philip Morris
6 年Hi. I appreciate your unique set of ideas over asset allocation, however if not Modern Portfolio Theory then what according to you can be a better basis of judgement for portfolio construction and risk diversification? Secondly I believe that it is in human nature to keep evolving and trying new set of combinations to get the desired outcomes, therefore through portfolio construction if an investor is able to get rid of relative risk then he has the means to achieve the desired outcome. Hence absolute risk of a security may not even be his concern. Thirdly the assumption that an investor was not willing to take the risk that he actually took because of MPT is very ambiguous according to me, because risk appetite is very volatile and highly subjective.
Partner at Pine Island New Energy Partners
6 年Hi, I think the criticism of asset allocation or MPT presented here is misplaced. My view: 1) The broad failure of MBS in the GFC, abetted by multiple layers of structuring, was the mother of all agency problems. The originators, the structured finance guys, etc., weren't underwriting per se, they were selling something for more than they bought it, they weren't fiduciaries, and they retained too little risk to care. The rating agencies became too codependent with structured finance and created models based on incredibly limited, and arguably irrelevant, data histories. The advisors of the various end purchasers were, for the most part, buying a rated bond without further examination. 2) In any case, Markowitz nowhere specified any particular method for creating expected returns or expected joint covariances of assets - he merely posited a model, evidence forthwith, and showed how to get the output with the input. Anyway, perhaps we can agree to disagree just on this particular point.