Buying time promotes happiness

Ashley V. Whillansa,1, Elizabeth W. Dunnb, Paul Smeetsc, Rene Bekkersd, and Michael I. Nortona
a
Harvard Business School, Harvard University, Cambridge, MA 02163; bDepartment of Psychology, University of British Columbia, Vancouver, BC, Canada
V6T 1Z4; cDepartment of Finance, Maastricht University, 6200 MD Maastricht, The Netherlands; and dCenter for Philanthropic Studies, Vrije Universiteit
Amsterdam, 1081 HV Amsterdam, The Netherlands

Edited by Susan T. Fiske, Princeton University, Princeton, NJ, and approved June 13, 2017 (received for review April 19, 2017)

Around the world, increases in wealth have produced an unin- sample of working Americans living in the United States (n =
tended consequence: a rising sense of time scarcity. We provide 1,260), adults in Denmark (n = 467), and Canada (n = 326), and
evidence that using money to buy time can provide a buffer both a nationally representative sample (n = 1,232) and a sample
against this time famine, thereby promoting happiness. Using of millionaires (n = 818) in The Netherlands. See Table 1 for
large, diverse samples from the United States, Canada, Denmark, sample demographics. In all samples, respondents completed two
and The Netherlands (n = 6,271), we show that individuals who questions about whether—and how much—money they spent
spend money on time-saving services report greater life satisfac- each month to increase their free time by paying someone else to
tion. A field experiment provides causal evidence that working complete unenjoyable daily tasks. In addition, respondents rated
adults report greater happiness after spending money on a time- their satisfaction with life (SWL) and reported their annual
saving purchase than on a material purchase. Together, these results household income, the number of hours they work each week, age,
suggest that using money to buy time can protect people from the marital status, and the number of children living at home (SI
detrimental effects of time pressure on life satisfaction. Appendix). In the Canadian and Dutch surveys (n = 2,376), re-
spondents also completed a measure of time stress (4), allowing us
time | money | happiness | well-being to test the prediction that buying time mitigates the negative ef-

SOCIAL SCIENCES
fects of time stress on life satisfaction.
Here we report the meta-analytic effects across samples (16);
I n recent decades, incomes have risen in many countries (1, 2),
potentially exacerbating a new form of poverty: from Germany
to Korea to the United States, people with higher incomes report
results for individual samples are provided in Fig. 1 and SI Ap-
pendix. Across samples (n = 4,469), 28.2% of respondents spent
greater time scarcity (3). Feelings of time stress are in turn linked money to buy themselves time each month [meanamount =
to lower well-being, including reduced happiness, increased anxi- $147.95 US dollars (USD) for respondents who reported buying
ety, and insomnia (4–6). Time stress is also a critical factor un- time]. Respondents who spent money in this way reported
derlying rising rates of obesity: lacking time is a primary reason greater life satisfaction, d = 0.24, P < 0.001, 95% CI (0.18, 0.31).
that people report failing to eat healthy foods or exercise regularly This relationship was positive within each sample and reached
(7, 8). In theory, rising incomes could offer a way out of the “time statistical significance for the nationally representative sample of
famine” of modern life (9), because wealth offers the opportunity working Americans, adults in Canada and Denmark, and mil-
to have more free time, such as by paying more to live closer to lionaires in The Netherlands (Fig. 1). This effect held controlling
work. However, some evidence suggests that wealthier people for our key set of covariates (n = 3,983), d = 0.22, P < 0.001, 95%
spend more time engaging in stressful activities, such as shopping CI (0.15, 0.29) and was not moderated by income or wealth,
and commuting (10). Experimental research shows that simply Z = −0.35, P = 0.729, 95% CI (−0.08, 0.06): people from across
leading people to feel that their time is economically valuable the income spectrum benefitted from buying time. These results
induces them to feel that they do not have enough of it (11). also held when we controlled for an alternative set of covariates
A great deal of attention has been devoted to reducing financial where we replaced household income with log income and added
scarcity, but there is relatively little rigorous research examining an age-squared variable (SI Appendix, Tables S6–S23b). These
how to reduce feelings of time scarcity, which in fact may offer a results provide initial evidence for a robust link between buying
particularly difficult challenge given that time, unlike money, is time and life satisfaction across diverse samples.
inherently finite. Could allocating discretionary income to buy free
time—such as by paying to delegate common household chores, Significance
like cleaning, shopping, and cooking— reduce the negative effects
of the modern time famine, thereby promoting well-being? The Despite rising incomes, people around the world are feeling in-
growth of the sharing economy has made time-saving services in- creasingly pressed for time, undermining well-being. We show
creasingly accessible, but no empirical research has tested whether that the time famine of modern life can be reduced by using
using such services enhances happiness. money to buy time. Surveys of large, diverse samples from four
From our theoretical perspective, buying time should protect countries reveal that spending money on time-saving services is
people from the negative impact of time stress on life satisfaction. linked to greater life satisfaction. To establish causality, we show
This conceptualization draws on the social support literature, in that working adults report greater happiness after spending
which research on the “buffering hypothesis” has demonstrated that money on a time-saving purchase than on a material purchase.
receiving social support can protect people from experiencing the This research reveals a previously unexamined route from
negative consequences of stress (12). That is, the typical relationship wealth to well-being: spending money to buy free time.
between stress and reduced well-being is attenuated for individuals
who are able to access social support (13–15). We suggest that Author contributions: A.V.W., E.W.D., and M.I.N. designed research; P.S. contributed to
buying time may provide an alternate mechanism to receiving the the design of studies 5 and 6; P.S. and R.B. collected data for studies 5 and 6; A.V.W.
support needed to cope with daily demands, such that the re- analyzed the data; and A.V.W., E.W.D., and M.I.N. wrote the paper.
Downloaded at Indonesia: PNAS Sponsored on October 24, 2020

