THE REFLECTION EFFECT FOR HIGHER-ORDER RISK PREFERENCES

THE REFLECTION EFFECT FOR HIGHER-ORDER RISK PREFERENCES

Han Bleichrodt and Paul van Bruggen*

Abstract—Higher-order risk preferences are important determinants of eco-
nomic behavior. We apply insights from behavioral economics: we measure
higher-order risk preferences for pure gains and losses. We find a reflec-
tion effect not only for second-order risk preferences, as did Kahneman
and Tversky (1979), but also for higher-order risk preferences: we find risk
aversion, prudence and intemperance for gains and much more risk-loving
preferences, imprudence and temperance for losses. These findings are at
odds with a universal preference for combining good with bad or good
with good, which previous results suggest may underlie higher-order risk
preferences.

I.

Introducción

WHILE risk aversion has been the cornerstone of the eco-

nomic analysis of decision making under risk since the
1950s, only relatively recently have higher-order risk pref-
erences been receiving the attention they arguably deserve.
Although prudence as a concept has been used in the anal-
ysis of intertemporal risk preferences since Leland (1968),
Sandmo (1970), and Drèze and Modigliani (1972), specifi-
cally as a precautionary savings motive, the use of the term
comes from Kimball (1990). Temperance was introduced
even more recently, as a concept by Pratt and Zeckhauser
(1987) and as a term by Kimball (1992). Model-free defi-
nitions of prudence and temperance proposed by Eeckhoudt
and Schlesinger (2006) are illustrated in figure 1.1 Prudence is
equivalent to aversion to downside risks (Menezes, Geiss, &
Tressler, 1980), and temperance relates to whether the pres-
ence of background risk makes a person more risk averse
(Gollier & Pratt, 1996).

The importance of these preferences has been shown in
such different contexts as insurance in the presence of back-
ground risks (fei & Schlesinger, 2008), auctions (Es˝o &
Blanco, 2004), and probabilistic insurance (Eeckhoudt &
Gollier, 2005; Peter, 2017). Prudence also plays an impor-
tant role in explaining several macroeconomic consumption
puzzles through the precautionary savings motive it induces
(Caballero, 1990). Evidence of higher-order risk preferences
has been found with field data (Browning & Lusardi, 1996;

Received for publication October 25, 2018. Revision accepted for publi-

cation July 27, 2020. Editor: Brigitte C. Madrian.

∗Bleichrodt: Erasmus School of Economics and University of Alicante,
Departamento de Economía; Van Bruggen: Tilburg University, Departamento
of Economics.

We are grateful to Aurélien Baillon, Olivier l’Haridon, Gijs van de Kuilen,
and Peter Wakker for their comments on the manuscript; the editor, Brigitte
Madrian, and two anonymous referees for their suggestions and comments;
and Jingni Yang and Victoria Granger for helping us conduct the experiment.
The experiment reported in this paper was partially funded by contributions
from the Erasmus Research Institute of Management and the Tinbergen
Instituto.

A supplemental appendix is available online at https://doi.org/10.1162/

rest_a_00980.

1Higher-order risk preferences have been extended to multivariate risk
preferences (Eeckhoudt, Rey, & Schlesinger, 2007), to ambiguity (Baillon,
2017), and to time (Ebert, 2020). For an introduction to higher-order risk
preferences, see Eeckhoudt and Schlesinger (2013).

Carroll & Samwick, 1998), as well as in experiments (see the
review article by Trautmann & Van de Kuilen, 2018).

en este documento, we apply insights from behavioral economics
to the study of these higher-order risk preferences. Daniel
Kahneman called reference dependence “the core of prospect
theory” (kahneman, 2003, pag. 1457). Yet reference depen-
dencia, which leads to a reflection of risk aversion over the
gain and loss domain, has not been thoroughly investigated
or controlled for in existing studies. We measure higher-order
risk preferences while directly controlling the reference point
to separate gains and losses. We control the reference point
by giving subjects an endowment before the experiment.2

Separating gains and losses allows us to investigate
whether higher-order risk preferences, like risk aversion, re-
flect between the gain and loss domains. This is important
for two reasons. The first is external validity. Preferences
measured under gains in an experiment may be a poor pre-
dictor for choices that naturally involve losses, such as in-
surance and self-protection decisions. Decision makers may
purchase insurance that involves nonperformance risk despite
a prediction based on prudence observed under gains that
they would not demand such insurance (Eeckhoudt & Gol-
lier, 2005). Asimismo, prevention efforts are closely related to
prudence, and basing health policy on estimates of prudence
under gains may be suboptimal.

The second reason is that the domain will influence deci-
sions dependent on higher-order risk preferences. In a finan-
cial crisis, when their portfolios are deeply in the loss domain,
investors may come to prefer downside risks, and greater
volatility on the financial markets may lead investors to de-
mand insurance against unrelated risks where under more
normal circumstances, greater volatility on financial markets
has no such effect.

Separating gains and losses means we measure preferences
without loss aversion, which affects choices for lotteries that
mix gains and losses. Por ejemplo, if subjects take the high-
est possible outcome they are sure to get (the MaxMin) como
the reference point, loss aversion will bias responses toward
risk aversion, prudence, and temperance.3 Thus, we can test
the hypothesis of a preference for combining good with bad
or for combining good with good without the confounding
effects of loss aversion. Eeckhoudt, Schlesinger, and Tsetlin
(2009) and Crainich, Eeckhoudt, and Trannoy (2013) espectáculo
that these simple preferences may underlie higher-order risk
preferences, in which case decision makers who are risk

2Deck and Schlesinger (2010) and Maier and Rüger (2012) also had treat-
ments to investigate the influence of the domain on higher-order risk pref-
erences. We discuss these studies in section V.

3With such a reference point, the worst outcome between two lotteries is
perceived as a loss. This is always an outcome of the option that indicates
risk-loving preferences, imprudence, or intemperance. Loss aversion then
decreases the relative attractiveness of this option.

La revista de economía y estadística., Julio 2022, 104(4): 705–717
© 2020 The President and Fellows of Harvard College and the Massachusetts Institute of Technology. Publicado bajo una atribución Creative Commons 4.0
Internacional (CC POR 4.0) licencia.
https://doi.org/10.1162/rest_a_00980

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

mi
d
tu
/
r
mi
s
t
/

yo

a
r
t
i
C
mi

pag
d

F
/

/

/

/

1
0
4
4
7
0
5
2
0
3
3
3
7
3
/
r
mi
s
t
_
a
_
0
0
9
8
0
pag
d

.

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
7
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

706

THE REVIEW OF ECONOMICS AND STATISTICS

FIGURE 1.—DEFINITIONS OF RISK AVERSION, PRUDENCE, AND TEMPERANCE

Risk aversion (arriba), prudence (abajo a la izquierda), and temperance (bottom right), based on Eeckhoudt and Schlesinger (2006), for all k, r > 0 and all zero-mean random variables ˜ε, ˜ε1, and ˜ε2.

averse should be temperate, whereas those who are risk lov-
ing should be intemperate, and both should be prudent.

We also measure higher-order risk preferences with lotter-
ies for which the branches in figure 1 involve probabilities
menor que 0.5.4 We extend the definitions of Eeckhoudt
and Schlesinger (2006) to such small probability lotteries
and show that the results of Eeckhoudt et al. (2009) apply to
a ellos.

We find a reflection effect for higher-order risk prefer-
ences: under gains, the majority of choices are risk averse,
prudent, and intemperate, whereas under losses, the majority
of choices are risk loving, imprudent, and temperate. The im-
prudence we find for losses and the combination of risk aver-
sion and intemperance for gains is at odds with the hypothesis
that a universal preference for combining good with bad or
good with good underlies higher-order risk preferences. Nosotros
find similar behavior for small probability lotteries and the
usual 50-50 lotteries.

In section II, we discuss the definitions of higher-order risk
preferences in detail and how these may be derived from pref-
erences for combining good with bad or good with good. Nosotros
also extend the definitions of higher-order risk preferences
so they can be used with smaller probabilities. We present
the design of our experiment in section III. In section IV, nosotros
present our results, which we discuss and relate to previous
findings in section V. Section VI concludes.

