In the mid 1990s, I conducted and subsequently published a widely cited research study — jointly with Dr. Steven Posner, who was then working as a derivatives quant for Goldman Sachs — on the nature and magnitude of mortality and expense (M&E) fees charged within variable annuities.
As most readers already know, these products are economically similar to mutual funds but classified as insurance products because of various explicit and implicit insurance guarantees; hence they also benefit from tax-sheltered inside buildup. At the time we conducted this research, most VA policies were only offering a basic guaranteed return-of-premium (RoP) death benefit, which meant that at the market’s worst you got your money back, and only if you died. These guaranteed minimum death benefits, or GMDBs as they are abbreviated, were the only relevant financial economic features that distinguished VA policies from their mutual fund cousins.
Yes, during the mid-1990s some companies started offering variable annuities with more lucrative minimum investment returns or maximum anniversary guarantees. These enhanced GMDBs promised that, in the event of death, beneficiaries would be guaranteed at least the premium deposit, increased by up to 7 percent per annum, or a death benefit equal to the best historical anniversary value. But then again, you had to die to get these benefits and many policies were surrendered and lapsed well before the policyholder ever died. More importantly, at the time of our analysis the bulk of VA assets — and especially the policies inside tax-sheltered qualified plans such as 401(k) or IRA accounts — only offered a plain vanilla (return of premium) GMDB.
Back to the above-mentioned study, we examined a comprehensive database which Morningstar graciously provided to us, which included over 400 VA policies and 7,000 investment sub-accounts. We compared the M&E fees being charged — which ranged from 7 to over 140 basis points per annum — to our theoretical model values. These so-called model values were intended to be a proxy for the wholesale manufacturing costs of these guarantees in the derivative markets. Our primary motivation was academic and intellectual curiosity as to whether consumers were getting their money’s worth on the
GMDBs offered within VAs. To a derivatives trader (or theoretician, such as myself), the embedded options were quite similar to equity put options that grant the holder a right but not an obligation to sell an underlying portfolio at a fixed price on a given date. In the VA case, the maturity date of the option was random — namely, the date of death. We called these securities Titanic options.
And alas our main conclusion was that — if the M&E fee was only meant to cover true risk — the typical VA policyholder was being grossly overcharged for this so-called protection and peace of mind. We found that the basic return-of-premium GMDB was worth no more than 5 to10 basis points of assets per annum. By the term “worth” we meant that it would only cost the insurance company backing the guarantee 5 to 10 basis points to reinsure or hedge their exposure to this risk. Even the more lucrative — although at that time rare — stepped-up, ratcheted and rolled-up GMDBs were worth no more than 60 basis points at most, and that was only if you were old (and male) enough to be close to death, risk tolerant enough to invest aggressively, and persistent enough not to surrender (or 1035-exchange) your policy prior to death.
In addition to these results, our published study reported on a number of other peculiarities in the VA market. As an example, we found that older (unhealthy) males who invested more aggressively were receiving a guarantee that was much more valuable to them relative to younger (healthy) females who invested conservatively. This was because the odds of dying during a bear market — which was the only way the insurance guarantee would pay off — were much higher for the former group compared to the latter. Yet both groups were paying the exact same level of M&E fees. It was akin to traditional life insurance being sold to young and old, healthy and sick, for the exact same premium. Likewise, M&E fees were being charged independently of the actual asset allocation of the underlying sub-accounts, even though the chances of a bond fund or money market fund being underwater at a random time of death was close to zero. We couldn’t help but wonder why weren’t sub-account risk fees being pro-rated by the true risk? Anyway, none of this made much sense to us at the time, especially given our training as financial economists where markets and prices are supposed to reflect costs and benefits. So, we threw our hands up and declared: “Why would anyone buy this?”
And, on the topic of dying, (non-qualified) VA policies do not garner a basis step-up at death. Thus, in stark contrast to most other investments such as mutual funds, exchange-traded funds or even individual stocks, variable annuities were converting lightly taxed capital gains into heavily taxed ordinary income gains. Indeed, you don’t need a Ph.D. in finance to realize that this is the wrong direction on the tax-arbitrage diagrams. Yes, there might have been some unique situations in which the tax deferral might make sense, but you had to work really hard to come up with such scenarios. Either way, that wasn’t the focus of our research.
These findings — which were eventually published after scientific review in the prestigious Journal of Risk and Insurance in 2001 and subsequently quoted numerous times in publications ranging from The Wall Street Journal and Newsweek to Reader’s Digest — were seized upon by investor advocates, financial commentators, regulators and plaintiff lawyers as evidence that variable annuities were overpriced, oversold and unsuitable. At the time, I was quite surprised at the attention this report garnered, since the article itself was full of equations and regressions, which normally don’t travel beyond the ivory tower.
In fact, I actually ended up taking the witness stand in a number of related lawsuits and regulatory actions to opine that a promise of getting your money back when you die was “kind of pointless” and at the very least could be replicated using cheaper forms of life insurance.
Indeed, I still stand behind those results, even if it means that I occasionally come face-to-face with disgruntled insurance industry executives who believe that our results were misguided, for one reason or another. Remember, I never said that variable annuities were evil, dangerous or unsuitable. Our basic position was that for many investors, a similar financial economic outcome could be achieved at a lower cost.
However — and this is the point of this article — in the last few years that I have been observing this industry, I am seeing an enormous shift in the way VA policies are being designed, priced and marketed to the public. It is now time for me to update my official position on these instruments.
In fact, after spending quite a bit of time poring over some of the more recent designs as well as talking to actuaries, regulators and advisors, I’m not even sure these instruments deserve the old (and maligned) name of variable annuities.
This isn’t your grandmother’s variable annuity!
Regardless of what you want to call these increasingly heterogeneous products, it seems the relative value pendulum has swung in the opposite direction. I can no longer claim that you are being overcharged for these guarantees or that you can achieve similar goals at a lower cost. It would be very difficult and expensive to bake a living benefit in your kitchen.