The Impact of Payment Frequency on Consumer Spending and Subjective Wealth Perceptions
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
The Impact of Payment Frequency on
Consumer Spending and Subjective Wealth
Perceptions
Downloaded from https://academic.oup.com/jcr/advance-article/doi/10.1093/jcr/ucab052/6373900 by guest on 30 December 2021
WENDY DE LA ROSA
STEPHANIE M. TULLY
Payment frequency is a fundamental yet underexplored feature of consumers’
finances. As higher payment frequencies are becoming more prevalent, consumers
are receiving more frequent yet smaller paychecks. An analysis of income and ex-
penditure data of over 30,000 consumers from a financial services provider demon-
strates a naturally occurring relationship between higher payment frequencies and
increased spending. A series of lab studies support this finding, providing causal
evidence that higher (vs. lower) payment frequencies increase spending. The effect
of payment frequency on spending is driven by changes in consumers’ subjective
wealth perceptions. Specifically, higher payment frequencies reduce consumers’
uncertainty in predicting whether they will have enough resources throughout a pe-
riod, increasing their subjective wealth perceptions. As such, situational factors that
reduce prediction uncertainty for those paid less frequently (e.g., the timing of con-
sumers’ expenses, income levels) moderate the impact of payment frequency. The
effects of payment frequency on subjective wealth and spending can occur even
when objective wealth favors those with lower payment frequencies. More broadly,
the current work underscores a need to understand how timing variations in con-
sumers’ income impact their perceptions, behaviors, and general well-being.
Keywords: payment frequency, subjective wealth, consumer spending, consumer
finance, uncertainty
P ayment frequency is a fundamental feature of consum-
ers’ finances. An increase in the number of people
who hold multiple jobs, lower payroll processing costs,
Wendy De La Rosa (wendyde@wharton.upenn.edu) is an Assistant
Professor of Marketing at The Wharton School, University of
and payroll technology advancements have made it in-
Pennsylvania, 3730 Walnut Street, Philadelphia, PA 19104, USA. creasingly common for consumers to receive more frequent
Stephanie M. Tully (smtully@stanford.edu) is an Assistant Professor of (albeit smaller) paychecks. Indeed, the percentage of US
Marketing and Business School Trust Faculty Scholar 2020-2021 at the
Stanford Graduate School of Business, Stanford University, 655 Knight
employers that will increase their employees’ payment fre-
Way, Stanford, CA 94305, USA. This research is based on the lead author’s quency is expected to quadruple to 20% by 2023 (Gartner
dissertation. Please address all correspondence to Wendy De La Rosa. The Research 2019). Nevertheless, despite this growing popu-
authors wish to thank Jennifer Aaker, Itamar Simonson, Jeffrey Pfeffer, larity, it is largely unknown how higher payment frequen-
Paul Brest, Dan Ariely, John Lynch, Jonathan Levav, Carey Morewedge,
Fred Feinberg, Broderick Turner, Kalinda Ukanwa, Jonathan Jean-Pierre, cies (i.e., smaller, more frequent paychecks) will impact
Martha Felipe, and the entire review team for their helpful feedback during consumers’ perceptions and behaviors compared to lower
the preparation of this article. Supplementary materials are included in the payment frequencies (i.e., larger, less frequent paychecks).
web appendix accompanying the online version of this article.
Prior research has demonstrated that higher payment fre-
quencies impact when consumers spend. Compared to
Editor: J. Jeffrey Inman
lower payment frequencies, higher payment frequencies
lead to more distributed consumption patterns throughout
Associate editor: Manoj Thomas the month (Berniell 2018; Parsons and Van Wesep 2013;
Shapiro 2005; Stephens 2003, 2006; Stephens and
Advance Access publication 22 September 2021
Unayama 2011). In the current work, we examine whether
VC The Author(s) 2021. Published by Oxford University Press on behalf of Journal of Consumer Research, Inc.
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (https://creativecommons.org/licenses/by-nc/4.0/), which permits
non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com
Vol. 00 2021
DOI: 10.1093/jcr/ucab052
12 DE LA ROSA AND TULLY
getting paid more frequently impacts consumers in ways Kamenica 2010). In the current work, we examine payment
that extend beyond merely the timing of their consumption. frequency, a ubiquitous aspect of consumers’ finances.
We suggest that higher payment frequencies lead to in-
creased spending. Specifically, we posit that higher pay- Payment Frequency and Spending
ment frequencies decrease consumers’ uncertainty in
Consumers’ income can vary across three dimensions:
predicting whether they will have enough resources
level (the amount of income earned), structure (the rate of
throughout a period, increasing consumers’ subjective
payment per unit of effort), and timing (any variations in
Downloaded from https://academic.oup.com/jcr/advance-article/doi/10.1093/jcr/ucab052/6373900 by guest on 30 December 2021
wealth perceptions. These higher subjective wealth percep-
the temporal patterns of when pay is disbursed to employ-
tions lead consumers to increase their spending.
ees for a given level and structure) (Parsons and Van
The current research uses real-world spending data and
Wesep 2013). We focus on an essential aspect of payment
controlled lab studies to examine the impact of payment
timing, which is payment frequency. Payment frequency
frequency on spending. First, we analyze banking transac-
does not refer to the frequency at which consumers earn
tions of more than 30,000 consumers from a financial serv- their income but rather the frequency at which they receive
ices provider. Using natural variation in payment their income. Traditionally, payment frequency has been
frequency, we find a naturally occurring relationship be- described as the recurring cycle by which employers pay
tween higher payment frequencies and increased spending their employees (e.g., weekly, bi-weekly, or monthly). In
(study 1). Next, we test the causal impact of payment fre- the current research, we define payment frequency from a
quency on spending using an online life simulation. consumer’s perspective, reflecting the number of times a
Consistent with the real-world spending analysis, we dem- consumer receives income within a given period. This
onstrate that higher payment frequencies lead to more more comprehensive definition allows payment frequency
spending than lower payment frequencies (study 2). We to reflect common payment frequencies when consumers
then show that the effect of payment frequency on spend- are paid cyclically by one employer (e.g., weekly, bi-
ing is driven by consumers’ subjective wealth perceptions weekly), as well as payment frequencies that may result
(study 2). These subjective wealth differences stem from from multiple sources of income, irregular work, or incon-
differences in consumers’ uncertainty in predicting sistent pay schedules (Golden 2015; Rothwell 2019; U.S.
