On TIME , MOOD & PREFERENCES: A Great paper

Electronic copy available at: http://ssrn.com/abstract=1491247
 all the right rserved to the Authors, courtesy of the Authors and SSRN . 

Happiness and Time Preference: The Effect of Positive Affect in a Random-
Assignment Experiment
John Ifcher
Santa Clara University
Homa Zarghamee
Santa Clara University
Abstract
We conduct a random-assignment experiment to investigate whether
positive affect impacts time preference, where time preference denotes a
preference for present over future utility. Our result indicates that,
compared to neutral affect, mild positive affect significantly reduces
subjects’ time preference. Affects were induced using short film clips.
Our result has implications for the effect of happiness on time preference
and the role of emotions in economic decision making in general. Finally,
we reconfirm the ubiquity of time preference and start to explore its
determinants.
(JEL D01, D90, D60)
Keywords: positive affect, happiness, time preference, random-assignment
experiment, mood-inducement

We wish to thank Abigail Brown, Rachel Croson, Linda Kamas, Michael Kevane,
Silvana Krasteva, Erin Krupka, Eugenio Proto, Tanya Rosenblat, William Schulze, Joel
Sobel, William Sundstrom, Neslihan Uler, two anonymous referees, and seminar
participants at various locations. We also wish to thank Marianne Farag and Anders
Loven-Holt who provided excellent research assistance. Financial support from Santa
Clara University’s Presidential Research Grant and Leavey Grant is gratefully
acknowledged.
Electronic copy available at: http://ssrn.com/abstract=1491247
1
Initially ignored—and subsequently thought to be disruptive—emotions have historically
received a bum rap from decision researchers (Isen 2008; and Loewenstein and Lerner
2003). A great deal of recent research, however, has shown that emotions—and in
particular mild positive affect—can have important, and often salutary, effects (Isen
2008). For example, positive affect has been shown to increase cognitive flexibility (Isen
2008); reciprocity in “gift-exchange” games (Kirchsteiger et al 2006); work effort and
productivity (Erez and Isen 2005; and Oswald et al 2008); loss aversion (Isen et al 1988);
and risk-aversion (Isen and Geva 1987); and to decrease spending and willingness to pay
(Cryder et al 2008; and Lerner et al 2004). We conduct a random-assignment experiment
to investigate whether positive affect impacts time preference, where time preference
denotes a preference for present over future utility (Frederick et al 2002). Our results
indicate that mild positive affect significantly reduces subjects’ time preference.
This is important in economics, which has a large literature devoted to modeling and
estimating parameters of time preference (Andersen et al 2008; Benhabib and Bisin 2005;
Coller and Williams 1999; Coller et al 2003; Gul and Pesendorfer 2001; Harrison and
Lau 2005; Harrison et al 2002; Laibson 1997; Loewenstein and Prelec 1992;
O’Donoghue and Rabin 1999; Rubinstein 2003; Shapiro 2005; Thaler and Shefrin 1981;
and Warner and Pleeter 2001). Further, time preference has been shown to be
commonplace, is believed to lead to self-control problems, and may increase the
likelihood of negative outcomes—for example, overconsumption, obesity, addiction,
reduced human capital accumulation, and diminished retirement saving (Ameriks et al
2007; Benhabib et al 2007; Bernheim and Rangel 2004; Frederick et al 2002; and Kirby
et al 1999). Further, Mischel et al (1989) found that preschoolers’ ability to delay
gratification was a strong predictor of SAT scores over a decade later.
Initial empirical evidence for our hypothesis comes from the General Social Survey
(GSS), which in three waves (1973, 1974, and 1976) includes both a measure of selfreported
happiness (“Taken all together, how would you say things are these days—
would you say that you are very happy, pretty happy, or not too happy?”) and time
preference (agreement or disagreement with the statement, “Nowadays, a person has to
Electronic copy available at: http://ssrn.com/abstract=1491247
2
live pretty much for today and let tomorrow take care of itself.”). We find that happier
respondents are less likely to agree with the “live for today” statement than are less happy
respondents. This holds even after controlling for covariates that have been shown to be
related to happiness (see Appendix A for regression results). While this empirical
evidence provides loose support for our result, it has a number of shortcomings. Most
notably, the direction of causation cannot be identified. Further, this measure of time
preference is imperfect at best. While in this paper we will study the impact of mild
positive affect—rather than happiness—on time preference, our methodology will
address both of these shortcomings.1
To explore the relationship between mild positive affect and time preference, we run a
random-assignment experiment in which subjects’ time preference is measured after their
mood has been manipulated. Specifically, subjects are randomly assigned to the
treatment group—positive affect—or the control group—neutral affect. The appropriate
affect is induced using short video clips. Time preference is measured using a standard
matching procedure, in which subjects report the present value of a future payment and
truthful responses are incentivized.
Our results indicate that subjects in the treatment group exhibit significantly lower time
preference than do subjects in the control group. This result is robust to various
specification checks. The next section of this paper provides a brief overview of the
literature; the third and fourth sections describe the experimental design and the
econometric specifications, respectively; and the fifth and sixth sections present and
discuss the results.
Literature Review
The impact of mild positive affect on decision making has been extensively studied over
the past 20 years. Researchers, most notably Alice Isen, have conducted randomassignment
experiments which demonstrate that mild positive affect can have beneficial
1 The degree to which one can generalize from our result to the effect of happiness on time preference is
explored in the discussion.
3
effects on decision making, increasing cognitive flexibility, work effort, helpfulness, and
creativity (recent reviews include Isen 2008; and Lyubomirsky et al 2005). Moreover,
mild positive affect does not appear to impede decision making as many would expect it
to. For example, there is no experimental evidence that mild positive affect causes
individuals to be impulsive, thoughtless, or overly optimistic (Isen 2007).