II. Theoretical Background

Eeckhoudt and Schlesinger (2006) define higher-order risk
preferences through simple lottery pairs. Dejar [X, y] denote the
lottery that gives outcome x with probability 0.5 and outcome
y with probability 0.5. Dejar (cid:3) denote the decision maker’s
preference relation. Risk aversion is defined as the prefer-
ence [−k, −r] (cid:3) [0, −k − r] for all wealth levels and for all
k, r > 0. Eeckhoudt and Schlesinger (2006) name such an at-
titude risk apportionment of order 2. A decision maker is risk

4Ebert and Wiesen (2011, 2014) use skewed zero-mean lotteries ( ˜ε in fig-
ura 1), which therefore involve probabilities smaller than 0.5, but the effect
of smaller probabilities there is confounded by a preference for skewness.

loving if the reverse preferences hold. To define prudence
(risk apportionment of order 3), the fixed deduction −r is
replaced by a zero-mean nondegenerate random variable ˜ε
(see figure 1). The decision maker is prudent if [−k, ˜ε] (cid:3)
[0, ˜ε − k] for all wealth levels, for all k > 0, and for all zero-
significar, nondegenerate random variables ˜ε. The decision maker
is imprudent if the reverse preferences hold. Temperance (riesgo
apportionment of order 4) is defined by replacing −k by an-
other independent random variable. The decision maker is
temperate if [ ˜ε1, ˜ε2] (cid:3) [0, ˜ε1 + ˜ε2] for all wealth levels and
for all zero-mean, nondegenerate, and independent random
variables ˜ε1 and ˜ε2. If the reverse preferences hold, the risk at-
titude is called intemperance. Risk attitudes of orders higher
than 4 can be defined through similar procedures, but we do
not study those attitudes in this paper.

Eeckhoudt et al. (2009) show how stochastic dominance
preferences lead to risk apportionment of any order. Ellos
show that if ˜xi dominates ˜yi, i = a, b, through ith order
stochastic dominance, then the 50-50 lottery [ ˜xa + ˜yb, ˜xb +
˜ya] dominates [ ˜ya + ˜yb, ˜xa + ˜xb] a través de (a + b)th stochas-
tic dominance. Stochastic dominance preferences thus im-
ply a preference for combining the “good” lottery with the
“bad” lottery and contain risk apportionment preferences as
defined by Eeckhoudt and Schlesinger (2006) as a special
caso. A preference for combining good with bad thus leads
to a combination of risk aversion, prudence, and temperance,
Por ejemplo.

Crainich et al. (2013) apply this logic to decision makers
who prefer combining good with good and show that this
leads to a combination of risk-loving, prudent, and intemper-
ate preferences. It follows that indifference toward combining
good with bad and good with good leads to risk neutrality,
prudence neutrality, and temperance neutrality. The results
above also hold when all final outcomes are only in the do-
main of gains (relative to some reference point) or of losses.
We can therefore test for these implications separately for
both domains to see if there are behavioral differences be-
tween them.

Because we also study preferences over lotteries involv-
ing probabilities different from 50-50, we need to extend the

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

mi
d
tu
/
r
mi
s
t
/

yo

a
r
t
i
C
mi

pag
d

F
/

/

/

/

1
0
4
4
7
0
5
2
0
3
3
3
7
3
/
r
mi
s
t
_
a
_
0
0
9
8
0
pag
d

.

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
7
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

THE REFLECTION EFFECT FOR HIGHER-ORDER RISK PREFERENCES

707

above definitions. Dejar (pag: X, pag: y, 1 − 2p: z) denote the lottery
that gives outcomes x and y with probability p and outcome
z with probability 1 − 2p. Risk aversion is then defined as

(pag: −k, pag: −r, 1 − 2p: C) (cid:3) (pag: 0, pag: −k − r, 1 − 2p: C),

prudence as

(pag: −k, pag: ˜ε, 1 − 2p: C) (cid:3) (pag: 0, pag: ˜ε − k, 1 − 2p: C),

and temperance as

(pag: ˜ε1, pag: ˜ε2, 1 − 2p: C) (cid:3) (pag: 0, pag: ˜ε1 + ˜ε2, 1 − 2p: C)

for all p ∈ [0, 1], all k, r > 0, all c, all independent zero-mean
risks ˜ε, ˜ε1, and ˜ε2 and all wealth levels. Risk-loving pref-
erences, imprudence, and intemperance are defined as the
reverse preferences. Under expected utility, these extended
definitions are equivalent to the usual definitions of risk aver-
sión, prudence, and temperance (necessity follows from the
independence axiom and sufficiency follows from their def-
inition). Además, the results of Eeckhoudt et al. (2009)
apply to these extended definitions: si [ ˜xa + ˜yb, ˜xb + ˜ya] dominación-
inates [ ˜ya + ˜yb, ˜xa + ˜xb] a través de (a + b)th stochastic dom-
inance, then the probabilistic mixture (λ: [ ˜xa + ˜yb, ˜xb +
˜ya], 1 − λ: C) also dominates (λ: [ ˜ya + ˜yb, ˜xa + ˜xb], 1 −
λ: C) a través de (a + b)th stochastic dominance (a proof is in
appendix A).

III. Diseño

We measure higher-order risk preferences in three treat-
mentos, two involving gains (with all outcomes positive addi-
tions to the initial payment) and one involving losses (con
all outcomes negative additions to the initial payment). A
induce a strong reference point, subjects face both gains and
losses relative to their initial endowment; we therefore have
a within-subject design for testing gains and losses. The two
gain treatments involve a between-subject design (subjects
were assigned randomly to either gain treatment). El 50-
50 gain treatment involves the usual 50-50 lotteries, y el
small probability gain treatment involves the small proba-
bility lotteries discussed in section II. The small probability
treatment allows us to offer the possibility of sizable gains.
For the loss treatment, we measure preferences using the
usual 50-50 lotteries. Small probabilities cannot be used to-
gether with large losses, because losses exceeding the ini-
tial endowment would lead to negative earnings. For each of
the three treatments, we measure three higher-order risk at-
titudes (risk aversion, prudence, and temperance). De este modo, nosotros
have nine treatment-risk attitude pairs in total.

To study the effects of reference-dependence, it is impor-
tant to control the reference point. Para tal fin, subjects were
given a 15 euro endowment at the start of the experiment.
They were told that this endowment was their payment for
participating in the experiment, that they could gain addi-
tional money or lose part of it, and that it was equal to the

expected value of participating. Throughout our analysis, nosotros
assume that subjects take the initial endowment as their refer-
ence point. This is a common assumption in the literature and
is consistent with a reference point based on rational expec-
taciones (K˝oszegi & Rabin, 2006) or based on the status quo.
Baillon, Bleichrodt, and Spinu (2020) find evidence that a siz-
able fraction of subjects take the status quo as their reference
punto, and Etchart-Vincent and l’Haridon (2011) find that be-
havior is similar under losses from an initial endowment and
losses out of subjects’ own pockets.

A screenshot of one of the tasks is in appendix B, y
the tasks are listed in table 6 in appendix C. Lotteries are
presented in compound form in all tasks. We use compound
lotteries because they most clearly present the choice as be-
tween combining good with bad or good with good. Haering,
Enrique, and Mayrhofer (2020) find that prudence and tem-
perance are stronger for compound lotteries than for reduced-
form lotteries. Deck and Schlesinger (2017) investigate pre-
senting lotteries in compound form and in reduced form and
find that while aggregate patterns are not much different,
individuals have different preferences between the different
formats.

Probabilities are presented as drawing a colored token from
a bag with 100 colored tokens. To avoid mixing gains and
losses, outcomes are chosen in such a way that relative to
the initial payment, they are always negative in the loss treat-
ment and always positive in the gain treatments. Hay
no zero outcomes (relative to the initial 15 euro) to prevent
possible effects such as loss and zero avoidance from influ-
encing responses. Subjects were presented with twelve tasks
for each risk attitude in a given treatment. The relatively large
number of tasks allows us to distinguish subjects who are
indifferent (or confused) from those who have a clear atti-
tude without having to measure willingness to pay. Measur-
ing willingness to pay would complicate the procedure while
the binary choices are already quite complex, especialmente para
temperance.

The experiment was performed in the ESE-econlab of the
Erasmus School of Economics. Subjects were randomly se-
lected from the ESE-econlab subject pool, which consists of
people who have registered to participate in experiments and
invited to sign up for sessions through an automated email
sistema. El 245 subjects who participated in the experiment
hecho 17,640 choices.5 Upon entering the lab, subjects were
given an envelope containing 15 euros and assigned a seat
in a cubicle. They started the experiment and left the lab at
the same time and made their choices on a computer. En el
instructions, they were informed that one of their choices in
the experiment would be implemented for real, which would
result in winning additional money or losing part of their ini-
tial payment, and that their expected earnings were equal to

5Del 245 subjects, 122 participated in sessions in January 2017, cual
was replicated with the other 123 subjects in May 2019. As the data from
both samples were very similar and no significant differences were found,
we do not distinguish between the two data sets in the main text. A discussion
is in the online appendix.

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

mi
d
tu
/
r
mi
s
t
/

yo

a
r
t
i
C
mi

pag
d

F
/

/

/

/

1
0
4
4
7
0
5
2
0
3
3
3
7
3
/
r
mi
s
t
_
a
_
0
0
9
8
0
pag
d

.