lationship between time stress and reduced life satisfaction should The authors declare no conflict of interest.
be attenuated among people who use money to access more time. This article is a PNAS Direct Submission.
Freely available online through the PNAS open access option.
Results 1
To whom correspondence should be addressed. Email: awhillans@hbs.edu.
As an initial test of this hypothesis, we surveyed Mechanical Turk This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
workers in the United States (n = 366), a nationally representative 1073/pnas.1706541114/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1706541114 PNAS | August 8, 2017 | vol. 114 | no. 32 | 8523–8527

 Table 1. Demographic characteristics of survey respondents in studies 1–6
US Mechanical US Denmark Canadian Netherlands Netherlands
Sample Turk representative adults adults representative millionaires US Qualtrics

n 366 1,260 467 326 1,232 818 1,802

Median (range), SWL 6.00 (0–10) 7.00 (0–10) 8.00 (2–10) 7.50 (1.5–10) 8.00 (1–10) 8.00 (1–10) 7.00 (0–10)
% Buying time (1 = yes) 15.8 22.0 23.0 26.6 21.2 60.3 50.0*
Median amount spent/mo, USD† 61–80 201–300 114.6 75–150 149.30 213.29 80–99
% Female — 47 39 — 48 16 60
Median age, y 30 43 51–60 36 51 68 35–44
Median, household income 30–35K 75–85K 57K 56–74K 11K 879K 60–75K
or wealth, USD (mean = 160K) (mean = 2.4M)
Median (range) no. children 1 (1–6) 1 (0–6+) — 0 (0–3) 0 (0–3) 0 (0–11) 1 (1–4+)
living at home
Median (range) no. hours worked 40 (0–100) 40 (1–90) — 40 (0–100) 37 (0–40+) — 40 (0–120)
at main job each week
% Married or partnership 51 66 62 55 53 76 68

All money reported in USD. We used a currency converter tool (OANDA) to estimate the amount in USD based on March 2017 exchange rates. Thus, the
USD values for the Denmark and The Netherlands samples are estimates that are subject to historical fluctuations.
*This study used a broader definition.
†
This represents median money spent for respondents who buy time.

