No topic has gathered more interest since the financial crisis of 2008 than the topic broadly called “tail risk management.” The term and its practice have been open to much interpretation; this phenomenon of initial confusion is not particularly different from the growing pains experienced by many other market sectors. Mutual funds, hedge funds, even ETFs at the very beginning of their life cycle operated without much uniformity or proper reference indexes. As the market for tail-hedging solutions evolves, it will become critical that the end-user at least have a framework within which to evaluate the potential and realized costs and benefits of particular practices. We believe that to add value over time, tail risk management has to be active rather than purely passive; thus, a proper benchmarking framework is not simply a luxury but a necessity. The purpose of this article is to start to lay out exactly such a framework, which we have evolved over almost a decade of implementation.
Defining A Hedge Mandate
As discussed in much detail elsewhere,1 a small set of inputs or guidelines is the natural starting point for defining a tail-hedge mandate. In our view, the minimal set consists of the following:
- Basis risk
The first step is quantifying exposures. Our analysis of the long-term history of many different types of assets shows that a small set of risk factors drives the returns of these assets. The two major secular exposures are the equity beta and the interest-rate or duration exposure. In addition, over cyclical periods, factors like liquidity, currency exposure, momentum and monetary policy also play important and significant roles. In our practice, we first try to quantify the exposures of each underlying portfolio to these key factors, both for normal and stressed periods. Interestingly, both our research and the work of others show that even very diversified portfolios exhibit similar exposures to the key risk factors, with equity beta as the dominant risk exposure.
The second step is to define what we have called the “attachment” level (taking a term from the reinsurance industry), which is not very different than the deductible one would have in a policy for automobile or earthquake insurance. The closer the attachment level is to the current value of the portfolio, the higher one should expect the cost of the tail risk protection. Generally, we believe that broadly diversified portfolios should have an attachment level anywhere from 10 to 15 percent below the current portfolio value.
This brings us to the important question of cost. We generally do not believe that tail hedging can be done efficiently in a perfectly costless manner over short-term horizons. Yes, there are structures (especially exotics) that purport to reduce the cost, or in many cases even eliminate the cost, but usually they consist of embedded sales of options that one would frequently rather not sell. Instead of this hidden discount, we believe that an explicit cost target is essential both to thinking of tail risk management as an asset allocation decision and as a commitment that one can continue to support in periods where fat-tail events do not occur. Because of the natural difficulty in forecasting the time and form of the next tail event, we believe that tail hedging is an “always on” part of any risky investment portfolio. Our empirical and theoretical research validates the belief that over longer periods (three to five years), tail hedging is generally self-financing when one accounts for both the ability to tilt portfolios more aggressively and following a systematic approach to rebalancing in the presence of such hedges.
Finally, one has some freedom to replace what might be expensive direct hedges with relatively cheaper indirect hedges, taking advantage of the tendency for correlations to increase, especially when extreme events happen. This cheapening comes with a trade-off, that the indirect hedges will not perform as well as the direct hedges conditional on the extreme event happening. To quantify this basis risk, we specify a level of confidence within which the likely outcomes of the actual portfolio are likely to fall relative to the direct hedge through simulations. The performance of a particular hedge program should be quantified in terms of the trade-off between basis risk and cost savings relative to a low- or no-basis risk benchmark.