Research that explores the impact of positive affect on self-control is most relevant to our
study, since the psychological concept of a lack of self-control overlaps with the
economic concept of time preference. For example, Tice and Bratslavsky (2000)
reported that psychologists generally understand a lack of self-control to be distinguished
by a proclivity toward short-term gains, even in the face of long-term costs. Three recent
psychological studies are of particular interest: First, Baumeister et al (1998) found that
mild positive (negative) affect increases (decreases) the proportion of time subjects spent
studying for a non-incentivized test in the laboratory; second, Isen and Reeve (2005)
demonstrated that mild positive affect increases intrinsic motivation for interesting tasks;
and third, Isen (2007) reported that mild positive affect replenishes will power.2 These
results are taken to indicate that mild positive affect increases forward-looking thinking,
the ability to stay on task, and self-control. In summary, this line of research is far
smaller and less well corroborated than other research regarding the impact of mild
positive affect on decision making, leading Isen and Reeve (2005) to conclude that, “the
topic of the role of positive affect in development of self-control seems a promising one
for investigation.”
A related strand of economic literature considers the effect of self-control on subjective
well-being. This research attempts to identify the impact of poor self-control by
comparing the happiness of individuals who make decisions that are thought to result
from a lack of self-control to those who do not make such decisions. For example, Frey
et al (2007) found that heavy TV watchers report being less happy; and Stutzer (2007)
2 Six more studies considering the effect of affect on self-control have been conducted exclusively with
young children (Fry 1975; Mischel et al 1968; Mischel et al 1973; Schwarz and Pollack 1977; Seeman and
Schwarz 1974; and Underwood et al 1973). For developmental reasons, the results of these papers cannot
be generalized to adults (Bucciol et al 2009).
4
found that obesity reduces the subjective well-being of individuals who report having
limited self-control. Further, Gruber and Mullainathan (2005) found that cigarette taxes
increase happiness among individuals prone to smoke. The authors interpreted the tax as
a self-control device, which suggests that a lack of self-control may cause unhappiness.
These studies appear to indicate that a lack of self-control reduces happiness. By
extension this suggests that time preference may affect positive affect. While this
direction of causation—the opposite of that which will be tested in this study—may hold,
it is outside of the scope of the paper to explore this possibility.3
Finally, this research also adds to a small literature regarding the determinants of time
preference. The two most well-established results in this literature are that (a) the
magnitude of the future payment, and (b) the length of time over which discounting
occurs are important determinants of time preference (Kirby and Marakovic 1996; and
Thaler 1981). More recently, it has been shown (a) that high cognitive load increases
time preference (Benjamin et al 2006; and Hinson et al 1999); and (b) that individuals
with greater cognitive skills, as measured by IQ tests, exhibit lower time preference
(Burks et al 2009). Thus, this study also adds to the literature by identifying a new
determinant of time preference.
Experimental Design
We examined the effect of positive affect on time preference in a laboratory experiment
conducted at Santa Clara University. In brief the experimental procedure was as follows
(additional details are provided below): First, subjects read and signed an informed
consent form. Second, subjects were instructed (a) that they would be answering 30
time-preference questions, (b) that their payment would be based on one of these
questions, (c) that the payment question would be determined randomly at the end of the
session, and (d) that a mechanism would be used to provide an incentive for truthful
responses. Third, the mood inducement procedure was administered. Fourth, subjects
3 An additional paper, Wertheim and Schwarz (1983), found that self-reported depression was correlated
with a preference for immediate (delayed) benefits (costs). This study is correlational and does not provide
any evidence regarding a causal relationship in either direction.
5
answered the 30 time-preference questions. Fifth, subjects answered questions regarding
their subjective well-being. Sixth, payments were determined. Seventh, subjects
answered questions regarding their mood and demographic characteristics. In total, the
experiment lasted approximately 45 minutes, and subjects received an average of $24 for
their participation.
Participants
Sixty-nine undergraduate students were recruited from introductory English courses that
all Santa Clara students are required to take; these courses were chosen in an attempt to
avoid potential disciplinary bias.
Participant instructions
After completing the informed consent form, subjects received detailed written
instructions, which were also read aloud by the experimenter. The instructions
introduced the time-preference questions and informed the subjects that the experimental
procedure had been designed to encourage truthful responses. Specifically, subjects were
told that the 30 time-preference questions would be of the following form: “What amount
of money, $p, if paid to you today would make you indifferent to $m paid to you in t
days?” Five values of m, {$11.34, $18.31, $24.28, $32.84, $51.71}, and six values of t,
{1 day, 3 days, 7 days, 14 days, 28 days, 56 days}, were used. We chose abstruse values
for m to discourage the use of heuristics, such as answering one-half of m for each p. The
questions were randomized so that there was no pattern in the order of the values of m or
t; this was done to avoid anchoring. The time-delay, t, was not extended beyond two
months so that all possible payouts would occur within the academic term. Also, values
of t were chosen to avoid weekends and school holidays.