Based on the buffering hypothesis, we predicted that buying assessed respondents’ spending on groceries as an index of
time would protect people from the negative impact of time stress nondurable spending (17) and we assessed respondents’ spend-
on life satisfaction. To examine this hypothesis, we entered time- ing on material and experiential purchases (18). If respondents
saving purchases, time stress, and a Purchase × Time Stress in- who spend money on time-saving purchases are happier only
teraction into a regression predicting life satisfaction. Across because they have more discretionary income, then controlling
samples, there was a significant interaction between time-saving for these other spending indicators should eliminate the bene-
purchases and time stress, Z = 3.85, P < 0.001, 95% CI (0.06, ficial effect of time-saving purchases.
0.20). Deconstructing this interaction, time stress was associated Faced with our broader definition of time-saving purchases,
with lower life satisfaction among respondents who did not spend 50% of respondents reported spending money in this way each
money on time-saving purchases (n = 1,504), B = −0.18, Z = 8.93, month (meanamount = $80–$99 USD for respondents who
P < 0.001, 95% CI (−0.22, −0.14). In contrast, for respondents who reported buying time). In this study, the majority of respondents
spent money on time-saving purchases (n = 804), the negative reported spending money to buy themselves out of cooking,
effect of time pressure on life satisfaction was relatively weak, shopping, and household maintenance.
B = −0.03, Z = 1.46, P = 0.144, 95% CI (−0.08, 0.01). These results Next, we tested our main hypotheses, which we preregistered
suggest that using money to buy time indeed buffers people from through the Open Science Framework (OSF; https://osf.io/fvpg2/).
the negative effects of time stress on well-being. Consistent with our previous findings, respondents who spent
Presenting all respondents with identical spending questions money on time-saving purchases reported greater life satisfaction,
allowed us to document comparable results across diverse sam- β = 0.24, P < 0.001. These results were unchanged, controlling for
ples. However, a limitation of this approach is that our results our predetermined set of covariates, β = 0.23, P < 0.001. These
may depend on the wording of this specific question. To over- results also held controlling for how much money respondents
come this limitation, we presented a new sample of working spent on nondurable expenses (groceries), material goods, and
adults in the United States (n = 1,802) with a broader definition experiential purchases, β = 0.15, P < 0.001.
of time-saving purchases to encompass any way in which re- In addition, we found evidence for our preregistered buffering
spondents could spend money that would provide more free time hypothesis: entering time-saving purchases, time stress, and their
(SI Appendix). Although the results above suggest that total in- interaction into a regression predicting life satisfaction revealed
come does not drive our results, it is possible that the decision to a significant Purchase × Time Stress interaction, B = 0.22, P <
spend money on time-saving purchases might reflect, in part, 0.001. Deconstructing this interaction, time stress was associated
respondents’ level of discretionary income. As a result, we also with lower life satisfaction among respondents who did not spend
money on time-saving purchases (n = 901), B = −0.17, P < 0.001,
95% CI (−0.25, −0.10). For respondents who spent money on
time-saving purchases (n = 901), the negative effect of time
pressure on life satisfaction was not significant, B = 0.05, P =
d score of outsourcing predicting SWL

0.9

0.227, 95% CI (−0.03, 0.12).

0.7
0.36 0.41 Across seven studies with over 6,000 respondents, spending
0.5
0.36
0.29
0.20
money to buy time was linked to greater life satisfaction, and the
0.2
0.11 typical, detrimental effect of time stress on life satisfaction was
0.3 0.05
0.06
attenuated among individuals who used money to buy time. We
0.1
suggest that this broad correlational link stems in part from the
cumulative day-to-day benefits that are caused by the reductions in
Downloaded at Indonesia: PNAS Sponsored on October 24, 2020

-0.1 time stress that such purchases provide. To document this causal
pathway, we conducted a 2-wk within-subjects experiment, exam-
-0.3
United States United States Denmark Canadian Netherlands Netherlands ining whether spending money on time-saving services caused
Mturk Representative Panel Adults Representative Millionaires
reductions in time pressure and improvements in daily mood.
Fig. 1. The effect of time-saving purchases on life satisfaction across Using a within-subjects design, we recruited working adults
studies 1–6. from Vancouver, Canada to spend two payments of $40 on two

8524 | www.pnas.org/cgi/doi/10.1073/pnas.1706541114 Whillans et al.

 consecutive weekends (n = 60). On one weekend, participants Indirect Effect: 0.11(0.06) [0.02, 0.24]
were randomly assigned to spend $40 on a purchase that would
Subjective Time
save them time. On the other weekend, to control for the ex- B = -0.74** Pressure B = -0.15**
perience of receiving and spending a windfall, participants were
assigned to spend $40 on a material purchase. We chose this
comparison because material purchases are unlikely to save time,
which was confirmed by our manipulation checks (SI Appendix, Time-saving Positive
Purchase Affect
Additional Methodological Details & Results for study 8). We B = 0.29**
counterbalanced the order of the spending weeks: results did not B = 0.12, ns
differ by order. After making each purchase, participants re-
Note. All B’s represent unstandardized regression coefficients obtained through bootstrapping
ceived a phone call at 5:00 PM and reported their feelings of using 10,000 resamples. The range in brackets represents the 95% CI of the indirect effect.
positive affect, negative affect, and time stress on that day.
We preregistered our hypotheses through OSF (https://osf.io/ambdk/). †p < .10, *p < .05, **p < .01
We hypothesized that people would report greater daily well-being Fig. 2. The effect of time-saving purchases on end-of-day positive affect
after making a time-saving (vs. material) purchase, and that these through time pressure.
benefits would be explained by reduced feelings of time-stress that day.
Consistent with our preregistered hypotheses, a paired-samples
t test showed that participants reported greater end-of-day positive within-subjects experiment, making a time-saving purchase caused
affect after making a time-saving purchase (mean = 4.00, SD = improvements in daily mood.
0.64) than after making a material purchase (mean = 3.71, SD = Why does buying time promote happiness? Our experiment
0.81), t(59) = 2.57, P = 0.007, 95% CI (0.10, 0.48), d = 0.33. After provides the clearest window into this process, by demonstrating
making a time-saving (vs. material) purchase, participants also that people felt less end-of-day time pressure when they purchased
reported lower levels of negative affect, t(59) = −2.45, P = 0.009, time-saving services, which explained their improved mood that