whether they will have sufficient resources throughout a Bureau of Labor Statistics 2019). Since payment frequency
period (studies 3–5). As such, situational factors that re- does not impact consumers’ total income, those with lower
duce differences in prediction uncertainty, such as consum- payment frequencies have larger and less frequent pay-
ers’ income level (study 1) and the timing of consumers’ checks, whereas those with higher payment frequencies
expenses (study 4), moderate the effect of payment fre- have smaller and more frequent paychecks.
quency on spending and subjective wealth. Finally, we Recent employment and technological changes are mak-
demonstrate that the effects of higher payment frequencies ing the study of payment frequency increasingly important.
persist even when those with lower payment frequencies First, the rise of gig economy platforms has made it easier
are objectively more wealthy (studies 3 and 5) or when ac- for consumers to have multiple sources of income. Data
cess to higher payment frequency is optional, and consum- suggest that approximately a quarter of US workers have
ers must request additional paychecks (study 6). more than one job (Gallup 2018). That number increased
to 29% in 2019, reflecting a 21% increase (Rothwell
THEORETICAL FRAMEWORK 2019). As consumers increase their income streams and re-
ceive paychecks from multiple sources, their payment fre-
Prior research has demonstrated that consumers readily quency also increases. Second, the ability to transfer funds
attend to and automatically encode the frequency of events electronically and automatically has reduced some of the
(Hasher and Zacks 1979, 1984). Processing frequency in- cost and logistical considerations that previously dictated
formation seems to be a fundamental human ability, as employers’ payment frequency decisions (Stell 2016). In
even kindergartners have been shown to automatically en- addition, an increase in competition has led many payroll
code frequency information (Hasher and Zacks 1979). providers to decrease their pricing and eliminate the elec-
Frequency information can influence consumers’ judgment tronic transfer fees associated with making a payroll trans-
and decision-making, especially in complex situations fer (Wakefield Research 2019). Thus, employers can now
(Alba et al. 1994, 1999; Alba and Marmorstein 1987). For choose a wider variety of payment frequency options for
example, when product prices vary widely, the frequency their employees, including higher payment frequencies like
of price discounts influences consumers’ price estimates daily pay. For example, Walmart, the largest employer in
and their product choices (Alba et al. 1999). Thus, the fre- the United States, allows its workers to receive their in-
quency at which an event occurs may be particularly im- come daily (Corkery 2017). Given the greater flexibility of
portant in the financial domain, where consumers often employers to select payment frequencies based on factors
make difficult and complex decisions (Iyengar and other than logistical considerations, a natural question thatJOURNAL OF CONSUMER RESEARCH 3
arises is whether and how payment frequency impacts Sharma and Alter 2012), planning behavior (Fernbach et
consumers. al. 2015), spending and borrowing decisions (Ailawadi et
The existing literature on payment frequency primarily al. 2001; Karlsson et al. 2005; Sussman and Shafir 2012;
focuses on how payment frequency impacts the timing of Tully et al. 2015), and post-consumption behaviors like
expenditures (Berniell 2018; Parsons and Van Wesep word-of-mouth (Paley et al. 2019).
2013; Shapiro 2005; Stephens 2003, 2006; Stephens and
Unayama 2011). For example, Stephens and Unayama Payment Frequency and Subjective Wealth
(2011) demonstrated that Japanese retirees exhibited
Downloaded from https://academic.oup.com/jcr/advance-article/doi/10.1093/jcr/ucab052/6373900 by guest on 30 December 2021
greater consumption smoothing when they received one re- Although subjective wealth perceptions are typically im-
tirement paycheck every two months instead of one pay- pacted by the level of one’s financial resources, subjective
check every three months. In other words, consumers wealth perceptions are not simply a measure of objective
spread their expenditures throughout the year more evenly wealth (Gasiorowska 2014; Netemeyer et al. 2018; Sharma
when paid once every two months instead of every three and Alter 2012; Sussman and Shafir 2012; Tang et al.
months. In this work, rather than considering when con- 2004; Tully et al. 2015; Zauberman and Lynch 2005).
sumers spend, we examine whether payment frequency Indeed, consumers with objectively similar levels of finan-
impacts how much consumers spend. cial resources can vary in their perceptions of subjective
Some research suggests that payment frequency should wealth (Sussman and Shafir 2012). Instead, wealth percep-
not impact consumers’ spending. Such an outcome is in tions are a subjective assessment of the sufficiency of fi-
line with the Permanent Income Hypothesis (Friedman nancial resources relative to a benchmark, typically one’s
1957). The Permanent Income Hypothesis suggests that spending needs (Berman et al. 2016; Fernbach et al. 2015;
barring any liquidity constraints, consumers spend money Paley et al. 2019; Tully et al. 2015; Zauberman and Lynch
based on their expected lifetime earnings rather than cur- 2005). Although some research focuses on perceptions of
rent earnings. Thus, the Permanent Income Hypothesis financial oversufficiency (i.e., financial slack; Berman et
would argue that because payment frequency does not im- al. 2016; Zauberman and Lynch 2005) and other research
pact consumers’ total income, payment frequency should focuses on perceptions of financial insufficiency (i.e., fi-
not impact consumers’ spending. nancial constraints, financial deprivation; Fernbach et al.
Other research suggests that higher payment frequencies 2015; Paley et al. 2019; Sharma and Alter 2012; Tully et
should reduce consumers’ spending because getting paid al. 2015), in the current work, we refer to perceptions of
more frequently results in smaller amounts of money per sufficiency across the continuum as subjective wealth
paycheck. Research shows that consumers evaluate money perceptions.
and costs on a relative basis (Buechel and Morewedge In evaluating their subjective wealth, consumers often
2014; Kassam et al. 2011; Thaler 1985). More specifically, predict whether their financial resources are sufficient rela-
Morewedge, Holtzman, and Epley (2007) showed that con- tive to their spending needs (Berman et al. 2016; Fernbach
sumers spend less when thinking about a small account et al. 2015; Paley et al. 2019; Tully et al. 2015; Zauberman
(e.g., the money in their wallet) as compared to a large and Lynch 2005). Consumers can experience uncertainty
account (e.g., the money across their financial accounts) in making this prediction. Indeed, prior research has shown
because purchases made from a small account feel more that consumers often express uncertainty when predicting
expensive than purchases made from a large account. their future financial resources (Ben-David et al. 2018;
Thus, if consumers focus on their average paycheck size, Dominitz and Manski 1997a, 1997b). Such feelings of un-
which by definition decreases as one’s payment frequency certainty may be an important driver of subjective wealth
increases, then higher payment frequencies may make perceptions. Compared to those with lower prediction un-
costs feel larger and reduce one’s spending. certainty over their finances, consumers with higher pre-
Despite the aforementioned possibilities, in the current diction uncertainty have been shown to safeguard their
work, we argue that higher payment frequencies increase financial resources (Ben-David et al. 2018; Caldwell,
consumers’ spending by increasing consumers’ subjective Nelson, and Waldinger 2021) and believe they will need
wealth perceptions. Subjective wealth perceptions reflect €
more financial resources in the future (Ulkümen, Thomas,
consumers’ assessments about the sufficiency of their fi- and Morwitz 2008). These findings provide suggestive evi-
nancial resources. These perceptions are important for dence that greater prediction uncertainty decreases con-
researchers to understand as they predict a wide range of sumers’ subjective wealth perceptions.
