The Effect of Risk Aversion and Loss Aversion on Equity‐Linked Life Insurance With Surrender Guarantees
DOI | http://doi.org/10.1111/jori.12297 |
Date | 01 September 2020 |
Author | Christian Hilpert |
Published date | 01 September 2020 |
© 2019 American Risk and Insurance Association. Vol. 87, No. 3, 665–687 (2020).
DOI: 10.1111/jori.12297
THE EFFECT OF RISK AVERSION AND LOSS AVERSION ON
EQUITY‐LINKED LIFE INSURANCE WITH SURRENDER
GUARANTEES
Christian Hilpert*
ABSTRACT
We price equity‐linked life insurance with surrender guarantees and ac-
count for risk preferences in the form of risk‐averse and loss‐averse pol-
icyholders in continuous time. Risk‐averse policyholders surrender their
policy for higher equity index values. Compared to optimally surrendered
policies, this behavior creates substantial policy value losses. In contrast,
loss‐averse policyholders surrender once the surrender benefit realizes a
gain but keep under‐performing policies. This disposition effect reduces
the policy value relative to both optimally surrendered policies and policies
surrendered by risk‐averse policyholders. Insurers in competitive markets
need to estimate their policyholders’risk preferences accurately.
INTRODUCTION
Equity‐linked life insurance products combine a classical term life insurance and a
savings component, which invests the policyholder’s premiums in equity. Usually,
they include a capital guarantee plus a minimum interest rate guarantee and an
option to surrender the policy.
1
Policy surrender is a significant risk for insurers as
*Christian Hilpert is at the Lingnan College, Sun Yat‐sen University, Guangzhou 510275, PR
China. Hilpert can be contacted via e‐mail: martin@mail.sysu.edu.cn. We thank Knut Aase
(discussant), Stefan Ankirchner, An Chen, Chunli Cheng, Keith Crocker (the editor), Kon-
stantin Neinstell, Philipp Schaper (discussant), Judith Schneider, Petra Steinorth (discussant),
Alexander Szimayer, and two anonymous referees for helpful comments and discussions.
Additionally, we thank seminar participants of the 41st Simposio de la Asociación Española
de Economía, the 43rd Annual Seminar of the European Group of Risk and Insurance
Economists, the American Risk and Insurance Association Annual Meeting 2016, the 2016
China International Congress on Insurance and Risk Management, the Annual Congress of
the German Insurance Science Association 2016, the 8th Conference in Actuarial Science &
Finance, the 18th Congress on Insurance: Mathematics and Economics, and seminar partici-
pants in Bonn, Hamburg, and Odense for helpful discussions. Financial Support of the
German Research Foundation through the Bonn Graduate School of Economics and the
research training group “Heterogeneity, Risk, and Dynamics in Economic Systems”is
gratefully acknowledged.
665
identified by European Union regulators (Eling and Kochanski, 2013). Empirically,
the surrender rate depends on the policyholder’s economic situation and the pol-
icy’s performance that depends on an underlying equity index or fund. Kuo, Tsai,
and Chen (2003)link the surrender rate to both interest and unemployment rates.
Eling and Kiesenbauer (2013)show the relevance of both policy and policyholder
characteristics for policyholders’surrender decisions.
This paper analyzes the impact of policyholders’risk preferences as on their sur-
render behavior and policy values. We model the surrender decision of a re-
presentative policyholder in the spirit of Albizzati and Geman (1994). The policy-
holder employs mental accounting, as introduced by Thaler (1980, 1985), that is, he
isolates the surrender decision from other portfolio components. We model the
surrender decision as a continuous‐time stopping problem for a policyholder with
both risk‐averse and loss‐averse expected‐utility preferences. We derive the implied
surrender behavior and show that the policyholders’risk preferences strongly in-
fluence surrenders and the policy’s value. For loss‐averse policyholders, a bench-
mark value called the reference point separates payments into gains and losses as in
Tversky and Kahneman (1992). Together with the diminishing utility of gains and
losses implied by risk aversion, the reference point creates an S‐shaped utility
function. If the reference point is zero, we obtain classical constant relative risk‐
aversion preferences, that is, power utility, as a special case.
This paper shows that risk‐averse and loss‐averse policyholders’surrender behavior
deviates from the value‐maximizing strategy that is characterized as an American
option exercise decision. Risk‐averse policyholders surrender for higher equity
index levels compared to the American solution in order to limit the policy risk.
Loss‐averse policyholders surrender systematically different compared to purely
risk‐averse policyholders: because of loss aversion, they do not realize surrender
benefits lower than their reference point. In contrast, they surrender mildly profit-
able policies relative to their reference point to secure gains. Shefrin and Statman
(1985)label this effect as disposition effect. Barberis and Xiong (2009)show that
individual investors commonly display this effect. It influences the policy’s value:
While the early surrender of purely risk‐averse policyholders reduces their con-
tract’s value by up to 15 percent, loss‐averse policyholders also fall short of this
policy value. Surrender behavior substantially affects insurers. Our numerical
analysis shows that misspecification of policyholders’risk preferences leads to
significant mispricing of the policy of up to 20 percent.
Our results are robust with respect to surrender fees and the specification of mor-
tality. Realistic surrender fees strongly discourage surrender because of high costs,
in particular, at early times. The disposition effect, however, is robust with respect to
surrender fees. We specify the mortality as either stochastic mortality in Lee and
1
There are different technical ways of terminating a life insurance contract. One option is to
hand the contract back to the insurance company and collect a surrender benefit, another is
to keep the insurance that the already invested capital provides, but set all future premiums
to zero. In this paper, we consider the first version. The second version is sometimes referred
to as “lapse of the contract.”
666 THE JOURNAL OF RISK AND INSURANCE
To continue reading
Request your trial