SOCIAL SCIENCES
95% CI (−0.45, −0.09), d = 0.33 and lower feelings of time stress, day. According to the broaden-and-build theory (20), improve-
t(59) = −2.76, P = 0.004, 95% CI (−1.18, −0.29), d = 0.36 (Table 2). ments in daily mood should promote greater life satisfaction over
Moreover, these results could not be explained by other purchase time. Consistent with this idea, our correlational studies show that
characteristics, such as the extent to which the purchases were ex- people who spend money on time-saving purchases report greater
ceptional, useful, or high in status (SI Appendix, Table S26). life satisfaction. Importantly, time pressure had little or no nega-
We also predicted that the beneficial effects of time-saving tive effect on life satisfaction for individuals who used money to
purchases on daily mood would be mediated by reduced feelings buy time. Taken together, our findings suggest that using money to
of time stress on that day. Our preregistered within-subjects buy time may reduce feelings of time pressure on a given day and
mediational analyses (19) showed that the relationship between provide a cumulative benefit by serving as a buffer against the
time-saving purchases and daily mood was indeed explained by deleterious effects of time pressure on overall life satisfaction.
reductions in perceived time stress (Fig. 2). That is, participants At first glance, it may seem surprising that there was no direct
reported higher positive affect after making time-saving pur- relationship between buying time and feelings of time pressure in
chases, B = 0.29, SE = 0.11, P = 0.013, 95% CI (0.06, 0.52), and our survey studies. Although we detected this link in our experi-
significantly lower feelings of time pressure after making time- mental study, this direct relationship should be difficult to observe
saving purchases, B = −0.74, SE = 0.27, P = 0.008, 95% CI in correlational designs: buying time should reduce time stress, but
individuals with higher levels of preexisting time stress should also
(−1.27, −0.20). After controlling for time stress, time-saving
be more likely to buy time, potentially cancelling out the overall
purchases were no longer a significant predictor of positive af-
relationship between these variables. The identical pattern has
fect, B = 0.18, SE = 0.11, P = 0.108, 95% CI (−0.04, 0.41). Upon
been observed in research on social support: social support re-
testing the significance of the indirect effect using bootstrap
duces feelings of stress, but individuals experiencing higher levels
estimation with 10,000 samples, the indirect coefficient was sig-
of stress are more likely to seek out social support (12). Fur-
nificant, B = 0.11, SE = 0.06, 95% CI (0.02, 0.24). In sum, time- thermore, in correlational research, there has been substantial
saving compared with material purchases increased positive support for the buffering hypothesis: social support has been
affect by reducing feelings of time stress; the same pattern of found to moderate the association between stress and well-being.
results held for negative affect (SI Appendix, Figs. S2 and S3). Among individuals with relatively little social support, stress tends
Discussion to show a negative association with well-being. Among those with
higher levels of social support, stress tends to have a weak or zero
Across several distinct samples, including adults from Canada, association with well-being (12–14). Our research therefore pro-
the United States, Denmark, The Netherlands, and a large vides evidence that a similar buffering process occurs in the con-
sample of Dutch millionaires, buying time was linked to greater text of support purchased through the market economy.
life satisfaction. These results held controlling for a wide range Buying time may serve as a buffer against the negative effects of
of demographics, as well as for the amount that respondents time stress in part by enhancing perceived control. People often
spent on groceries and material and experiential purchases each complain of being in a time bind not only because they are objectively
month. These results were not moderated by income, suggesting busy, but also because they perceive a lack of control over their time
that people from various socioeconomic backgrounds benefit from (21, 22). Ironically, spending too much money on time-saving services
making time-saving purchases. Furthermore, in a preregistered could undermine perceptions of personal control, by leading people
to infer that they are unable to handle any daily tasks, potentially
reducing well-being. Consistent with this possibility, we found ex-
Downloaded at Indonesia: PNAS Sponsored on October 24, 2020