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
7
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

708

THE REVIEW OF ECONOMICS AND STATISTICS

the amount they had been given. The instructions furthermore
contained an explanation of the possible outcomes of a lot-
tery similar to one of the more complicated lotteries used in
the experiment and subjects were asked three comprehension
questions that had to be answered correctly before they could
proceed to the incentivized tasks. Además, subjects were
asked to answer a (nonincentivized) practice question before
getting to the incentivized tasks to allow them to become
familiar with the interface.

Each subject was randomly assigned to either the small
probability gain treatment or the 50-50 gain treatment by the
software, and all participated in the loss treatment. Every sub-
ject thus participated in two treatments, the loss treatment and
one of the gain treatments, y 121 subjects participated in
el 50-50 gain treatment, 124 subjects in the small probability
gain treatment, and all of these subjects (245 en total) en el
loss treatment. The tasks measuring a particular higher-order
risk attitude for a given treatment were presented together,
and the order of the tasks within such a block of tasks was
randomized, as was the order of the blocks themselves. De este modo,
whether a subject first faced a loss or a gain task was deter-
mined randomly. The location, left or right, of the option that
indicates risk apportionment was randomized for each sub-
ject and each task. Subjects could go back to previous tasks
in the same block of questions and change their choices if
they so wanted.6 Subjects could continue to the next task
only after making a choice in the current one, which they
could do by clicking on the lottery they preferred. To indi-
cate the choice they had made, the selected option was then
highlighted. After participating in the treatments, subjects
were asked some background questions (on gender, edad, ya-
nacionalidad, degree program, and income); they were explicitly
allowed not to answer these questions.7

After answering all questions in the experiment, subjects
were asked one by one to come to the front desk to play one
of their choices for real. Subjects were asked to roll a six-
sided die to select one of the six blocks of tasks, three of
which measured preferences under losses and three of which
measured preferences under gains, and a twelve-sided die
to select according to which of the twelve choices in that
block they would be paid. The subject would then draw a
colored plastic token from an opaque bag with a composition
of tokens corresponding to that described in the selected task.
Composing the bag was done in full view of the first subject
in each session for whom that bag was needed, and subjects
who came thereafter could inspect the bag if they wished.
Some subjects had to draw tokens from more than one bag.
Depending on the final outcome, the subject would then be
paid in addition to their initial €15 payment or have to give
up part of it. The average earnings of subjects were €14.80,
and the total duration of a session, including the payment
procedimiento, was less than one hour.

6In the experiment, 2.5% of all choices were revised by subjects.
7Only gender was asked about in the January 2017 sessions.

IV. Resultados

A. Aggregate Behavior, Nonparametric Methods

Cifra 2 shows the number of times subjects chose in
agreement with risk aversion, prudence, and temperance in
the loss treatment and in the gain treatments. For the two gain
tratos, the patterns are very similar: we find risk aver-
sion and prudence in both gain treatments, which is consistent
with the usual findings in the literature, and weak (if signifi-
cant) intemperance, which is consistent with the findings of
Deck and Schlesinger (2010) and Baillon, Schlesinger, y
Van de Kuilen (2017) but not with the results of Ebert and
Wiesen (2014), Deck and Schlesinger (2014), and Noussair,
Trautmann, and Van de Kuilen (2014), who find modest tem-
perance in the aggregate.

Cifra 2 suggests a reflection effect for higher-order risk
attitudes: responses are markedly different for the loss treat-
ment compared to the gain treatments. For losses, we find
much more risk-loving preferences, leading to risk neutrality
in the aggregate, as well as imprudence and much temper-
ance. This reflection effect is also visible in table 1, cual
shows the proportion of choices consistent with risk appor-
tionment in each of the three treatments.8 To test the reflection
efecto, we perform a Wilcoxon signed rank test. Esto indica
that the differences in the number of risk-averse choices be-
tween the loss treatment and the 50-50 gain treatment and
the small probability gain treatment are significant, as are the
differences in the number of prudent choices between the loss
treatment and the gain treatments and the greater frequency
of temperate choices in the loss treatment compared to the
gain treatments, all with p-values smaller than 0.001.

For all risk attitudes in all treatments, the mode is either
twelve or zero choices consistent with risk apportionment
(risk aversion, prudence, or temperance), meaning fully con-
sistent choices for or against risk apportionment, and in most
casos, the second-most-common outcome is twelve choices
for the reverse higher-order risk attitude. This consistency is
reassuring considering the involved choices subjects need to
make when measuring higher-order risk preferences.