outcomes (Ailawadi, Neslin, and Gedenk 2001; Fernbach, We suggest that because payment frequency inherently
Kan, and Lynch 2015; Frank 1999; Karlsson et al. 2005; changes the resource inflows and outflows that consumers
Paley, Tully, and Sharma 2019; Shah, Mullainathan, and experience throughout a period, payment frequency
Shafir 2012; Sharma and Alter 2012; Sussman and Shafir impacts consumers’ uncertainty over predicting their re-
2012; Tully, Hershfield, and Meyvis 2015). Wealth percep- source sufficiency and thus their subjective wealth percep-
tions impact consumers’ attention (Shah et al. 2012; tions. Consider the temporal patterns of consumers’4 DE LA ROSA AND TULLY
resources as a function of their income and expenses. If higher payment frequencies increase consumers’ per-
Consumers incur expenses very frequently, with the aver- ceptions of subjective wealth, then higher payment fre-
age consumer incurring approximately 70 expenses per quencies should also lead to more spending. Indeed,
month (Greene and Stavins 2018). Therefore, compared to consumers’ subjective wealth perceptions have been shown
those with higher payment frequencies, those with lower to predict their spending decisions, above and beyond their
payment frequencies experience larger and more frequent objective wealth (Karlsson et al. 2005). Thus, changes to
daily decreases in their overall resource levels, as expenses subjective wealth perceptions are likely to influence con-
occur very frequently with no income to offset them. For
Downloaded from https://academic.oup.com/jcr/advance-article/doi/10.1093/jcr/ucab052/6373900 by guest on 30 December 2021
sumers’ spending. More formally, we hypothesize:
example, a consumer with a weekly payment frequency
typically experiences a resource increase four times per H4: Higher payment frequencies will increase consumers’
month and a resource decrease on all other days when there spending compared to lower payment frequencies.
is an expense, resulting in a general pattern of resource H5: Subjective wealth perceptions will mediate the effect of
decumulation. In contrast, a consumer with a daily pay- higher payment frequencies (vs. lower payment frequencies)
ment frequency will experience smaller and less frequent on spending.
daily resource decreases as their income offsets expenses We test these hypotheses across six studies (and four
as they occur. This reduced pattern of resource decumula- supplemental studies in the web appendix). Data from
tion resulting from higher payment frequencies ought to re-
study 1 are proprietary and the legal data sharing agree-
duce consumers’ uncertainty in predicting their resource
ment prohibit the dissemination of these data. All other
sufficiency throughout a period. Thus, higher (vs. lower)
data including data from our web appendix studies, as well
payment frequencies should lead to lower prediction uncer-
as the relevant pre-registrations, can be found in Research
tainty over their resource sufficiency, and consequently,
higher subjective wealth perceptions. More formally, we Box #231 (https://researchbox.org/231).
propose the following hypotheses:
STUDY 1: PAYMENT FREQUENCY AND
H1: Higher payment frequencies will increase consumers’
perceptions of their subjective wealth compared to lower REAL-WORLD SPENDING
payment frequencies.
In study 1, we explored the relationship between natural
H2: Consumer’s prediction uncertainty over their resource variations in consumers’ payment frequency and their
sufficiency will mediate the effect of higher payment fre-
spending. To do so, we analyzed a large dataset from a fi-
quencies (vs. lower payment frequencies) on subjective
nancial services provider, which included consumers’ in-
wealth perceptions.
come and expenditure data. We hypothesized that higher
We have argued that higher payment frequencies in- payment frequencies would be associated with increased
crease consumers’ subjective wealth perceptions by de- spending.
creasing consumers’ prediction uncertainty over whether
they will have enough resources throughout a period. If
true, then the effect of payment frequency on subjective
Data
wealth should depend on differences in prediction uncer- We received data from a financial services provider that
tainty. Thus, situational factors that reduce differences in gathered consumers’ income and expense transactions
prediction uncertainty, such as consumers’ income or their across their debit and credit cards for 2014. The data in-
expense profile, should attenuate the effect of payment fre- clude both credits (income) as well as debits (expenses).
quency on subjective wealth. For example, the impact of The dataset contained the amount, date, and currency for
payment frequency on spending should be attenuated at each transaction for 30,963 consumers. It also included a
very high levels of income, as consumers with very high tag for whether the transaction was a credit or a debit. A
incomes likely face little to no uncertainty in predicting summary of the dataset can be found in web appendix A.
their resource sufficiency. As an additional example, if the We identified consumers for whom analysis of their in-
timing of expenses is such that a consumer with lower pay- come and expense transactions was possible. As such, we
ment frequency can assess whether they will have enough excluded 132 consumers with foreign exchange transac-
money throughout a period with the same level of uncer- tions as the dataset did not provide enough information to
tainty as a consumer with higher payment frequency, then convert foreign exchange transaction amounts to the home
the impact of payment frequency on subjective wealth currency. Further, we excluded three consumers who had
should also be attenuated. In sum: missing transaction-level data, such as the amount of the
H3: Situational factors that reduce differences in prediction transaction. Thus, the final dataset contained income and
uncertainty will attenuate the impact of payment frequency expense transactions for 30,828 consumers, accounting for
on subjective wealth perceptions. over 5.3 million transactions.JOURNAL OF CONSUMER RESEARCH 5
Results Log ðSpendingÞit ¼ b1 Payment Frequencyit þ b2
Spending. We first analyzed the number of consumers’ Log ðIncomeÞit þ ai þ it (5)
expenditures as a function of their payment frequency Log ðSpendingÞit ¼ b1 Payment Frequencyit þ b2
and income. To do so, we conducted a series of fixed- Log ðIncomeÞit þ Montht þ ai
effect regressions in which we regressed the number þ it
of expenditures that consumers made each month on (6)
Downloaded from https://academic.oup.com/jcr/advance-article/doi/10.1093/jcr/ucab052/6373900 by guest on 30 December 2021
their payment frequency, using the following model
specifications: Again, consistent with our hypothesis, all models identi-
fied payment frequency as a significant predictor of total
Number of Expendituresit ¼ b1 Payment Frequencyit spending, with higher payment frequencies predicting
more total spending (model 6: b ¼ 0.05, t(20389) ¼ 21.81,
þ ai þ it
p < .001, Cohen’s f2 ¼ .03, see table 1 for all models).
