Judgment and Decision Making, vol. 1, no. 1, July 2006, pp. 23-32.
A psychological law of inertia and the illusion of loss aversion
David Gal1
Graduate School of Business
Stanford University
Abstract
The principle of loss aversion is thought to explain a wide range
of anomalous phenomena involving tradeoffs between losses and
gains. In this article, I show that the anomalies loss aversion
was introduced to explain - the risky bet premium, the endowment
effect, and the status-quo bias - are characterized not only by
a loss/gain tradeoff, but by a tradeoff between the status-quo
and change; and, that a propensity towards the status-quo in
the latter tradeoff is sufficient to explain these phenomena.
Moreover, I show that two basic psychological principles - (1)
that motives drive behavior; and (2) that preferences tend to
be fuzzy and ill-defined - imply the existence of a robust and
fundamental propensity of this sort. Thus, a loss aversion principle
is rendered superfluous to an account of the phenomena it was
introduced to explain.
Keywords: inertia, loss aversion, endowment effect, status-quo
bias, risky choice, reference-dependent preferences
An object at rest remains at rest and an object in motion
remains in motion unless acted upon by an outside force.
- Newton's First Law of Motion (law of inertia)
1 Introduction
Research has shown that people tend to evaluate outcomes not
in terms of their impact on an individual's resulting state of
wealth, but in terms of changes from a reference state (e.g.,
Kahneman & Tversky, 1979). Moreover, evidence has been interpreted
to imply that people are loss averse: negative changes (i.e.,
losses) from a reference state are thought to loom larger than
positive changes (i.e., gains) of equivalent magnitude (e.g Kahneman
& Tversky, 1979; Tversky & Kahneman, 1991). This principle,
named loss aversion, is commonly considered the most robust and
important finding of behavioral decision theory, and has been
widely hailed (Camerer, 2005) and cited as a "seemingly ubiquitous
phenomenon" (Novemsky & Kahneman, 2005).
This seeming ubiquity is evident in the economics and finance
literature, where loss aversion has been cited, inter alia, to
account for the equity premium puzzle (Benartzi & Thaler, 1995),
the disposition effect (O'Dean, 1998), and the inability of risk-aversion
based on wealth to explain people's unwillingness to accept small
even bets (Rabin & Thaler, 2001). In the marketing literature,
loss aversion has similarly been cited widely to account, inter
alia, for the endowment effect (e.g., Sen & Johnson, 1997; Strahilevitz
& Loewenstein, 1998), the compromise effect (Simonson & Tversky,
1992), and an observed asymmetry in the price elasticity of demand
(Putler, 1992; Hardie, Johnson, & Fader, 1993).
The principle of loss aversion was first introduced by Kahneman
and Tversky (1979) to account for the finding that experimental
subjects required a premium over expected value to accept a bet
offering an even chance of a gain or loss ("the risky
bet premium"). Subsequently, the principle was extended to
the context of riskless choice: Thaler (1980) coined the term
"endowment effect" to refer to the finding that
randomly assigned owners of an object appear to value the object
more than randomly assigned non-owners of the object. For
instance, in one well-known series of endowment effect
experiments, Kahneman, Knetsch and Thaler (1990) found that
randomly assigned owners of a mug required significantly more
money to part with their possession (around $7) than randomly
assigned buyers were willing to pay to acquire it (around $3).
Kahneman et al. (1990, 1991) and Tversky and Kahneman (1991)
attributed this result to loss aversion: owners' loss of the mug
loomed larger than buyers' gain of the mug. "The status
quo bias" - individuals' tendency to prefer to remain at the
status-quo - is similarly attributed to loss aversion: It is
assumed that the loss of the status-quo option looms larger than
the gain of an alternative option (e.g., Kahneman et al., 1991).
For instance, in one empirical demonstration of the status-quo
bias, Samuelson and Zeckhauser (1988) showed that individuals
participating in a hypothetical investment choice task were more
likely to choose to invest an inheritance in a particular
investment option (out of four) when that option was presented as
the status-quo (i.e., when they were informed that the money from
the inheritance was already invested in that option).
Remarkably for a principle that is so pervasive, the principle
of loss aversion is not derived from any theory of behavior or
more basic psychological principles, but is an ad hoc principle
introduced to account for a range of phenomena involving tradeoffs
between losses and gains that are anomalous in the context of
the classical choice paradigm. The absence of an accepted psychological
theory to account for loss aversion has led to a paradoxical
situation: loss aversion is cited as the explanation for phenomena
associated with loss/gain tradeoffs (e.g., the endowment effect,
status-quo bias, risky bet premium) and, circuitously, the
same phenomena are cited as evidence for the existence of loss
aversion.
This is not to say that loss aversion lacks a potentially
plausible psychological basis. Indeed, a number of researchers
have attempted to uncover an underlying psychological mechanism
that could explain a loss/gain asymmetry. Posited psychological
mechanisms for loss aversion include the proposition that the
hedonic impact of losses is greater than that of gains (e.g.,
Bar-Hillel & Neter, 1996), that people's locus of attention
tends to be focused on losses more than on gains (Carmon &
Ariely, 2000), and - through studies with either animals or
fMRI - that a loss/gain asymmetry is cognitively hard-wired.
