Risk Budgeting: Newer Approaches than the ‘New Approach’ of 2000
|Oct 10th, 2012 | Filed under: Book review, Risk management, Today's Post | By: cfaille||
The 2012 edition of Risk Budgeting is a re-working of a book of the same name from more than a decade ago. The subtitle of the 2000 edition was “A New Approach to Investing.” The subtitle to this one is “Risk Appetite and Governance in the Wake of the Financial Crisis.” The differences then, right on the covers, indicate both that the idea of risk budgeting is no longer so new and that some of its particulars have had to be re-assessed.
The editor of both editions, Leslie Rahl, founder and managing partner of Capital Market Risk Advisors, and a former director of ISDA, says in her introduction to the second that when the first edition was published, the buy side of the investment world was only gradually adopting the Value at Risk metric from its sell side colleagues. Much of that book was a quite positive exposition of this new tool VaR. That has changed.
Two of the essays in this edition take sustained looks at VaR, and they come at the issue from rather different perspectives. Stephen Rahl, of Rahl Capital Management, thinks the problems with VaR run very deep. It is not simply that VaR as classically formulated presumes a Bell curve with the very narrow tails that implies (although that is one of his criticisms, it is by now pretty much everybody’s criticism). Other problems are: that VaR treats the past as the guarantor of the future, and that it arbitrarily identifies variance with risk.
Risk is not Variance
Let’s pause on that latter point. It isn’t too difficult to re-jigger the presumed curve (changing kurtosis if we may use the jargon) so that more of the variance is the consequence of more extreme deviations from the norm than the bell (Gaussian) curve presumes. Then we can calculate VaR on the basis of the new curve. But … we’re still treating variance as risk. Why?
For many assets, such as option contracts, the connection between variance and risk depends critically upon which side of the deal you’re on. Are you the buyer or the seller? As Stephen Rahl explains: “because of the nature of options contracts, the BUYER’s potential losses are capped to the premiums paid, but the BUYER’s potential gains are infinite. The SELLER, on the other hand, has potential gains that are capped to the premiums received, but faces potentially infinite losses. In other words, the BUYER is structurally immune to blow-up risk, while he SELLER is perpetually exposed to blow-up risk. And yet any variance analysis will treat these two portfolios exactly the same.”
Later in the book, Richard Horwitz, Erin Simpson, and Terry Smith take a somewhat kindlier tone toward VaR than does Rahl. [Horwitz is a managing director at Capital Market Risk Advisors. Simpson and Smith are with Risk Fundamentals.]
A Five-Step Plan
Horwitz at al. say that though VaR “is a much-derided methodology,” reactive rejection thereof is too simplistic. They suggest for example that VaR is a more sophisticated approach to portfolio risk than one of the approaches sometimes preferred to it, stress testing, understood as the analysis of the sensitivity of returns to specific stressing scenarios. While VaR “integrates the volatilities, frequencies and correlations of returns,” stress testing “deals with only the first of these.”
Horwitz et al. also acknowledge that it would be impossible for sell-side institutions simply to proceed as if VaR had never been dreamt of, for VaR has in effect gotten itself written into the regulatory rule books.
But VaR mostly drops out of their approach as they proceed, and their chapter ends with a proposed five step process for risk budgeting, devised seemingly with the buy-side especially in mind. The five steps are as below.
- Establish objectives: will your fund be passively or actively managed? What is its time horizon?
- Decide on tactics: how frequently to rebalance? How to hedge currency exposures? How much sector concentration?
- Execute day to day: Investment decisions have to be implemented with the objectives and tactics in mind: in security selection, risk controls, sector concentrations, and so on
- Measure performance: The attribution of an entity’s returns “must be measured consistently with the way in which the organization initially analyzed risk in constructing the portfolio and in which risks were measured.”
- Compensate performers: managers have long been compensated on the basis of returns of course, but a more sophisticated concept is emerging of compensation based on risk-adjusted returns.
All in all, this is a fascinating and insightful book, whence much can be learned, for risk professionals and for their bosses, chief counsels, and chief risk officers.
Christopher Faille is a Jamesian pragmatist. William James has taught him, for example, that "you can say of a line that it runs east, or you can say that it runs west, and the line per se accepts both descriptions without rebelling at the inconsistency."