(1) Though the number of days a consumer receives a de-
Number of Expendituresit ¼ b1 Payment Frequencyit posit in a month is arguably the most straightforward
þ b2 Log ðIncomeÞit þ ai means of identifying payment frequency, we performed a
þ it series of robustness checks operationalizing payment fre-
(2) quency in different ways. Specifically, we used the follow-
Number of Expendituresit ¼ b1 Payment Frequencyit ing alternative operationalizations of payment frequency:
þ b2 Log ðIncomeÞit (1) the number of days in a month a consumer received a
þ Montht þ ai þ it deposit that was explicitly labeled as income (e.g., “Salary/
(3) Paychecks,” “Wages Paid”) and (2) the number of individ-
ual deposits a consumer received throughout a month.
Payment frequency was operationalized as the number Furthermore, we identified consumers who were in the
of days in which a consumer received income in a month. dataset for at least three months and calculated the average
Payment frequency was positively correlated with total in- number of days each consumer received income per month.
come (log-transformed) in this dataset (r ¼ .61, p < .001). As such, we relied on between-person (rather than within-
Thus, model 2 includes total income (log-transformed) re- person) variability in consumer’s monthly payment fre-
ceived by consumer i in month t. Model 3 includes month quency and spending. Across all three of these robustness
fixed effects to account for differences in total spending checks, payment frequency predicted both the number of
throughout the year. Across all models, we included a con- purchases and the amount of spending (log-transformed)
sumer-level fixed effect, ai, to account for consumer-level (see web appendix B for more details).
heterogeneity. As such, we relied on within-person vari- Payment Frequency and Income. We have suggested
ability in consumer’s monthly payment frequency and that the relationship between payment frequency and
spending. All standard errors were clustered at the con- spending should depend on situational factors that impact
sumer level. These models were run using the felm func- prediction uncertainty. Since prediction uncertainty should
tion in the lfe R package. As hypothesized, all models decrease as one’s income increases, we next analyzed
found that consumers’ payment frequency was a significant whether consumers’ monthly income level moderated the
predictor of the number of expenditures consumers made, relationship between payment frequency and spending.
such that higher payment frequencies predicted a greater Indeed, we found a significant interaction between con-
number of expenditures (model 3: b ¼ 3.35, t(20389) ¼ sumers’ payment frequency and their income level on the
18.50, p < .001, Cohen’s f2 ¼ .06, see table 1 for all number expenditures (b ¼ 0.55, t(20388) ¼ 6.26, p < .001)
models).1 as well as their total spending (b ¼ 0.03, t(20388) ¼ 22.18,
We then performed the same analysis with consumers’ p < .001), such that the relationship between payment fre-
total spending amount during the month as our dependent quency on spending was attenuated at higher-income lev-
variable (log-transformed). As such, we had the following els, see table 2 and web appendix B for more details).
model specifications:
Discussion
Log ðSpendingÞit ¼ b1 Payment Frequencyit þ ai þ it
This study found that payment frequency predicted
(4) spending. Higher payment frequencies were associated
with higher spending both in terms of the number of expen-
1 The relationship between payment frequency and spending is sig- ditures and the amount of spending. The relationship be-
nificant using a Poisson regression or a negative binomial regression. tween payment frequency and spending was robust to the6 DE LA ROSA AND TULLY
TABLE 1
THE OBSERVED RELATIONSHIP BETWEEN PAYMENT FREQUENCY AND SPENDING (STUDY 1)
Dependent variable
Number of expenditures Log(spending)
(1) (2) (3) (4) (5) (6)
Downloaded from https://academic.oup.com/jcr/advance-article/doi/10.1093/jcr/ucab052/6373900 by guest on 30 December 2021
Payment frequency 3.771*** 3.525*** 3.350*** 0.071*** 0.048*** 0.046***
(0.186) (0.178) (0.181) (0.002) (0.002) (0.002)
Log(income) 1.569*** 1.527*** 0.143*** 0.143***
(0.220) (0.217) (0.008) (0.008)
Month fixed effects No No Yes No No Yes
Consumer-level fixed effects Yes Yes Yes Yes Yes Yes
Observations 51,230 51,230 51,230 51,230 51,230 51,230
Note: þp < .01; *p < .05; **p < .01; ***p < .001.
TABLE 2
THE OBSERVED INTERACTION BETWEEN PAYMENT FREQUENCY AND INCOME ON SPENDING (STUDY 1)
Dependent variable
Number of expenditures Log(spending)
(1) (2) (3) (4)
Payment frequency 2.072** 2.058** 0.232*** 0.232***
(0.787) (0.775) (0.013) (0.013)
Log(income) 0.505* 0.500* 0.090*** 0.090***
(0.208) (0.206) (0.007) (0.007)
Interaction term (payment frequency Log(income)) 0.564*** 0.545*** 0.028*** 0.028***
(0.088) (0.087) (0.001) (0.001)
Month fixed effects No Yes No Yes
Consumer-level fixed effects Yes Yes Yes Yes
Observations 51,230 51,230 51,230 51,230
Note: þp < .01; *p < .05; **p < .01; ***p < .001.
inclusion of multiple controls, as well as various operation- we created a life simulation where participants earned in-
alizations of payment frequency. The results indicate that come, incurred expenses, and made a series of binary
getting paid every workday as opposed to once a week spending decisions. We varied the frequency with which
would increase monthly spending by approximately $20. participants got paid such that some participants were paid
Furthermore, the results suggest that the relationship of weekly (lower payment frequency), whereas others were
payment frequency on spending may be attenuated at high- paid daily (higher payment frequency). We expected
income levels. higher (vs. lower) payment frequency to result in more
Of course, as with any correlational data, it is not possi- spending, and for these differences in spending to be medi-
ble to establish causality. Furthermore, while we try to con- ated by consumers’ subjective wealth perceptions.
trol for consumers’ monthly income, we recognize that the
amount of money deposited into a consumer’s account is
Method
an imperfect measure of their entire financial situation.