A common feature of these attempts to uncover a psychological
mechanism for loss aversion is the premise that a fundamental
loss/gain asymmetry in fact exists, and that this asymmetry is
reflected in the phenomena it purports to explain. In contrast,
in the present research, I do not attempt to explain the existence
of a loss/gain asymmetry, but to challenge the notion that a
reference-dependent asymmetry is necessary to explain these phenomena
at all. In particular, I recognize that the phenomena most commonly
cited as evidence for loss aversion - the status-quo bias, the
endowment effect, and the risky bet premium - are characterized
not only by a loss/gain tradeoff, but by a tradeoff between the
status-quo and change; and, that a propensity towards the status
quo in the latter tradeoff is sufficient to explain these phenomena.
Moreover, I show that two basic psychological principles - (1)
that motives drive behavior, and (2) that preferences tend to
be fuzzy and ill-defined - imply the existence of a robust and
fundamental propensity of this sort. Thus, a propensity to remain
at the status-quo - i.e., inertia - is not simply an alternative
account to loss aversion for these phenomena, but one that renders
the introduction of a loss aversion principle superfluous.
The remainder of this article is organized as follows: First,
I discuss the implication of the nature of behavior and preferences
for a propensity to remain at the status-quo. Subsequently, I
compare this inertia account with the loss aversion account for
the status-quo bias, the endowment effect, and the risky bet
premium. I conclude with the argument that the existence of a
basic behavioral tendency to favor the status-quo over change
renders the loss aversion principle superfluous to an account
of the phenomena it was introduced to explain, and that the principle
should therefore be abandoned.
2 A psychological law of inertia
In this section, I argue that a propensity to remain at the status
quo, rather than a fundamental loss/gain asymmetry, offers the
most parsimonious account for the phenomena loss aversion was
introduced to explain. That is, a propensity to remain at the
status-quo logically follows from basic, well-founded psychological
principles, whereas loss aversion is an auxiliary principle,
introduced ad hoc to account for seemingly anomalous phenomena.
In this section I show how psychological insights into the nature
of behavior and preferences imply a robust tendency for people
to remain at the status-quo in two parts: First the need for
psychological motives to drive behavior implies that people will
tend to remain at the status-quo when they have no clear preference
between the status-quo and an alternative (or `change') option.
In addition, the fuzzy and ill-defined nature of preferences
implies that people will often have unclear preferences between
options, and hence, that a propensity to remain at the status
quo is likely to be a robust effect.
2.1 Motive-driven behavior
In the classical choice paradigm (Von Neumann & Morgenstern,
1944) of precise and well-defined preferences, individuals making
a choice between two options, A and B, are thought to either
(1) prefer A to B, (2) prefer B to A, or (3) be indifferent between
A and B (i.e., to prefer A and B exactly the same). A particular
preference ordering is assumed to be independent of context,
the description of the problem, or the procedure used to elicit
the preferences. Therefore, in the classical choice paradigm,
preference for A or B should be the same regardless of whether
option A or option B is the status-quo option. This implies
that individuals who prefer option A to option B should choose
option A regardless of whether it is the status-quo option or
not; and, likewise, that individuals who prefer option B to option
A should choose option B regardless of whether it is the status
quo option or not.
However, a question arises as to what individuals who are
indifferent between Option A and Option B should do. That is,
what option should be chosen by individuals who have the same
exact valuation for Options A and B? Although such a situation is
not addressed in the economic literature, it seems clear that
individuals who are indifferent between two options should choose
the status quo. For instance, even in the absence of transaction
costs, we should not be surprised, in the context of precise and
well-defined preferences, if people "prefer" to keep the dollar
in their pocket rather than exchange it for another dollar. This
is because, at the most basic level, economists and psychologists
alike recognize that people's behavior is directed in accordance
with psychological motives (i.e., reasons, drive states, goals,
incentives, etc.).
From this basic notion, it follows that people will not act to
alter the status-quo unless they are impelled to do so by some
motive.2 Moreover, we can surmise that
the possibility of becoming better off - but not
equally as well off - can provide the necessary motive
to impel people to change the status quo (see Figure 1A). This
discussion is formalized as follows:
Psychological Law of Inertia: A person will tend to
maintain the status-quo unless impelled to alter the status-quo
by a psychological motive to do so.
Corollary: The possibility of becoming better off -
but not equally as well off - can provide the necessary motive
to impel a person to alter the status-quo.
As highlighted by the discussion to this point, the need for
a psychological force, or motive, to alter the status-quo implies
that people can be expected to manifest a preference to remain
at the status-quo when they are indifferent between options.
However, such an effect is unlikely to be very robust in the
context of precise and well-defined preferences, because such
precise preferences make the likelihood of indifference between
two nonidentical options extremely slight. For instance, if an
individual is indifferent between options A and B, then the classical
choice assumption of monotonic preferences implies that the individual
should prefer option A and a penny to option B. Thus, given a
large pool of individuals, it is likely that only a minimal percentage
of participants will value both options exactly the same, and
these individuals will thus have only a minimal impact on any
outcome that aggregates responses over this pool of individuals.