Table 2. Mean (SD) for end-of-day mood after time-saving and

ploratory evidence that the beneficial relationship between buying
material purchases (n = 60 within-subjects study)
time and life satisfaction may be curvilinear, reversing at the highest
Affect measure Time-saving Material levels of spending on time-saving purchases (SI Appendix, Table S2).
Future research could also explore the boundaries of the re-
End-of-day negative affect 1.38 (0.43) 1.65 (0.75)
lationship between buying time and life satisfaction across the in-
End-of-day perceived time pressure 3.49 (1.64) 4.22 (1.72)
come spectrum. Across studies, we found no consistent evidence

Whillans et al. PNAS | August 8, 2017 | vol. 114 | no. 32 | 8525

 that the benefits of buying time are limited to relatively wealthy Study 1. We first conducted an initial exploratory study assessing the re-
people. If anything, within our United States samples, we observed lationship between buying time and SWL with Americans recruited from
a stronger relationship between buying time and life satisfaction Amazon’s Mechanical Turk (n = 366). Respondents completed the two life
satisfaction items (see Table 1 for reliabilities). Respondents then completed
among less-affluent individuals (SI Appendix, Table S5). Our
the two time-saving purchase questions of interest, demographic items, and
studies include relatively few people at the lowest rungs of the several tertiary measures (SI Appendix, Table S1).
income spectrum, however, leaving open the supposition that the
benefits of buying time would not emerge for individuals struggling Study 2. We examined the relationship between buying time and SWL in a
to meet their own basic needs. representative sample of working Americans living in the United States through
Despite the potential benefits of buying time, many respondents the GfK Knowledge Networks Panel. Over an 11-d fielding period, 1,275 re-
allocated no discretionary income to buying time, even when they spondents completed our survey. Fifteen respondents did not complete our key
could afford it: just under half of the 818 millionaires that we variables of interest; thus, our primary analyses are based on 1,260 respondents.
surveyed spent no money outsourcing disliked tasks. Our initial Panel members respond to an average of two online surveys per month and
surveys used a narrow definition of buying time (“outsourcing receive small cash rewards and prizes for survey completion (www.gfk.com).
GfK uses equal probability sampling to recruit potential panel members by mail
disliked tasks”), but even when we broadened our definition in our
and phone and provides respondents in noninternet households with free in-
preregistered survey study, half of our respondents still reported ternet access. This allows GfK to recruit a statistically representative sample of
not using any money on “purchases that save time.” Of course, the United States population. We selectively recruited GfK respondents who
some participants may have failed to accurately remember or reported being working and who were 19 y of age or older at the time of
construe how their recent purchases might have saved them time. completing the initial GfK demographic profile. Consequently, our respondents
Therefore, we asked an additional sample of 98 working adults approximate a representative sample of working adults over 19 in the United
how they would spend a windfall of $40 (using the identical States. It is worth noting that some respondents may not have had US citi-
prompt and the identical participant population from our exper- zenship, and that this study did not include younger or unemployed individuals.
iment), and asked them to describe the intention of the purchase. Respondents first reported their SWL. Respondents then completed the two
buying-time questions and the key demographic covariates of interest (SI Ap-
Only 2.0% spontaneously reported that they would make a time-
pendix, Table S1). Respondents also completed measures outside the scope of
saving purchase (see SI Appendix, Additional Results for study 9). the current investigation (see ref. 26 for more information).
These low rates may further vary as function of culture and gender.
Within many cultures, women may feel obligated to complete Study 3. We then examined the relationship between buying time and SWL in
household tasks themselves, working a “second-shift” (23) at home, a sample of Danish adults. These data were collected by the Happiness Re-
even when they can afford to pay someone to help. In recent decades, search Institute in Copenhagen. In this survey, respondents first reported
women have made gains, such as improved access to education, but their SWL using items from studies 1 and 2. Next, respondents completed our
