Beware of Backtested RILA Performance

Registered Index-Linked Annuities (RILAs) — also commonly referred to as index variable annuities, structured annuities, or buffered annuities — are an increasingly popular product.

RILAs can be viewed as an evolution of fixed indexed annuities (FIAs), to some extent. FIAs tend to be relatively one-dimensional, where the annuitant can't lose money (i.e., the initial premium) but have limited upside. With RILAs, there is significantly more upside and downside potential, where the actual risk exposures vary notably by strategy.

Given the complexities associated with RILAs, the financial advising community is increasingly looking at ways to better understand their risks and where RILAs can potentially fit in a portfolio. One way to do this would be to simply apply the current RILA attributes (such as caps or participant rates) to historical market returns.

While this might seem reasonable, it can actually result in an incredibly flawed analysis.

The Problem With Historical Returns

RILAs are built using financial options (i.e., calls and puts). For example, a buffer RILA is built by selling an out-of-the-money (OTM) put option (at the buffer level), buying an at-the-money (ATM) call option, and selling an OTM call option. The higher the price of the OTM put option, the higher bond yields, and the lower relative cost of the OTM call option will all increase the potential upside of the product.

Understanding the pricing mechanics of RILAs is incredibly important when thinking about how the attractiveness has evolved over time and why you can't simply use historical returns to backtest them. For example, right now, it's possible to get a 20% buffer RILA with a 6-year term with an unlimited cap and 100% participation rate.

For those unfamiliar, participation rate is the percentage of the gains of the underlier that would be achieved for the respective term of the product. For example, with a 100% participation rate the annuitant would receive 100% of the gains of the respective index, up to the assumed cap, which for this analysis is assumed to be unlimited.

What's not necessarily clear, though, is how the attributes of today's RILAs compare to those that would have been purchased ten or twenty years ago. To provide some perspective on this, I put together an analysis where I estimate the participation rates for a RILA with a 20% buffer where the underlier is assumed to be the S&P 500 from January 1871 to October 2022.

For the analysis, options are priced using the Black-Scholes model with a constant implied volatility of 25% (both over time and across strike prices), where dividend yields and 10-year government bond yield data obtained from Robert Shiller's website, for terms of one, three and six years.

Capturing Current Market Factors

The primary reason participation rates (as well as caps, if that was the primary upside level) are so much lower today (vs. the long-term historical average) is the relatively lower bond yield environment. Lower bond yields result in less revenue available to generate the upside (i.e., purchase the OTM call options), and to the extent bond yields exceed dividend yields can also negatively impact options pricing dynamics.