A: Assuming precise and well-defined preferences:
B: Assuming fuzzy and ill-defined preferences:
Figure 1: Relative preference for Option A over Option B with increasing
absolute attractiveness of Option B.
2.2 Fuzzy and ill-defined preferences
In the previous subsection, I argued that a change to the status
quo requires a motive to alter the status-quo, and, accordingly,
there is a tendency to remain at the status-quo when people are
indifferent between options. However, I also acknowledged that
the classical notion of precise and well-defined preferences
implies that indifference between nonidentical options is quite
rare, and thus any proclivity towards the status-quo is unlikely
to be a robust effect.
However, in recent times, the classical notion of precise and
well-defined preferences has been challenged by a great deal
of evidence, which has shown that preferences tend to be fuzzy
and ill defined, and that they are often constructed on an ad
hoc basis (for review, see Slovic, 1995; Bettman, Luce, & Payne,
1998). Moreover, evidence suggests that people are unable to
precisely assess the value of options and outcomes in an absolute
sense (e.g., Hsee, 1996; Nowlis & Simonson, 1997). For instance,
Kivetz and Simonson (2002) have shown that people tend to use
the relative effort of others as a reference to judge the absolute
amount of effort associated with a frequency reward program.
More specifically, they showed that if a "deal" is relatively
better for person X than for the average individual, it will
be extremely attractive - to the point where it might be preferred
over a dominated option. In one experiment, Kivetz and Simonson
offered diners a reward program in which they could receive a
free meal at a dining hall after having paid for a certain number
of meals. In a between subject design, they found that sushi
lovers would actually prefer a reward program which required
the purchase of 10 sandwiches and 10 sushi platters to a program
which required only the purchase of the 10 sandwiches. Although
the former option was dominated by the latter, sushi lovers perceived
a relative advantage in that they would likely have eaten the
sushi anyway. Based on this relative advantage, sushi lovers
inferred that they were getting a "bargain" in an absolute
sense.
Research on choice deferral also suggests that people are unable
to precisely judge the value of options in an absolute sense.
For instance, Dhar (1997) finds that when two options are rated
similarly in terms of their attractiveness, people are likely
to defer choice, rather than choose one of the two options. This
suggests that people are unable to precisely judge the absolute
attractiveness of the options and, accordingly, do not have a
precise ordering of preferences over the options that would allow
them to justify choosing one option over the other.
If we extend this reasoning to a choice between any two options,
A and B, then we can surmise that people may be indifferent - i.e.,
have no clear preference - between options A and B, and also
have no clear preference between option A plus a penny and option
B - and even between option A plus a dollar and option B. That
is, fuzzy and ill-defined preferences imply a fuzzy range of
absolute attractiveness values for option A, that are not
clearly differentiated in terms of relative attractiveness
to option B - and vice versa (see Figure 1B). For example, if
option A is a monetary amount and option B is a mug, it is possible
that values of A between roughly $3 and roughly $7 will not
feel sufficiently different from the value of the mug to induce
a clear preference between the monetary amount and the mug. Similarly,
if option A is $5 and option B is a mug, people may have no
clear preference both between $5 and a mug with a standard handle
and between $5 and a mug with a fancy handle - even if they strongly
prefer the fancy handle relative to the standard handle.
Thus, the recognition that preferences tend to be fuzzy and ill-defined
suggests that people will often have unclear preferences between
two options. Accordingly, we can expect that people will often
lack a motive to alter the status-quo.
3 Inertia versus loss aversion
In the previous section, I argued that the nature of behavior
and preferences imply a fundamental behavioral proclivity to
prefer the status-quo to change. In this section, I show that
a propensity to remain at the status-quo can account for the
status-quo bias, the endowment effect, and the risky bet premium - the
phenomena most widely cited as evidence for loss aversion - and
do so in a more logically consistent manner.
3.1 Status-quo bias and endowment effect
As discussed earlier in this article, experimental evidence has
demonstrated that people have a tendency to remain at the status
quo. Proponents of loss aversion assert that a status-quo propensity
is a consequence of a loss/gain asymmetry (i.e., a reference-dependent
asymmetry in favor of losses), whereas the proposed inertia account
asserts that any such asymmetry is auxiliary to an explanation
of the basic behavioral propensity to remain at the status-quo.
Instead, the inertia account asserts that the status-quo bias
logically follows from the basic principle that behavior is directed
in accordance with psychological motives.
Clearly, the inertia account is more parsimonious than the loss
aversion account; however, is there a way to also compare the
descriptive validity of the inertia and loss aversion accounts
as explanations for the status-quo bias? We can consider a thought
experiment of the extreme case where an individual has precisely
identical valuations for the status-quo option and an alternative
option: would an individual exchange the dollar bill in her pocket
for another, essentially identical, dollar bill absent some external
motivation (e.g., a desire to comply with an experimenter's request)
to do so? An inertia account predicts that, absent an external
motive, an individual will "choose" to retain her dollar bill
rather than exchange it for a different dollar bill because of
the absence of any motive to exchange dollar bills.
