## Access

You are not currently logged in.

Access your personal account or get JSTOR access through your library or other institution:

# The Denomination Effect

Priya Raghubir and Joydeep Srivastava
Journal of Consumer Research
Vol. 36, No. 4 (December 2009), pp. 701-713
DOI: 10.1086/599222
Stable URL: http://www.jstor.org/stable/10.1086/599222
Page Count: 13
Item Type
Article
References

# The Denomination Effect

Priya Raghubir
Joydeep Srivastava *
John Deighton served as editor and Brian Ratchford served as associate editor for this article.

Labeled the “denomination effect,” study 1 shows in three field studies that the likelihood of spending is lower when an equivalent sum of money is represented by a single large denomination (e.g., one $20 bill) relative to many smaller denominations (e.g., 20$1 bills). In two of the three field studies, individuals spent more once the decision to spend had been made. Study 2 then shows that consumers deliberately choose to receive money in a large denomination relative to small denominations when there is a need to exert self‐control in spending. Study 3 further shows that the denomination effect is contingent on individual differences in people’s desire to reduce the pain of paying associated with spending. The results suggest that the denomination effect occurs because large denominations are psychologically less fungible than smaller ones, allowing them to be used as a strategic device to control and regulate spending.

Keywords: Behavioral Decision Theory, Judgment and Decision Making, Economic Psychology, Experimental Design and Analysis (ANOVA)

Money is any token that functions as a socially and legally acceptable medium of exchange for goods and services. Standard economic theories of the consumer generally assume that money affects purchase decisions through the budget constraint (e.g., income) or that individuals’ spending decisions are influenced by the amount of money available to them. Accordingly, money is used as a unit of account or common measure of value for measuring and comparing the worth of different goods and services. Although the importance of the amount of money is recognized in most economic theories, these theories are generally silent regarding how different representations of an identical amount of money influence spending decisions. The principle of descriptive invariance (Kahneman and Tversky 1979; Tversky, Sattath, and Slovic 1988) suggests that decisions and preferences ought to be invariant across different presentations of the same objective stimuli—in this case, the same amount of money. The extent to which consumer spending and saving decisions differ as a function of how a specific amount of money is represented would thus challenge a basic tenet of most economic theories.

Previous research demonstrates that the normative principle of descriptive invariance is commonly violated in the domain of money (e.g., Gourville 1998; Raghubir and Srivastava 2002; Shafir, Diamond, and Tversky 1997). For example, Gourville (1998) showed that people evaluate a transaction more positively when an identical amount of money is framed as “pennies a day” ($1 a day) rather than aggregately ($365 a year). Shafir et al.’s (1997) money illusion suggests that people think predominantly in terms of nominal rather than real monetary values. In a related investigation, Raghubir and Srivastava (2002) showed that individuals’ valuation of a product in a foreign currency was biased toward the nominal value of the price of the product—its face value, with inadequate adjustment for the exchange rate (see also Wertenbroch, Chattopadhyay, and Soman 2007). Research has also shown that people tend to spend more when using different payment modes, such as a credit card or a gift certificate, than when using cash (Raghubir and Srivastava 2008).

More relevant to the current investigation, Mishra, Mishra, and Nayakankuppam (2006) provide yet another demonstration of the principle of descriptive invariance by documenting a “bias for the whole,” in which a large denomination of money (e.g., one $100 bill) led to lower spending intentions than an equivalent amount of money in smaller denominations (five$20 bills). Mishra et al. (2006) argue that people perceive higher value when money is in the form of a large single denomination because of the greater fluency experienced in processing the large denomination relative to many small denominations. The greater processing fluency translates into positive affect toward the money, which leads people to overvalue the whole, thereby making them less likely to spend it compared to an equivalent amount in smaller parts. This article adds to this growing body of literature by showing that individuals’ spending decisions and decisions to receive money are influenced by the denomination of an identical sum of money. Labeled the “denomination effect,” this article offers an alternative explanation by showing that the effect occurs due to the relatively large denominations being perceived as less fungible. Leveraging the lower perceived fungibility of larger denominations, individuals strategically choose large versus small denominations when they wish to exert self‐control in spending.

This article extends the previous research in several ways. Given the large literature on the discrepancy of behavioral intentions from actual behavior (e.g., Ajzen, Brown, and Carvajal 2004), we demonstrate that the likelihood of spending is lower when money is represented by a single large denomination (e.g., one $20 bill) relative to when it is represented by many smaller denominations (e.g., 20$1 bills), holding total amount constant, for actual purchase decisions rather than spending intentions as in Mishra et al. (2006). In addition to the obvious advantage of studying actions over intentions, actual spending decisions allow an examination of how much is spent once the decision to spend has been made. Importantly, this article offers an alternative explanation for the denomination effect that is rooted in ideas of self‐control and regulation. Individuals often use various tactics and mechanisms that impose constraints or alter incentives to control their own short‐term behavior for a long‐term benefit (Baumeister 2002; Baumeister, Heatherton, and Tice 1994; Hoch and Loewenstein 1991). We suggest that individuals are less likely to spend and more likely to receive money in a large denomination relative to many smaller denominations as a way to curb or control their spending.

The systematic influence of denomination on spending decisions has important implications from a consumer welfare perspective as well as a monetary policy perspective. While a few studies have examined the influence of denomination on macrolevel measures such as price inflation and monetary policy (e.g., Chen 1976), these studies typically consider aggregate‐level, interdependent factors, making it difficult to establish causal relationships. There is thus a need for research at the microlevel that examines how and why the denomination of money affects individuals’ decisions to spend and receive money.

## Conceptual Background

An extensive literature in cognitive psychology and behavioral decision theory, documenting violations of the principle of descriptive invariance, shows that alternative representations of the same objective monetary stimuli lead to marked differences in judgments and decisions (e.g., Gourville 1998; Raghubir and Srivastava 2002, 2008; Shafir et al. 1997; Shefrin and Thaler 1988; Thaler 1985; Tversky et al. 1988). In documenting a bias for the whole, Mishra et al. (2006) argue that a single denomination allows easier information processing or greater processing fluency, which leads to positive affect and thereby lower purchase intentions relative to many smaller denominations. They demonstrated that the effect was attenuated when fluency was primed through another task, when a high familiarity with the bills was induced, and when participants were informed that the background music could bias their feeling toward the money they were given.

Mental accounting ideas, in which people are modeled as creating separate source‐use expense and income ledgers in their mind, also provide clues for why individuals may differentially spend an equivalent sum of money based on its denomination (Thaler 1985). Shefrin and Thaler (1988) suggest that people tend to categorize income into different mental accounts (defined in terms of amount and frequency of payments), which then affects their propensity to spend. Similarly, Heath and Soll (1996) showed that people have separate mental accounts for expense categories, such as food and entertainment, with spending behavior driven by the available surplus or deficit in each category. As such, a large denomination of money may be placed into a mental account of “real money,” whereas an equivalent amount in smaller denominations may be placed into a “petty cash” or “loose change” account. To the extent that there are different norms governing the spending patterns within the two accounts, with tighter controls for the spending of real money versus petty cash, money in large denominations should be less likely to be spent (or more likely to be saved) relative to smaller denominations. Moreover, to the extent that individuals are aware of the differences in the likelihood of spending or saving money as a function of its denomination, they may choose to receive or save a specific denomination over another as a tactic to control their impulse to spend.