The data appear to become noisier as we go from mea-
suring risk aversion, to measuring prudence, to measuring
temperance, at least for gains. Taking the distance from six
choices for the option satisfying risk apportionment as an or-
dinal measure of consistency,9 we find a negative Spearman
rank correlation between the risk order and consistency for
50-50 gains (ρ −0.186, p-value <0.001) and small probabil- ity gains (ρ −0.134, p-value 0.001). The small probability gains data are more consistent than the 50-50 gains data: the average distance from random choice is 4.71 for 50-50 gains and 5.05 for small probability gains for second-order tasks 8When measuring types of preferences, as in this study, reflection indicates a reversal from one type to the opposite type, that is, from risk apportionment to the reverse preference, or vice versa. 9This is, on average, the number of choices made for the options satisfying risk apportionment by a decision maker choosing randomly. l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . e d u / r e s t / l a r t i c e - p d f / / / / 1 0 4 4 7 0 5 2 0 3 3 3 7 3 / r e s t _ a _ 0 0 9 8 0 p d . f b y g u e s t t o n 0 7 S e p e m b e r 2 0 2 3 THE REFLECTION EFFECT FOR HIGHER-ORDER RISK PREFERENCES 709 FIGURE 2.—HISTOGRAMS OF CHOICES IN THE THREE TREATMENTS l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . e d u / r e s t / l a r t i c e - p d f / / / / 1 0 4 4 7 0 5 2 0 3 3 3 7 3 / r e s t _ a _ 0 0 9 8 0 p d . f b y g u e s t t o n 0 7 S e p e m b e r 2 0 2 3 TABLE 1.—AGGREGATE CHOICES BY TREATMENT AND HIGHER-ORDER RISK PREFERENCE TABLE 2.—RANK CORRELATIONS BETWEEN RISK-AVERSE, PRUDENT, AND TEMPERATE CHOICES Risk-Averse Choices (%) Prudent Choices (%) Temperate Choices (%) 49.3 80.6*** 87.8*** 42.8*** 60.3*** 71.1*** 69.4*** 42.9*** 46.0*** Losses (N = 245) 50-50 gains (N = 121) Small probability gains (N = 124) Asterisks indicate a significant difference from 50% (binomial test) at the ∗∗5% and ∗∗∗1% level. (Mann-Whitney U test, p-value 0.059), respectively, 4.05 and 4.74 for third-order tasks (p-value 0.008), and 3.68 and 4.36 for fourth-order tasks (p-value 0.015). For losses, we find no relation between consistency and the risk apportionment order of the tasks (ρ 0.050, p-value 0.177). The percentage of choices of the risk apportionment option are indicated in table 6 for each task. For most treatment-risk attitude pairs, the majority of choices is for or against risk apportionment across all tasks. A preference for combining good with bad or good with good leads to three possible combinations of risk attitudes (see section II): those who are risk averse, prudent, and tem- perate; those who are risk loving, prudent, and intemperate; and those who are risk neutral, prudence neutral, and temper- ance neutral. At the aggregate level, we do not find evidence in support of a preference for combining good with bad or good with good underlying higher-order risk preferences. In the gain treatments, which induce the most risk aversion, we find the least temperance, and for losses we find imprudence in the aggregate. Losses Prudent choices Temperate choices 50-50 gains Prudent choices Temperate choices Small probability gains Prudent choices Temperate choices p-values are in parentheses. Risk-Averse Choices Prudent Choices 0.184 (0.004) −0.056 (0.386) −0.031 (0.737) −0.061 (0.505) 0.399 (<0.001) 0.181 (0.045) −0.018 (0.775) 0.164 (0.072) 0.136 (0.133) B. Individual Behavior, Nonparametric Methods Within treatments, we can test for the correlation between the various higher-order risk preferences. Spearman corre- lation coefficients and p-values are reported in table 2 for the three treatments. A preference for combining good with bad or good with good predicts that the number of temperate choices is positively correlated with the number of risk-averse choices, but the number of prudent choices is not. There is a significant correlation between temperance and risk aversion only for the small probability gain treatment, and there it is dwarfed by the correlation between prudence and risk aver- sion. We also find a significant correlation between prudence and risk aversion for losses. 710 THE REVIEW OF ECONOMICS AND STATISTICS A caveat to the tests of the correlations in table 2 is that large numbers of indifferent subjects would push any positive correlation toward 0. Many subjects chose the risk-averse, prudent, or temperate option no more than two times (out of twelve) or at least ten times, and the probability of do- ing so when choosing randomly is slightly less than 4%. We therefore classify subjects accordingly. The remaining sub- jects, who chose the risk-averse, prudent, or temperate option between three and nine times, we classify as risk neutral, pru- dence neutral, and temperance neutral.10 The frequencies of types are reported in table 7 in appendix D. Using this classification, we also find little evidence that fourth-order risk attitudes depend on second-order risk atti- tudes. For losses and 50-50 gains, the patterns of temperance do not appear to depend on second-order risk preferences, and a Fisher’s exact test (p-value 0.610, respectively 0.645) cannot reject that the patterns are the same. The distribu- tion of fourth-order risk preferences does depend on second- order risk preferences for the small probability gain treatment (p-value 0.001), but this is mostly driven by the relatively large number of temperance-neutral subjects among the group of risk-neutral individuals.11 C. Aggregate Behavior, Maximum Likelihood Estimation The previous results are based on an informal argument that choices at the extremes are unlikely to be the result of a subject choosing randomly between the options. To model this explicitly, we perform maximum likelihood estimation (MLE). We estimate a mixed binomial distribution for each order of risk apportionment, with πs the proportion of deci- sion makers who satisfy risk apportionment for that particular order, πo the proportion of decision makers with the opposite risk attitude, and πn the proportion of neutral or indifferent decision makers. We allow for types with strict preferences to make errors. In any task, subjects who satisfy risk apportionment have a probability η < 1/2 of choosing the option that does not in- dicate risk apportionment, which is allowed to differ from the error rate δ < 1/2 for those who have the opposite pref- erence. We assume that the probability that an indifferent decision maker chooses the option that indicates risk appor- tionment is equal to 1/2.12 The probability of choosing the option indicating risk apportionment x out of twelve times can be represented by the following density for a given order, 10This classification is consistent with the maximum likelihood estima- tions of the proportion of each type, which we present below. We also considered the choice patterns of perfectly consistent subjects only. These looked qualitatively similar. 11When excluding neutral types, the difference is not statistically signifi- cant (p-value 0.117). Maximum likelihood estimation also does not indicate significance. See section IVD. 12The position (left or right) of the lottery indicating risk apportionment was randomized for each task, so even an indifferent subject who always chooses the left or right option would choose the option indicating risk apportionment with a probability of 0.5. TABLE 3.—ESTIMATED PARAMETER VALUES FOR DIFFERENT HIGHER-ORDER RISK ATTITUDES Order Treatment 2 3 4 Losses 50-50 gains Small probability gains Losses 50-50 gains Small probability gains Losses 50-50 gains Small probability gains ˆπs 0.32 0.79 0.85 0.27 0.46 0.63 0.57 0.18 0.31 ˆπn 0.35 0.13 0.13 0.31 0.34 0.19 0.28 0.50 0.29 ˆπo 0.34 0.08 0.02 0.43 0.20 0.18 0.14 0.32 0.39 ˆη 0.04 0.07 0.05 0.04 0.08 0.05 0.06 0.05 0.04 ˆδ 0.06 0.04 0.00 0.05 0.01 0.03 0.02 0.03 0.05 πs indicates the share of subjects satisfying risk apportionment (risk aversion, prudence, and temperance for orders 2, 3, and 4), mistakenly choosing the reverse option with probability η. πo indicates those with the opposite attitude and error rate δ, and πn the neutral subjects. Underlines indicate the mode. TABLE 4.—LIKELIHOOD RATIOS AND p-VALUES FOR THE TEST THAT THE SHARE OF SUBJECTS WHO SATISFY RISK APPORTIONMENT IS EQUAL TO THE SHARE WHO DO NOT Order Treatment 2 3 4 Losses 50-50 gains Small probability gains Losses 50-50 gains Small probability gains Losses 50-50 gains Small probability gains LR 0.17 79.24 90.37 8.87 10.80 30.29 64.17 4.42 0.99 p-Value 0.676 <0.001 <0.001 0.003 0.001 <0.001 <0.001 0.036 0.319 l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . e d u / r e s t / l a r t i c e - p d f / with πs + πn + πo = 1: (cid:2) (cid:3)(cid:4) f (x) = 12 x πs(1 − η)xη12−x + πn (cid:5) + πoδx(1 − δ)12−x . 