their life satisfaction has declined (24); increasing uptake of time- buying-time questions, demographics, and other tertiary questions (SI Ap-
saving services may provide a pathway toward reducing the harmful pendix, Table S1).
effects of women’s second-shift. To this end, some organizations have
begun to reward employees with just such time-saving services. As Study 4. We also examined the relationship between buying time and SWL
among a sample of working adults from Vancouver, Canada. We recruited
part of a recent initiative, Stanford University conducted a small pilot
participants from public places, including train stations and public parks.
study in which doctors were rewarded with vouchers for time-saving Consistent with study 1, we targeted 300 participants and recruited more
services. Doctors who received these vouchers reported better work– participants than planned because the research assistants worked set data-
life balance (25) and retention rates increased, suggesting that orga- collection schedules. As discussed in text, in studies 4–6 we also included a
nizations may benefit from rewarding employees with time. More measure of time stress. This allowed us to examine whether buying time
broadly, in the face of an increasing time famine, organizations and helped to protect people from the impact of time stress on SWL. Respon-
policymakers could move beyond their focus on promoting financial dents completed the identical SWL items from our previous studies. Next,
affluence to promoting time affluence as well. respondents reported on their feelings of time stress (4). Respondents then
completed the two buying-time questions of interest. Respondents also
Materials and Methods provided demographic information (SI Appendix, Table S1).
Studies 1–6. Is buying time linked to happiness? To provide an initial test of this
question, we conducted six correlational studies with large and diverse samples, Study 5. Studies 5 and 6 were collected as part of a larger study examining
including Americans recruited from Amazon’s Mechanical Turk (study 1), a philanthropy in The Netherlands (27). In this study, we used data from a
representative sample of working Americans living in the United States (study 2), nationally representative sample of Dutch adults. This sample was recruited
adults living in Denmark (study 3), adults living in Canada (study 4), a repre- via TNS NIPO, one of the leading survey agencies in The Netherlands (27).
sentative sample of adults living in The Netherlands (study 5), and a large sample In this study, respondents completed the one-item Cantril Ladder. Next,
of millionaires (study 6). Below we provide additional details about each study. participants reported their feelings of time stress (4). Respondents then com-
See SI Appendix, Table S1 for more information about the measures used in each pleted several measures that are outside the scope of the current investigation
study. See SI Appendix, Table S2 for the regression results reported separately for and completed the two key buying-time questions. At the end of the survey,
each study. Across studies, unless otherwise noted, respondents rated their life respondents answered our key demographic questions of interest.
satisfaction using the identical two items. First, respondents answered the
question, “Taking all things together, how happy would you say you are?” on a Study 6. We also recruited a sample of high net-worth Dutch adults relying on a
scale from 0 = Not at all to 10 = Extremely. Next, respondents completed the database constructed by Elite Research based on public records (28). The
Cantril Ladder. For this question, respondents were asked to report where they methods and questions used in study 6 were identical to those used in study 5.
currently stand in life on a ladder spanning from the worst possible to the best
possible life imaginable (0 = Bottom Rung to 10 = Top Rung). We selected these Study 7. After completing studies 1–6, we conducted a preregistered study
questions because they are brief measures that are used extensively in large-scale examining the relationship between time-saving purchases and SWL (https://
survey research (24). The datasets used in studies 4–6 included several items from osf.io/9kc9g/). We recruited working Americans over the age of 19 y through
a recently validated measure of time stress (4); participants rated their agreement Qualtrics, a professional survey company. Qualtrics oversampled to meet
with statements, such as “I feel pressed for time today” on a Likert scale from 1, their internal standards for data quality. Because Qualtrics’ internal standards
Downloaded at Indonesia: PNAS Sponsored on October 24, 2020