In contrast, a loss aversion account makes no such prediction.
This is because it is assumed (quite reasonably) that people
do not typically view an exchange of identical items as a tradeoff
between a loss and a gain. For instance, Kahneman (2003) has
stated that loss aversion should not be expected to apply in
an exchange of five $1 bills for a $5 bill. Similarly, Novemsky
and Kahneman (2005, p. 123), in highlighting one of several proposed
"boundaries of loss aversion," surmise that "[A] shopper is
unlikely to experience loss aversion when giving up a good for
a nearly identical one."
Although a thought experiment is likely sufficient, I conducted
a simple experiment to confirm that the outcome of a choice between
two essentially identical options will significantly favor
the status-quo option (as predicted by the inertia account, but
not by the loss aversion account). In a between subject design,
110 participants - undergraduates at a large west coast university - were
asked to imagine that they owned a quarter minted in either Denver
or Philadelphia. They were then asked whether - given a choice - they
would choose to switch their coin with a coin minted in the other
city, assuming insignificant time and effort involvement for
the switch. Over 85% of participants in either condition chose
to retain their original coin, consistent with the inertia account
of the status-quo bias.
Although the experiment described above involved goods that had
well-defined relative valuations (i.e., their valuations were
equal), people typically do not have well-defined
relative valuations for goods. Therefore, we can expect that
there will be many pairs of options for which people will have no
relative preference for one option over the other (as between two
quarters) - and hence no reason or motive to alter the
status-quo.
Thus, the inertia account can explain a propensity towards the
status-quo both when a status-quo option and an alternative
option have equivalent valuations and when they do not.
Conversely, a loss aversion account is descriptively consistent
with a propensity towards the status-quo in cases where the
status-quo option and an alternative option are not equivalent,
but it provides no insight into why such a propensity persists
when option values are equivalent.
3.1.1 Endowment effect
As discussed earlier in this article, the endowment effect is
the name for the finding that randomly assigned owners of an
object appear to value their possession more than randomly assigned
non-owners.
Because the status-quo bias and endowment effect are such similar
phenomena, the logic regarding inertia as an explanation of the
status-quo bias in the previous subsection extends fairly trivially
to the endowment effect. For instance, using the Kahneman et
al., (1990) example of buyers and sellers with divergent reservation
prices for a mug, it is clear that sellers view ownership of
the mug as the status-quo and non-ownership (plus receipt of
payment) as the change option. For buyers, the status-quo and
change options are reversed (see Figure 1B).
Moreover, when one of the two options is a variable monetary
sum as in the Kahneman et al. (1990) experiments, measures of
maximum willingness to pay (WTP) to acquire a good and minimum
willingness to accept (WTA) to part with a good can be thought
of as rough approximations to the fuzzy boundaries of the fuzzy
indifference range depicted in Figure 1B. This is highlighted
in Figure 2, which can be viewed as an instance of Figure 1B
where option B is a variable monetary sum. WTP represents the
lower boundary because at higher valuations there is either (a)
indifference between the monetary sum and the good, or (b) the
monetary sum is preferred to the good. Therefore, there is no
motive for the individual to pay any more money for the good
than the lower boundary of the fuzzy indifference range. Similar
logic applies to WTA as the upper boundary of the indifference
range.
Figure 2: Relative preference for a Good A over a variable monetary sum.
One potential challenge to the inertia account of the endowment
effect arises from the findings of Dubourg, Jones-Lee, and Loomes
(1994). Dubourg et al. (1994) found that the gap between their
experimental participants' WTP and WTA persisted even after accounting
for "imprecise preferences." Specifically, Dubourg et al. attempted
to define participants' WTP and WTA as confidence intervals rather
than as point estimates. They defined the upper end of the WTP
interval as "the smallest amount a respondent definitely would
not pay" for a good and the lower end of the WTP interval
as "the largest amount a respondent definitely would pay."
Similar elicitation procedures were used to obtain the upper
and lower ends of respondents' WTA interval. Dubourg et al. hypothesized
that if imprecise preferences were the source of the endowment
effect, then the WTP and WTA range would overlap, but participants'
point estimates of their WTP might trend toward the lower WTP
bound and their point estimates of their WTA might trend toward
the upper bound of the WTA interval leading to a WTP/WTA gap.
Instead, they found that the entire WTP interval tended to be
well below the entire WTA interval. Thus, they surmised that
imprecise preferences could not wholly account for the WTP/WTA
gap.
Although the notion of imprecise preferences in Dubourg et al.'s
account sounds similar to the notion that people often lack clear
relative preferences between options, the manner in which Dubourg
et al. operationalize a range of imprecise preferences does not
equate to the fuzzy indifference range described by the inertia
account. Indeed, the inertia account predicts that
measures of WTP should be below measures of WTA because it is the
gap between WTP and WTA that represents the fuzzy indifference
range (i.e., the range over which people do not have a clear
preference for the money or the good and hence do not trade due
to the absence of a motive to trade.) Dubourg et al.'s use of
different elicitation methods to obtain a range for each of WTP
and WTA merely demonstrates that the borders of the
indifference range should not be thought of as clear demarcations
between indifference and a clear preference for one option over
another, but as fuzzy and imprecise. Accordingly, different
elicitation methods of WTP and WTA should be expected to yield
different values for the borders of the fuzzy indifference range.