12−x 1 2 x 1 2 / / / 1 0 4 4 7 0 5 2 0 3 3 3 7 3 / r e s t _ a _ 0 0 9 8 0 p d . f b y g u e s t t o n 0 7 S e p e m b e r 2 0 2 3 The error rates are estimated separately for each higher- order risk preference because the lotteries become increas- ingly complex as the order of the risk preference increases, and it is important to be able to distinguish risk preferences from a tendency toward random choice. Noisy behavior may also be captured by the proportion of decision makers with neutral risk attitudes: a decision maker who is confused or inattentive may simply choose (almost) randomly. The pa- rameter values, estimated using numerical methods, are pre- sented in table 3. The estimated proportions support the results from the nonparametric analysis. For losses, there are slightly more risk lovers than risk averters; there is also much imprudence and much temperance. For 50-50 gains and small probabil- ity gains, there is much risk aversion, much prudence, and some intemperance. Likelihood ratio tests, reported in table 4, show that seven of these differences are statistically sig- nificant. Only the risk-loving attitude for losses and intem- perance for small probability gains are not significantly more common than the opposite attitude; in these cases, we cannot reject the null hypothesis that πs = πo against the alternative hypothesis that πs (cid:5)= πo. THE REFLECTION EFFECT FOR HIGHER-ORDER RISK PREFERENCES 711 TABLE 5.—LIKELIHOOD RATIOS AND p-VALUES TESTING THE REFLECTION EFFECT USING MLE FIGURE 3.—MLE, COMBINATIONS OF SECOND- AND FOURTH-ORDER RISK ATTITUDES AND LOSS TREATMENT Order Treatment 2 3 4 50-50 gains Small probability gains 50-50 gains Small probability gains 50-50 gains Small probability gains LR 54.38 76.04 19.19 38.41 35.02 31.19 p-Value <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 Null hypothesis is that the proportion of those who satisfy risk apportionment relative to those who have the opposite attitude (πs/πo) is the same for losses and the specified gain treatment. Estimated error rates: ˆυ = 0.04 (risk-averse subjects), ˆθ = 0.06 (risk loving), ˆω = 0.06 (temperate), and ˆζ = 0.02 (intemperate subjects). FIGURE 4.—MLE, COMBINATIONS OF SECOND- AND FOURTH-ORDER RISK ATTITUDES, AND GAIN TREATMENTS l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . e d u / r e s t / l a r t i c e - p d f / / / / 1 0 4 4 7 0 5 2 0 3 3 3 7 3 / r e s t _ a _ 0 0 9 8 0 p d . f b y g u e s t t o n 0 7 S e p e m b e r 2 0 2 3 /πl Finally, to test the reflection effect, we test whether the proportion of those who satisfy risk apportionment relative to those who have the opposite preference is the same for losses and gains. Table 5 reports likelihood ratio estimates /πg and p-values for the null hypothesis that πl o, s where πl o is the proportion of subjects who satisfy risk s apportionment for a given order relative to the proportion of subjects with the opposite attitude for losses, and πg /πg o s has the same meaning but for either gain treatment. The re- sults show that the greater proportion of risk-loving, impru- dent, and temperate (relative to risk-averse, prudent, and in- temperate) subjects for losses than for gains are all highly significant. = πg s /πl o D. Individual Behavior, Maximum Likelihood Estimation To investigate a preference for combining good with bad or good with good further, we perform a maximum likelihood estimation where we test the frequencies of combinations of second- and fourth-order risk attitudes. Subjects are risk averse and temperate if they have a preference for combin- ing good with bad, risk loving and intemperate if they have a preference for combining good with good, and risk neutral and temperance neutral otherwise. There are nine possible combinations of second- and fourth-order risk attitudes. We denote these different types as πa,b, a, b ∈ {s, n, o}, where a indicates whether the second-order risk attitude satisfies risk apportionment of order 2 (risk aversion), matches the oppo- site preference (risk-loving preferences), or is neutral toward risk apportionment of order 2 (risk neutral) and b indicates whether the fourth-order risk attitude satisfies fourth-order risk apportionment (temperance), matches the opposite atti- tude (intemperance), or is neutral (temperance neutral). We again allow for errors, which differ across types and or- ders of risk attitudes. A risk-averse subject is assumed to have a probability υ of mistakenly choosing the riskier option, a risk-loving subject has a probability θ of mistakenly choosing the safer option, and a risk-neutral subject chooses either op- tion with probability 1/2. Temperate subjects are assumed to mistakenly choose the intemperate option with probability ω, intemperate subjects mistakenly choose the temperate option with probability ζ , and temperance-neutral subjects choose either option with probability 1/2. The estimated proportions are presented in figure 3 for losses and in figure 4 the gain treatments. 712 THE REVIEW OF ECONOMICS AND STATISTICS If subjects have a preference for combining good with bad or good with good, then the symmetric types (πs,s, πn,n, πo,o) should be most common. This means that most of the mass should be on the diagonals from the top left to the bottom right in figures 3 and 4. The data do not show such a pattern. For losses, the most common type in fact combines temper- ance with a risk-loving attitude. For 50-50 gains, the two most frequent types combine risk aversion with temperance neu- trality and risk aversion with intemperance. For small prob- ability gains, the most common type combines risk aversion with temperance, in agreement with a preference for combin- ing good with bad, but the second most common type, with a share of 31% of the subjects, combines risk aversion with intemperance, and the difference in the proportions of these two types is only 1 percentage point. To formally test the hypothesis of a preference for com- bining good with bad or good with good for gains and losses, we test whether πs,s + πo,o = πs,o + πo,s against the alter- native that πs,s + πo,o (cid:5)= πs,o + πo,s using a (log) likelihood ratio test. We do not include the proportion of neutral types in the test as subjects may be classified as indifferent be- cause they are confused, distracted, or inattentive, and the extent of this may depend on the order of the risk prefer- ence measured. Neither for losses (LR 1.11, p-value 0.292), nor for 50-50 gains (LR 2.01, p-value 0.157), nor for small probability gains (LR 0.18, p-value 0.674) can we reject that πs,s + πo,o = πs,o + πo,s. Thus, we do not find evidence that risk aversion is combined more often with temperance than intemperance or that risk lovers are more likely to be intem- perate than temperate for gains or losses. E. Background Characteristics and Robustness In this section, we briefly report on tests to investigate the influence of various factors on responses.13 First, we inves- tigate the effects of background characteristics. As all sub- jects participated in a gain and the loss treatment, background characteristics cannot drive the reflection effect we find, but nonetheless it may be interesting to see whether they play a role. Gender in particular has been found to influence risk preferences in various studies (see Croson & Gneezy, 2009, for a review). Other characteristics such as age, nationality, income, or degree program may also play a role, although ex- perimental samples tend to be quite homogeneous, making it difficult to detect such effects. We do not find robust evidence of background characteristics influencing responses. See the online appendix for details. We also perform two robustness checks testing for ses- sion effects and ordering effects. The results of a Kruskal- Wallis test do not indicate that there are important session effects. ANOVA with variables indicating whether subjects were presented with gain or loss tasks first, whether the first block measuring a given risk attitude did so for losses or for gains, and the position (as the first, second, and so on) of 13We thank an anonymous referee for this suggestion. each block of questions does not indicate significant ordering effects. The details are provided in the online appendix. V. Discussion The aggregate pattern of risk aversion, prudence, and slight intemperance we find for pure gains in our experiment mir- rors earlier findings of Deck and Schlesinger (2010) and Baillon et al. (2017). Deck and Schlesinger (2014), Ebert and Wiesen (2014), and Noussair et al. (2014) also find risk aversion and prudence in the aggregate, but they also find moderate temperance rather than intemperance. Deck and Schlesinger (2014) point out that the modest intemperance of Deck and Schlesinger (2010) can be explained if there were