Strongly Disagree to 7, Strongly Agree. To capture this construct in our confir- were not part of our preregistered inclusion criteria, we used the full sample
matory studies, we included the top-three highest loading items from a more that met our preregistered inclusion criteria (n = 1,802). First, respondents
frequently used measure, the time affluence subscale of the Material Affluence completed the identical SWL items from studies 1–5 (α = 0.85). Respondents
and Time Affluence scale (MATAS) (11). On the same seven-point scale, participants also completed three measures assessing time stress (α = 0.74). As a measure of
rated their agreement with statements such as “There have not been enough discretionary income, respondents were asked to report how much money
minutes in the day.” See SI Appendix, Table S3 for the specific items used they spent on groceries each week. Respondents then reported whether they
across studies. spent any money in a typical month on material purchases for themselves, on

8526 | www.pnas.org/cgi/doi/10.1073/pnas.1706541114 Whillans et al.

 experiences, and on time-saving purchases. We used the identical wording for the OSF. Because our hypotheses were directional, we planned to use one-
the material and experiential questions from previously published research tailed tests. All results also remain significant using two-tailed tests.
(18). We randomized the order of the material, experiential, and time-saving Measures. In this study, we assessed subjective well-being by asking participants
purchase questions. If respondents spent money in these categories in an aver- to complete the 12-item scale of positive and negative experience after making
age month, respondents were then asked how much money they spent in a typical their purchase each week (material purchase week positive affect: α = 0.88,
month on each type of purchase. Finally, respondents completed our demographic material purchase week negative affect: α = 0.88; time purchase week positive
questions-of-interest and completed questions for an unrelated study. The full affect: α = 0.86, time purchase week negative affect: α = 0.77). To assess
survey items for this study are available through the OSF (https://osf.io/vr9pa/). participants’ feelings of time affluence each week, we asked participants to
complete four items from the MATAS (material purchase week MATAS: α =
Study 8. We also examined whether buying time causally promoted happi- 0.88, time purchase Week MATAS: α = 0.87). As a manipulation check, we
ness. To examine this question, we recruited a sample of working adults from asked participants to report how many minutes that their purchases saved. We
Vancouver, Canada (n = 60). We preregistered our data-collection stopping rule, also asked participants to report on whether they felt as if their purchase had
methods, and a priori predictions through the OSF (https://osf.io/ambdk/). Dur- cost or saved time using one-item scale ranging from −3 = Cost a lot of time
ing 1 wk of the study, participants were instructed to spend a $40 payment on a overall to 3 = Saved a lot of time overall. We then asked participants to report
material purchase for themselves. During another week of the study, partici- how they had spent any free time that resulted from making each purchase. It
pants were instructed to spend a $40 payment on a purchase that would save is possible that the time-saving and material purchases could differ in various
them time. We counterbalanced the order of the spending weeks; results did ways. Thus, we asked participants to respond to a standard set of purchase
not differ by order. After making each purchase, participants received a phone characteristics from previously published research (18). Each week, participants
call at 5:00 PM, and they were asked to report on their feelings of positive were asked to report how exceptional, useful, helpful, fun, and high in social
affect, negative affect, and time-stress. As a manipulation check, we asked status each of the purchases were. Because time-saving purchases could result
participants to report how much time they had saved by making each purchase. in participants feeling higher in social status themselves, participants were also
Participants also completed several questions about their purchases each week. asked to rate their subjective social status each week.
At 9:00 AM on their scheduled spending day, participants received an
e-transfer from the research assistant in charge of the study. At 5:00 PM, a Study 9. We provided the experimental instructions to a new set of working
trained research assistant called the participant and asked them several adult participants (n = 98) who were recruited using the identical re-
questions about their positive and negative affect, their current feelings of

SOCIAL SCIENCES
cruitment strategies as in study 8. We asked participants to report what they
time affluence, and their spending experience. We also asked participants to would purchase if they received a $40 windfall in the upcoming week. We
email their receipts to our study team to confirm that they had spent in a way also asked participants to report on the intention of this purchase and on
consistent with their spending guidelines. Participants were not allowed to the recipient of this purchase. Two of the authors used participants’ re-
complete week 2 of the study if they did not correctly follow the week sponses to code whether each purchase could be defined as a material
1 instructions: eight participants spent incorrectly in week 1 and were ex- purchase, a time-saving purchase, an experiential purchase, or a prosocial
cluded. Of these eight participants, six participants failed to spend any money
purchase (each purchase could be classified as more than one type of pur-
on their scheduled spending day (four of the six participants were assigned to
chase). For the full results of this study, see SI Appendix, Additional Results.
the time-saving purchase) and two participants spent incorrectly (one in the
We obtained informed consent from all respondents before their par-
time-saving and one in the material purchase condition).
ticipation. These studies were approved by the research ethics boards at the
Sample size considerations. Because of budgetary constraints, we could only
University of British Columbia and the Harvard Business School. All study data
collect a maximum of 90 completed observations. This sample size would provide
and study materials are available through the OSF (https://osf.io/vr9pa/).
95% power to detect a small effect (d = 0.30) of purchase type on well-being.
Because of the high cost of this research ($80 per participant), we performed
sequential analyses. This procedure is common in medical research and allows ACKNOWLEDGMENTS. The authors thank Jessie Pow for her comments on
an earlier draft of this manuscript. The authors also thank Kjartan Sveistrup
researchers to examine the data as it is being collected, without inflating type
Andsbjerg and Meik Wiking at The Happiness Research Institute for providing
1 errors. We performed a one-sided interim analysis after collecting 60 partici- the data for study 3. A.V.W. received funding from the Public Scholar Initiative
pants. Using this approach, we collected 60 participants and assessed whether at the University of British Columbia, the Social Sciences and Humanities Re-
the key analyses fell below our preregistered boundary conditions of 0.0465/ search Council of Canada, the Society for Personality and Social Psychology
1.6794. Our interim analyses met this predetermined threshold. We terminated Heritage Foundation, and the Society for Personality and Social Psychology
data collection at n = 60, as per our preregistered stopping rule posted through Q&Pay Initiative.