This is highlighted by the short lines on either side of the long
line in Figure 2. The short lines represent possible ranges for
WTP and WTA as found by Dubourg et al. (1994), whereas the long
lines between them represent particular point estimates of WTP
and WTA.
3.1.2 Degree of preference clarity
Although the principle of loss aversion is agnostic about the
magnitude of the loss aversion coefficient (Daniel Kahneman,
personal communication, 2004), several researchers have sought
to address this question empirically. In general, most researchers
have concluded that the "coefficient of loss aversion" is somewhere
around 2 (e.g., Tversky & Kahneman, 1992). However, other researchers
have found that the degree of loss aversion depends on the degree
of similarity between options being evaluated. For instance,
unlike in the quarters experiment presented earlier in this section,
Chapman (1998) found that a majority of experimental participants
were willing to trade items that they owned for identical
items. Moreover, Chapman showed that participants were more willing
to trade identical items than similar items and similar items
than dissimilar items. However, Chapman was able to obtain these
results only when she offered participants a nickel for the act
of trading in order to cover participants' "transaction costs."
The inertia account introduced in this article provides insight
into this finding whereas the loss aversion account is silent.
Specifically, the requirement for an incentive - in the form of
a nickel - to induce transactions for identical and similar items
is in accord with the inertia account. When items being traded
are identical or very similar, relative preferences are well-defined - i.e.,
there is a relatively narrow range of absolute values of the
options over which there is no clear preference between the options
(i.e., a narrow indifference range in Figure 1B) - and hence, even
a slight increase in the value of the alternative option (e.g.,
an extra nickel) will be a sufficient enough incentive for participants
to alter the status-quo.
In other words, a participant asked to trade item X for item X
will have no motive to do so; however a participant asked to
trade item X for item X + 5 cents can recognize that item X +
5 cents is clearly better than item X, and hence has a motive to
execute the trade (absent transaction costs). On the other hand,
if a participant has no clear preference between two dissimilar
items, X and Y, then she is also unlikely to have a clear
preference between item X + 5 cents and item Y (see Figure 1B),
and therefore is likely to lack a motive to alter the status-quo
with or without a nickel incentive.
3.1.3 Do people like the status-quo?
In recent research, Moshinsky and Bar-Hillel (2005) found that
participants tended to evaluate public policy options more favorably
when they were presented as the status-quo option than when they
were not, a phenomenon they dubbed, the "status-quo label bias."
Moshinksy and Bar-Hillel (2005) argued that this finding constituted
support for loss aversion. This is an interesting assertion,
because it is diametrically opposed to the findings and arguments
of Loewenstein and Kahneman (1991) and Kahneman et al. (1991).
Loewenstein and Kahneman (1991) found that despite the persistence
of an endowment effect, experimental participants did not rate
the attractiveness of endowed options more favorably than the
same options when they were not endowed. Kahneman et al. (1991)
interpreted this finding to imply that the endowment effect does
not "enhance the appeal of the good one owns, only the pain
of giving it up." Thus, while the status-quo label bias may
be - under certain circumstances - a complimentary contributor
to a status-quo bias, evidence for a status-quo label bias does
not appear to support the loss aversion account over the inertia
account of the status-quo bias.
3.2 Risky bet premium
The status-quo bias and endowment effect phenomena involve a
loss/gain tradeoff that is entangled with a status-quo/change
tradeoff. That is, the status-quo option is always associated
with potential loss, whereas the change option is always associated
with potential gain.
At first inspection, the risky bet premium phenomenon does not
appear to involve a status-quo/change tradeoff. There appears
to be only a tradeoff between the potential for loss associated
with taking the bet and the potential for gain associated with
taking the bet. However, upon closer inspection, the risky bet
premium phenomenon is actually quite similar to the endowment
effect and status-quo bias phenomena. In particular, in deciding
whether to accept a single risky bet, the status-quo option is
not taking the bet, whereas the change option is taking
the bet. That is, the decision to accept a single risky bet can
be thought of as a choice between two options, A and B, where
Option A is not taking the bet (i.e., the status-quo) and Option
B is taking the bet (i.e., the change option). This is depicted
in Figure 3, which can be viewed as an instance of Figure 1B
where option A is the status-quo and Option B is the status-quo
plus a risky bet.
Figure 3: Relative preference for status-quo (SQ) over SQ plus a risky
bet with 50% chance of losing $c and 50% chance of Winning
$X.
The loss aversion account of the risky bet premium ignores the
status-quo/change tradeoff. It asserts that people demand a premium
over expected value to accept an even bet because the potential
for loss associated with taking the bet looms larger than the
potential for gain associated with taking the bet. This, of course,
is premised on the belief that, objectively, people should judge
the value of risky prospects according to their expected value
(e.g., Arrow, 1965).