### Self‐Control and the Denomination Effect

Extensive research has examined consumers’ use of self‐control to avoid hedonic temptations such as smoking, overeating, and overspending (e.g., Baumeister 2002; Hoch and Loewenstein 1991; O’Guinn and Faber 1989; Rook 1987; Wertenbroch 1998). Hoch and Loewenstein (1991) conceptualized the self‐control problem as the psychological conflict between desire and willpower, with the outcome depending on the relative strength of these opposing forces. Ideas of self‐control and self‐regulation are particularly important for buying and spending impulses and decisions. Individuals commonly face spending decisions in which they have to balance the short‐term impulse to spend against a long‐term benefit (e.g., saving or holding out for a more useful purchase). Schelling (1978) viewed this struggle as an intrapersonal conflict between two selves of an individual—one myopic and the other farsighted (see also Thaler and Shefrin 1981). Often the short‐term impulse wins out, resulting in self‐control failures. For example, consumers sometimes report experiencing a loss of self‐control, resulting in excessive purchases and postpurchase regret (O’Guinn and Faber 1989; Rook 1987).

According to Schelling’s (1978) notion of “egonomics,” or the art of self‐management, the farsighted self anticipates that its myopic twin will not be able to resist a drink at a party and thus increases the cost of consuming alcohol by precommitting to abstinence by swallowing the drug Antabuse, which causes nausea and vomiting if alcohol is consumed. Precommitment involves any mechanism that imposes constraints or alters incentives for future behavior (Hoch and Loewenstein 1991; Thaler and Shefrin 1981). The classic example is Ulysses’ instructions to his crew to bind him to the mast so that he could hear the Sirens without jumping to his death. Less drastic and nonbinding precommitment rules that individuals commonly use to control their short‐term impulses are setting spending limits (Ariely and Wertenbroch 2002) and mental accounts (Thaler 1985). In our context, a large denomination may serve as a precommitment mechanism to exert self‐control and willpower to overcome the urge to spend, relative to many small denominations.

The idea of precommitment has an important implication. Given that exerting self‐control and willpower requires energy resources (Baumeister 2002), individuals will incur a psychological cost if the precommitment rule is somehow violated, resulting in self‐control failure (Hoch and Loewenstein 1991). For example, the psychological cost after excessive spending may be reflected in regret and a feeling of failure (O’Guinn and Faber 1989). However, once the cost is incurred to overcome the self‐imposed constraint or goal, there may be a “what‐the‐hell” effect such that the short‐term impulse behavior is accentuated (Cochran and Tesser 1996). It is important to note that the goal does not have to be explicit—a what‐the‐hell effect may be observed for implicit goals. The implication is that once the decision to spend has been made, the amount spent is likely to be higher for a large denomination relative to many small denominations (Soman and Cheema 2004). Since the large denomination did not help in exerting self‐control and, thereby, did not lead to not spending at all, individuals are less sensitive to the total amount spent once the decision to spend has been made.

Baumeister et al. (1994) identified three causes of self‐control failure: uncertain or conflicting goals, failing to monitor and keep track of one’s behavior, and depletion of resources that allow self‐control to operate (see also Baumeister 2002). Since, over time, individuals realize that self‐control often fails, they use various tactics and mechanisms to overcome desire or short‐term impulses (Hoch and Loewenstein 1991). In the present context, money in the form of a large denomination is easier to manage as it facilitates relatively better monitoring and tracking of behavior (one’s spending) than an equivalent amount in many smaller denominations. Given that self‐control failures in spending are more likely to occur when individuals’ monitoring and tracking capabilities are attenuated, individuals are less likely to spend money that is in the form of a large denomination relative to many small denominations.

Individual trait differences in the ability to exercise self‐control also exist, particularly in the domain of spending (Rick, Cryder, and Loewenstein 2008). Rick et al. (2008) showed that individuals differ in the extent to which they spend, such that “tightwads” spend less than they would ideally like, and “spendthrifts” spend more than they would ideally like. At first blush this individual difference may suggest that spendthrifts are in greater need to control their spending than are tightwads and should accordingly be more likely to use denomination as a precommitment strategy to curb their spending. However, a deeper examination of the reasons underlying individual differences in spending suggests otherwise. The reason that tightwads are more controlled and spend less is that they experience a higher pain of paying relative to spendthrifts who are more impulsive and spend more. It is the difference in the pain of paying associated with spending rather than the normative level of spending itself that should drive tightwads’ and spendthrifts’ need for a precommitment strategy. In our context, this implies that tightwads would like to reduce the potential pain of paying by strategically precommitting to not spending in a situation in which they expect to be tempted to spend. However, spendthrifts, who experience relatively little pain of paying, have less of a need to use denomination choice as a precommitment strategy. Thus, although tightwads spend less than they would ideally like, and relatively less than spendthrifts, and are arguably less in need of self‐control, the higher anticipatory pain of paying associated with spending makes them more inclined to use a precommitment strategy in order to minimize the pain.

### Overview of Studies

Three studies that test the denomination effect and the extent to which self‐control reasons underlie the denomination effect are reported. Study 1 begins the investigation by examining the denomination effect in actual purchase decisions as opposed to purchase intentions. Three replications of the denomination effect with different populations and purchase situations, as well as significant amounts of money, demonstrate the robustness and generalizability of the effect. Results show that once the decision to spend is made, the amount spent is higher in the large denomination condition than in the small denomination condition. Although this result suggests that self‐control reasons underlie the effect, studies 2 and 3 more directly examine this issue. Study 2 demonstrates that people prefer to receive money in a large denomination rather than several small denominations in contexts in which they need to exercise self‐control in spending. Study 3 further shows that individual difference in the anticipatory pain of paying moderates the denomination effect. Tightwads choose to receive money in a large denomination as a precommitment device when the need for self‐control is high, and they do so strategically (i.e., are aware of the reasons for their choice and do so intentionally).

## Study 1: The Denomination Effect and Actual Spending

Study 1 has three main objectives. First, in contrast to behavioral intentions (Mishra et al. 2006), this study examines the effect of denomination on actual purchase decisions. Second, this study explores the robustness and generalizability of the denomination effect across different participants and amounts of money. Third, this study examines the amount spent once the decision to spend has been made (see Chandon and Wansink [2002] for an analysis of purchase incidence and quantity given incidence in a consumption context). The amount spent, after the decision to spend, allows us to explore the extent to which individuals exhibit the what‐the‐hell effect (Cochran and Tesser 1996). If the denomination effect is driven by self‐control reasons, the what‐the‐hell effect suggests that the amount spent once the decision to spend has been made is likely to be higher when individuals have a large denomination relative to many small denominations of an equivalent amount of money. In contrast, to the extent that the denomination effect is perceptually driven (i.e., a single large denomination is valued more than small denominations), the amount spent once the spending decision is made is likely to be lower or no different in the large denomination condition. Based on the documented bias for the whole (Mishra et al. 2006), tests pertaining to the likelihood of spending use one‐tailed tests, whereas those examining the amount conditional on spending use two‐tailed tests. The study method and results are summarized in table 1.

Table 1
Study 1: Description and Results
 Study 1a Study 1b Study 1c Denomination effect Denomination manipulation $4\times \$ .25$versus$1 $5\times \$ 1$versus$5 $\mathrm{RMB}\,\,50+2\times 20+10$ versus 100 … Sample characteristics Undergraduate students thanked for study participation Drivers at a gas station in Omaha Female homemakers in Xiangtan, China … Sample size 89 75 150 … Experimental context Laboratory Field Simulated field … Task Purchase of candy Purchase of nongas items at convenience store at gas station Purchase of soap, shampoo, bedding, and pots and pans … Results—small versus large denominations: Percent purchasing 62.79 versus 26.09 24 versus 16 90.7 versus 80 … Amount spent conditional on purchase being made $.82 versus$1.17 $4.13 versus$1.72 RMB 56.76 versus 67.67 Consistent with what‐the‐hell effect Amount spent overall $.56 versus$.30 $.99 versus$.28 RMB 51.47 versus 54.13 Aggregate effect contingent on the strength of denomination and what‐the‐hell effects Postpurchase satisfaction … … 3.60 versus 3.23 Smaller denominations are associated with greater satisfaction