Subjects were informed that their payment would be based on one of the 30 timepreference
questions, that the payment-question would be determined randomly at the
end of the session, and thus, that they should answer each question as if their payment
depended on it. Subjects were then introduced to the Becker-DeGroot-Marschak (BDM)
mechanism and informed that it would be used to provide an incentive for truthfully
6
answering the 30 time-preference questions (Becker et al 1964; and Benhabib et al
2007).4 Finally, subjects were guided through numerous examples in an attempt to
ensure that they understood the process.5
Subjects were also informed of the payment-pickup process, which was designed
carefully in an attempt to prevent subjects from preferring payment on the day of the
session simply because it was easier or less risky. Specifically, subjects were informed
that they would have to walk across campus to retrieve their payment regardless of the
payment date; the session was conducted on the south end of campus, and the paymentpickup
location was on the north end of campus. This was done in an attempt to ensure
that decisions were not impacted by “transaction costs” or relative doubt about the
legitimacy of future pickups.6 Next, subjects were informed of the anonymous and blind
nature of the payment-process. They were told (a) that one person would prepare the
payment-envelopes, (b) that a second person would distribute the sealed paymentenvelopes,
(c) that neither would know their identify, only their subject identification
number, and (d) that the envelope-distributor would not know the payment amount. This
procedure was implemented in an attempt to minimize the potentially confounding role of
subject-experimenter reciprocity.
Mood-inducement procedure
We attempted to manipulate subjects’ mood by showing them a short film clip. The use
of film clips to induce moods is common in psychological and, increasingly, economic
4 Using the BDM mechanism, the amount and timing of the payment were determined as follows: m balls
numbered 1 through m were placed in a spinner, and one was chosen. If the number R on the drawn ball
was less than or equal to p, then the subject was paid $m in t days. If the number drawn exceeded p, the
subject was paid $R on the day of the experimental session. Finally, note that since the values of m were
fractional, the number of balls placed in the spinner was rounded to the next highest integer. For example,
if m = $11.34, then balls numbered 1 through 12 were placed in the spinner.
5 Subjects did not answer the 30 time-preference questions at this time. Rather, they answered them
immediately following the mood inducement procedure. This order of events eliminated the possibility that
the induced mood could have been moderated, if not nullified, by reading the participant instructions.
6 These same two concerns were addressed in Andreoni and Sprenger (2010) in a different manner. To
equalize “transaction costs,” the authors only used subjects who had a campus mail address and delivered
all payments through the campus mail system. Further, they made an arrangement with the campus mail
staff to ensure that all payments arrived on the scheduled date. To minimize concerns about not getting
paid, subjects received Professor Jim Andreoni’s business card and were encouraged to contact him if there
was any problem with their payment. Finally, subjects were paid with a personal check from Professor Jim
Andreoni that was drawn from the university credit union.
7
experiments (Gross and Levenson 1995; Kirchsteiger et al 2006; Oswald et al 2008; and
Rottenberg et al 2007). Westerman et al (1996) evaluated eleven mood-inducement
procedures. They found that the use of film or story was the most effective means of
inducing positive affect.
In our experiment, half of the subjects (34 of 69) were randomly assigned to the treatment
group and watched a film clip intended to induce positive affect. The other half of the
subjects (35 of 69) were assigned to the control group and watched a film clip intended to
induce neutral affect. Except for the variant film clip, the experimental protocol was
identical for the control and treatment groups.
Our choice of film clips followed Gross and Levenson (1995). There, over 200 film clips
were evaluated for their efficacy in inducing each of seven different affects. Nearly 500
subjects were asked to rate on a scale from 0 to 8 the greatest intensity of each affect they
felt during the course of the film clip. The strongest affect felt during the clip was
dubbed the target affect for that film. The authors then reported the two most effective
clips for inducing each of the affects.
In our experiment, the positive-affect film clip was a short montage of stand-up comedy
bits from the 2002 “Robin Williams – Live on Broadway.” This choice followed directly
from Gross and Levenson (1995), in which one of the two most successful positiveaffect-
inducing film clips was Robin Williams’s 1986 “A Night at the Met.” We opted
for the 2002 montage primarily because it was more current and put to rest any concerns
of the humor being out-dated. The neutral-affect film clip was also one commonly used
by psychologists and featured tranquil images of landscapes and wildlife in Denali
National Park, Alaska (for example, Rottenberg et al 2007).
Time-preference questions and completing the session
Immediately following the mood-inducement procedure, subjects answered the 30 timepreference
questions. Then, subjects answered questions about their subjective wellbeing.
Next, the payment question was determined, and the BDM mechanism was
8
implemented. Finally, subjects answered questions regarding their mood and
demographic characteristics. When all surveys were completed, subjects received
instructions for retrieving their payments and exited the experimental session; the
instructions indicated the payment amount, and when and where to pick up the paymentenvelope.
Econometric Specifications
In analyzing the relationship between positive affect and time preference, we consider a
model of the form:
p = βH + δ mIM (m)
m Σ
+ α
t IT (t)
t Σ
+ φ mt IM (m) × IT (t) + λKkIK (k) +ε
k Σ
K Σ
t Σ
m Σ
(1)
where H is a dummy for the positive-affect treatment, M is the set of all future payment
amounts m, T is the set of all time delays t (in days), and p is the subjective present value
of $m in t days. The indicator functions IM(m) and IT(t) take the value of one for a given
m and t, respectively, and zero otherwise; this specification allows for all possible linear
and interacted effects of time-delay and future payments on discounting. The model also
includes demographic controls. K denotes the set of controls, and for a given control, k is
the set of possible values. For example, for K = gender, k ∈{male, female}, and
IGender(male) equals one if the subject is male, and zero otherwise.
In this specification, the amount $p necessary today to be indifferent to $m in t days is
considered to be a function of m, t, and m*t; that is, the regression is fully saturated with
respect to the effect of time and future payments on present value. The appeal of this
econometric specification is that it imposes minimal ex ante structure and does not
depend on the distribution of p. Thus, this parametrization does not restrict us to a
specific model of discounting and fits the conditional expectation function for p perfectly
(Angrist and Pischke 2009).
