Through the years I have run hundreds of thousands of hypothetical return calculations for indexed annuities—and that is not hyperbole.
Hypothetical output, based on past index movements, can give you an idea of future returns—if there are no moving parts. That means the participation rate, caps and spread are locked in and cannot change during the 5-, 7-, 10- or 12-year surrender period of the product—and the past repeats. Sadly, the past doesn’t repeat, except in very broad sweeps with irregular timing, which means even if an annuity’s participation rate, caps and spreads are known, this hypothetical modeling cannot predict future results. The only thing this modeling does show (if you have no moving parts) is how the annuity would have performed in past periods. Plus, if a hypothetical model does not have all of the crediting parts locked in for the term it cannot even predict the past.
One problem with models in which the parts may move is that almost no one moves them. Influenced by a cognitive bias called naive extrapolation, most models take the current day’s rates, apply them to past periods, and then conclude that is how annuities would have performed. Thus, hypothetical illustrations done today use caps of 4 to 5 percent; those done a few months earlier used caps of 2 to 3 percent; and illustrations calculated a few years ago assumed caps would never, ever, be lower than 8 percent. Reality has shown how foolish it is to assume that everything will stay the same in the financial world—and that includes using a static cap.
Every modeling example I’ve seen from a carrier or agent assumes this never-changing world, but because the world does change (as do annuity caps) the output generated is meaningless garbage. Saying “let’s assume the caps from 1996 to 2001 were 3 percent”—when in reality they averaged 9 percent—means your assumption is invalid, as are your results. Sadly, even academic studies looking at annuity returns (or studies trying to appear academic) usually suffer from this bias, making their conclusions worthless.
Very few studies have attempted to adjust to the real world by looking at historic annuity pricing factors and then deciding what the index annuity rate or cap should have been. So they’ll note that in 1997 the average bond yield was “X” and the implied volatility of the index option was “Y” and conclude that the cap would have been 14 percent. This works better than assuming nothing ever changes, yet, like all past-based predictions, this assumption is handicapped by the reality of the past. What this means is although one could have done a hypothetical model using the guaranteed minimum caps and rates, no one did because no one would have believed that 2 percent caps would ever be a reality 5, 10 or 15 years ago.
Taking into account past pricing factors does let you know what caps and rates should have been. I have copies of hypothetical illustrations that assumed participation rates of 200 to 300 percent (without caps) for certain years in the early 1980s when bond yields were in the double digits. Can you imagine telling an annuity owner “the index went up only 6 percent, but you earned 15 percent interest”? Sounds absurd, and this points to the fundamental reason why indexed annuity illustrations cannot predict the future or the past.
Unlike an indexed mutual fund or exchange traded fund where next year’s return is index performance minus a previously known fee, the degree to which an indexed annuity participates in the index return next year is determined by the annuity carrier. Although the return cap is largely based on the realities of a carrier’s investment portfolio returns and the price of hedging the cap on the annuity policy, there is a human factor. This human factor influences and may override the financial ones in setting the cap for the coming year.
Sometimes the human factor provides a more generous cap than finances dictate. I have witnessed this happening during the last couple of years at several carriers when caps were renewed at a higher level than economic factors justified. Sometimes the human factor results in a stingier cap than finances allowed, something I witnessed in the middle of the last decade. The reasons why the annuity carrier may decide to ignore their own financial pricing model are varied, but are often due to what the competition is doing. Thus, if your pricing model permits you to offer a 250 percent participation rate (that 15 percent interest mentioned earlier) and all of your competitors are paying 100 percent, you might offer 110 percent of index performance (but not 250 percent).
Does this mean that creating hypothetical illustrations of indexed annuity performance (and indexed life performance as well) is a waste of time? If you’re trying to tell someone how their annuity (or IUL policy) will perform in the future, the answer is “yes.” However, this doesn’t mean you should not do hypothetical modeling.
What the models will tell you is whether a carrier’s current rates and caps are out of whack with the rest of the field. What I mean is if you run illustrations for 10 carriers and one has an average return 3 percent higher than the others, you need to find out why. It could be that the one carrier has lower costs or that its crediting method performs better. A more likely answer is that the current cap is overly generous and will be “adjusted” over time.
What it comes down to with indexed annuities or indexed life insurance is that your return is far more dependent on how the carrier treats you when they renew those participation rates, caps and spreads than on the financial factors.
Find a carrier you trust and throw away the spreadsheets.