unusually many risk lovers in their sample but could not verify this explanation as Deck and Schlesinger (2010) collected no information on second-order risk attitudes. The aggregate intemperance in our sample does not appear to be caused by intemperate risk lovers. The number of risk lovers is small in the gain treatments, and we do not find evidence that fourth-order risk preferences are a function of second- order risk preferences. Baillon et al. (2017) also have few risk lovers in their sample while finding intemperance. The respective temperance or intemperance is quite weak at the aggregate level in all studies, suggesting that the observed discrepancies may simply be the consequence of differences in the makeup of samples combined with modest effects from differences in presentation. Although a preference for combining good with bad or good with good seems to explain higher-order risk prefer- ences for mixed lotteries, our results indicate this is not the case for pure gains or losses. We do not find evidence at the individual level that risk aversion is combined with tem- perance, and the imprudence we find for losses, as well as the combination of risk aversion and intemperance for gains, is evidence against the hypothesis. The differences between earlier findings and ours may be explained by loss aversion. This is a question that deserves attention in future studies. We find clear reference dependence of higher-order risk preferences. As per the usual findings, the loss frame in- duces much more risk-loving preferences. We find that pru- dence and temperance are also affected: preferences shift from prudence and intemperance under gains to imprudence and temperance under losses. Thus, we have a full rever- sal of higher-order risk attitudes: risk aversion, prudence, and intemperance under gains and much more risk-loving preferences, imprudence, and temperance under losses. This is consistent with reflection (Kahneman & Tversky, 1979), where risk preferences are reversed under losses. The differ- ent findings suggest it should be worthwhile to investigate ref- erence dependence further in the context of higher-order risk preferences. Our results contrast with those by Deck and Schlesinger (2010) and Maier and Rüger (2012), who find no influence of a loss frame on higher-order risk preferences. Neither of those studies directly controls the reference point, which is l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . e d u / r e s t / l a r t i c e - p d f / / / / 1 0 4 4 7 0 5 2 0 3 3 3 7 3 / r e s t _ a _ 0 0 9 8 0 p d . f b y g u e s t t o n 0 7 S e p e m b e r 2 0 2 3 THE REFLECTION EFFECT FOR HIGHER-ORDER RISK PREFERENCES 713 needed to separate gains and losses. Deck and Schlesinger (2010) rewrite lotteries so fixed payments are presented as fixed deductions. Maier and Rüger (2012) have subjects re- turn to the lab and lose part of their earlier winnings, but they cannot control what happens between sessions. Whether a reference point is induced successfully is ultimately an em- pirical question. Reproducing reference dependence of risk aversion demonstrates that outcomes intended to be gains and losses are perceived as such, and neither study does so.14 Prospect theory (Tversky & Kahneman, 1992) has been suggested as an explanation of findings of higher-order risk preferences (Deck & Schlesinger, 2010). Reference depen- dence is an important component of prospect theory, which can explain the differences in higher-order risk preferences between the gain and loss treatments. The other important component of prospect theory is inverse-S-shaped probability weighting, which leads to risk-loving behavior for small prob- ability gains. We do not find the predicted behavior: choices in the small probability gain treatment closely resemble those in the 50-50 gain treatment, meaning, in particular, that we observe clear risk aversion. A possible explanation for the ob- served risk aversion in the small probability gain treatment is that the probability weighting function is convex rather than inverse-S shaped. Convex probability weighting for gains has been found in some experiments, including Van de Kuilen and Wakker (2011). Table 8 in appendix E shows the predicted choices of prospect theory based on the functional form used in Tver- sky and Kahneman (1992) and the parameter values they es- timate. These imply inverse-S probability weighting. For our tasks, the predictions are the same for all tasks in a given treatment measuring a specific risk attitude. Besides predict- ing that people are risk loving for our small-probability treat- ment, prospect theory predicts prudence15 and intemperance for losses. It predicts prudence for both losses and gains be- cause the prudent option has both a better best and worst out- come than the imprudent option, the probabilities of which are overweighted with inverse-S probability weighting. Con- vex probability weighting for losses can accommodate the imprudence we find, as this leads to underweighting the worst outcome. As empirical findings point to a fourfold pattern of risk attitudes, it would have been interesting to include a small probability loss treatment. We did not do this for practical considerations. Another interesting extension would be to 14Using hypothetical choices, Attema, l’Haridon, and Van de Kuilen (2019) find risk aversion and prudence for gains and risk neutrality and prudence neutrality for losses. Brunette and Jacob (2019) claim to find im- prudence for losses, but their results do not support this. Their loss tasks are the negative of their gain tasks, but this reverses which option is prudent. Hence, their modal preference is prudence for both gains and losses. They also pay out two tasks, one of which may involve gains, so prospects are mixed rather than pure losses, and they do not reproduce reference depen- dence of risk aversion. 15The utility function that Tversky and Kahneman (1992) use is prudent (under expected utility), but this does not drive the results: with linear utility, the same predictions follow. test for the effects of different endowments. We leave this for future work. Our results indicate that it is important for policy recom- mendations to consider whether one is measuring higher- order risk preferences for gains, losses, or both. For exam- ple, when investigating how demand for insurance with non- performance risk (a type of probabilistic insurance) relates to prudence, it will be important to measure prudence for losses, the domain in which insurance decisions are taken. The imprudence we find for losses indicates that demand for probabilistic insurance may be greater than expected based on the prudence found for gains, and policies aimed at reduc- ing nonperformance risk (such as increasing trust in insurer solvency) may not increase demand by as much as expected based on an assumption of prudent consumers. In a similar vein, prudence reduces the attractiveness of prevention efforts, which reduce but do not take away en- tirely the probability of a loss. Imprudence means that deci- sion makers undertake more prevention efforts than predicted based on prudence found for gains. This suggests opportuni- ties for nudging; for example, by framing outcomes as losses, people may be more willing to engage in action to reduce the probability of catastrophic climate change. The influence of background risks on risk-taking behav- ior may also interact with the domain in unexpected ways: a financial crisis may at the same time lead to increased volatil- ity of stocks and shift the domain of investment decisions of people with significant wealth tied to stocks to losses. Our results indicate this would lead to greater temperance, which means such people would become more risk averse and may, for example, respond by purchasing insurance against unre- lated losses (such as health insurance) or by making more defensive investment decisions even for assets unaffected by the increased volatility of stocks. VI. Conclusion It is well established that second-order risk attitudes for gains are the mirror image of those for losses. In an exper- iment, we observe the reflection effect also for higher-order risk preferences. We find prudence for gains but imprudence for losses and intemperance for gains but temperance for losses. This reflection affects the external validity of higher- order risk preferences measured under gains only and has behavioral implications when choices involve losses. The recent literature has found evidence for the hypothesis that higher-order risk preferences are generated by a prefer- ence for combining good with bad or good with good. Such a preference implies risk aversion should be combined with temperance and that all decision makers should be prudent. We find that