Our choice of the fully-saturated model follows from the underwhelming support we
found for the exponential, fixed-cost, and hyperbolic models of discounting, which we
9
estimated and tested using the techniques described in Benhabib et al (2007).7 Those
authors also rejected the exponential and hyperbolic models, though they found support
for fixed-cost discounting.
Finally, OLS is used to estimate equation (1). Corrected standard errors are calculated by
clustering observations by subject; this is necessary since subjects answered 30 questions
and their responses are unlikely to be independent.
Results
Demographic characteristics of subjects
There is no significant difference between the values of any of the demographic
characteristics for the treatment and control groups, so random assignment is valid ex
post (see Table 1). Roughly half of the subjects are female. The division of students
across colleges roughly mimics the university’s population. Santa Clara University has
three colleges: arts and sciences, business, and engineering. Only a small number of
subjects are in the engineering college, the rest are evenly split between arts and sciences
and business. Roughly half of the subjects report that they practice a religion. Given that
Santa Clara is a Jesuit university, it is not surprising that the most heavily represented
religion is Christianity, accounting for three-quarters of the subjects. Almost all other
subjects consider themselves Atheists, with only two students identifying a different
religion. The next demographic factor we control for is race. Almost all of the subjects
consider themselves white (63%), Asian (20%), or Hispanic (14%), with only two
subjects identifying with other racial/ethnic categories. Finally, subjectswere asked to
give their best estimate of their family income. The mean response falls in the $100,000 -
$150,000 category, while the median is in $150,000 - $200,000 category. Few students
report family income of less than $80,000 or greater than $500,000.
The measure of long-term happiness comes from the question, “Taken all together, how
would you say things are these days—would you say that you are…,” where possible
7 Please contact the authors for complete results.
10
responses ranged from 1 (completely unhappy) to 7 (completely happy). This measure is
similar to the ones used in the GSS and the World Values Survey (WVS), each of which
has been used extensively in the happiness-economics literature as a measure of longterm
happiness. The GSS, however, uses a 3-point scale and the WVS a 4-point scale.
We expanded the scale to 7 points to increase sensitivity. This question was asked after
mood-inducement and the time-preference questions, but because of its broad scope, is
hypothesized to be unaffected by mood-inducement (Kahneman et al 1999). Indeed,
there is no statistical difference in happiness for the treatment and control groups. Mean
responses, reported with other descriptive statistics in Table 1, are indistinguishable (ttest
p-value = 0.923), as is the response-distribution (chi-square p-value = 0.353). Thus,
it appears that long-term happiness is independent of short-term affective shocks.
Subjects exhibit time preference
As in Benhabib et al (2007), discounting will be represented by D, the factor that when
multiplied by m yields the dollar amount p necessary today to make one indifferent to $m
in t days: D = p/m. Discounting may itself depend on the levels of m and t, so D = D(m,
t). Without this structure, comparisons of p across different values of m would be
meaningless, since p’s domain is defined by m. This structure will be relinquished in the
regression analysis, where equation (1) will be estimated using the absolute level of p as
the dependent variable. Table 2 presents observed values of D for all (m, t)-
combinations; panel (a) pools all subjects, and panels (b) and (c) present the treatment
and control groups, respectively.
Subjects consistently discount the distant future more heavily than they do the near
future. Holding m constant, D clearly trends downward in all panels; this is visualized in
Figure 1(a) for the pooled data. In panels (a) and (c) of Table 2, there are few exceptions
to the downward trend. The monotonicity of the negative relationship between D and t is
violated more frequently in the positive-affect treatment (see panel (b)), but in general the
relationship holds.
11
The relationship between future payment m and the discount factor D, holding t constant,
does not follow so consistent a pattern. For virtually no time-delay t is there
monotonicity in D with respect to m. Nor is there a clear non-monotonic trend (for
example, in panel (a), for t = 1, D has two interior peaks, for t = 14, there appears to be an
inverted U-shape, and for t = 56, a U-shape).
Mood-inducement is successful
Mood-inducement had the intended effect on affect. In the final questionnaire before
leaving the experimental session, subjects were asked whether watching the film clip
made them happier, sadder, or neither happier nor sadder. Controlling for payment,
watching the positive-affect clip made subjects significantly more likely to answer
happier.8
Positive affect reduces time preference
Preliminary support for positive affect reducing time preference can be seen in panel (d)
of Table 2, which reports the difference between DTreatment (m, t) and DControl (m, t). This
difference is positive for each (m, t)-combination; the positive-affect inducement
unambiguously reduces time preference. This is visualized in panel (b) of Figure 1,
which separately illustrates the discount factors for the control and treatment groups over
time-delay t; the average discount factor for each group is emphasized.
Estimating equation (1) without demographic controls, we see further evidence that
positive affect reduces time preference. Controlling for future-payment and time-delay,
watching a positive- instead of neutral-affect film clip increases the present value of
future payments by $2.20 on average (see column (1) of Table 3). This result is
marginally statistically significant, with a p-value = 0.054. Further, the magnitude is
8 We included this simple check to confirm that the film clips had the intended impact; however, the use of
film clips in general and our choice of clips in particular is well-established and -tested in the literature
(Gross and Levenson 1995; Rottenberg et al 2007; and Westerman et al 1996). Thus, the affect check in
corroboration with prior research provides strong evidence of successful mood-inducement. Our affect
check does, however, suffer from the following shortcoming: it was administered after subjects answered
the 30 time-preference questions and after the payments were determined. The alternative—administering
the affect check immediately following the film clip—was even less attractive, since the induced affect
could have been dampened, or nullified, by the affect check itself (Isen and Gorgoglione 1983).
12
quite substantial. Even for the largest value of m, $51.71, the increase in the present
value of the future payment is over four percent. Since we are not studying the effects of
time and monetary sums on time preference, the coefficients of the m and t dummies and
their interactions are not presented.