### Study 1a

Eighty‐nine undergraduate business students from two U.S. universities were assigned at random to either a small denomination (four quarters) or a larger denomination ($1 bill) condition. Data are aggregated across the two schools as there were no differences. The cover story was that students were being thanked for their participation in an experimental session. They were told that they could keep the money or buy candy with it. There were two choice sets. In one condition the choice set consisted of Trident gum ($1 per packet). In a replicate condition, in addition to Trident, Wrigley’s gum (25¢ each) and a box of Altoids ($2 per box) were also available. Prices used were those charged by the university coffee shop. The primary dependent measure was whether participants spent the dollar (or any part) to purchase candy. #### Results. Across the two denomination conditions, approximately 44% (39/89) of the participants chose to purchase candy. Consistent with the denomination effect, participants were more likely to spend when they were given four quarters than when they were given a$1 bill. While about 63% (27/43) of the participants chose to purchase candy in the small denomination condition, only 26% (12/46) chose to purchase candy in the large denomination condition ($\chi ^{2}( 1) =12.16$ , $p< .001$ ). Given the significant difference in purchase likelihood, it is not surprising that the average amount spent in the small denomination condition was almost twice the amount spent in the large denomination condition ($M{'}\mathrm{s}\,=\$ 0.56$and$0.30; $F( 1,\, 87) =3.76$ , $p< .05$ ).

The amount spent once the decision to spend had been made (or conditional on purchase incidence) was, however, significantly higher in the large denomination condition ($M=\$ 1.17$) relative to the small denomination condition$( M=\$0.82$ ; $F( 1,\, 37) =5.99$ , $p< .05$ ). This finding is inconsistent with the idea that a large single denomination is perceptually valued more than many smaller denominations. Instead, the finding is consistent with the what‐the‐hell effect (Cochran and Tesser 1996), whereby people spend more in the large denomination condition once the decision to spend has been made.

In sum, participants were less likely to spend when they were given a large denomination relative to when they were given an identical amount in small denominations. However, once the decision to spend was made, the amount spent was higher in the large versus small denomination condition. A limitation of this study is that it was conducted with a student population using a relatively small sum of money. A replication with 24 elementary‐ and middle‐school children during Halloween involved giving them $1 (a substantial amount of money for the subject population) and measuring their likelihood of exchanging their money for candy. The results showed that all who received quarters purchased candy, whereas less than half the children in the$1‐bill condition did so (5/11, or 45.45%; $\chi ^{2}=9.46$ , $p< .01$ ). Despite the replication, studies 1b and 1c examine the denomination effect with significantly larger sums of money.

Another limitation is that the denomination conditions were confounded with the form of money, as the large denomination was in the form of a paper bill, whereas the small denominations were in the form of coins. Study 1b examines the denomination effect with a larger sum of money with an adult population in a more realistic shopping scenario. The design also allowed an examination of whether the difference in spending behavior is due to the difference in form (coins vs. paper bills) or the denomination of money.

### Study 1b

#### Method.

Seventy‐five adult drivers at a gas station in the Midwest were individually asked to participate in a short survey with three questions on gas usage (used as the cover story), for which they were given $5 in one of three forms: five$1 bills, five $1 coins, or one$5 bill. Holding the monetary form constant across the $5 bill and the five$1 bills allows us to test the effect of denomination on the amount spent with a larger amount. Holding the denomination constant at $1 across two types of forms (bills and coins) allows us to control for any effect of the form of the money. Note that people may retain the relatively uncommon$1 coin as a souvenir.

They were told, “If you would like, you may use this money to buy something from the convenience store at the gas station.” If they elected to go to the store, when they exited the store they were asked to provide their cash receipt from the store and that was used to calculate the amount they spent on gas and non‐gas‐related products, the manner in which they paid (cash or credit card), and whether they generated change in the transaction. Given that participants had to pay for the gas they filled, the dependent variable of interest is not whether they would spend but the likelihood of making a non‐gas‐related purchase.