correlations between the number of risk-averse choices and the number of temperate choices are small and insignificant. Furthermore, the imprudence we find as the ma- jority preference for losses, and the simultaneous preference for risk aversion and intemperance on the aggregate level in the gain domain, are at odds with this hypothesis for pure gains and pure losses. l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . e d u / r e s t / l a r t i c e - p d f / / / / 1 0 4 4 7 0 5 2 0 3 3 3 7 3 / r e s t _ a _ 0 0 9 8 0 p d . f b y g u e s t t o n 0 7 S e p e m b e r 2 0 2 3 714 THE REVIEW OF ECONOMICS AND STATISTICS Appendix A Proof That Probabilistic Mixture Preserves then Stochastic Dominance We follow the definition of stochastic dominance of Eeck- houdt, Schlesinger, and Tsetlin (2009). It is assumed that all random variables have bounded supports contained in [l, u]. With F (0)(x) the cumulative distribution function of distri- (cid:6) l F (i−1)(t )dt for i ≥ 1. Dis- bution F , they define F (i)(x) ≡ tribution F weakly dominates distribution G through Nth- order stochastic dominance if (a) F (N−1)(x) ≤ G(N−1)(x) for all l ≤ x ≤ u and (b) F (i)(u) ≤ G(i)(u) for i = 1, . . . , N. x We will show that if distribution F dominates distribution G through Nth order stochastic dominance, then the proba- bilistic mixture F (cid:9) of distribution F and distribution H dom- inates probabilistic mixture G(cid:9) of distribution G and distribu- tion H through the Nth-order stochastic dominance. The cumulative distribution function F (cid:9)(0)(x) can be written as λF (0)(x) + (1 − λ)H (0)(x), λ ∈ [0, 1]. Similarly, G(cid:9)(0)(x) ≡ λG(0)(x) + (1 − λ)H (0)(x). Then (cid:7) F (cid:9)(1)(x) = λ l x F (0)(t )dt + (1 − λ) (cid:7) x l H (0)(t )dt = λF (1)(x) + (1 − λ)H (1)(x) and G(cid:9)(1)(x) = λ (cid:7) x l G(0)(t )dt + (1 − λ) (cid:7) x l H (0)(t )dt = λG(1)(x) + (1 − λ)H (1)(x) for all l ≤ x ≤ u. Furthermore, if F (cid:9)(i−1)(x) = λF (i−1)(x) + (1 − λ)H (i−1)(x), F (cid:9)(i)(x) = λ (cid:7) x l F (i−1)(t )dt + (1 − λ) (cid:7) x l H (i−1)(t )dt = λF (i)(x) + (1 − λ)H (i)(x), and if G(cid:9)(i−1)(x) = λG(i−1)(x) + (1 − λ)H (i−1)(x), then G(cid:9)(i)(x) = λ (cid:7) x l G(i−1)(t )dt + (1 − λ) (cid:7) x l H (i−1)(t )dt = λG(i)(x) + (1 − λ)H (i)(x). Thus, by induction, F (cid:9)(i)(x) = λF (i)(x) + (1 − λ)H (i)(x) and G(cid:9)(i)(x) = λG(i)(x) + (1 − λ)H (i)(x) for all i ≥ 0 and all l ≤ x ≤ u. If distribution F dominates distribution G through Nth- order stochastic dominance, that is, F (N−1)(x) ≤ G(N−1)(x) for all l ≤ x ≤ u and F (i)(u) ≤ G(i)(u) for i = 1, . . . , N, then F (cid:9)(N−1)(x) ≤ G(cid:9)(N−1)(x) for all l ≤ x ≤ u and F (cid:9)(i)(u) ≤ G(cid:9)(i)(u) for i = 1, . . . , N, and distribution F (cid:9) stochasti- cally dominates distribution G(cid:9) through Nth-order stochastic dominance. l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . e d u / r e s t / l a r t i c e - p d f / / / / 1 0 4 4 7 0 5 2 0 3 3 3 7 3 / r e s t _ a _ 0 0 9 8 0 p d . FIGURE 5.—A SCREENSHOT OF ONE OF THE PRUDENCE TASKS SUBJECTS WERE FACED WITH IN THE 50-50 GAIN TREATMENT Appendix B Task Screenshot f b y g u e s t t o n 0 7 S e p e m b e r 2 0 2 3 All outcomes are in addition to the €15 subjects had been given on entering the lab. THE REFLECTION EFFECT FOR HIGHER-ORDER RISK PREFERENCES 715 Order 2 Losses Treatment Task 2 50-50 gains 2 Small probability gains 3 Losses 3 50-50 gains 3 Small probability gains Appendix C Tasks TABLE 6.—TASKS USED IN THE EXPERIMENT Risk Apportionment Option [−8, −9] [−7, −8] [−8, −10] [−7, −9] [−5, −6] [−6, −8] [−7, −9] [−6, −7] [−6, −8] [−7, −9] [−6, −10] [−6, −8] [8,9] [7,9] [9,11] [7,9] [4,5] [5,7] [6,7] [7,9] [6,8] [8,9] [8,9] [6,7] (0.16: [70, 46], 0.84: 2) (0.18: [38, 28], 0.82: 2) (0.14: [60, 48], 0.86: 2) (0.12: [35, 32], 0.88: 1) (0.10: [110, 90], 0.90: 1) (0.10: [40, 35], 0.90: 2) (0.08: [30, 24], 0.92: 1) (0.08: [66, 62], 0.92: 2) (0.06: [58, 44], 0.94: 2) (0.04: [180, 144], 0.96: 1) (0.02: [70, 54], 0.98: 1) (0.20: [50, 40], 0.80: 1) [−9, −6 + [4, −4]] [−10, −4 + [3, −3]] [−8, −5 + [4, −4]] [−6, −4 + [2, −2]] [−10, −5 + [4, −4]] [−9, −4 + [3, −3]] [−9, −7 + [5, −5]] [−8, −4 + [3, −3]] [−9, −5 + [4, −4]] [−8, −4 + [2, −2]] [−7, −2 + [1, −1]] [−10, −3 + [2, −2]] [4, 11 + [3, −3]] [3, 9 + [2, −2]] [5, 8 + [4, −4]] [5, 10 + [3, −3]] [3, 8 + [1, −1]] [5, 9 + [4, −4]] [6, 12 + [5, −5]] [6, 10 + [5, −5]] [5, 10 + [4, −4]] [4, 6 + [3, −3]] [2, 6 + [1, −1]] [3, 6 + [2, −2]] (0.14: [20, 50 + [15, −15]], 0.86: 1) (0.16: [22, 80 + [16, −16]], 0.84: 1) (0.18: [30, 70 + [25, −25]], 0.82: 1) (0.10: [30, 60 + [24, −24]], 0.90: 2) (0.10: [54, 96 + [50, −50]], 0.90: 2) (0.08: [60, 100 + [46, −46]], 0.92: 1) (0.06: [42, 88 + [38, −38]], 0.94: 2) (0.04: [34, 64 + [20, −20]], 0.96: 2) (0.04: [60, 125 + [45, −45]], 0.96: 2) Reverse Option [−4, −13] [−1, −14] [−4, −14] [−3, −13] [−1, −10] [−3, −11] [−4, −12] [−1, −12] [−1, −13] [−2, −14] [−2, −14] [−2, −12] [1,16] [2,14] [4,16] [3,13] [1,8] [2,10] [3,10] [4,12] [3,11] [5,12] [3,14] [1,12] (0.16: [100, 16], 0.84: 2) (0.18: [60, 6], 0.82: 2) (0.14: [86, 22], 0.86: 2) (0.12: [52, 15], 0.88: 1) (0.10: [180, 20], 0.90: 1) (0.10: [70, 5], 0.90: 2) (0.08: [48, 6], 0.92: 1) (0.08: [120, 8], 0.92: 2) (0.06: [90, 12], 0.94: 2) (0.04: [280, 44], 0.96: 1) (0.02: [110, 14], 0.98: 1) (0.20: [80, 10], 0.80: 1) [−6, −9 + [4, −4]] [−4, −10 + [3, −3]] [−5, −8 + [4, −4]] [−4, −6 + [2, −2]] [−5, −10 + [4, −4]] [−4, −9 + [3, −3]] [−7, −9 + [5, −5]] [−4, −8 + [3, −3]] [−5, −9 + [4, −4]] [−4, −8 + [2, −2]] [−2, −7 + [1, −1]] [−3, −10 + [2, −2]] [11, 4 + [3, −3]] [9, 3 + [2, −2]] [8, 5 + [4, −4]] [10, 5 + [3, −3]] [8, 3 + [1, −1]] [9, 5 + [4, −4]] [12, 6 + [5, −5]] [10, 6 + [5, −5]] [10, 5 + [4, −4]] [6, 4 + [3, −3]] [6, 2 + [1, −1]] [6, 3 + [2, −2]] (0.14: [50, 20 + [15, −15]], 0.86: 1) (0.16: [80, 22 + [16, −16]], 0.84: 1) (0.18: [70, 30 + [25, −25]], 0.82: 1) (0.10: [60, 30 + [24, −24]], 0.90: 2) (0.10: [96, 54 + [50, −50]], 0.90: 2) (0.08: [100, 60 + [46, −46]], 0.92: 1) (0.06: [88, 42 + [38, −38]], 0.94: 2) (0.04: [64, 34 + [20, −20]], 0.96: 2) (0.04: [125, 60 + [45, −45]], 0.96: 2) Choices (%) 44.5 49.8 44.9 44.5 54.3 46.9 50.6 53.9 49.8 50.2 49.8 52.2 83.5 85.1 81.0 82.6 76.0 81.0 77.7 81.0 78.5 70.2 84.3 86.0 92.7 85.5 84.7 84.7 93.5 91.1 91.1 89.5 86.3 87.1 75.8 91.1 44.1 38.4 45.3 44.9 45.7 42.0 48.2 40.4 44.9 42.0 35.5 41.6 63.6 56.2 63.6 57.9 52.9 67.8 63.6 62.8 62.8 66.1 47.1 59.5 69.4 67.7 67.7 78.2 78.2 67.7 75.8 68.5 70.2 l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . e d u / r e s t / l a r t i c e - p d f / / / / 1 0 4 4 7 0 5 2 0 3 3 3 7 3 / r e s t _ a _ 0 0 9 8 0 p d . f b y g u e s t t o n 0 7 S e p e m b e r 2 0 2 3 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 716 THE REVIEW OF ECONOMICS AND STATISTICS Order Treatment Task 4 Losses 4 50-50 gains 4 Small probability gains 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 TABLE 6.—CONTINUED. Risk Apportionment Option (0.02: [55, 150 + [30, −30]], 0.98: 2) (0.20: [40, 60 + [28, −28]], 0.80: 1) (0.12: [44, 75 + [32, −32]], 0.88: 1) −8 + [[2, −2], [4, −4]] −7 + [[1, −1], [4, −4]] −6 + [[2, −2], [3, −3]] −9 + [[2, −2], [2, −2]] −7 + [[1, −1], [5, −5]] −7 + [[3, −3], [3, −3]] −9 + [[2, −2], [3, −3]] −8 + [[1, −1], [2, −2]] −7 + [[2, −2], [3, −3]] −8 + [[2, −2], [3, −3]] −6 + [[1, −1], [1, −1]] −8 + [[1, −1], [3, −3]] 7 + [[2, −2], [4, −4]] 7 + [[3, −3], [3, −3]] 5 + [[1, −1], [2, −2]] 5 + [[1, −1], [3, −3]] 8 + [[2, −2], [3, −3]] 9 + [[2, −2], [6, −6]] 8 + [[3, −3], [4, −4]] 8 + [[2, −2], [5, −5]] 10 + [[3, −3], [6, −6]] 10 + [[4, −4], [5, −5]] 8 + [[1, −1], [6, −6]] 5 + [[2, −2], [2, −2]] (0.08: 70 + [[20, −20], [45, −45]], 0.92: 1) (0.10: 70 + [[28, −28], [30, −30]], 0.90: 2) (0.10: 65 + [[24, −24], [36, −36]], 0.90: 2) (0.12: 32 + [[14, −14], [14, −14]], 0.88: 2) (0.12: 60 + [[22, −22], [26, −26]], 0.88: 2) (0.14: 75 + [[25, −25], [30, −30]], 0.86: 1) (0.16: 35 + [[10, −10], [20, −20]], 0.84: 1) (0.06: 100 + [[30, −30], [50, −50]], 0.94: 1) (0.04: 85 + [[34, −34], [40, −40]], 0.96: 2) (0.02: 68 + [[18, −18], [42, −42]], 0.98: 1) (0.20: 55 + [[5, −5], [38, −38]], 0.80: 1) (0.08: 58 + [[14, −14], [34, −34]], 0.92: 2) Reverse Option (0.02: [150, 55 + [30, −30]], 0.98: 2) (0.20: [60, 40 + [28, −28]], 0.80: 1) (0.12: [75, 44 + [32, −32]], 0.88: 1) −8 + [0, [2, −2] + [4, −4]] −7 + [0, [1, −1] + [4, −4]] −6 + [0, [2, −2] + [3, −3]] −9 + [0, [2, −2] + [2, −2]] −7 + [0, [1, −1] + [5, −5]] −7 + [0, [3, −3] + [3, −3]] −9 + [0, [2, −2] + [3, −3]] −8 + [0, [1, −1] + [2, −2]] −7 + [0, [2, −2] + [3, −3]] −8 + [0, [2, −2] + [3, −3]] −6 + [0, [1, −1] + [1, −1]] −8 + [0, [1, −1] + [3, −3]] 7 + [0, [2, −2] + [4, −4]] 7 + [0, [3, −3] + [3, −3]] 5 + [0, [1, −1] + [2, −2]] 5 + [0, [1, −1] + [3, −3]] 8 + [0, [2, −2] + [3, −3]] 9 + [0, [2, −2] + [6, −6]] 8 + [0, [3, −3] + [4, −4]] 8 + [0, [2, −2] + [5, −5]] 10 + [0, [3, −3] + [6, −6]] 10 + [0, [4, −4] + [5, −5]] 8 + [0, [1, −1] + [6, −6]] 5 + [0, [2, −2] + [2, −2]] (0.08: 70 + [0, [20, −20] + [45, −45]], 0.92: 1) (0.10: 70 + [0, [28, −28] + [30, −30]], 0.90: 2) (0.10: 65 + [0, [24, −24] + [36, −36]], 0.90: 2) (0.12: 32 + [0, [14, −14] + [14, −14]], 0.88: 2) (0.12: 60 + [0, [22, −22] + [26, −26]], 0.88: 2) (0.14: 75 + [0, [25, −25] + [30, −30]], 0.86: 1) (0.16: 35 + [0, [10, −10] + [20, −20]], 0.84: 1) (0.06: 100 + [0, [30, −30] + [50, −50]], 0.94: 1) (0.04: 85 + [0, [34, −34] + [40, −40]], 0.96) (0.02: 68 + [0, [18, −18] + [42, −42]], 0.98: 1) (0.20: 55 + [0, [5, −5] + [38, −38]], 0.80: 1) (0.08: 58 + [0, [14, −14] + [34, −34]], 0.92: 2) Choices (%) 64.5 71.8 73.4 72.2 68.2 70.6 67.3 61.6 71.4 71.4 70.2 73.5 73.9 63.3 69.4 45.5 43.0 41.3 47.9 43.0 40.5 42.1 41.3 43.0 42.1 39.7 45.5 46.0 45.2 43.5 50.8 44.4 44.4 48.4 46.8 46.0 46.0 44.4 46.8 Brackets indicate equal probabilities; all outcomes are in euros. Choices reported for the option satisfying risk apportionment. Appendix D Frequency Tables Appendix E Prospect Theory Predictions TABLE 7.