In column (2), controls are added for gender, race, family income, religion, regular
religious service attendance, and college. Treatment is now statistically significant,
increasing present value by $2.01 on average. In columns (3) and (4), the analyses of
columns (1) and (2), respectively, are repeated controlling for self-reported, long-term
happiness. The main effect is stable regardless of the specification; thus, the results
appear to be robust. Finally, we estimate equation (1) conditional on p < m; that is, we
exclude observations in which subjects did not discount. With the restricted sample, the
magnitude and the significance of the treatment effect increase (see Columns (5) through
(8) of Table 1). The positive-affect inducement now increases the present value of a
future payment by an average of $3.00 in each of the four restricted specifications.
In summary, our experiment demonstrates that mild positive affect significantly increases
the present value of a future payment, a result that is robust to a number of econometric
specification-checks. The magnitude of the increase depends on the value of the future
payment and the specification, ranging from 4% of the future payment (specification (4),
m = $51.71) to 27% (specification (8), m = $11.56).
Discussion
Our findings indicate that mild positive affect significantly reduces time preference, that
is, increases the present value of a future payment. Since this result is derived from a
random-assignment experiment, the relationship between positive affect and time
preference is presumably causal with positive affect reducing time preference. Below,
we discuss, in turn, mechanisms that may explain the result and its external validity.
13
Potential mechanisms
A possible economic explanation of our result is that positive affect substitutes for
material income today. At first glance, this may seem plausible, but not upon closer
inspection. Our measure of time preference is the amount of money today that produces
indifference to a future payment. According to the hypothesis above, positive affect
would substitute for money today; hence subjects in the positive-affect treatment should
require less money today to be indifferent to a future payment, assuming neutralized
expected future affect. However, subjects in the positive-affect group require more, not
less, money today to be indifferent to a future payment.
The psychological literature, where the bulk of affect-research has been reported,
discusses two additional potential mechanisms that may explain why mild positive affect
decreases time preference. First, Isen (2008) states that mild positive affect increases
cognitive flexibility by broadening focus and attention, promoting openness to
information, and enabling improved integration of information. Thus, the observed
reduction in time preference may result from a more thorough consideration of broader
(including future) net benefits. In addition, support for this mechanism comes from what
the psychological literature dubs the “dopamine hypothesis”: that the release of dopamine
(the neurotransmitter linked to rewards) in areas of the brain responsible for cognitive
flexibility is the mediating factor in the effect of positive affect on behavior (Ashby et al
1999; and Isen 2008).
Second, the depletion and/or replenishment of will power may help explain our result.
As mentioned in the literature review, mild positive affect has been shown to replenish
will power (Isen 2007). Further, evidence abounds that exercising self-control negatively
impacts one’s ability to exert self-control in subsequent tasks (see Vohs and Heatherton
2000). Thus, if watching the neutral-affect clip required more will power than watching
the positive-affect clip, then the neutral-affect inducement should deplete subjects’ will
power more than the positive-affect inducement. Consequently, post-inducement,
subjects in the treatment group should have their will power replenished and subjects in
14
the control group should have their will power depleted. This might explain why subjects
in the treatment group exhibit greater patience for a future payment.
External validity
In this section, we will discuss what can and cannot be extrapolated from our result and
then turn to policy implications. To understand what can be extrapolated, it is important
to recognize that the effects of affect do not lie on a continuum from intense positive
affect to intense negative affect. In particular, the “continuum” is violated in three
significant ways. First, the effect of intense positive affect is not necessarily an
amplification of the effect of mild positive affect (Isen 2007). Thus, we cannot assume
that intense positive affect would result in a further reduction of time preference. Second,
negative affect does not necessarily have the opposite effect of positive affect. The
independence of positive and negative affect is well-established in psychology (Bradburn
1969; Diener and Emmons 1984; Isen 2007; Lyubomirsky et al 2005; and Watson et al
1988). Moreover, neural research suggests the two are unrelated. Each is associated with
a distinct neurotransmitter: dopamine with positive affect and serotonin with negative
affect. In other words, it should not be inferred from our result that negative affect would
necessarily increase time preference. Third, two affects may have variant effects, even if
they are both positive or negative. For example, anger and sadness, while both negative
affects, may have different effects on decision making. In general, negative affects are
more likely to have variant effects and positive affects are less likely to have variant
effects (Isen 1987).9 Thus, our result may not be specific to the particular positive affect
induced—amusement—and may apply to a wide range of positive affects.
We now turn to the relationship between positive affect and happiness, that is, the
relationship between being in a happy mood and being happy. In our experiment, we
successfully improved subjects' mood, but were unable—not surprisingly—to change
subjects' underlying, long-term happiness. Thus, can we validly claim that happier
individuals should exhibit lower time preference?
9 We are considering the effect of negative affects on time preference in a separate paper, specifically
focusing on the effects of sadness, anger, and fear. Of these treatments, we have already conducted the
sadness-inducement and are now turning to the others.
15
The distinction between positive affect and happiness, and what this distinction means for
studying the impact of happiness, have been considered by Konow and Earley (2008) and
Lyubomirsky et al (2005). In the latter, four possible strategies for identifying the impact
of happiness are discussed. The least valid is a correlational study due to the inherent
difficulty of demonstrating causation. Conversely, the ideal would be a randomassignment
experiment, in which underlying happiness is manipulated. Such an
experiment, however, is neither feasible nor ethical. Hence, Lyubomirsky and her
colleagues argue that the best feasible approach is a random-assignment experiment in
which positive affect is induced. They argue that such an approach is a valid way to
identify the impact of being happy since happier individuals experience more frequent
positive affect (Diener and Seligman 2002; and Lyubomirsky et al 2005). In fact, Diener,
Sandvik, and Pavot (1991) state that, “frequent positive affect is both necessary and
sufficient to produce the state we call happiness, whereas intense positive experience is
not.” Thus, our result suggests that happier individuals should exhibit less time
preference.
