What Makes Annuitization More Appealing?

Featured in print Bulletin on Aging & Health

Relatively few U.S. households annuitize their private pension balances at retirement or purchase annuities with other assets. In defined benefit (DB) pension plans that offer the choice of receiving benefits as an annuity or a lump sum, between 50 and 75 percent of benefits are taken as a lump sum, even though receiving an annuity is the default option. In defined contribution (DC) plans, only 10 percent of those leaving their jobs after age 65 choose to annuitize.

Economists have long pointed out that the benefit of buying insurance against the risk of outliving one's assets should create strong demand for annuities. Numerous rational motives have been suggested to explain the puzzlingly low demand for annuities, including bequest motives, uncertain healthcare expenses, high annuity prices (due in part to adverse selection, the tendency of those with longer expected lives to purchase annuities at higher rates), the existence of means-tested government benefits such as Medicaid, and the level of mandatory annuitization through Social Security and some DB plans.

In a new paper, researchers John Beshears, James Choi, David Laibson, Brigitte Madrian, and Stephen Zeldes ask What Makes Annuitization More Appealing? (NBER Working Paper 18575). In posing this question, the authors take no stand on how much of the "annuitization puzzle" remains after accounting for these rational motives. Instead, the authors explore how annuity demand responds to annuity product design and choice architecture.

To study these issues, the authors field two large surveys in which they elicit hypothetical annuitization choices from individuals aged 50 to 75. As the authors note, the use of such surveys has both advantages and disadvantages. On the one hand, they are able to ask questions that directly measure preferences, including preferences for products that may not exist on the market today. On the other hand, the answers people give do not translate into choices that affect their actual outcomes, so the results may not correspond to the choices people would make in real life.

The authors have five major findings. First, the three considerations respondents report as being most important for their decision about whether or not to annuitize include: a desire to "make sure I have enough income later in life," a desire for "flexibility in the timing of my spending," and being "worried about [the] company not being able to pay me in the future." The authors point out that insurance companies are currently banned from mentioning government insurance of annuity payments, a policy that may reduce annuity demand. Other issues, including worries about inflation, the desire to invest one's own money, and the desire to prevent overspending were intermediate level concerns, while the desire to give money to children and concerns about dying early were less relevant.

Second, the authors find that a majority of individuals choose partial annuitization when it is offered. Furthermore, offering partial annuitization increases both the share of respondents who say they would annuitize and the average percentage of pension balances that is annuitized. The U.S. Treasury Department has proposed a new regulation that would make it easier for DB plans to offer partial annuitization - the authors' findings suggest that this proposal would increase annuitization in plans that already offer a lump sum withdrawal option.

Third, the authors find that for a given present value of payments, respondents prefer flat or rising real payments over time to declining payments. This finding makes it all the more surprising that there are few inflation-indexed annuities available on the market. The authors also find that highlighting the effects of inflation on fixed nominal payments increases the demand for annuities with a cost of living adjustment.

Fourth, the authors test how various frames affect annuity demand. They find that framing that focuses on flexibility and control or on investment attributes decreases the demand for annuities. Other frames are found to have little effect on annuity demand, including explaining that the annuity being offered is a better deal than what could be purchased on the private market (as would generally be true for an annuity offered through a DB plan), presenting the total expected undiscounted lifetime payments from the annuity, explaining that an annuity provides insurance against outliving one's savings, and explaining that an annuity transfers money from states where one is dead and the value of money is low to states where one is alive and the value of money is high.

Finally, the authors find that a majority of respondents prefer an annuity that pays an annual bonus in a month of their choosing over a traditional annuity with a uniform monthly payment. This is consistent with respondents highlighting flexibility in the timing of spending as an important factor in their annuity decision.

The authors conclude by noting that their findings have implications for annuity product design and choice architecture. First, to increase annuity demand, annuity providers could design products that give beneficiaries more flexibility and control. This could include, for example, giving annuitants control over the payout stream within each year or over asset allocation (as already occurs in the variable annuity market), though the authors caution that there is a tradeoff between flexibility and complexity. Providing greater access to partial annuitization for DB plan balances, using frames that downplay investment, and adopting regulations that reduce fear that providers will be unable to meet their annuity obligations are other policies that may increase annuity demand.

The authors acknowledge funding from the U.S. Social Security Administration through grants to the RAND/Dartmouth/Wharton Financial Literacy Center and the NBER as part of the SSA Retirement Research Consortium (grants FLR09010202-02 and #5RRC08098400-04-00), the TIAA-CREF Institute, and the National Institutes of Health (grants P01-AG005842, P30-AG034532, and R01-AG021650). At least one author has disclosed a financial relationship of potential relevance for this research. Further information is available online at http://www.nber.org/papers/w18575.