—FREQUENCIES OF COMBINATIONS OF SECOND- AND FOURTH-ORDER RISK ATTITUDES TABLE 8.—PROSPECT THEORY PREDICTIONS BASED ON PARAMETER ESTIMATES OF TVERSKY AND KAHNEMAN (1992) Attitude Losses Risk averse Risk neutral Risk loving Total 50-50 gains Risk averse Risk neutral Risk loving Total Small probability gains Risk averse Risk neutral Risk loving Total Temperate Temperance Neutral Intemperate Total 40 43 49 132 14 6 2 22 38 1 0 39 25 30 21 76 42 13 3 58 27 11 0 38 13 14 10 37 29 7 5 41 37 6 4 47 78 87 80 245 85 26 10 121 102 18 4 124 Order Treatment Predicted Behavior 2 3 4 Losses 50-50 gains Small probability gains Losses 50-50 gains Small probability gains Losses 50-50 gains Small probability gains Risk loving Risk averse Risk loving Prudent Prudent Prudent Intemperate Temperate Intemperate REFERENCES Attema, A. E., O. l’Haridon, and G. Van de Kuilen, “Measuring Multi- variate Risk Preferences in the Health Domain,” Journal of Health Economics 64 (2019), 15–24. 10.1016/j.jhealeco.2018.12.004, PubMed: 30658248 Baillon, A., “Prudence with Respect to Ambiguity,” Economic Journal 12:604 (2017), 1731–1755. 10.1111/ecoj.12358 l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . e d u / r e s t / l a r t i c e - p d f / / / / 1 0 4 4 7 0 5 2 0 3 3 3 7 3 / r e s t _ a _ 0 0 9 8 0 p d . f b y g u e s t t o n 0 7 S e p e m b e r 2 0 2 3 THE REFLECTION EFFECT FOR HIGHER-ORDER RISK PREFERENCES 717 Baillon, A., H. Bleichrodt, and V. Spinu, “Searching for the Refer- ence Point,” Management Science 66:1 (2020), 93–112. 10.1287/ mnsc.2018.3224 Baillon, A., H. Schlesinger, and G. Van de Kuilen, “Measuring Higher Or- der Ambiguity Preferences,” Experimental Economics 21:2 (2017), 233–256. 10.1007/s10683-017-9542-3, PubMed: 29720888 Browning, M., and A Lusardi, “Household Saving: Micro Theories and Micro Facts,” Journal of Economic Literature 34:4 (1996), 1797– 1855. Brunette, M., and J. Jacob, “Risk Aversion, Prudence and Temperance: An Experiment in Gain and Loss,” Research in Economics 73 (2019), 174–189. 10.1016/j.rie.2019.04.004 Caballero, R. J., “Consumption Puzzles and Precautionary Savings,” Journal of Monetary Economics 25:1 (1990), 113–136. 10.1016/ 0304-3932(90)90048-9 Carroll, C. D., and A. A. Samwick, “How Important Is Precautionary Sav- ing?” this REVIEW 80:3 (1998), 410–419. Crainich, D., L. Eeckhoudt, and A. Trannoy, “Even (Mixed) Risk Lovers Are Prudent,” American Economic Review 103:4 (2013), 1529–1535. 10.1257/aer.103.4.1529 Croson, R., and U. Gneezy, “Gender Differences in Preferences,” Jour- nal of Economic Literature 47:2 (2009), 448–474. 10.1257/jel.47.2 .448 Deck, C., and H. Schlesinger, “Exploring Higher Order Risk Effects,” Review of Economic Studies 77:4 (2010), 1403–1420. 10.1111/ j.1467-937X.2010.00605.x ——— “Consistency of Higher Order Risk Preferences,” Econometrica 82:5 (2014), 1913–1943. 10.3982/ECTA11396 ——— “On the Robustness of Higher Order Risk Preferences,” Journal of Risk and Insurance 85:2 (2017), 313–333. 10.1111/jori.12217 Drèze, J. H., and F. Modigliani, “Consumption Decisions under Uncer- tainty,” Journal of Economic Theory 5:3 (1972), 308–335. Ebert, S., “Decision Making When Things Are Only a Matter of Time,” Operations Research 68:5 (2020). 10.1287/opre.2019.1923 Ebert, S., and D. Wiesen, “Testing for Prudence and Skewness Seeking,” Management Science 57:7 (2011), 1334–1349. 10.1287/mnsc.1110 .1354 ——— “Joint Measurement of Risk Aversion, Prudence, and Temperance,” Journal of Risk and Uncertainty 48:3 (2014), 231–252. 10.1007/ s11166-014-9193-0 Eeckhoudt, L., and C. Gollier, “The Impact of Prudence on Optimal Prevention,” Economic Theory 26:4 (2005), 989–994. 10.1007/ s00199-004-0548-7 Eeckhoudt, L., B. Rey, and H. Schlesinger, “A Good Sign for Multivariate Risk Taking,” Management Science 53:1 (2007), 117–124. 10.1287/ mnsc.1060.0606 Eeckhoudt, L., and H. Schlesinger, “Putting Risk in Its Proper Place,” American Economic Review 96:1 (2006), 280–289. 10.1257/ 000282806776157777 ——— Higher-Order Risk Attitudes” (pp. 41–57), in G. Dionne, ed., Hand- book of Insurance (New York: Springer Science + Business Media, 2013). Eeckhoudt, L., H. Schlesinger, and I. Tsetlin, “Apportioning of Risks via Stochastic Dominance,” Journal of Economic Theory 144:3 (2009), 994–1003. 10.1016/j.jet.2008.11.005 Es˝o, P., and L. White, “Precautionary Bidding in Auctions,” Econometrica 72:1 (2004), 77–92. Etchart-Vincent, N., and O. l’Haridon, “Monetary Incentives in the Loss Domain and Behavior toward Risk: An Experimental Comparison of Three Reward Schemes Including Real Losses. Journal of Risk and Uncertainty 42:1 (2011), 61–83. 10.1007/s11166-010-9110-0 Fei, W., and H. Schlesinger, “Precautionary Insurance Demand with State- Dependent Background Risk,” Journal of Risk and Insurance 75:1 (2008), 1–16. 10.1111/j.1539-6975.2007.00245.x Gollier, C., and J. W. Pratt, “Risk Vulnerability and the Tempering Ef- fect of Background Risks,” Econometrica 64:5 (1996), 1109–1123. 10.2307/2171958 Haering, A., D. Heinrich, and T. Mayrhofer, “Exploring the Consistency of Higher Order Risk Preferences,” International Economic Review 61:1 (2020), 283–320. 10.1111/iere.12424 Kahneman, D., “Maps of Bounded Rationality: Psychology for Behavioral Economics,” American Economic Review 93:5 (2003), 1449–1475. 10.1257/000282803322655392 Kahneman, D., and A. Tversky, “Prospect Theory: An Analysis of Decision under Risk,” Econometrica 47:2 (1979), 263–292. 10.2307/1914185 Kimball, M. S., “Precautionary Saving in the Small and in the Large,” Econometrica 58:1 (1990), 53–73. 10.2307/2938334 ——— “Precautionary Motives for Holding Asset” (pp. 158–163), in P. K. Newman, M. Milgate, and J. Eatwell, eds., The New Palgrave Dic- tionary of Money and Finance (London: Macmillan, 1992). K˝oszegi, B., and M. Rabin, “A Model of Reference-Dependent Prefer- ences,” Quarterly Journal of Economics 121:4 (2006), 1133–1165. Leland, H. E., “Saving and Uncertainty: The Precautionary Demand for Saving,” Quarterly Journal of Economics 82:3 (1968), 465–473. 10.2307/1879518 Maier, J., and M. Rüger, “Experimental Evidence on Higher-Order Risk Preferences with Real Monetary Losses,” working paper (2012). Menezes, C., C. Geiss, and J. Tressler, “Increasing Downside Risk,” Amer- ican Economic Review 70:5 (1980), 921–932. Noussair, C. N., S. T. Trautmann, and G. Van de Kuilen, “Higher Order Risk Attitudes, Demographics, and Financial Decisions,” Review of Economic Studies 81:1 (2014), 325–355. 10.1093/restud/rdt032 Peter, R., “Optimal Self-Protection in Two Periods: On the Role of Endoge- nous Saving,” Journal of Economic Behavior and Organization 137 (2017), 19–36. 10.1016/j.jebo.2017.01.017 Pratt, J. W., and R. J. Zeckhauser, “Proper Risk Aversion,” Econometrica 55:1 (1987), 143–154. 10.2307/1911160 Sandmo, A., “The Effect of Uncertainty on Saving Decisions,” Review of Economic Studies 37:3 (1970), 353–360. 10.2307/2296725 Trautmann, S. T., and G. Van de Kuilen, “Higher Order Risk Attitudes: A Review of Experimental Evidence,” European Economic Review 103 (2018), 108–124. 10.1016/j.euroecorev.2018.01.007 Tversky, A., and D. Kahneman, “Advances in Prospect Theory: Cumulative Representation of Uncertainty,” Journal of Risk and Uncertainty 5:4 (1992), 297–323. 10.1007/BF00122574 Van de Kuilen, G., and P. P. Wakker, “The Midweight Method to Measure Attitudes toward Risk and Ambiguity,” Management Science 57:3 (2011), 582–598. 10.1287/mnsc.1100.1282 l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . e d u / r e s t / l a r t i c e - p d f / / / / 1 0 4 4 7 0 5 2 0 3 3 3 7 3 / r e s t _ a _ 0 0 9 8 0 p d . f b y g u e s t t o n 0 7 S e p e m b e r 2 0 2 3THE REFLECTION EFFECT FOR HIGHER-ORDER RISK PREFERENCES image

Descargar PDF