MULTIPLE REFERENCE POINTS AND THE DEMAND FOR PRINCIPAL‐PROTECTED LIFE ANNUITIES: An EXPERIMENTAL ANALYSIS

• Published date: 01 January 2016
• Author: Christian Knoller
• DOI: http://doi.org/10.1111/jori.12101
• Date: 01 January 2016

MULTIPLE REFERENCE POINTS AND THE DEMAND FOR

PRINCIPAL-PROTECTED LIFE ANNUITIES:ANEXPERIMENTAL

ANALYSIS

Christian Knoller

ABSTRACT

We conducted an experiment in which participants were confronted with

an experimental annuitization decision. Previous research has argued in

favor of the hypothesisthat a combination of mental accounting and prospect

theory can explain why annuities containing a capital guarantee are

preferred to standard annuities. However, from this perspective people

wouldnot annuitize their assets at all, butrather invest the money in a risk-free

alternative. Recent research has also suggested a “cushion effect.” When all

possibleoutcomes of two options are above a certain goal, this goalserves as a

cushion in case of unfavorable outcomes. Hence, individuals might have a

higher propensity to exhibit risk-seeking behavior. We find that individuals

were indeedmore willing to choose the annuity option if it containeda capital

guarantee and that individuals using this guarantee as a cushion were even

more willing to choose the annuity. Thus, the cushion effect can partially

explain the high demand for guarantee features in annuity contracts.

INTRODUCTION

Life annuities generate a predefined income stream to the end of the policyholder’s

life. Unlike other investment products, they provide insurance against the individual

Christian Knoller is member of the Munich Risk and Insurance Center, Ludwig-Maximilians-

Universit€

at Munich, Schackstraße 4/III, 80539 Munich, Germany. In addition to his academic

work, Christian Knoller works for Allianz SE. Christian Knoller can be conducted via e-mail:

knoller@bwl.lmu.de. Financial support from the research funding program “LMUexcellent” is

gratefully acknowledged. For providing laboratory resources I kindly thank the Munich

Experimental Laboratory for Economic and Social Sciences (MELESSA) of the Ludwig-

Maximilians-Universit€

at Munich. I am thankful for helpful comments from participants at the

MELESSA Brownbag Seminar, the Hamburg-M€

unchen-Hohenheim Insurance Economics

Colloquium, the Doctoral Student “Insurance” Conference at the University of Erlangen-

N€

urnberg, the 2011 DVfVW meeting, the 2011 ARIA meeting, the participants of the 2012 CEAR

Behavioral Risk Management Workshop, as well as constructive comments from three referees.

In particular, I am indebted to Andreas Richter and Glenn Harrison for valuable input. The

conclusions are those from the author and do not necessarily reflect the opinions of Allianz SE.

All remaining errors are mine.

© 2015 The Journal of Risk and Insurance. Vol. 83, No. 1, 163–179 (2016).

DOI: 10.1111/jori.12101

163

longevity risk of outliving one’s assets. Yaari (1965) shows that under quite restrictive

assumptions, expected utility-maximizing investors will find it optimal to completely

annuitize their wealth. However, annuity markets are rather thin all over the world,

which is puzzling in the light of this optimality result. There are a number of potential

factors, such as incomplete annuity markets, transaction costs, bequest motives,

adverse selection in annuity markets, and existing annuities from social security, that

might be able to explain the annuity puzzle within an expected utility framework

(Brown, 2007). However, Davidoff et al. (2005) show that the main result still holds

after relaxing many of Yaari’s original assumptions. They conclude that “the absence

of annuitization outside of Social Security and defined-benefit pensions cannot easily

be explained within a rational life-cycle model” and suggest that “lack of annuity

demand may arise from behavioral considerations.”

Hu and Scott (2007) examine the demand for annuities from a behavioral economics

perspective, referring to the literature on mental accounting (Thaler, 1999) and

framing (Tversky and Kahneman, 1981; Read et al., 1999). From this perspective, an

immediate annuity purchased at the beginning of retirement is evaluated in a

separate mental account. It is not considered as insurance against longevity risk but

rather as an investment. The annuity is viewed as a bet on the policyholder’s own life,

since the level of total repayment depends on when he dies. The bet is then viewed as

only being profitable if he lives long enough to recoup the invested amount. Due to

loss aversion, the potential losses in the case of an early death outweigh the possible

gains when he lives longer than expected and hence the annuity is rejected.

1

Brown et al. (2008) find evidence that the framing of annuities and other financial

products can influence individuals’ investment decisions. These products were

presented by them to roughly half of the participants in a survey using a consumption

frame that highlighted the impact on consumption over lifetime. The insurance

characteristic of the annuity concerning longevity risk therefore became salient. In

this treatment, annuities were strongly preferred to other types of financial products.

For the other half of the survey participants, the annuity and the alternative financial

products were presented in an investment frame focusing on risk and return. Savings

accounts and other financial products were strongly preferred to annuities in this

framing. Gazzale and Walker (2009) argue that the “hit-by-bus” concern, the worry

about dying soon after the annuitization date and thus losing money to the insurance

company, might enforce the aversion against annuities. They find evidence that many

people face a “risk-ordering bias” that might even strengthen the loss aversion effect.

In their incentivized experiment, participants had to choose between a lump sum

payment and an experimental annuity option. The number of participants choosing

the annuity was lower when they had to sequentially survive each period than when

the number of survived periods was determined by a single draw from a known

distribution.

Many individuals purchase capital guarantee features along with their annuities.

Period-certain annuities ensure that the policyholder or beneficiary will receive

payments at least for a predefined number of years, even if the policyholder dies

1

See also Benartzi et al. (2011) for potential behavioral economics obstacles to annuitization.

164 THE JOURNAL OF RISK AND INSURANCE