Even though frequent mild positive affect likely has important long-term consequences, a
natural question arises: How long does a one-time affective shock last? Isen et al (1976)
and Isen and Gorgoglione (1983) estimate that the positive affect induced by a free gift,
or by reading statements with a positive valence, lasts between 10 and 20 minutes.
Further, Isen and Gorgoglione (1983) state that positive affect induced by film may last
longer.
While the effect of a one-time affective shock may be brief, there are still important
policy implications that arise from our research. For example, our affect-inducement
procedure mimics an “every day experience,” namely, viewing enjoyable or amusing
audio-visual media. The ubiquity of such media—as well as other stimuli that may
induce mild positive affect—implies that individuals’ time preference may be perturbed
regularly. Further, individuals may not even be aware of the affective shocks, let alone
the impact that such shocks are having on decision making. For example, in the post16
experiment survey no subject suspected a link between the film clip and the timepreference
questions. Lastly, an important implication for other researchers is that affect
should be neutralized at the beginning of time preference experiments. Indeed,
uncontrolled affect may be partially responsible for the highly variant estimates of
discount rates reported in the time preference literature (Frederick et al 2002).
Interestingly, our result implies that others can use affective shocks to affect our time
preferences. In fact, the popular book Nudge argues that the government should act with
“libertarian paternalism” when behavior can be easily manipulated for the better (Thaler
and Sunstein 2008); clearly, such tactics may also be used by parties with less benevolent
motives.
Finally, evidence from the positive psychology literature suggests that people can
increase their own positive affect (Snyder and Lopez 2005). Insofar as self-induced
positive affect has the same influence on time preference as a positive affective shock,
our results suggest that people may be able to control their own time preference, which
could lead to improved decision-making.
In summary, our experiment adds to the mounting evidence that positive affect influences
behavior in economically important ways. However, we can only speculate upon how
this effect comes to be. In addition, this study adds to the behavioral economics literature
by identifying a determinant of time preference.
17
References
Ameriks, John, Andrew Caplin, John Leahy, and Tom Tyler. 2007. “Measuring Self-
Control Problem.” American Economic Review, 97(3): 966-72.
Andersen, Steffen, Glenn W. Harrison, Morten I. Lau, and E. Elisabet Rutstrom. 2008.
“Eliciting Risk and Time Preferences.” Econometrica, 76(3): 583-618.
Andreoni, James, and Charles Sprenger. 2010. “Risk Preferences Are Not Time
Preferences.” Working Paper.
Angrist, Joshua D., and Jorn-Steffen Pischke. 2009. Mostly Harmless Econometrics: An
Empiricist’s Companion. Princeton: Princeton University Press.
Ashby, F. Gregory, Alice M. Isen, and And U. Turken. 1999. “A Neuropsychological
Theory of Positive Affect and its Influence on Cognition.” Psychological Review, 106(3):
529-50.
Baumeister, Roy E., Bratslavsky, Ellen, and Dianne M. Tice. 1998. “Procrastination and
Affect Regulation.” Manuscript in preparation.
Becker, Gordon M., Morris H. Degroot, and Jacob Marschak. 1964. “Measuring utility
by a single-response sequential method.” Behavioral Science, 9(3): 226-32.
Benhabib, Jess, and Alberto Bisin. 2005. “Modelling Internal Commitment Mechanisms
and Self-Control: A Neuroeconomics Approach to Consumption-Saving Decision.”
Games and Economic Behavior, 52(2): 460-92.
Benhabib, Jess, Alberto Bisin, and Andrew Schotter. 2007. “Present-bias, Quasi-
Hyperbolic Discounting, and Fixed Costs.” Working paper.
Benjamin, Daniel J., Sebastian A. Brown, and Jesse M. Shapiro. 2006. “Who is
'Behavioral'? Cognitive Ability and Anomalous Preferences.” SSRN Working Paper No.
675264.
Bradburn, Norman M. 1969. The Structure of Psychological Well-Being. Chicago: Aldine
Publishing Company.
Bucciol, Alessandro, Daniel Houser, and Marco Piovesan. 2009. “Temptation at Work:
A Field Experiment on Willpower and Productivity.” George Mason University,
Interdisciplinary Center for Economic Science Working Paper 1013.
Bernheim, Douglas, and Antonio Rangel. 2004. “Addiction and Cue-Conditioned
Decision Processes.” American Economic Review, 94(5): 1558-90.
18
Burks, Stephen V., Jeffrey P. Carpenter, Lorenz Goette, and Aldo Rustichini. 2009.
“Cognitive Skills Affect Economic Preferences, Strategic Behavior, and Job
Attachment.” Proceedings of the National Academy of Sciences, 106(19): 7745-50.
Coller, Maribeth, Glenn W. Harrison, and E. Elisabet Rutstrom. 2003. “Are Discount
Rates Constant? Reconciling Theory and Observation.” Working Paper 3-31, Department
of Economics, College of Business Administration, University of Central Florida.
Coller, Maribeth, and Melonie B. Williams. 1999. “Eliciting Individual Discount Rates.”
Experimental Economics, 2(2): 107-27.
Cryder, Cynthia E., Jennifer S. Lerner, James J. Gross, and Ronald E. Dahl. 2008.
“Misery is not Miserly: Sad and Self-Focused Individuals Spend More.” Psychological
Science, 19(6): 525-30.