In reality, however, this presumption is a simplification made
so that risky prospects can be incorporated into rational (i.e.,
mathematical) theories of choice. There is, in fact, no demonstrably
objective formula by which people should judge risky prospects.
Instead, people should judge risky prospects according to their
goals: they should weigh their desire to avoid potential loss
against their desire to acquire potential gains according to
their preference for the tradeoff between them.
However, because this preference is likely to be fuzzy and ill-defined,
there is likely to be a fuzzy range of values for a prospect
for which people will not have a clear preference between the
prospect and the status-quo. For instance, people may have no
clear preference between the status-quo and a prospect featuring
a 50% chance of losing $100 and a 50% chance of gaining $100;
and, they may also not have a clear preference between the status
quo and a 50% chance of losing $100 and a 50% chance of gaining
$150. That is, they may not have a clear sense that either bet,
on balance, will tend to make them better off than not taking
the bet. Thus, people are likely to demand a premium to accept
an even bet when they have no preexisting psychological motive
to alter the status-quo.
To distinguish these competing accounts, I conducted a simple
experiment, where participants faced a choice between risky prospects
that featured a tradeoff between potential loss and potential
gain, but no clear tradeoff between the status-quo and change.
Specifically, experimental participants (133 undergraduates at
a large west coast university) were asked to allocate a hypothetical
monetary sum between a risk-free ("safe") option and an even
bet. The problem featured no clear tradeoff between the status
quo and change because the problem featured no clear status-quo
option. The problem appeared as follows:3
Allocation Task:
Assume you have $100 that you want to invest and that the available
options are the two investment options below. How would you allocate
your money between the 2 options?
Investment Option A
You will make 3% on your investment for sure.
Investment Option B
You will double your investment with a 50% chance.
You will lose your investment with a 50% chance.
Of my $100, I would invest $____ in Option A |
and $____ in Option B.
|
After a series of unrelated tasks, participants were also asked - as
in previous risky bet premium experiments - to indicate the premium
they would require to accept a single risky bet. The problem
appeared as follows:
Single Risky Bet Task:
Suppose you were offered a risky bet that offered a 50%
chance of losing $100 and a 50%
chance of winning X. What is the least X would have to
be for you to be willing to take this bet?
X would have to be $____.
In the single risky bet task - consistent with prior findings - less
than 2% of participants were willing to accept an even bet,
and the rest tended to require a significant premium to accept
the bet (median value of X was $500).
In contrast, in the allocation task, 23% of participants allocated
the entire monetary sum to the `even bet' option, 55% of participants
allocated some of the monetary sum to the `even bet' option and
some to the `safe' option, and only 23% of participants allocated
the entire monetary sum to the `safe' option. Thus, the results
of these two tasks showed a robust requirement for a premium
to accept an even bet only when participants were faced with
a status-quo/change tradeoff (i.e., in the single risky bet task).
Indeed, this is the first experiment to show that a large percentage
of experimental participants - nearly 80% - are willing to accept
an even bet, a finding which challenges the most basic prediction
of loss aversion (i.e., that people are unwilling to accept even
bets.)
3.2.2 Discussion
The results of this experiment show that people demand a premium
over expected value to accept a single bet with even odds of
a gain or loss, but do not necessarily demand such a premium
when allocating funds across assets with different levels of
risk (i.e., in a task with greater ecological validity). At first
blush, one concern is that the allocation task may have prompted
a demand effect, whereby participants assumed that the task was
intended to elicit allocations to both of the options. However,
the finding that nearly half of participants allocated the entire
monetary sum to a single option, and that of those participants,
half allocated the entire sum to the risky option, suggests that
demand effects cannot explain the large share of funds allocated
to the risky option.
Another initial concern is that the sums participants were asked
to allocate were small. It is possible that participants would
have allocated a greater share of the monetary sum to the safe
option had participants been asked to allocate a sum that constituted
a larger share of their budgets or wealth. However, risky bet
premium experiments are typically conducted with small sums of
money, because it has been recognized that larger sums will lead
to an increasing impact of wealth, budgets and other classical
economic variables on participants' decision making (e.g., Kahneman
& Tversky, 1979). Indeed, Rabin and Thaler (2001) argue that
it is the fact that people are so risk averse with such small
sums of money that provides the greatest support for the existence
of loss aversion. Thus, the findings of this experiment are highly
inconsistent with the loss aversion account, but consistent with
the proposed inertia account of the risky bet premium.
However, the fact that the evidence from this experiment is consistent
with the inertia account does not imply that the evidence strongly
supports the inertia account. I have argued that the main difference
between the allocation task and the single risky bet task is
the absence of a clear status-quo option - and hence of a tradeoff
between the status-quo and change - in the allocation task. An
alternative account, however, is that the manipulations between
tasks (e.g., temporal distance and choice vs. willingness-to-pay)
simply led to a dramatic shift in risk preference between tasks:
participants were risk-seeking in the allocation tasks and, a
few minutes later, dramatically risk-averse in the single risky
bet task.