Thus, the following study uses a more controlled lab envi- This study was pre-registered on AsPredicted.org
ronment to examine the causal link between payment fre- (https://aspredicted.org/hn2am.pdf). Four hundred and five
quency and spending. participants completed the study on Prolific Academic in
exchange for monetary compensation. Participants were in-
STUDY 2: PAYMENT FREQUENCY’S formed that they would play a life simulator game in which
IMPACT ON SPENDING AND they would work, incur expenses, and make spending deci-
SUBJECTIVE WEALTH sions just as they would in real life. All participants were
given the same starting balance in their checking account
Study 2 aimed to examine the impact of payment fre- ($850). Participants could spend more than was in their
quency on spending in a more controlled setting. To do so, checking account, but to increase realism they wereJOURNAL OF CONSUMER RESEARCH 7
informed that if their balance ever went negative, they Results
would incur a $35 overdraft fee (the most common amount
Four participants failed the attention check and were ex-
charged for overdrafting; see Bankrate 2018). Participants
cluded from all analyses, leaving a final sample of 401 par-
were asked to make decisions as they would in their every-
ticipants (Mage ¼ 32.07, 48% female). Overall, participants
day life.
All participants received the same total amount of in- reported making decisions as they would in real life, with
come during the five weeks ($2,800). However, the dis- the median response rating being a 7 out of 7 (M ¼ 6.63,
bursement of these funds varied by condition. At the SD ¼ 0.70).
Downloaded from https://academic.oup.com/jcr/advance-article/doi/10.1093/jcr/ucab052/6373900 by guest on 30 December 2021
beginning of the study, participants in the high payment Spending. We first examined our primary dependent
frequency condition read that they would receive $140 per measure: the number of times a participant chose the more
day, every Monday through Friday. In contrast, those in expensive option across the 15 decisions in the life simula-
the lower payment frequency condition read that they tion. Consistent with our hypothesis, participants in the
would receive $700 per week, every Friday. As participants higher payment frequency condition chose the more expen-
worked and earned income, participants were informed of sive options significantly more often than those in the
their earnings (i.e., “You earned $700 this week”). lower payment frequency condition (Mhigher payment frequency
Participants incurred bills (e.g., rent, phone bill, utilities) ¼ 6.17, SD ¼ 2.55 vs. Mlower payment frequency ¼ 5.16, SD ¼
and made 15 binary spending decisions (e.g., whether to 2.09), t(399) ¼ 4.35, p < .001, Cohen’s d ¼ .43.
eat out or not, whether to go to a concert or not, whether to As a secondary dependent measure, we examined
buy expensive or cheap sneakers) throughout the life simu- whether payment frequency impacted the total amount of
lation (see web appendix C for a complete list of decisions money consumers’ spent overall. Aggregating the amount
and a short video preview of the simulation). For each of of money spent across the 15 decisions, participants in the
the 15 decisions, there was a more expensive option and a higher payment frequency condition spent more money
less expensive option. For example, in one decision, con- than those in the lower payment frequency condition
sumers could choose to eat out or cook dinner at home (Mhigher payment frequency ¼ $404.35, SD ¼ $122.26 vs.
(more expensive vs. no expense decision), and in another
Mlower payment frequency ¼ $363.55, SD ¼ $107.73), t(399) ¼
decision, they could choose to buy an expensive or cheap
3.55, p < .001, Cohen’s d ¼ .35.
pair of jeans (more expensive vs. less expensive). The
number of times a person chose the more expensive option Subjective Wealth Perceptions. We combined the four
across the 15 decisions served as our primary dependent questions assessing participants’ subjective wealth percep-
measure. Participants were always shown their current tions into a single index, reverse-coding the last two ques-
checking account balance at the bottom of every screen in tions (Cronbach’s a ¼ 0.77). As expected, participants in
order to control for any differences in the potential saliency the higher payment frequency condition reported higher
of their checking accounts (e.g., “Current Checking subjective wealth perceptions relative to those in the lower
Account Balance: $850”). payment frequency condition (Mhigher payment frequency ¼
After participants finished the life simulation, they were 2.86, SD ¼ 1.22 vs. Mlower payment frequency ¼ 2.24, SD ¼
asked to write down any reflections they had regarding 1.01), t(399) ¼ 5.53, p < .001, Cohen’s d ¼ .55. Thus,
their experience. Next, participants answered four ques- even though participants in the daily condition ended the
tions assessing their subjective wealth perceptions through- life simulation with objectively less money (since they
out the simulation: Based on your experience in the life spent more), they felt that they had more financial
simulation, how often did you. . . (1) feel like you had ex- resources.
cess money?, (2) feel like you had more than enough We then examined whether perceptions of subjective
money?, (3) feel like you had a low checking account bal- wealth mediated the effect of payment frequency on the
ance?, and (4) make a decision you did not want to make number of times participants chose the more expensive op-
because you had a low checking account balance? (all 7- tion. To do so, we utilized the bootstrapping mediation
point scales, 1 ¼ Never, 7 ¼ Always). method outlined in Hayes (2017) (PROCESS, Model 4). In
Participants were then given a list of three decisions, line with our predictions, subjective wealth perceptions
where two were decisions they made during the simulation, significantly mediated the effect of payment frequency on
and one was not. They were asked to select which of the the number of expensive decisions (indirect effect ¼ 0.27,
three decisions listed was not a decision they had to make 95% CI [0.12, 0.46], 10,000 resamples).
during the life simulation. This measure served as an atten-
tion check to filter out participants who had not paid atten-
tion throughout the simulation. Finally, participants
Discussion
reported to what extent they made decisions as they would In study 2, participants paid more frequently spent more
have made in real life on a 7-point scale (1 ¼ not at all, than participants who were paid less frequently. Moreover,
7 ¼ very much) and shared their demographic information. this study found that the effect of payment frequency on8 DE LA ROSA AND TULLY
spending was explained by differences in consumers’ per- participants made the same expenditures totaling $2,600.
ceptions of their subjective wealth. In this study, partici- Participants in the higher payment frequency condition re-
pants were allowed to spend as much as they wanted, even ceived $100 each day, while participants in the lower pay-
if it meant having a negative checking account balance. ment frequency condition received $700 each week. As
However, the overdraft fee may have created the sense of such, all participants earned $2,800 in the simulation.
having liquidity constraints. Thus, across two additional Each day, participants saw their income, expenses, and
web appendix studies, we examine the effect of payment checking account balance. Importantly, participants in the
frequency on spending in contexts where lack of liquidity daily pay condition started the simulation with $20 while
Downloaded from https://academic.oup.com/jcr/advance-article/doi/10.1093/jcr/ucab052/6373900 by guest on 30 December 2021
is not a concern. These results demonstrate that higher pay- participants in the weekly pay condition started the simu-
ment frequency increases spending even in the absence of lation with $500. Thus, participants in the daily pay con-
any overdraft fees or when account balances and spending dition had a lower average daily balance ($322 vs. $513)
opportunities are constructed such that participants can and a lower minimum balance than the weekly pay condi-
never run out of money (web appendices D and E). tion ($20 vs. $44; see web appendix F for daily account
balances).
