SLCG Economic Consulting's Logo


Variable Annuity Fees Linked to the VIX -- Part I

We've discussed the CBOE Volatility Index -- known as the VIX-- many times before. Essentially, the VIX is a very complex calculation of the expected future variance of the S&P 500 (see the full calculation methodology), and is popularly known as the 'investor fear gauge'. The VIX is not a tradeable asset, but there are VIX options and futures contracts, and those contracts serve as the basis for several VIX-related exchange-traded products (TVIX, XIV, VXXto name a few). The VIX is very often misunderstood and we think that investments linked to the VIX are almost always too complex and risky for retail investors.

We've also talked about variable annuities, which are investments that have been the subject of regulatory scrutiny over the past few years due to their complexity and potentially insufficient disclosures of fees. Given that context, we were surprised to see that the CBOE has a white paper that describes linking variable annuity fees to the VIX. The authors note that the insurance giant AIG has a patent application pending for this design.

Before we get into the details of that paper, the most obvious concern about this proposition is how such a fee could be fairly disclosed to potential annuity purchasers. Variable annuity fees can be complex to begin with, and understanding the VIX requires serious study and some quantitative background. For example, one critical feature of the VIX is that it tends to be negatively correlated to the S&P 500. Therefore, investors should at the very least understand what negative correlation is, what it means, and how it comes about, which is a subject of academic research including our own(PDFs). We think it's worth asking how retail investors could be expected to understand the implications of the VIX on their variable annuity fees, when the VIX itself is so complex.

But there are other concerns as well. Linking fees (an outflow) to the VIX is, from the investors' point of view, similar to a short position on the VIX. Because VIX returns are negatively correlated with S&P 500 returns, that means that the fees will likely spike when the S&P 500 crashes. Therefore, an investor in an S&P 500 linked variable annuity will be paying lower fees when the S&P 500 is rising and more in fees when the value of that annuity is declining. This would also be difficult to explain to most retail investors.

The paper contains an illustrative example that compares a fixed fee of 1% to a VIX-linked fee. The authors find that the variable VIX-linked fee is roughly equivalent to the 1% fixed fee if the variable fee, in basis points, is approximately 4.8 times the VIX level.* An important point though is that this fee was determinedex post--in other words, after the fact. An insurer would likely charge a different fee based upon different expectations of the VIX during the term of a variable annuity since the fee needs to be determined at the inception of the contract (ex ante). Again, it would be difficult for an investor to determine if such a multiplier fairly priced his or her annuity.

The reason variable annuity issuersmight be interested in this approach is that the VIX also indirectly measures the cost of the hedging position most variable annuity issuers use (a rolling call option strategy). By linking the fees to the VIX, the issuers could charge higher fees when their hedge is more expensive to obtain. While that may help solve a problem for the issuers, it is unclear what benefit that would have to investors, especially at the cost of greatly increased complexity.

In our next post, we'll dive a little deeper into just how this VIX-based variable annuity fee works and how it might actually be implemented in a real annuity contract. Stay tuned!


* This multiplier shouldn't be a surprise since the average VIX level was around 20.5 for the period considered by the authors (1990 to 2012). So if we want the average daily fee to be equal to the average daily fee in the fixed account (100 basis points), then we would need a multiplier of about 4.88.