1. Deaton A (2008) Income, health, and well-being around the world: Evidence from the 16. Lipsey MW, Wilson DB (2011) Practical Meta-Analysis (Sage, Thousand Oaks, CA).
Gallup World Poll. J Econ Perspect 22:53–72. 17. Headey B, Muffels R, Wooden M (2004) Money Doesn’t Buy Happiness or Does It? A
2. Easterlin RA, McVey LA, Switek M, Sawangfa O, Zweig JS (2010) The happiness- Reconsideration Based on the Combined Effects of Wealth, Income, and Consumption
income paradox revisited. Proc Natl Acad Sci USA 107:22463–22468. (IZA, Institute for the Study of Labor, Bonn, Germany) Discussion Paper no. 1218.
3. Hamermesh DS, Lee J (2007) Stressed out on four continents: Time crunch or yuppie 18. Van Boven L, Gilovich T (2003) To do or to have? That is the question. J Pers Soc
kvetch? Rev Econ Stat 89:374–383.
Psychol 85:1193–1202.
4. Kasser T, Sheldon KM (2009) Time affluence as a path toward personal happiness and
19. Montoya AK, Hayes AF (2017) Two-condition within-participant statistical mediation
ethical business practice: Empirical evidence from four studies. J Bus Ethics 84:243–255.
analysis: A path-analytic framework. Psychol Methods 22:6–27.
5. Roxburgh S (2004) ‘There just aren’t enough hours in the day’: The mental health
20. Fredrickson BL (2004) The broaden-and-build theory of positive emotions. Philos
consequences of time pressure. J Health Soc Behav 45:115–131.
Trans R Soc Lond B Biol Sci 359:1367–1378.
6. Höge T (2009) When work strain transcends psychological boundaries: An inquiry into
21. Goodin RE, Rice JM, Parpo A, Eriksson L (2008) Discretionary Time: A New Measure of
the relationship between time pressure, irritation, work-family conflict and psycho-
somatic complaints. Stress Health 25:41–51. Freedom (Cambridge Univ Press, United Kingdom).
7. Banwell C, Hinde S, Dixon J, Sibthorpe B (2005) Reflections on expert consensus: A 22. Schor J (2008) The Overworked American: The Unexpected Decline of Leisure (Basic
case study of the social trends contributing to obesity. Eur J Public Health 15:564–568. Books, New York).
8. Strazdins L, et al. (2011) Time scarcity: Another health inequality. Environ Plann A 43: 23. Croft A, Schmader T, Block K (2015) An underexamined inequality: Cultural and
545–559. psychological barriers to men’s engagement with communal roles. Pers Soc Psychol
9. Perlow LA (1999) The time famine: Toward a sociology of work time. Adm Sci Q 44:57–81. Rev 19:343–370.
10. Kahneman D, Krueger AB, Schkade D, Schwarz N, Stone AA (2006) Would you be 24. Stevenson B, Wolfers J (2009) The paradox of declining female happiness. Am Econ J
happier if you were richer? A focusing illusion. Science 312:1908–1910. Econ Policy 1:190–225.
11. DeVoe SE, Pfeffer J (2011) Time is tight: How higher economic value of time increases 25. Fassioto M (2016) Increasing joy in medicine. International Conference on Physician Health,
feelings of time pressure. J Appl Psychol 96:665–676. 10 September (American Medical Association, Canadian Medical Association, British Med-
12. Cohen S, Wills TA (1985) Stress, social support, and the buffering hypothesis. Psychol
ical Association, Boston).
Downloaded at Indonesia: PNAS Sponsored on October 24, 2020

Bull 98:310–357.
26. Whillans AV, Weidman AC, Dunn EW (2016) Valuing time over money is associated
13. DeLongis A, Folkman S, Lazarus RS (1988) The impact of daily stress on health and
with greater happiness. Soc Psychol Personal Sci 7:213–222.
mood: Psychological and social resources as mediators. J Pers Soc Psychol 54:486–495.
27. Bekkers R, Boonstoppel E, De Wit A (2017) Giving in The Netherlands Panel Survey -
14. Pow J, King DB, Stephenson E, DeLongis A (2017) Does social support buffer the ef-
fects of occupational stress on sleep quality among paramedics? A daily diary study. User Manual, (Center for Philanthropic Studies, Vrije Universiteit Amsterdam, Am-
J Occup Health Psychol 22:71–85. sterdam), Version 2.7.
15. Collins NL, Feeney BC (2000) A safe haven: An attachment theory perspective on support 28. Smeets P, Bauer R, Gneezy U (2015) Giving behavior of millionaires. Proc Natl Acad
seeking and caregiving in intimate relationships. J Pers Soc Psychol 78:1053–1073. Sci USA 112:10641–10644.

Whillans et al. PNAS | August 8, 2017 | vol. 114 | no. 32 | 8527