STUDY 3: OBJECTIVE VERSUS After the simulation, participants were asked six ques-
SUBJECTIVE WEALTH tions measuring their subjective wealth perceptions on a
101-point scale (0 ¼ not at all, 100 ¼ very much): As you
Study 2 demonstrated that higher payment frequency were going through the simulation, to what extent did you
increases consumer spending by increasing subjective feel like you (1) had a lot of money? (2) had more than
wealth perceptions. Study 3 aimed to examine the link be- enough money? (3) had excess money? (4) lacked money?
tween payment frequency and subjective wealth percep- (5) did not have enough money, and (6) were going to run
tions. In particular, study 3 was designed to disentangle out of money? Participants then answered four questions
differences in subjective wealth from differences in objec- regarding their prediction uncertainty on a 101-point scale
tive wealth. Because getting paid more frequently often (0 ¼ disagree, 100 ¼ agree): My daily income and expenses
results in receiving funds earlier, those paid more fre- made (1) it easy to predict whether I would have enough
quently often have greater accumulated wealth on any money throughout the simulation, (2) it difficult to predict
given day compared to those paid less frequently. Indeed, whether I would have enough money throughout the simu-
in the previous study, while total income and possible lation, (3) me feel confident predicting whether I would
expenditures were held constant across conditions, partici- have enough money throughout the simulation, and (4) me
pants in the higher payment frequency condition had a feel uncertain predicting whether I would have enough
higher average daily account balance than participants in money throughout the simulation.
the lower payment frequency condition. To isolate the im- Participants were then asked if they believed they had
pact of payment frequency on subjective wealth, in this taken part in a similar study in the past. We asked partici-
study, those paid less frequently were endowed with more pants to answer this question honestly and assured them
money than those paid more frequently. that their answer would not affect their compensation for
Moreover, study 3 examined why, if not for differences the study. Next, participants were asked three attention
in objective wealth, payment frequency increases subjec- check questions: (1) how often they were paid, (2) how
tive wealth. We have suggested that getting paid more fre- much they were paid per paycheck, and (3) the typical
quently decreases consumers’ uncertainty in predicting range for their daily expenses. Finally, participants
whether they will have enough resources throughout a pe- reported their demographic information.
riod, which leads to higher subjective wealth perceptions.
To examine this explanation for changes to subjective Results
wealth, we measured subjective wealth, as well as consum- Thirty-one participants reported taking a similar study in
ers’ prediction uncertainty. the past and nine additional participants failed our attention
checks. Consistent with the exclusion criteria used in all of
Method our lab studies, these participants were excluded from
One hundred and fifty-two participants completed the all analyses, leaving a final sample of 112 participants
study on Amazon Mechanical Turk in exchange for mone- (Mage ¼ 36.59, 41% female).
tary compensation. This study was similar to the previous Subjective Wealth Perceptions. We combined the six
life simulation, except that the life simulation included 28 subjective wealth perception measures into an index
“days.” To isolate the impact of payment frequency from (Cronbach’s a ¼ 0.93).2 Consistent with our hypothesis,
the impact of objective wealth, all expenses were held con-
stant. Specifically, all participants saw the same daily 2 We conducted a principal component analysis to understand
expenses, and no expenses were optional such that all whether our subjective wealth and prediction uncertainty measuresJOURNAL OF CONSUMER RESEARCH 9
but in contrast to what would be predicted by objective frequencies are more likely to lead to a pattern of daily re-
wealth levels, participants in the higher payment frequency source decumulation, which should increase consumers’
condition reported higher subjective wealth perceptions prediction uncertainty. When the frequency and timing of
than participants in the lower payment frequency condition expenses enable consumers with lower payment frequen-
(Mhigher payment frequency ¼ 44.40, SD ¼ 25.55 vs. Mlower pay- cies to avoid patterns of resource decumulation, differences
ment frequency ¼ 34.57, SD ¼ 24.17), t(110) ¼ 2.09, p ¼ in prediction uncertainty and subjective wealth across pay-
.039, Cohen’s d ¼ .40. ment frequencies should be attenuated. We tested this mod-
eration in study 4.
Downloaded from https://academic.oup.com/jcr/advance-article/doi/10.1093/jcr/ucab052/6373900 by guest on 30 December 2021
Prediction Uncertainty. We combined the four predic-
In study 4, we also examined two alternative explana-
tion uncertainty measures into an index (Cronbach’s a ¼
tions for the effect of payment frequency on subjective
0.92). Consistent with our hypothesis, participants in the wealth. One alternative explanation is that because con-
higher payment frequency condition indicated feeling less sumers prefer receiving segregated (vs. aggregated) gains
uncertainty than participants in the lower payment fre- (Thaler 1985, 1999), consumers who receive higher pay-
quency condition in predicting whether they would have ment frequencies—and thus receive income in their pre-
enough resources throughout the simulation (Mhigher payment ferred manner—may feel better about their financial
frequency ¼ 35.29, SD ¼ 23.16 vs. Mlower payment frequency ¼ situation, resulting in greater subjective wealth. Another al-
50.34, SD ¼ 28.63), t(110) ¼ 3.05, p ¼ .003, Cohen’s ternative explanation is that segregated gains are miscalcu-
d ¼ .58. lated to be larger than their equivalent sum. Both of these
Mediation. As predicted, the impact of payment fre- explanations suggest that expense frequency should be ir-
quency on subjective wealth perceptions was mediated by relevant in moderating the effect of payment frequency on
differences in prediction uncertainty (indirect effect ¼ subjective wealth perceptions.
8.22, 95% CI [2.55, 13.89], 10,000 resamples).