Davis, James Allen, and Tom W. Smith. 2007. General Social Surveys, 1972-2006
[machine readable data file] /Principal Investigator, James A. Davis; Director and Co-
Principal Investigator, Tom W. Smith; Co-Principal Investigator, Peter V. Marsden;
Sponsored by National Science Foundation. NORC ed. Chicago: National Opinion
Research Center [producer]; Storrs, CT: The Roper Center for Public Opinion Research,
University of Connecticut [distributor].
Diener, Ed, and Robert A. Emmons. 1984. “The Independence of Positive and Negative
Affect.” Journal of Personality and Social Psychology, 47(5): 1105-17.
Diener, Ed, Ed Sandvik, and William Pavot. 1991. “Happiness Is the Frequency, Not the
Intensity, of Positive Versus Negative Affect.” In Subjective Well Being: An
Interdisciplinary Perspective, ed. Fritz Strack, Michael Argyle, and Norbert Schwarz,
119-40. Oxford: Pergamon Press.
Diener, Ed, and Martin E.P. Seligman. 2004. “Beyond Money: Toward an Economy of
Well-Being.” Psychological Science in the Public Interest, 5(1): 1–31.
Erez, Amir, and Alice M. Isen. 2002. “The Influence of Positive Affect on the
Components of Expectancy Motivation.” Journal of Applied Psychology, 87(6): 1055-67.
Frederick, Shane, George Loewenstein, and Ted O’Donoghue. 2002. “Time Discounting
and Time Preference: A Critical Review.” Journal of Economic Literature, 40(2): 351-
401.
Frey, Bruno S., Christine Benesch, and Alois Stutzer. 2007. “Does watching TV make us
happy?” Journal of Economic Psychology, 28(3): 283-313.
Fry, P.S. 1975. “Affect and Resistance to Temptation.” Developmental Psychology,
11(4): 466-72.
19
Gross, James J., and Robert W. Levenson. 1995. “Emotion Elicitation Using Films.”
Cognition and Emotion, 9(1): 87-108.
Gruber, Jonathan, and Sendhil Mullainathan. 2005. "Do Cigarette Taxes Make Smokers
Happier." Advances in Economic Analysis & Policy, Advances in Economic Analysis &
Policy, 5(1): article 4.
Gul, Faruk, and Wolfgang Pesendorfer. 2001. “Temptation and Self-Control.”
Econometrica, 69(6): 1403-35.
Harrison, Glenn W., and Morten Igel Lau. 2005. “Is the Evidence for Hyperbolic
Discounting in Humans Just an Experimental Artefact?” Behavioral & Brain Sciences,
28(5): 657.
Harrison, Glenn W., Morten I. Lau, and Melonie B. Williams. 2002. “Estimating
Individual Discount Rates for Denmark: A Field Experiment.” American Economic
Review, 92(5): 1606-17.
Hinson, John M., Tina L. Jameson, and Paul Whitney. 2003. “Impulsive decision making
and working memory.” Journal of Experimental Psychology: Learning, Memory, and
Cognition, 29(2): 298-306.
Isen, Alice M. 2008. “Some Ways in Which Positive Affect Influences Decision Making
and Problem Solving.” In Handbook of Emotions, ed. Michael Lewis, Jeannette M.
Haviland-Jones, and Lisa Feldman Barrett, 548-73. New York: The Guilford Press.
Isen, Alice M. 2007. “Positive Affect, Cognitive Flexibility, and Self-Control.” In
Persons in Context: Building a Science of the Individual, ed. Yuichi Shoda, Daniel
Cervone, and Geraldine Downey, 130-47. New York: The Guilford Press.
Isen, Alice M., Margaret Clark, and Mark F. Schwartz. 1976. “Duration of the Effect of
Good Mood on Helping: Footprints on the Sands of Time.” Journal of Personality and
Social Psychology, 34(3): 385-93.
Isen, Alice M., and Nehemia Geva. 1987. “The Influence of Positive Affect on
Acceptable Level of Risk: The Person with a Large Canoe has a Large Worry.”
Organizational Behavior and Human Decision Processes, 39(2): 145-54.
Isen, Alice M., and Joyce M. Gorgoglione, 1983. “Some Specific Effects of Four Affect-
Induction Procedures.” Personality and Social Psychology Bulletin, 9(1): 136-43.
Isen, Alice M., Thomas E. Nygren, and Gregory F. Ashby. 1988. “Influence of Positive
Affect on the Subjective Utility of Gains and Losses: It is just not Worth the Risk.”
Journal of Personality and Social Psychology, 55(5): 710-17.
20
Isen, Alice M., and Johnmarshall Reeve. 2005. “The Influence of Positive Affect on
Intrinsic and Extrinsic Motivation: Facilitating Enjoyment of Play, Responsible Work
Behavior, and Self-Control.” Motivation and Emotion, 29(4): 295-323.
Kahneman, Daniel, Ed Diener, and Norbert Schwarz. 1999. Well-being: The Foundations
of Hedonic Psychology. New York: The Russel Sage Foundation.
Kirby, Kris N., and Nino Marakovic. 1996. “Delay-Discounting Probabilistic Rewards:
Rates Decrease as Amounts Increase.” Psychonomic Bulletin & Review, 3(1): 100-4.
Kirby, K.N., Nancy M. Petry, and Warren K. Bickel. 1999. “Heroin Addicts Have Higher
Discount Rates for Delayed Rewards Than Non-Drug-Using Controls.” Journal of
Experimental Psychology: General, 128(1): 78-87.
Kirchsteiger, Georg, Luca Rigotti, and Aldo Rustichini. 2006. “Your morals might be
your moods.” Journal of Economic Behavior and Organization, 59(2): 155-72.