A logistic regression was conducted on the likelihood of spending on a non‐gas‐related purchase, including denomination (1 or 5), form ($\mathrm{bill}\,=1$ ), the amount spent on gas, the frequency with which gas was purchased, the price paid for gas, and whether a credit card or cash was used in the transaction as predictor variables. The regression was significant ($\chi ^{2}( 6) =35.95$ , $p< .001$ ), revealing a marginal effect of denomination ($\beta =-.70$ , $\mathrm{Wald}\,=3.20$ , $p< .07$ ) and price of fuel ($\beta =-.32$ , $\mathrm{Wald}\,=4.71$ , $p< .05$ ), as well as whether a credit card was used ($\beta =7.21$ , $\mathrm{Wald}\,=10.57$ , $p< .001$ ; all other $p{'}\mathrm{s}\,> .25$ ). Specifically, consistent with the denomination effect, controlling for monetary form (i.e., across the five‐$1‐bills and one‐$5‐bill conditions), the likelihood of a non‐gas‐related purchase was higher when five $1 bills were given (24%) relative to when a single$5 bill was given (16%). The likelihood of spending when five $1 coins were given was the lowest of all three conditions (12%, or 3/25). The low level of spending is probably due to the fact that these coins are relatively low in circulation and were thus retained as souvenirs. Note that if one does not control for form of the money and combines the two five‐$1 conditions, then there is no significant difference between this combined condition and the $5 condition. Given the difference in non‐gas‐related purchase likelihood, it is not surprising that a regression on the amount spent on non‐gas‐related purchases as a function of denomination (1 or 5), form of money ($\mathrm{bill}\,=1$;$\mathrm{coin}\,=0$), and four covariates (amount spent on gas, frequency with which gas was purchased, price paid for gas, and whether a credit card was used) was significant ($R^{2}_{a}=.30$;$F( 6,\, 62) =5.91$,$p< .001$). In particular, the amount spent on non‐gas‐related purchases was significantly lower in the large versus small denomination condition ($\beta =-.22$,$t=-2.26$,$p< .05$). The average amount spent on non‐gas‐related expenses was higher in the five‐$1‐bills condition ($M=\$ .99$) and the five‐$1‐coins condition ($M=\$ .60$) relative to the single‐$5‐bill condition ($M=\$ .28$). The covariate of whether a credit card was used was also significant ($\beta =1.78$,$t=3.87$,$p< .001$; all other$p{'}\mathrm{s}\,> .15$). A similar regression on the amount spent conditional on purchase incidence (subset of individuals who made a non‐gas‐related purchase) was not significant ($F( 6,\, 4) =1.61$,$p> .34$). Although F‐statistics with higher degrees of freedom in the numerator than in the denominator are inherently unstable, it is worth noting that the amount spent conditional on having spent was in the opposite direction than what is predicted by the what‐the‐hell effect. The four individuals who spent and received a$5 bill spent an average of $1.72 on nongas expenditures, whereas the three who received five$1 coins spent an average of $4.32, and the six who received five$1 bills spent an average of $4.13. The low power due to the small sample size suggests these results need to be replicated before one can draw an inference from them. In sum, the likelihood of a non‐gas‐related purchase is higher for those given small denominations relative to those given a large denomination. Study 1c further examines the robustness of the denomination effect using a significantly larger amount of money in a more controlled shopping scenario. ### Study 1c #### Method. One hundred and fifty women in China, between the ages of 25 and 45, were recruited by a marketing research agency as study participants with the stated intention of completing a survey. Interviewers were locally recruited and were blind to the hypothesis. The women were taken individually to a room where, as a token of appreciation for completing the survey, they were handed an envelope containing RMB 100 in one of two forms: a single RMB 100 bill or five bills (one RMB 50 bill, two RMB 20 bills, and two RMB 5 bills). They were told that they could keep the money or purchase one or more of four products: soap, shampoo, bedding, and pots and pans. All products were of good quality, priced at about the average price in the market, and chosen to be appealing to the sample. The instructions emphasized that if the participants did not spend the entire amount, they would get to keep the change. All participants were asked to indicate their satisfaction with their decision and then rate the extent to which they needed each of the products ($1=\mathrm{do}\,$not need it at all;$2=\mathrm{do}\,$not need it very much;$3=\mathrm{neutral}\,$;$4=\mathrm{need}\,$it somewhat; and$5=\mathrm{need}\,$it very much). They also evaluated the prices of each of the products using a 5‐point scale, with higher numbers indicating affordability. All participants answered demographic questions regarding their personal and household incomes, education, family size, and marital status. Note that based on the monthly income of this sample, RMB 100 represented a significant amount of money. About 18.7% ($n=28$) of the women earned less than RMB 300 per month, 65% ($n=97$) earned between RMB 301 and 600 per month, and only 16.7% ($n=25$) earned more than RMB 600 per month. There were no differences in monthly incomes across the two denomination conditions. The family structure was also similar, with the average household size being about 3.3 in both conditions. Although soap was rated the most needed ($M=4.31$on a 5‐point scale)—followed by shampoo ($M=4.15$), bedding ($M=3.15$), and pots ($M=2.70$)—these ratings did not vary across the two denomination conditions ($F( 4,\, 145) =1.23$,$p> .30$). The price of soap was rated most favorably ($M=3.00$), followed by bedding ($M=2.57$), shampoo ($M=2.45$), and pots ($M=2.26$), but these ratings did not differ across the two denomination conditions ($F( 4,\, 145) =1.27$,$p> .30$). #### Results. A logistic regression on the likelihood of purchase, using denomination as the explanatory variable and incorporating participants’ evaluations of the prices of the four items as covariates, was significant ($\chi ^{2}( 5) =19.55$,$p< .005$; Nagelkerke pseudo$R^{2}=.22$). As predicted, the effect of the denomination condition was significant ($\chi ^{2}( 1) =5.27$,$p< .05$), as was the effect of the price covariates (shampoo:$\chi ^{2}( 1) =13.67$,$p< .001$; pots:$\chi ^{2}( 1) =4.70$,$p< .05$). When the women were given RMB 100 in small denominations, 9.3% (7/75) of the women did not spend, while this proportion doubled to 20% (15/75) when they were given the RMB 100 as a single bill (Fisher’s exact test$p< .05$). These data provide support for the denomination effect when the amount of money is significant. The amount of money spent given the decision to spend (conditional on purchase incidence) was significantly higher in the large denomination condition relative to the small denomination condition ($M{'}\mathrm{s}\,=67.67$and 56.76;$F( 1,\, 126) =5.42$,$p< .05$). As in study 1a, this result is consistent with the what‐the‐hell effect. A linear regression showed that the satisfaction ratings were contingent on the denomination of the money, the evaluations of prices of the four items, and the amount that people had spent ($R^{2}_{a}=.43$;$F( 6,\, 143) =19.34$,$p< .01$). Respondents reported a higher level of satisfaction when they had been provided the RMB 100 in smaller denominations ($M=3.60$) as compared to when they had been provided the RMB 100 as a single large bill ($M=3.23$;$\beta =.51$,$t=2.56$,$p< .05$). The coefficients associated with the price of pots and the amount spent were also significant ($\beta {'}\mathrm{s}\,=.43$and .03, respectively;$t=2.92$and 9.51, respectively;$p< .005$for both). In sum, women in the large denomination condition exhibited lower purchase incidence, spent more once the decision to spend had been made, but indicated lower postpurchase satisfaction relative to women in the small denomination condition. The higher satisfaction associated with smaller denominations is inconsistent with the perceptual fluency argument in which higher denominations are perceived to have greater value (Mishra et al. 2006). This result, however, can be interpreted in terms of the pain associated with paying. That is, the women may have been less satisfied when they received a large bill and spent it (as compared to when they received many small bills and spent them), as it was more painful to spend. Together, the results of studies 1a–1c extend the previous finding on the bias for the whole (Mishra et al. 2006). First, study 1 demonstrates the denomination effect with actual purchase decisions rather than purchase intentions. Second, the findings across different denominations and populations, involving significant amounts of money, attest to the robustness and generalizability of the effect. Third, in contrast to the perceptual explanation, we find that once the decision to spend has been made, individuals spend more and are less happy about it. This finding suggests that a large denomination may serve as a device to curb or control spending and that failure in self‐control leads to lower satisfaction. Studies 2 and 3 directly examine the extent to which self‐control reasons underlie the denomination effect. ## Study 2: Moderating Role of the Need for Self‐Control ### Method Seventy‐nine undergraduate students participated in a hypothetical decision‐making task in which need for self‐control in spending was manipulated by providing participants a spending history and a savings goal. Participants read the following scenario adapted from Soman and Cheema (2004): Imagine that you have recently graduated from college. A few months ago you accepted a job that is to your liking. Your monthly take home salary (after taxes) is$3500.

Your monthly non‐discretionary expenses like rent, food and groceries, utilities, phone bill, transportation, car insurance, and other essential expenses add up to $2500. This leaves you with$1000 each month that you can spend on discretionary expenses.

As you were leaving college to enter the workforce, everyone at school and home had advised you to start saving for the future—the words “it’s never too early to save for your future” keep echoing in your head. Taking a tip from a popular money management website, you decide to set a personal savings goal for yourself.

In particular, you decide that of the $1000 that remains after your essential expenses, you would save$400 each month. In other words, you would like to spend no more than $600 on discretionary expenses. In the last few months, you have been fairly successful in meeting this goal, achieving it in more than 70% of the months. It is now near the end of month and you are looking over your monthly expenses. This month, you have already spent ($700/$600) of the$1000 that you can spend on discretionary expenses, which is ($100 more than/exactly) your target spending amount. Thus, you have saved ($300/$400) this month. Participants were then told: Just then a friend calls to tell you that a local market research firm is looking for people in your age group for a focus group study they are conducting. The focus group would involve you along with other folks in your age group to talk about a new product idea that a firm is considering launching in the market. The focus group study will take about an hour and you will be paid$100 for your participation. You decide to participate in the focus group.

Your friend calls again to tell you that a group of your friends is getting together to go to the new mall that has recently opened after the focus group study. The new shopping mall has a variety of stores including coffee shops, restaurants, book stores, music stores, and of course apparel and clothing stores. You promise your friend that you will call back soon. You have to make a decision soon.

Note that when participants have already spent $700, they are$100 short of achieving their savings goal. However, the $100 payment from the focus group study provides a mechanism to achieve the savings goal only if they exercise self‐control and no part of the$100 is spent. In contrast, when participants have spent $600 and have already achieved their savings goal, the$100 payment from the focus group is akin to an unexpected windfall that can be spent freely. In other words, the previous spending along with the focus group payment implies a high need for self‐control in the under‐savings‐goal condition and relatively low (or no) need for self‐control in the savings‐goal‐achieved condition.

After reading the scenario, participants were told that the market research firm gave them a choice in terms of how they would like to be paid the $100. Participants could choose either a single$100 bill or five $20 bills. In order to make the choice vivid, images of the denominations were clearly shown. On a separate page, they were then asked to respond to the following question: “How likely are you to go to the shopping mall with your friends after the focus group?” ($1=\mathrm{very}\,$unlikely;$9=\mathrm{very}\,$likely). Participants were also asked to respond to four items that measured their affect toward receiving the$100. The items were adapted from the affect scale used by Pham et al. (2001). The four items were “receiving the $100 made me feel happy,” “I had unpleasant feelings about receiving the$100,” “receiving the $100 made me feel good,” and “receiving the$100 made me feel bad.” An average of the four items (the second and fourth items were reverse scaled) was used as a measure of affect (Cronbach’s $\alpha =.79$ ). Finally, two 7‐point scales were averaged ($r=.78$ ) to assess the efficacy of the need for self‐control manipulation: “I will have to think twice before spending the $100” and “to not spend the$100 from the focus group study, I need to control myself” ($1=\mathrm{not}\,$ at all; $7=\mathrm{very}\,$ much).