Method
Discussion This study was pre-registered on AsPredicted.org
In study 3, participants paid more (vs. less) frequently (https://aspredicted.org/np53c.pdf). Five hundred and
felt greater subjective wealth. This effect emerged despite ninety-nine participants completed the study on Prolific
the fact that initial endowments differed across conditions Academic in exchange for monetary compensation. This
such that those paid more frequently had objectively less study used the same design as study 3 with the following
money than those paid less frequently. These subjective two exceptions. First, all participants received the same ini-
wealth differences were explained by differences in con- tial endowment amount ($500). Second, aside from manip-
sumers’ prediction uncertainty. ulating consumers’ payment frequency, we also
manipulated the frequency of consumers’ expenses. In the
higher expense frequency conditions, participants paid
STUDY 4: MODERATING THE EFFECT expenses every day. In contrast, in the lower expense fre-
OF PAYMENT FREQUENCY ON quency condition, participants’ expenses were aggregated
SUBJECTIVE WEALTH and paid once a week (on Fridays). Thus, this study used a
2 (payment frequency: higher vs. lower) 2 (expense fre-
Studies 2 and 3 demonstrate that payment frequency
quency: higher vs. lower) design. Importantly, the total
impacts subjective wealth perceptions (hypothesis 1).
amount of income and expenses was held constant across
Study 3 suggests that the effects of payment frequency are
all conditions.
not solely the result of objective wealth differences. After the simulation, participants answered the same
Instead, study 3 demonstrates that the effect of payment questions as in study 3.
frequency on subjective wealth perceptions is explained by
differences in consumers’ prediction uncertainty (hypothe-
Results
sis 2). If so, situational factors that reduce differences in
consumers’ prediction uncertainty should attenuate the ef- Thirty-seven participants reported taking a similar study
fect of payment frequency on subjective wealth percep- in the past and 34 additional participants failed our atten-
tions (hypothesis 3). tion checks. As pre-registered, these participants were ex-
According to our account, the frequency of expenses can cluded from all analyses, leaving a final sample of 528
impact consumers’ prediction uncertainty. Due to the high participants (Mage ¼ 32.06, 47% female).
frequency of expenses and the inability to offset these Subjective Wealth Perceptions. We combined the six
expenses as they occur, lower (vs. higher) payment subjective wealth measures into an index (Cronbach’s a ¼
loaded onto two different factors. Indeed, they did (see web appendix 0.89). We then regressed participants’ subjective wealth
G for more details). perceptions on their payment frequency (higher payment10 DE LA ROSA AND TULLY
frequency ¼ 1, lower payment frequency ¼ 1), expense Discussion
frequency (higher expense frequency ¼ 1, lower expense
Study 4 shows that differences in subjective wealth
frequency ¼ 1), and the interaction term between these across payment frequencies result from differences in
two factors. The model revealed a significant main effect experiencing resource decreases. Specifically, lower (vs.
of payment frequency (b ¼ 5.57, t(524) ¼ 5.42, p < .001) higher) payment frequencies typically result in larger and
and a significant main effect of expense frequency (b ¼ more frequent resource decreases, which increase consum-
6.05, t(524) ¼ 5.90, p < .001) on consumers’ subjec- ers’ uncertainty in predicting whether they will have
Downloaded from https://academic.oup.com/jcr/advance-article/doi/10.1093/jcr/ucab052/6373900 by guest on 30 December 2021
tive wealth perceptions. Importantly, these effects were enough resources throughout a period. As such, when con-
qualified by a significant interaction (b ¼ 4.10, t(524) ¼ sumers with lower payment frequencies did not experience
3.99, p < .001, Cohen’s f ¼ .17) (figure 1). When expenses larger, more frequent resource decreases than those with
were incurred on a daily basis, we replicated the effects in higher payment frequencies, the effect of payment fre-
previous studies, such that higher payment frequency led to quency on subjective wealth was attenuated.
higher subjective wealth perceptions compared to lower
payment frequency (b ¼ 19.32, t(524) ¼ 6.66, p < .001). In STUDY 5: PAYMENT FREQUENCY’S
contrast, when expenses were incurred on a weekly basis, IMPACT ON PREDICTION
payment frequency did not significantly impact subjective UNCERTAINTY, SUBJECTIVE WEALTH,
wealth perceptions (b ¼ 2.94, t(524) ¼ 1.01, p ¼ .311).
AND SPENDING
Prediction Uncertainty. We combined the four predic- Study 5 was designed to examine the full model, whereby
tion uncertainty measures into an index (Cronbach’s a ¼ higher payment frequencies decrease prediction uncertainty,
0.91). We then regressed participants’ prediction uncer- thereby increasing subjective wealth, and in turn, spending.
tainty on their payment frequency, expense frequency, and To do so, we returned to the version of the life simulation
the interaction term between these two factors using the where participants could make spending decisions (as in
same effect coding as in the previous analysis. The model study 2). Moreover, we aimed to ensure that differences in
revealed a significant main effect of payment frequency subjective wealth were not the result of differences in objec-
(b ¼ 3.16, t(524) ¼ 2.70, p ¼ .007) and a significant tive wealth by using a context in which higher payment fre-
main effect of expense frequency (b ¼ 4.46, t(524) ¼ 3.82, quencies result in a lower average daily account balance than
p < .001) on consumers’ prediction uncertainty. Again, lower payment frequencies. Instead of varying initial endow-
ments as in study 3, in study 5, participants were paid either
these effects were qualified by a significant interaction
daily or bi-weekly, but we changed the bi-weekly pay sched-
(b ¼ 5.28, t(524) ¼ 4.52, p < .001, Cohen’s f ¼ .20).
ule so that participants received their first bi-weekly pay-
When expenses were incurred on a daily basis, higher pay-
check after the first week and their last bi-weekly paycheck
ment frequency significantly decreased consumers’ predic- in the second to last week. Thus, total pay was held constant
tion uncertainty compared to lower payment frequency across conditions, but those paid bi-weekly had a higher av-
(b ¼ 16.87, t(524) ¼ 5.11, p < .001). In contrast, when erage daily balance than those paid daily. If differences in
expenses were incurred on a weekly basis, payment fre- spending result from differences in objective wealth levels,
quency did not significantly impact prediction uncertainty then those in the lower payment frequency condition should
(b ¼ 4.24, t(524) ¼ 1.28, p ¼ .200). spend more than those in the higher payment frequency con-
dition. However, if higher payment frequencies increase per-
Moderated Mediation. A moderated mediation analysis ceptions of subjective wealth by decreasing prediction
was conducted using PROCESS Model 7 (Hayes 2017). uncertainty, then those in the higher payment frequency con-
We found support for the expected moderated mediation dition should spend more than those in the lower payment
(indirect effect ¼ 4.86, 95% CI [2.71, 7.13], 10,000 resam- frequency condition despite the differences in objective
ples). The analysis showed that when consumers experi- wealth. Moreover, these differences in spending should be
enced daily expenses, their prediction uncertainty mediated explained by differences in subjective wealth perceptions
the effect of payment frequency on subjective wealth (indi- that result from differences in prediction uncertainty.
