From the August 2013 issue of Research Magazine • Subscribe!

Just How Puzzling Are Annuities?

At least since the ­seminal research of Menahem Yaari in 1965, financial economists have pondered why more people do not buy annuities to hedge longevity risk. Note Franco Modigliani’s famous statement when he received his Nobel Prize in 1985: “It is a well-known fact that annuity contracts, other than in the form of group insurance through pension systems, are extremely rare. Why this should be so is a subject of considerable current interest and debate.”

Income annuities remain widely unpopular yet would help to solve a variety of complex problems with which retirees struggle and which cannot be solved otherwise. Those who own annuities overwhelmingly say they are glad they do. Social Security, which is a lifetime annuity program, is an exceedingly popular program. Annuities provide more income than other investment choices and do so without market risk. And, subject to the claims-paying ability of the insurance carrier, an annuity will keep paying at the stated rate for as long as the annuitant lives. This quandary—why more people don’t buy annuities—is called the “annuity puzzle.”

Notwithstanding the above, Felix Reichling of the Congressional Budget Office and Kent Smetters of the University of Pennsylvania’s Wharton School have recently published a new working paper on annuity optimization that concludes that the low demand for annuities isn’t a puzzle at all. Instead, they claim that annuities are overused. Moreover, they suggest that, in the aggregate, the optimal level of annuity ownership may actually be negative, such that the true annuity puzzle may be why people buy them at all.

Reichling and Smetters’ research, while not yet formally published, will surely foster much reaction and debate. In a recent online interview with my friend and colleague Gil Weinreich, ­Research’s editor, Smetters hit the ­alleged annuity puzzle head on: “The point of our paper is mainly to simply ask: Does it actually make sense that people should buy annuities as much as conventional wisdom says? And the answer is ‘no’ for most Americans.”

According to Smetters, longevity risk is just another in a list of frequently uninsured risks that people suffer at various points in their lives, but not the biggest one: “For workers, the big risk they face is that they could become disabled. For retirees, it is the uninsured medical costs not covered by Medicare or even Medigap policies. When you have one of these shocks—disability or a long-term care need—at the very same time your longevity outlook goes down. Had you bought an annuity ahead of time, you then have this disability shock and you have a need for income at that point.”

The strengths of this working paper are its recognition of and emphasis upon the health care risks we all (and particularly retirees) face and the careful explanation of how a health shock devalues existing annuity payments by both demanding more resources and lowering life expectancy. On the other hand, such a shock increases the value of existing life insurance.

Yet the paper’s major weaknesses (or so it seems to me, at least tentatively) are (a) the suggestion that because of the ways that annuities and life insurance work at cross-purposes, they are an either-or proposition; and (b) the suggestion that it isn’t a good idea—when finances and circumstances permit—to insure both health and longevity risks. Moreover, the paper completely neglects the market risk (sequence risk) associated with failing to purchase an annuity. When a retiree’s portfolio suffers a significant drawdown relatively close to retirement, the likelihood of portfolio failure during the retiree’s lifetime increases dramatically. That risk can be mitigated substantially by purchasing an income annuity.

The paper also neglects the possibility of using an annuity truly designed for longevity risk (a “longevity” or “advanced life delayed” annuity) that only pays out if the annuitant reaches a certain advanced age, often age 85.

Smetters and Reichling are surely correct that people with insufficient means get little benefit from owning annuities. They are also correct that annuities can be purchased much more cheaply after a health shock. But since we are all prone to optimism bias (the expectation that things will turn out well even if not planned out well) and the planning fallacy (our tendency to overestimate our ability to impact and manage the future), and since we routinely and generally tend to underestimate our life expectancy, the need for some form of longevity insurance to make sure we don’t run out of money entirely is likely to be much greater than we tend to think.

Examining Biases

Now, since I have brought up our behavioral and cognitive biases, it’s also appropriate for me to acknowledge how and where I am prone to them. My skepticism of the Smetters and Reichling research may simply be motivated reasoning on my part. Motivated reasoning is reasoning influenced by financial or other interests that fall in line with a particular point of view. We aren’t surprised, for example, that tobacco company executives are far less likely than others to concede the health risks of smoking. Their reasoning is well motivated and well compensated.

While I receive no financial incentive for advocating for any particular financial product or type of product, I have been quite public in my call for greater use of annuitization to provide—together with Social Security and any pension income—a baseline income floor to cover essential retirement needs. My unwillingness to jump on the Smetters and Reichling bandwagon so far may simply reflect my inherent desire to protect my reputation and ego.

My skepticism may also reflect confirmation bias. Confirmation bias is our tendency to see what we want or expect to see. We like to think we act like judges, carefully and dispassionately evaluating all the evidence before reaching a conclusion. Unfortunately, we’re much more like lawyers, running around looking for anything that might support our pre-existing positions and that might undermine the opposition. It’s why Fox News viewers think those of MSNBC are nuts and vice versa.

My own research and the research of many that I admire and respect has long supported the idea that annuitization should be used far more often than it is. Therefore, my less than full-throated endorsement of the Smetters and Reichling thesis may not reflect a serious problem with their work. It may be nothing more than my own confirmation bias.

Sadly, the greatest risk we face in this area is bias blindness, our ability to see the biases of and in others while being blind to our own. It’s why we smile knowingly when the parents of a modestly gifted athlete see an athletic scholarship as virtually inevitable while remaining convinced of the obvious talents of our own children. It is really hard for any of us to change our minds when faced with evidence that contradicts or seems to contradict our firmly held convictions.

While I have read the Smetters and Reichling working paper carefully, I have not vetted it yet; the mathematical modeling that supports it is extensive and fairly complex. I am going to need to do a lot more work on the working paper itself, as well in trying to examine my own biases, to determine how persuasive I think it is. I look forward to the careful analysis and ongoing research of others to help me come to what I hope is a fair and accurate conclusion as to what the paper does and does not show. But it is surely a helpful addition to the literature examining the annuity puzzle, real or imagined.

Page 2 of 3
Single page view Reprints Discuss this story
This is where the comments go.