Konow, James, and Joseph Earley. 2008. “The Hedonistic Paradox: Is Homo
Economicus Happier?” Journal of Public Economics, 92(1): 1-33.
Laibson, David. 1997. “Golden Eggs and Hyperbolic Discounting.” Quarterly Journal of
Economics, 112(2): 443-77.
Lerner, Jennifer S., Deborah A. Small, and George Loewenstein. 2004. “Heart Strings
and Purse Strings.” Psychological Bulletin, 15(5): 337-341.
Loewenstein, George, and Jennifer S. Lerner. 2003. “The Role of Affect in Decision
Making”. In Handbook of Affective Sciences, ed. Richard J. Davidson, Klaus R. Scherer,
and H. Hill Goldsmith, 619-42. New York: Oxford University Press.
Loewenstein, George, and Drazen Prelec. 1992. “Anomalies in Intertemporal Choice:
Evidence and Interpretation.” Quarterly Journal of Economics, 107(2): 573-97.
Lyubomirsky, Sonja, Laura King, and Ed Diener. 2005. “The Benefits of Frequent
Positive Affect: Does Happiness Lead to Success?” Psychological Bulletin, 131(6): 803-
55.
Mischel, Walter, Brian Coates, and Antonette Raskoff. 1968. “Effects of Success and
Failure on Self-Gratification.” Journal of Personality and Social Psychology, 10(4): 381-
90.
Mischel, Walter, Ebbe B. Ebbesen, and Antonette Raskoff Zeiss. 1973. “Selective
Attention to the Self: Situational and Dispositional Determinants.” Journal of Peronality
and Social Psychology, 27(1): 129-42.
21
Mischel, Walter, and Ralph Metzner. 1962. "Preference for Delayed Reward as a
Function of Age, Intelligence, and Length of Delay Interval", Journal of Abnormal and
Social Psychology, 64(6): 425-31.
Mischel, Walter, Yuichi Shoda, and Monica L. Rodriguez. 1989. “Delay of Gratification
in Children.” Science, 244(4907): 933-8.
O'Donoghue, Ted, and Matthew Rabin. 1999. “Doing It Now or Later.” American
Economic Review, 89: 103-24.
Oswald, Andrew, Eugenio Proto, and Daniel Sgroi. 2008. “Happiness and Productivity.”
Working paper.
Rottenberg, Jonathan, Rebecca D. Ray, and James J. Gross. 2007. “Emotion Elicitation
Using Films.” In The Handbook of Emotion Elicitation and Assessment, ed. James A.
Coan and John J. B. Allen, 9-28. London: Oxford University Press.
Schwarz, J. Conrad, and Pamela R. Pollack. 1977. “Affect and Delay of Gratification.”
Journal of Research in Personality, 11(2): 147-64.
Seeman, Gloria, and J. Conrad Schwartz. 1974. “Affective State and Preference for
Immediate versus Delayed Reward.” Journal of Research in Personality, 7(4): 384-94.
Snyder, C. R., and J. Lopez, ed. 2005. Handbook of Positive Psychology. Oxford:
Oxford University Press.
Rubinstein, Ariel. 2003. “Economics and Psychology? The Case of Hyperbolic
Discounting.” International Economic Review, 44(4):1207-16.
Shapiro, Jesse M. 2005. “Is There a Daily Discount Rate? Evidence from the Food Stamp
Nutrition Cycle.” Journal of Public Economics, 89: 303-25.
Stutzer, Alois. 2007. “Limited Self-Control, Obesity, and the Loss of Happiness” IZA
Discussion Paper No. 2925.
Thaler, Richard H. 1981. “Some Empirical Evidence on Dynamic Inconsistency.”
Economics Letters, 8: 201-7.
Thaler, Richard H., and Hersh M. Shefrin. 1981. “An Economic Theory of Self Control.”
Journal of Political Economy, 89(2): 392-406.
Thaler, Richard H., and Cass R. Sunstein. 2008. Nudge: Improving Decisions About
Health, Wealth, and Happiness. New Haven: Yale University Press.
22
Tice, Dianne M, and Ellen Bratslavsky. 2000. “Giving in to Feel Good: The Place of
Emotion Regulation in the Context of General Self-Control.” Psychological Inquiry,
11(3): 149-59.
Underwood, Bill, Bert S. Moore, and D. L. Rosenhan. 1973. “Affect and Self-
Gratification.” Developmental Psychology, 8(2): 209-14.
Vohs, Kathleen D. and Todd F. Heatherton. 2000. “Self-Regulatory Failure: A Resource-
Depletion Approach.” Psychological Science, 11(3): 249-54.
Watson, David, Lee Anna Clark, and Auke Tellegen. 1988. “Development and
Validation of Brief Measures of Positive and Negative Affect: The PANAS Scales.”
Journal of Personality and Social Psychology, 54(6): 1063-70.
Wertheim, Eleanor H., and Schwarz, J. Conrad. 1983. “Depression, Guilt, and Self-
Management of Pleasant and Unpleasant Events.” Journal of Personality and Social
Psychology, 45(4): 884-9.
Warner, John T., and Saul Pleeter. 2001. “The Personal Discount Rate: Evidence from
Military Downsizing Programs.” American Economic Review, 91(1): 33-53.
Westerman, Rainer, Kordelia Spies, Gunter Stahl, and Friedrich W. Hesse. 1996.
“Relative Effectiveness and Validity of Mood Induction Procedures.” European Journal
of Social Psychology, 26(4): 557-80.
23

CAVEAT:  TABLES AND CHARTS HAVE BEEN OMITTED WE APOLOGISE WITH THE AUTHORS