Who Chooses Annuities? An Experimental Investigation of the Role of Gender, Framing, and Defaults

Full text

Page 1

“Who Chooses Annuities?
An Experimental Investigation of the Role of Gender, Framing and Defaults”

Julie R. Agnew, Lisa R. Anderson, Jeffrey R. Gerlach and Lisa R. Szykman

December 18, 2007
Julie R. Agnew (corresponding author), Assistant Professor of Economics and Finance, The
Mason School of Business, College of William and Mary, P.O. Box 8795, Williamsburg, VA
23187, phone (757) 221-2672, fax (757) 221-2937, e-mail: julie.agnew@mason.wm.edu.

Lisa R. Anderson, Professor of Economics, Department of Economics, College of William and
Mary, P.O. Box 8795, Williamsburg, VA 23187, phone (757) 221-2359, fax (757) 221-1175,
e-mail: lisa.anderson@wm.edu.

Jeffrey R. Gerlach, International Faculty Fellow, MIT Sloan School of Management, E48 501
238 Main Street, Cambridge, MA 02142, phone: (617) 253-7477, e-mail: jgerlach@mit.edu.

Lisa R. Szykman, Associate Professor of Marketing, The Mason School of Business, College of
William and Mary, P.O. Box 8795,Williamsburg, VA 23187, phone (757) 221-2908, fax (757)
221-2937, e-mail: lisa.szykman@mason.wm.edu.



AEA Session January 5, 10:15 am Hotel Riverside
Session Title: Savings and Investment Decisions: How Do Women Fare?
Session Chair: Olivia Mitchell
Discussants: Silvia Ardagna, Pascaline Dupas, Olivia Mitchell and Enrichetta Ravina

Page 2

“Who Chooses Annuities?
An Experimental Investigation of the Role of Gender, Framing and Defaults”
Julie R. Agnew, Lisa R. Anderson, Jeffrey R. Gerlach and Lisa R. Szykman*

Past research suggests that women are more risk averse and less financially literate than
men, and this is demonstrated in less risky asset investments. We contribute to the literature by
focusing on the role of gender in an increasingly important financial decision facing individuals
at retirement, the choice between purchasing an annuity (in this case, a fixed immediate lifetime
annuity) or investing their savings on their own. By using a controlled experiment, we eliminate
the role of adverse selection and unfair annuity pricing and are able to focus specifically on
gender. We also explore the role of defaults and framing, and whether women react differently to
these features. We find that women are more likely to choose the annuity and this is only partly
explained by differences in risk aversion and financial literacy. Furthermore, biases in a five-
minute presentation of information significantly affect choices in ways that differ across men and
women.

I. Why the Annuity Decision?
We focus on the annuity decision for several reasons. First, it is a relatively complicated
and increasingly important financial decision individuals must make as companies shift from
defined benefit plans to defined contribution plans. While research has shown that women prefer
less risky asset allocations, we know less about how women’s preferences differ from men’s in

*Contact Information: Julie Agnew, Mason School of Business, College of William and Mary, Williamsburg, VA
23187, julie.agnew@mason.wm.edu; Lisa Anderson, Department of Economics, College of William and Mary,
Williamsburg VA 23187, lisa.anderson@wm.edu; Jeffrey Gerlach, MIT Sloan School of Management, E48 501 238
Main Street, Cambridge, MA 02142, jgerlach@mit.edu and Lisa Szykman, Mason School of Business, College of
William and Mary, Williamsburg, VA 23187, lisa.szykman@mason.wm.edu. Financial support from FINRA
Investor Education Foundation (formerly NASD) is gratefully acknowledged.

1

Page 3

their decision to annuitize or not. Second, the annuity decision is not well understood and may
not be rational. The small size of the actual annuity market is inconsistent with theoretical
expectations even when rational extensions like adverse selection, pre-existing annuitization, risk
sharing by couples, high annuity prices, and bequest motives are added to the basic model.
Jeffrey R. Brown (2007) provides a thorough summary of the past literature, defines the “annuity
puzzle,” and suggests that future research should focus more on behavioral explanations. Our
study addresses a potential behavioral explanation and relies on a framing approach based on
loss aversion.1 Third, information about the annuity decision can be negatively framed to support
either choice, by focusing on the possibility of outliving resources when choosing the investment
option or the possibility of purchasing an annuity and not reaping the benefits because of dying
soon after. Thus, we can study how framing might influence participants either towards or away
from purchasing annuities, as well as look for gender-specific effects. Finally, a topic of recent
debate is whether annuitization should be the default distribution option in 401(k) plans.