However, such a dramatic change in risk preference between tasks
seems implausible. Moreover, despite the superficial difference
in risk preference expressed by participants across tasks - i.e.,
"risk-seeking" in the allocation task and "risk-averse" in
the single risky bet task - there was a correlation in the decisions
participants made between tasks. Those participants who required
the highest premiums in the single risky bet task (based on a
median split) tended to allocate a greater part of their hypothetical
monetary sum (64% vs. 46%) to the safe option in the allocation
task (t(131) = 2.71; p < 0.01). Thus, it would appear
that participants were, in fact, expressing a real and relatively
consistent underlying preference for risk - i.e., for the tradeoff
between potential loss and potential gain - across tasks, but
that this preference was being systematically shifted by an influence
unrelated to risk preference: a propensity to remain at the status
quo in the single risky bet task, but not in the allocation task.
4 Does loss aversion exist?
So far, in this article, I have argued that the notion that motives
drive behavior - together with the fuzzy and ill-defined nature
of preferences - necessarily implies a basic behavioral tendency
to remain at the status-quo, without the need for any other auxiliary
principle. I have also shown that this basic behavioral tendency
is sufficient for explaining the existence of a status-quo bias,
an endowment effect, and a risky bet premium, and that it provides
a more logically consistent account for these phenomena than
loss aversion.
Given this inertia account, what are the implications for the
existence of loss aversion? To be sure, the existence of inertia
does not preclude the possibility that other influences also
contribute to the complex phenomena investigated in this article.
Among those factors are anticipated regret, locus of attention,
and the status-quo label bias. Other research, however, casts
further doubt on the existence of a fundamental loss/gain asymmetry
by challenging the evidence for loss aversion in phenomena that
involve a loss/gain tradeoff but not a status-quo/change
tradeoff. For instance, the equity premium puzzle - the finding
that historical returns on stocks have significantly exceeded
those on bonds (beyond what could be explained by simple risk
aversion) - has previously been cited as evidence for loss aversion
(Benartzi & Thaler, 1995). However, Fama and French (2002) noted
that using historical data on returns alone is not very meaningful
for judging the forward-looking equity premium - i.e., the returns
investors could reasonably have expected at the time. Fama and
French (2002) estimated the forward-looking equity premium to
be substantially smaller than the realized equity premium, obviating
the need for a loss aversion explanation.4 Similarly, the
scanner panel data finding by Hardie et al. (1993) that demand
is more elastic for price increases than for price decreases
was challenged by a study by Bell and Lattin (2000), who found
no such asymmetry after controlling for the confounding influence
of heterogeneity in consumer price responsiveness.5
Even this evidence cannot disprove the existence of loss
aversion; but, the inability of researchers to find evidence for
loss aversion in these phenomena and its dispensability to an
account of the phenomena it was introduced to explain - as
highlighted by this article - do suggest that its existence may
well be moot. An analogy from cosmological physics serves to
highlight this point. At the end of the nineteenth and start of
the twentieth centuries cosmology faced an anomaly: Maxwell's
equations of electromagnetism required that light travel at a
constant rate, but Newtonian mechanics required all motion to be
relative. Hence, to resolve this anomaly, physicists posited the
existence of a `luminiferous ether,' a universal substance in
space; light was thus thought to move relative to the ether.
Then, in 1905, Einstein's theory of special relativity showed
that if time was not fixed - an observation subsequently
confirmed by empirical evidence - the presence of an ether was
no longer required. Thus, relativity did not preclude the
existence of the ether, but it did render its existence
superfluous to an explanation of the phenomenon it was introduced
to explain. Accordingly, the concept of an ether was abandoned.
Analogously, a basic behavioral tendency to remain at the status
quo does not disprove the existence of a fundamental loss/gain
asymmetry, but it does render its existence superfluous to an
account of the phenomena it was introduced to explain. Indeed,
given the fuzzy and ill-defined nature of preferences, and the
need for a motive to drive behavior, we should be surprised if
we did not observe a status-quo bias, an endowment effect,
and a risky bet premium. Therefore, like the ether, logic dictates
that the principle of loss aversion be abandoned.6
References
Arnott, R. D. & Bernstein, P. L. (2002). What risk premium is
"normal"? Financial Analysts Journal, 58, 64-85.
Arrow, K. (1965). Aspects of the theory of
risk-bearing. Helsinki: Yrjö Hahnsson Foundation.
Bell, D. & Lattin, J. (2000). Looking for loss aversion in
scanner panel data: the confounding effect of price response
heterogeneity. Marketing Science, 19, 185-201.
Bar-Hillel, M. & Neter, E. (1996). Why are people reluctant to
exchange lottery tickets? Journal of Personality & Social
Psychology, 70, 17-28.
Benartzi, S., & Thaler, R. (1995). Myopic loss aversion and the equity
premium puzzle. Quarterly Journal of Economics, 110, 73-92.
Bettman, J., Luce, M., & Payne, J. (1998). Constructive consumer
choice processes. Journal of Consumer Research, 25, 187-217
Camerer, C. (2005). Three cheers-psychological, theoretical,
empirical-for loss aversion. Journal of Marketing Research, 42,
129-134.
Carmon, Z., & Ariely, D. (2000). Focusing on the forgone: why
value can appear so different to buyers and sellers. Journal of
Consumer Research, 27, 360-370.