rect effect ¼ 3.88, 95% CI [2.32, 5.52], 10,000 resamples). We considered the possibility that the effect of payment
However, this mediation was not observed when consum- frequency on spending may be multiply determined. Thus,
in addition to measuring subjective wealth as the primary
ers experienced weekly expenses (indirect effect ¼ 0.98,
mechanism, we examined other potential mechanisms. One
95% CI [2.47, 0.51], 10,000 resamples).3
possibility is that due to the shorter time period between
paychecks, higher payment frequencies reduce the extent
3 For the curious mind, results are consistent and significant when
allowing for a direct effect of expense frequency on subjective wealth to which consumers plan for future expenses, which could
as well (PROCESS Model 8). increase spending. Another possibility is that, becauseJOURNAL OF CONSUMER RESEARCH 11
FIGURE 1
THE EFFECT OF PAYMENT FREQUENCY AND EXPENSE FREQUENCY ON SUBJECTIVE WEALTH PERCEPTIONS (STUDY 4)
100
*** NS
Downloaded from https://academic.oup.com/jcr/advance-article/doi/10.1093/jcr/ucab052/6373900 by guest on 30 December 2021
Subjective Wealth Perceptions
75
53.89
49.98 50.95
50
30.65
25
0
Higher Expense Frequency (Daily Expenses) Lower Expense Frequency (Weekly Expenses)
Higher Payment Frequency (Daily Pay) Lower Payment Frequency (Weekly Pay)
Notes: Expense frequency moderated the effect of payment frequency on consumers’ subjective wealth perceptions (study 4). Error bars represent 95% confidence
intervals. Significance levels: þp < .1; *p < .05; **p < .01; ***p < .001.
segregated gains have a larger effect on affect than aggre- hundred participants completed the study on Prolific
gated gains (Morewedge et al. 2007), spending differences Academic in exchange for monetary compensation.
across payment frequencies may be a function of affect Participants played a four-week life simulator where they
differences. Although the compensatory and retail therapy had to work and make spending decisions. All partici-
literature would suggest that negative affect increases pants started with an $875 checking account balance.
spending (Atalay and Meloy 2011), some research sup- Participants were then randomly assigned to either a
ports the possibility that positive affect increases spend- higher payment frequency condition or a lower payment
ing (Babin and Darden 1996). To examine these frequency condition ($140 daily, every Monday through
possibilities, in study 5, we measured planning behavior Friday vs. $1,400 bi-weekly, every other Friday), with
and affect. For each measure, we aimed to explore those in the lower payment frequency condition receiving
whether the potential mechanism was operating, and if so, their first bi-weekly paycheck on the first Friday of the
whether it was a better explanation for the effect of pay- simluation. This ensured that with no differences in
ment frequency on spending than subjective wealth differ- spending, the average daily balance in the lower payment
ences. Finally, we also measured participants’ financial frequency condition would be higher than the average
literacy and intertemporal discount rates to examine daily balance in the higher payment frequency condition
whether either of these factors moderated the effect of ($2,516 vs. $2,371; see web appendix H for complete
payment frequency on spending. details). In this life simulation, all expenses (including
bills) had a decision associated with them and partici-
pants made one spending decision per day (28 in total,
Method see web appendix I for the complete list of decisions).
This study was pre-registered on AsPredicted.org For each decision, one option was more expensive and
(https://aspredicted.org/zd5xz.pdf). One thousand two the other option was either less expensive or resulted in12 DE LA ROSA AND TULLY
no spending. As in study 2, the number of times a partici- Results
pant selected the more expensive option served as our pri-
Seventeen participants failed the attention check ques-
mary dependent measure. There were no overdraft fees in
tion and 66 participants reported previously participating
this study.
in a similar experiment. As pre-registered, these partici-
After going through the life simulation, participants com-
pants were excluded from all analyses, leaving a final sam-
pleted the subjective wealth and prediction uncertainty meas-
ple of 1,120 participants (Mage ¼ 32.71, 52% female).
ures from studies 3 and 4. They also answered the same
attention check question as in study 2. To measure planning Spending. First, we examined the number of times a
Downloaded from https://academic.oup.com/jcr/advance-article/doi/10.1093/jcr/ucab052/6373900 by guest on 30 December 2021
behavior, participants indicated their level of agreement with participant selected the more expensive option across the
the following three statements: During the life simulation. . . 28 decisions. Replicating the results of study 2, participants
(1) I actively tried to plan for large upcoming bills (rent, cell- in the higher payment frequency condition selected the
phone, TV and internet, health insurance, gas and electricity, more expensive option significantly more often than those
etc.), (2) I actively tried to budget for future expenses, and (3) in the lower payment frequency condition, (Mhigher payment
I consulted my checking account balance to budget how to frequency ¼ 17.00, SD ¼ 3.63 vs. Mlower payment frequency ¼
spend my money for the next few days (all on 7-point scales: 15.21, SD ¼ 3.30), t(1,118) ¼ 8.64, p < .001, Cohen’s d ¼
1 ¼ disagree, 7 ¼ agree). Participants were then asked to think .52. In addition, participants in the higher payment fre-
back to their experience in the life simulation and report their quency condition spent more money on their purchases
valence, arousal, and power during the simulation using the compared to participants in the lower payment frequency
3-item Self-Assessment Manikin scale (Bradley and Lang condition (Mhigher payment frequency ¼ $2,919.58, SD ¼
1994). $198.39 vs. Mlower payment frequency ¼ $2,816.52, SD ¼
In addition, we measured participants’ financial literacy $175.98), t(1,118) ¼ 9.19, p < .001, Cohen’s d ¼ .55. We
(8 items, Lusardi and Mitchell 2011) and intertemporal also analyzed each of the individual spending decisions
discount rates (16 item titration task) to explore whether across the 28 days (see figure 2 and web appendix K for
either of these factors moderated the effect of payment more details).
frequency on spending. As in prior studies, in order to Aside from having a lower average daily balance over-
ensure the naivety of our sample, we asked participants all, participants in the higher payment frequency condition
whether they believed they had taken this study in had a lower checking account balance than those in the
the past. Finally, participants completed demographic lower payment frequency condition on 20 out of the
questions. 28 days. Thus, we ran an additional analysis to understand
FIGURE 2
PARTICIPANTS SPENDING DECISIONS ACROSS DAY AND PAYMENT FREQUENCY CONDITION (STUDY 5)
+
Choosing the More Expensive Option
100%
** **
Percentage of Participants
*
75% ***
** *
+ ***
50%
*** *** ***
*** **
25% ***
**
0%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
Day (1−28)
Higher Payment Frequency (Daily Pay) Lower Payment Frequency (Bi−Weekly Pay)
Notes: Participants’ spending decisions across day and payment frequency condition (study 5). Higher (vs. lower) payment frequency led to more spending through-
out the period. Error bars represent 95% confidence intervals. Significance levels: þp < .1; *p < .05; **p < .01; ***p < .001.You can also read