II. Why Might Gender Matter?
While there is not a known psychological reason why gender should matter, it may be
that gender is proxying for other factors that influence financial decisions like risk aversion and
financial literacy. There is evidence that women are more risk averse than men in general
(Catherine Eckel and Philip Grossman forthcoming), and this translates to investing in less risky
assets in their retirement plans (Julie R. Agnew, Pierluigi Balduzzi and Annika Sunden 2003).
Olivia S. Mitchell et. al. (1999) show that as risk aversion increases, individuals are willing to
pay more for annuities. Given the evidence that women are more risk averse than men, we
hypothesize that they are more likely to choose annuities. However, if risk aversion is driving the

1 Our experiment is not designed to test for loss aversion.

2

Page 4

decision to choose annuities, after controlling for it gender differences in the annuity decision
may not be significant.
Differences in financial literacy between men and women may also explain differences in
their annuity decision. In most studies measuring financial literacy, women score lower than
men, and it is possible that literacy is driving the decision to annuitize. In fact, Peggy D. Dwyer,
James H. Gilkeson and John A. List (2002) find that the gender differences observed in mutual
fund investment decisions can be largely explained by differences in knowledge.

III. Defaults and Framing
The importance of defaults and framing to decision making has been established in
many contexts. In retirement research, the influence of defaults on financial decision making is
well documented. In fact, recent changes in 401(k) plans from voluntary enrollment to automatic
enrollment were driven by the finding that people are more likely to participate when they are
enrolled by default. While the role of information framing in choice, in general, is well known,
its role in financial retirement decisions specifically is not.
In this paper, we adopt a framing technique that has been shown to be effective in
improving preventative health behaviors. Negative framing (also known as “fear appeals”) has
been effective in increasing preventative health behaviors related to colon cancer, breast cancer,
sexually transmitted diseases and skin cancer (see for example, Lauren Block and Punam Anand
Keller 1995)
To test whether this framing technique can influence the annuity decision, we created an
experiment that either highlights the potential financial losses associated with the annuity choice
or the investment choice. If we are able to influence the decision based on the negative framing
of information, it is possible that these effects may also exist in the marketplace.

3

Page 5

In our experiment, we first showed our participants a five-minute slide show that
included one of three manipulated conditions. One slide show highlighted the negative features
of the investment option and provided the annuity as the solution to avoid the drawbacks. The
second slide show highlighted the negative features of the annuity option and provided the
investment option as the solution to overcome the drawbacks. The third slide show was a neutral
condition, in which neither option was favored. A summary of the information contained in each
of these manipulations is included in Appendix A, and the three slide shows are included in
Appendix B.

IV. Experimental Design
Risk preferences and financial literacy were collected at the beginning of each
experimental session using modifications of commonly used instruments. To capture risk
tolerance in the laboratory, we had participants complete the lottery choice experiment from
Charles A. Holt and Susan K. Laury (2002). Following the lottery experiment, subjects
completed a financial literacy survey adapted from several literacy tests used in previous
studies.2
Next participants were asked to play a “Retirement Game” which began with one of the
three slide shows discussed above. Then subjects were given unbiased instructions with specific
details about the experiment. In the retirement experiment subjects were given $60 to either
purchase an annuity or to invest in a self-chosen portfolio composed of a risk-free asset and our
simulated “market.”3 If participants chose the annuity, they received $16.77 for every round they
survived in the game. If the investment choice was selected, then subjects had to choose how

2 Appendix B provides more information on the lottery choice experiment and the financial literacy survey.
3 Unfair annuity pricing and adverse selection were avoided by making the annuity price actuarially fair and making
subjects aware of their identical survival probabilities over the six period game upfront.

4