Chapman, G., (1998). Similarity and reluctance to trade. Journal
of Behavioral Decision Making, 11, 47-58.
Dhar, R. (1997). Consumer preference for a no-choice option. Journal of
Consumer Research, 24, 215-231.
Dubourg, W. R., Jones-Lee, M. W., & Loomes, G. (1994). Imprecise
preferences and the WTP-WTA disparity. Journal of Risk
and Uncertainty, 9, 115-133.
Fama, E. F., & French, K. R. (2002). The equity
premium. Journal of Finance, 57, 637-660.
Hardie, B., Johnson, E. J., & Fader, P. (1993). Modelling loss
aversion and reference dependence effects on brand
choice. Marketing Science, 12, 378-394.
Hsee, C. K. (1996). The evaluability hypothesis: An explanation
for preference reversals between joint and separate evaluations
of alternatives. Organizational Behavior and Human Decision
Processes, 67, 247-257
Kahneman, D. (2003). A psychological perspective on
economics. American Economic Review (Proceedings), 93,
162-168.
Kahneman, D., Knetsch, J. L., & Thaler, R. (1990). Experimental
tests of the endowment effect and the Coase theorem. Journal of
Political Economy, 99, 1325-1348.
Kahneman, D., Knetsch, J. L., & Thaler, R. (1991). The endowment
effect, loss aversion, and status-quo bias. Journal
of Economic Perspectives, 5, 193-206.
Kahneman, D., & Tversky, A. (1979). Prospect theory: an analysis of
decision under risk. Econometrica, 47, 263-291.
Loewenstein, G., & Kahneman, D. (1991). Explaining the endowment
effect. Working Paper, Carnegie Mellon University (data reported
in Kahneman et al., 1991)
Moshinsky, A., & Bar-Hillel, M. (2005). Loss aversion and the
status-quo label bias. Working Paper, Hebrew University.
Nowlis, S. & Simonson, I. (1997). Attribute-task compatibility
as a determinant of consumer preference reversals. Journal of
Marketing Research, 34, 205-218.
Novemsky, N. & Kahneman, D. (2005). The boundaries of loss
aversion. Journal of Marketing Research, 42, 119-128.
Odean, T., (1998). Are investors reluctant to realize their
losses? Journal of Finance, 53, 1775-1798.
Putler, D, (1992). Incorporating reference price effects into a
theory of consumer choice.
Marketing Science, 11, 287-309.
Rabin, M., & Thaler, R. (2001). Risk Aversion. Journal of Economic
Perspectives, 15, 219-232.
Samuelson, W., & Zeckhauser, R. (1988). Status-quo bias in
decision making, Journal of Risk and Uncertainty, 1,
7-59.
Sen, S., & Johnson, E. J. (1997). Mere-possession effects
without possession in consumer choice. Journal of
Consumer Research, 24, 105-117.
Simonson, I., & Tversky, A. (1992). Choice in context: tradeoff
contrast and extremeness aversion. Journal of Marketing Research,
29, 281-295.
Slovic, P., (1995). The construction of preference. American
Psychologist, 50, 364-371.
Strahilevitz, M. A., & Loewenstein, G. (1998). The effect of
ownership history on the valuation of objects. Journal of
Consumer Research, 25, 276-289.
Thaler, R. (1980). Toward a positive theory of consumer
choice. Journal of Economic Behavior and Organization, 1, 39-60.
Tversky, A., & Kahneman, D. (1991). Loss aversion in riskless
choice: a reference-dependent model. Quarterly Journal of
Economics, 106, 1039-1061.
Tversky, A., & Kahneman, D. (1992). Advances in prospect theory:
cumulative representations of uncertainty. Journal of
Risk and Uncertainty, 5, 297-323.
Von Neumann, J., & Morgenstern, O. (1944). Theory of games and
economic behavior. Princeton, NJ: Princeton University Press.
Footnotes:
1I am grateful to Michal Maimaran, Baba
Shiv, Itamar Simonson, Christian Wheeler, Barbara Mellers,
the associate editor, and Jonathan Baron, the editor, for
comments on earlier drafts of this manuscript. Address
correspondence to: David Gal, 729 Escondido Rd. 12,
Stanford, CA 94305,
dgal@stanford.edu.
2Throughout this article, "alter the
status-quo" can be taken to include the similar case of
"reject the default option."
3The problem was the
first of six investment allocation type problems and participants
were asked to assume that the duration of the investment in each
problem was one year.
4Arnott and Bernstein
(2002) estimated that the forward looking equity premium at the
time of their study was zero or even negative
5I thank
Jim Lattin for reviewing this point.
6Abandoning
loss aversion does not imply abandoning other elements of Prospect
Theory's value function. None of the arguments made in this article
in any way challenge the separate coding of losses and gains
around a reference point, the concavity of the value function
in the domain of gains, or the convexity of the value function
in the domain of losses. This article merely challenges the notion
that the loss and gain curves are asymmetric (i.e., steeper in
the domain of losses than in the domain of gains).
File translated from
TEX
by
TTH,
version 3.67.
On 18 Jul 2006, 15:40.