### Results

#### Manipulation Check.

A one‐way ANOVA revealed a significant effect of the need for control such that the mean ratings were significantly higher in the high need for self‐control relative to the low need for self‐control condition ($M{'}\mathrm{s}\,=5.48$ and 4.24; $F( 1,\, 77) =11.64$ , $p< .001$ ). Thus, the need for self‐control manipulation worked as intended.

#### Denomination Choice.

Participants’ choice of how they would like to receive the $100 varied significantly across the two need for self‐control conditions ($\chi ^{2}( 1) =21.47$,$p< .0001$). Specifically, while 73.81% (31/42) of the participants in the low need for self‐control condition chose to receive the$100 in five $20 denominations, only 21.62% (8/37) of the participants chose to receive the$100 in small denominations in the high need for self‐control condition. The preference for the large denomination in the high need for self‐control condition suggests that choice of the large denomination may be a precommitment device so as to reduce the likelihood of self‐control failure. Interestingly, participants in the low need for self‐control condition appear to prefer the small denominations, perhaps due to their greater convenience in spending.

#### Likelihood of Going to the Mall.

As expected, participants’ likelihood of going to the mall with friends after the focus group was significantly higher in the low need for self‐control condition relative to the high need for self‐control condition ($M{'}\mathrm{s}\,=6.90$ and 4.57; $F( 1,\, 77) =23.20$ , $p< .0001$ ). For the participants who chose the smaller denomination, the likelihood of going to the mall did not vary across the high versus the low need for self‐control conditions ($M{'}\mathrm{s}\,=6.63$ and 7.41; $F( 1,\, 77) =2.34$ , $p> .13$ ), but for the participants who chose the large denomination, the likelihood of going to the mall was significantly lower in the high relative to the low need for self‐control condition ($M{'}\mathrm{s}\,=4.00$ and 5.45; $F( 1,\, 77) =3.25$ , $p< .05$ ). These data add support to the contention that in the high need for self‐control condition, participants chose the large denomination as a precommitment mechanism to reduce the likelihood of self‐control failure in spending.

#### Affect.

Participants’ affective responses did not vary across the high versus low need for self‐control conditions ($M{'}\mathrm{s}\,=6.43$ and 6.45; $F( 1,\, 77) =.03$ , NS). Unlike Mishra et al. (2006), the affective responses did not vary significantly with denomination choice as well as the two self‐control conditions, even after accounting for their choice of denomination (all $p{'}\mathrm{s}\,> .90$ ).

Unlike study 1 in which participants were randomly assigned to one of the two denomination conditions, study 2 examined participants’ choice of denomination under different conditions. Participants’ choice of denomination more directly allows an examination of the extent to which they precommit and choose a large denomination rather than an identical amount of money in small denominations, in conditions in which there is a greater versus lower need for self‐control. The results support the idea that people choose a larger denomination as a precommitment device to exert self‐control in spending. Study 3 examines the extent to which individual differences in spending moderate the denomination effect.

## Study 3: Individual Differences in Self‐Control

Study 3 directly examines whether choice of receiving money in a large denomination relative to many small denominations varies with individual differences in spending (Rick et al. 2008). Rick et al. (2008) categorized individuals as tightwads and spendthrifts. Tightwads spend less than spendthrifts and have a higher pain associated with paying. If the choice of denomination is based on the desire to control spending per se, then spendthrifts are clearly more in need of self‐control than are tightwads. However, if the choice of denomination is based on the desire to reduce the anticipated pain of paying, then tightwads should be more likely to use higher denominations as a strategic precommitment device to curb their spending. Thus, the goal of this study was to examine whether those who have a higher need to avoid the pain of paying (tightwads) or those who have a higher need to control spending (spendthrifts) are more likely to deliberately choose a large denomination under conditions in which they may be tempted to spend. We also examined whether our inability to replicate Mishra et al.’s (2006) results relating to affect could be due to respondents not receiving real money.

### Method

One hundred and nineteen undergraduate students were given the following scenario: “Imagine that you have just been informed that you have been short‐listed to take part in a marketing study sponsored by a large food manufacturer in the town where your parents live. You are planning to visit them over Thanksgiving, and the time the study will run fits very well into your schedule. The one‐hour participation pays well—you have been told that you will receive $100 for taking part in the study where you need to discuss your preferences for different snack foods.” Participants were assigned at random to one of two conditions in which need for self‐control was manipulated. In the high need for self‐control condition, they were told that they would deposit into their bank the$100 they received. The location of the nearest automated teller machine where they could deposit was described as located in a “shopping mall which has your favorite coffee shop, restaurant, book store, music store, and of course apparel stores.” In the relatively low (or no) need for self‐control, this information was absent.

They were then asked to imagine that they had completed the marketing study and had to make a choice of the denomination in which they wished to receive their remuneration: a single $100 bill or five$20 bills. In order to make the choice more vivid, images of the money in both denominations were provided. They were then asked to list the reasons for their choice as completely as possible. The reasons were coded by two judges into the following categories: utilitarian (e.g., convenient, easy to store), self‐control related (less likely to spend, resist spending it, avoidance of overspending), perception related (e.g., feels like more, looks cooler), and other (e.g., less risky). The interrater reliability was high ($r=.92$ ), and disagreements were resolved through mutual discussion. Participants were then asked to complete a six‐item affect scale that included items tapping positive and negative affect (see table 2).

Table 2
Study 3: Factor Analysis and Scale Reliability for Affect Scales
 $100$10 Negative affect Subjective perception Positive affect Negative affect Factor loadings: Receiving the $__ made me feel happy. .595 .864 I had unpleasant feelings about receiving$__. .902 .848 I felt good at receiving the $__. −.681 .809 The$__ made me feel bad. .887 .861 It is a quite a bit of money. .873 .715 I feel like I have a lot of money now. .877 .628 Scale reliability $\alpha =.76$ $\alpha =.73$ $\alpha =.74$ $\alpha =.80$

An additional three items tapped ease of spending the money to serve as a manipulation check to assess the efficacy of the self‐control manipulation. These were “I will have to think twice before spending the $100,” “I will have a hard time parting with the$100,” and “I can spend the $100 easily” (reverse coded). Participants then indicated their likelihood of spending all or part of their$100 ($1=\mathrm{not}\,$ at all likely; $7=\mathrm{very}\,$ likely) and estimated how much they expected to spend (using an open‐ended scale).

The tightwad‐spendthrift scale ($\alpha =.69$ ; $\mathrm{median}\,=3.88$ ) was administered and used to categorize respondents into tightwads or spendthrifts ($n=58$ and 61, respectively). All participants were then offered an additional $10 to participate in a follow‐up study. The$10 was given in one of three forms ($10 bill,$2\times \$5$ bills, or $1\times \$ 5\,\mathrm{bill}\,+5\times \$1$ bills). The cover story was to investigate “the relationship between numerical and verbal intelligence.” A subset of participants ($n=77$ ) who agreed to participate were asked, “To help us better understand the value of your time, could you use the following scales to indicate how you feel about receiving the $10 for this study.” The six‐item affect scale was administered, with the same three additional items tapping the ease of spending$10 and an additional two items tapping the convenience of carrying it.

### Results

Table 3 displays the dependent measures across the different experimental conditions.

Table 3
Study 3: Results by Dependent Measure
 Cell means Overall No self‐control needed High self‐control needed Tightwads Spendthrifts Tightwads Spendthrifts Tightwads Spendthrifts Choice of $100 24/58 (41.4%) 31/61 (50.8%) 5/25 (20.0%) 18/35 (51.4%) 19/33 (57.6%) 13/26 (50.0%) Intention to spend (on a 1–7 scale): Overall 4.43 5.41 4.24 5.46 4.58 5.35 Those who chose$100 4.08 5.26 3.20 5.50 4.32 4.92 Those who chose $5\times \$ 20$4.68 5.57 4.50 5.41 4.93 5.77 Estimate of amount ($) they will spend (open ended): Overall 23.86 41.69 22.80 39.66 24.67 44.42 Those who chose $100 18.92 38.39 12.00 37.50 20.74 39.62 Those who chose$5\times \$20$ 27.35 45.10 25.50 41.94 30.00 49.23 Affect toward the money: Negative affect 1.59 1.60 1.52 1.58 1.64 1.63 Subjective perception: 5.04 5.09 5.23 5.03 4.90 5.17 For $100 choice 4.57 4.75 3.33 4.96 4.89 4.79 For$5\times \$20$ choice 5.37 5.33 5.70 5.10 4.90 5.54 Manipulation check (ease of spending the money): Overall 3.82 3.15 3.53 2.98 4.04 3.38

#### Manipulation Check.

The three 7‐point scales were averaged ($\alpha =.69$ ) to assess the efficacy of the need for self‐control manipulation. A one‐way ANOVA revealed a significant effect of the need for self‐control, such that the mean ratings were significantly higher in the higher need for self‐control relative to the no need for self‐control condition ($M{'}\mathrm{s}\,=3.75$ vs. 3.21; $F( 1,\, 117) =4.99$ , $p< .05$ ; $\eta ^{2}=.04$ ).

#### Choice.

Overall, 46.2% (55/119) chose the $100 bill. A logistic regression on the choice of receiving$100 as a single bill (coded as 1) versus five $20 bills (coded as 0), self‐control (no information or low need for$\mathrm{self}\,- \mathrm{control}\,=1$; busy mall or high need for$\mathrm{self}\,- \mathrm{control}\,=0$), value on the tightwad‐spendthrift scale, and their interactions was significant ($\chi ^{2}=10.63$,$p< .05$; Nagelkerke pseudo$R^{2}=.11$). As expected, individuals with a higher tightwad‐spendthrift score were less likely to choose the large denomination ($\beta =-.72$,$\mathrm{Wald}\,=7.25$,$p< .05$), as were those in the high need for self‐control condition ($\beta =-4.56$,$\mathrm{Wald}\,=7.74$,$p< .005$). However, the interaction between these two factors was significant and positive ($\beta =.94$,$\mathrm{Wald}\,=6.03$,$p< .05$). The positive interaction reflects that the effect of self‐control on the choice of denomination is stronger for tightwads than for spendthrifts. To understand the pattern better, we did a median split between tightwads and spendthrifts and separately cross‐tabulated their choice of denomination in the two self‐control conditions. Among tightwads, 20% (5/25) chose to receive the single$100 bill in the low need for self‐control condition, but 57.6% (19/33) did so in the high need for self‐control condition ($\chi ^{2}=8.28$ , $p< .01$ ). However, an approximately equal percentage of spendthrifts chose the two denominations in both self‐control conditions (choice of $100 bill: 51.43% [18/35] and 50% [13/26] in the low and high need for self‐control conditions, respectively;$\chi ^{2}=.01$,$p> .90$). To examine the robustness of the results, we conducted a three‐way split on the tightwad‐spendthrift scale (43 tightwads, 54 spendthrifts, 22 neither). Consistent with the earlier pattern, tightwads’ choice of a single$100 bill increased from 20% (4/20) in the low need for self‐control condition to 56.5% (13/23) in the high need for self‐control condition ($\chi ^{2}=5.97$ , $p< .05$ ), whereas that of spendthrifts was unchanged (choice of $100 bill: 53.1% [17/32] and 45.5% [10/22] in the low and high need for self‐control conditions, respectively;$\chi ^{2}=.31$,$p> .50$). Participants in the middle (neither tightwads nor spendthrifts) were directionally more likely to choose the single$100 bill in the high need for self‐control condition (choice of $100 bill: 25% [2/8] and 64.3% [9/14] in the low and high need for self‐control, respectively;$\chi ^{2}=3.14$,$p< .07$). Unexpectedly, those categorized as neither tightwads nor spendthrifts were directionally more likely to choose the$100 bill (9/14, or 64.3%) in the high self‐control needed condition than those individuals characterized as tightwads (13/23, or 56.5%). This anomaly may be due to small sample sizes and should be replicated before conclusions are drawn.

#### Affect.

A factor analysis using varimax rotation of the six items revealed two factors (see table 2 for factor loadings and scale reliability). The first factor taps negative affect (“I had unpleasant feelings,” “I felt good,” and “the $100 made me feel bad”), and the second factor taps subjective perception (“receiving the$100 made me feel happy,” “it is quite a bit of money,” and “I feel like I have a lot of money now”).

A regression on the three‐item negative affect scale ($\alpha =.76$ ) using need for self‐control ($0=\mathrm{low}\,$ ; $1=\mathrm{high}\,$ ), the tightwad‐spendthrift scale value, and their interaction as predictor variables revealed no significant effects ($p{'}\mathrm{s}\,> .80$ for all). The same analysis on the three‐item subjective perception scale ($\alpha =.73$ ) also revealed null results ($p{'}\mathrm{s}\,> .30$ for all), failing to replicate Mishra et al.’s results relating to differences in affect as a function of denomination.

#### Spending Intentions.

Across conditions, individuals who chose the five $20 bills (53.8%, or 64/119) reported directionally higher spending intentions than the 55 individuals who chose the single$100 bill ($M{'}\mathrm{s}\,=5.09$ and 4.75; $F( 1,\, 117) =1.17$ , $p> .25$ ).

A regression on spending intentions using need for self‐control ($0=\mathrm{low}\,$ ; $1=\mathrm{high}\,$ ), the tightwad‐spendthrift scale value, and their interaction as predictor variables revealed a main effect of individual difference reflecting higher spending intentions as the tendency to be a spendthrift increased ($t=3.74$ , $p< .001$ , $\beta =.47$ ; $R^{2}_{a}=.15$ ; $F( 3,\, 115) =7.78$ , $p< .001$ ). This attests to the face validity of the spendthrift‐tightwad scale (Rick et al. 2008).

#### Estimates of Spending Amount.

Reflecting the pattern for spending intentions, an analysis on the estimates of spending revealed a significant main effect for individual difference, with greater estimates of spending amount, the higher the individual’s score on the tightwad‐spendthrift scale ($F( 3,\, 115) =6.45$ , $p< .001$ ; $\beta =.30$ , $t=2.34$ , $p< .001$ ), with no other effects significant.

To summarize, the results showed that tightwads (who have a higher pain of paying) are more likely to consciously choose a large denomination over several smaller denominations in situations in which they need to exercise self‐control in spending, whereas spendthrifts are more likely to spend and spend more than tightwads but are equally likely to choose a larger or smaller denominations as they do not have any pain associated with paying. Ironically, or perhaps, obviously, it is those who need to exercise self‐control (spendthrifts) who are less likely to do so.

#### Affect toward $10. An exploratory factor analysis with a varimax rotation on the six‐item affect scale for the$10 given to participants revealed a different factor structure than that seen with the six‐item affect scale based on their reactions to the hypothetical $100. Factor 1 was positive affect, and value included four items, and factor 2 comprised two items measuring negative affect (see table 2 for factor loadings). These items were averaged to form the positive affect and negative affect scales, respectively ($\alpha {'}\mathrm{s}\,=.74$and .80). A 3 ($10 bill, $2\times \$ 5$bills,$1\times \$5\,\mathrm{bill}\,+5\times \$ 1$bills) × 2 (spending tendency: tightwad, spendthrift) analysis on the affect scales, revealed marginal effects of denomination and a denomination by individual difference interaction on negative affect ($F( 2,\, 71) =3.03$and 2.99,$p< .06$for both;$\eta ^{2}{'}\mathrm{s}\,=.08$for both). There were no effects for positive affect. The means show that tightwads have the highest negative affect for the single$10 denomination ($M=1.85$ ), followed by two fives ($M=1.79$ ) and then one five plus five one’s ($M=1.67$ ). Spendthrifts, however, show an inverse U‐shaped pattern, with lowest negative affect for the $10 note ($M=1.25$) and highest negative affect for the two fives ($M=2.42$), with the five plus five ones in the middle ($M=1.46$). To examine the robustness of these effects, we repeated the analysis, splitting spendthrifts into three categories. A 3 ($10 bill, $2\times \$ 5$bills,$1\times \$5\,\mathrm{bill}\,+5\times \$ 1$bills) × 3 (spending tendency: tightwad, spendthrift, neither) analysis on the affect scales showed that the null effects on positive affect were robust. However, there was a spending tendency by denomination interaction for negative affect ($F( 2,\, 69) =2.90$,$p< .05$), reflecting that tightwads have the highest negative affect for the single$10 denomination ($M=2.00$ ), followed by two fives ($M=1.86$ ) and then one five plus five one’s ($M=1.29$ ). Spendthrifts, however, show an inverse U‐shaped pattern, with lowest negative affect for the $10 note ($M=1.11$) and highest negative affect for the two fives ($M=2.54$), with the five plus five ones in the middle ($M=1.58$). For those categorized as neither, the highest negative affect is toward the five ones and a five ($M=1.86$), followed by the one ten ($M=1.57$) and the two fives ($M=1.37$). Looking at the data differently, the$10 bill has the highest negative affect among those categorized as tightwads ($M=2.00$ ) as compared to spendthrifts ($M=1.11$ ), with those categorized as neither in the middle ($M=1.57$ ). This could provide a clue as to why tightwads were less likely to choose the single $100 bill in the vignette task—they do not like it per se, but they choose it only when they wish to exercise self‐control. It is possible that the mere presence of the large denomination cues their need to exercise self‐control, which is potentially aversive and, therefore, leads to the higher negative affect toward the larger denomination. Treating spending tendency as a continuous variable, we examined the correlation between the scale and positive and negative affect for each of the three denominations separately. Consistent with the results of the median split, there were no significant correlations for positive affect for any of the denominations, whereas there was a negative correlation between negative affect and the spendthrift scale for the$10 denomination.

These patterns are not consistent with Mishra et al.’s (2006) results using real money. In fact, they are reversed in the case of tightwads for whom higher denominations appear to generate more negative affect and no more positive affect. They reflect the results in study 1c (the China sample) of higher postpurchase satisfaction with smaller denominations. They are, instead, consistent with the idea that people who do not wish to exercise self‐control in spending (spendthrifts) spend it, whereas those who wish to exercise higher self‐control in spending (tightwads) try and save larger denominations of money for fear that they will spend it all. For both groups, it appears that effects are driven by negative affect (the pain of paying), rather than positive affect (perceptual fluency), although the direction of the effects is different. In sum, we do not find any evidence that denomination affects positive affect using actual money but find that it does affect negative affect (or the pain of paying) contingent on people’s spending tendencies.

## General Discussion

Consistent with the bias for the whole documented by Mishra et al. (2006) for spending intentions, this research demonstrates the denomination effect, wherein money in a large denomination is less likely to be spent relative to an equivalent amount in many smaller denominations, in actual purchase decisions with real money. While adding to the growing body of literature that documents the violation of the principle of descriptive invariance in the domain of money, this research reports the results of three studies that suggest that the need to exercise self‐control in spending provides a compelling explanation for the denomination effect.

In contrast to the perceptual fluency explanation (Mishra et al. 2006), this article offers an alternative explanation based on ideas of self‐control and regulation. Specifically, this article demonstrates that money is less likely to be spent when it is in the form of a single large denomination relative to several small denominations, and, therefore, consumers strategically choose to receive money in larger denominations under conditions in which they wish to save and exercise self‐control in spending. Study 1 demonstrates in three field studies that individuals are more likely to spend in actual purchase decisions, when they have money in many small denominations relative to a single large denomination. In two of the three studies (1a and 1c), conditional on the decision to purchase, the amount spent is higher in the large denomination condition than in the small denomination condition. This finding is consistent with the what‐the‐hell effect (Cochran and Tesser 1996). Individuals often precommit to exert self‐control, but in situations in which the precommitment is violated resulting in self‐control failure, a what‐the‐hell effect may occur in which the behavior that the individual wished to control is accentuated in the short term. The what‐the‐hell effect on the amount spent conditional on purchase suggests that self‐control reasons underlie the denomination effect.

Studies 2 and 3 provide a more direct test of the self‐control explanation in the context of receiving money. Study 2 shows that the denomination effect is contingent on contextual differences in the need to exert self‐control. Participants preferred to be paid in a single large denomination when the need to exert self‐control and save the $100 was relatively high. In contrast, when the need for self‐control was relatively low and the$100 seemed like a windfall, individuals preferred to be paid the \$100 in many small denominations. These findings support the contention that in the need for self‐control condition, the large denomination is used as a precommitment device to exert self‐control in spending, whereas when there is no need for self‐control, the many small denominations make it easier to spend. Study 3 further shows that the denomination effect is contingent on individual differences in the ability to exert self‐control in spending and the associated pain of paying. Using the spendthrift‐tightwad scale (Rick et al. 2008), we find that the need for self‐control affects denomination choice for tightwads but not for spendthrifts. Tightwads prefer to be paid in a large denomination when the need for self‐control is high but prefer to be paid in many small denominations when there is no need for self‐control. Spendthrifts, however, do not appear to prefer one denomination over another, regardless of the need for self‐control.

This pattern suggests that it is not the need to exert self‐control in spending (which is greater for spendthrifts vs. tightwads) but the need to avoid the pain of paying (which is greater for tightwads) that drives the choice of denomination as a strategic precommitment device. As such, these results lead to a greater understanding of the tendency to spend or save—tightwads may be tightwads because they fear spending even though they are aware that they spend less than they would like to, and spendthrifts may be spendthrifts because they do not fear spending even though they are aware that they spend more than they would like to. A direct implication of this result is that merely making people aware of their spending tendency may be inadequate at changing their behavior; however, reducing the pain of paying for tightwads may lead to them increasing their spending, and invoking the pain of paying for spendthrifts should be effective at helping them curb their spending. Future research could identify routes to increase the pain of paying beyond those that have been previously identified, such as decomposing expenses by category (Srivastava and Raghubir 2002) and increasing the salience of cash (Raghubir and Srivastava 2008). These findings have implications for both consumer welfare (e.g., aimed at reducing consumer debt) and public policy (e.g., aimed at increasing consumer spending during an economic slowdown).

The results suggest that large denominations are perceived to be less fungible than smaller ones and that they are preferred by people who would like to save as a strategic way to control spending. The studies add to the literature on the subjective value of money and the consequences of individual differences in self‐control.

This article adds to the growing body of literature that suggests that alternative representations of the same stimuli can lead to systematic differences in people’s valuation and choice (Tversky et al. 1988). In particular, this article adds to the research on the subjective valuation of money (Kahneman and Tversky 1979; Mishra et al. 2006; Prelec and Loewenstein 1998; Raghubir and Srivastava 2002; Shafir et al. 1997) that has demonstrated differences due to the denomination of money (Mishra et al. 2006), the currency in which prices are provided (Raghubir and Srivastava 2002), the temporal framing of those prices (Gourville 1998), and the mode of payment—cash or gift certificates (Raghubir and Srivastava 2008). In general, amassing evidence for violations of normative rules is critical for these empirical regularities to be recognized and eventually integrated into new theories.

Our results also contribute to the literature on self‐control and its consequences. We find evidence for the what‐the‐hell effect (Cochran and Tesser 1996) in which people end up spending more once they have made the decision to spend. This can lead to goals being counterproductive, as demonstrated by Soman and Cheema (2004), giving further credence to the saying “in for a penny, in for a pound.” The results are also similar in spirit to the shopping momentum effect in which an initial purchase provides the momentum for a second unrelated purchase (Dhar, Huber, and Khan 2007). As in shopping momentum, it is possible that the mind‐set in choosing a large denomination is deliberation based, but once the decision to spend has been made, the mind‐set changes to more implementation based, and this shift in mind‐set encourages more spending.

The notion of precommitment as a self‐control device brings to mind Ulysses’ instructions to bind him to the mast. However, other less drastic measures, such as trying to curb the impulse to consume vices by reducing the amount purchased and stocked (Wertenbroch 1998) and setting spending limits (Ariely and Wertenbroch 2002), are similar to the idea of a large denomination as a way to curb spending. Although money is equally fungible regardless of how it is physically represented, our research highlights that denomination of money affects spending and saving behavior and is used strategically as a psychological barrier to spending.

## References

1. Ajzen, Icek, Thomas C. Brown, and Franklin Carvajal (2004), “Explaining the Discrepancy between Intentions and Actions: The Case of a Hypothetical Bias in Contingent Valuation,” Personality and Social Psychology Bulletin, 30 (9), 1108–20.
2. Ariely, Dan and Klaus Wertenbroch (2002), “Procrastination, Deadlines, and Performance: Self‐Control by Precommitment,” Psychological Science, 13 (May), 219–24.
3. Baumeister, Roy F. (2002), “Yielding to Temptation: Self‐Control Failure, Impulsive Purchasing, and Consumer Behavior,” Journal of Consumer Research, 28 (March), 670–76.
4. Baumeister, Roy F., Todd F. Heatherton, and Dianne M. Tice (1994), Losing Control: How and Why People Fail at Self‐Regulation, San Diego, CA: Academic Press.
5. Chandon, Pierre and Brian Wansink (2002), “When Are Stockpiled Products Consumed Faster? A Convenience‐Salience Framework of Post‐purchase Consumption Incidence and Quantity,” Journal of Marketing Research, 39 (August), 321–35.
6. Chen, Chau‐Nan (1976), “Currency Denominations and the Price Level,” Journal of Political Economy, 84 (February), 179–84.
7. Cochran, Winona and Abraham Tesser (1996), “The ‘What the Hell’ Effect: Some Effects of Goal Proximity and Goal Framing on Performance,” in Striving and Feeling: Interactions among Goals, Affect and Self‐Regulation, ed. Leonard L. Martin and Abraham Tesser, Mahwah, NJ: Erlbaum, 99–120.
8. Dhar, Ravi, Joel Huber, and Uzma Khan (2007), “The Shopping Momentum Effect,” Journal of Marketing Research, 44 (3), 370–78.
9. Gourville, John T. (1998), “Pennies‐a‐Day: The Effect of Temporal Reframing on Transaction Evaluation,” Journal of Consumer Research, 24 (March), 395–408.
10. Heath, Chip and Jack Soll (1996), “Mental Budgeting and Consumer Decisions,” Journal of Consumer Research, 23 (June), 40–52.
11. Hoch, Stephen J. and George F. Loewenstein (1991), “Time‐Inconsistent Preferences and Consumer Self‐Control,” Journal of Consumer Research, 17 (March), 492–507.
12. Kahneman, Daniel and Amos Tversky (1979), “Prospect Theory: An Analysis of Decision under Risk,” Econometrica, 47 (March), 263–91.
13. Mishra, Himanshu, Arul Mishra, and Dhananjay Nayakankuppam (2006), “Money: A Bias for the Whole,” Journal of Consumer Research, 32 (March), 541–49.
14. O’Guinn, Thomas C. and Ronald J. Faber (1989), “Compulsive Buying: A Phenomenological Exploration,” Journal of Consumer Research, 16 (September), 147–57.
15. Pham, Michel T., Joel B. Cohen, John W. Pracejus, and G. David Hughes (2001), “Affect Monitoring and Primacy of Feelings in Judgment,” Journal of Consumer Research, 28 (September), 167–88.
16. Prelec, Drazen and George Loewenstein (1998), “The Red and the Black: Mental Accounting of Savings and Debt,” Marketing Science, 17 (Winter), 4–28.
17. Raghubir, Priya and Joydeep Srivastava (2002), “Effect of Face Value on Product Valuation in Foreign Currencies,” Journal of Consumer Research, 29 (December), 335–47.
18. ——— (2008), “Monopoly Money: The Effect of Payment Coupling and Form on Spending Behavior,” Journal of Experimental Psychology: Applied, 14 (3), 213–25.
19. Rick, Scott I., Cynthia E. Cryder, and George Loewenstein (2008), “Tightwads and Spendthrifts,” Journal of Consumer Research, 34 (April), 767–82.
20. Rook, Dennis W. (1987), “The Buying Impulse,” Journal of Consumer Research, 14 (September), 189–99.
21. Schelling, Thomas C. (1978), “Egonomics, or the Art of Self Management,” Proceedings of the American Economic Association, 68 (February), 290–94.
22. Shafir, Eldar, Peter Diamond, and Amos Tversky (1997), “Money Illusion,” Quarterly Journal of Economics, 112 (May), 341–74.
23. Shefrin, Hersh M. and Richard H. Thaler (1988), “The Behavioral Life Cycle Hypothesis,” Economic Inquiry, 26 (October), 609–43.
24. Soman, Dilip and Amar Cheema (2004), “When Goals Are Counterproductive: The Effects of Violation of a Behavior Goal on Subsequent Performance,” Journal of Consumer Research, 31 (June), 52–62.
25. Srivastava, Joydeep and Priya Raghubir (2002), “Debiasing Using Decomposition: The Case of Memory‐Based Credit Card Expense Estimates,” Journal of Consumer Psychology, 12 (3), 253–64.
26. Thaler, Richard H. (1985), “Mental Accounting and Consumer Choice,” Marketing Science, 4 (3), 199–214.
27. Thaler, Richard H. and Hersh M. Shefrin (1981), “An Economic Theory of Self‐Control,” Journal of Political Economy, 89 (2), 392–406.
28. Tversky, Amos, Shmuel Sattath, and Paul Slovic (1988), “Contingent Weighting in Judgment and Choice,” Psychological Review, 95 (July), 371–84.
29. Wertenbroch, Klaus (1998), “Consumption Self‐Control by Rationing Purchase Quantities of Virtue and Vice,” Marketing Science, 17 (4), 317–37.
30. Wertenbroch, Klaus, Amitava Chattopadhyay, and Dilip Soman (2007), “On the Perceived Value of Money: The Reference Dependence of Currency Numerosity Effects,” Journal of Consumer Research, 34 (June), 1–10.

## References

1. Ajzen, Icek, Thomas C. Brown, and Franklin Carvajal (2004), “Explaining the Discrepancy between Intentions and Actions: The Case of a Hypothetical Bias in Contingent Valuation,” Personality and Social Psychology Bulletin, 30 (9), 1108–20.
2. Ariely, Dan and Klaus Wertenbroch (2002), “Procrastination, Deadlines, and Performance: Self‐Control by Precommitment,” Psychological Science, 13 (May), 219–24.
3. Baumeister, Roy F. (2002), “Yielding to Temptation: Self‐Control Failure, Impulsive Purchasing, and Consumer Behavior,” Journal of Consumer Research, 28 (March), 670–76.
4. Baumeister, Roy F., Todd F. Heatherton, and Dianne M. Tice (1994), Losing Control: How and Why People Fail at Self‐Regulation, San Diego, CA: Academic Press.
5. Chandon, Pierre and Brian Wansink (2002), “When Are Stockpiled Products Consumed Faster? A Convenience‐Salience Framework of Post‐purchase Consumption Incidence and Quantity,” Journal of Marketing Research, 39 (August), 321–35.
6. Chen, Chau‐Nan (1976), “Currency Denominations and the Price Level,” Journal of Political Economy, 84 (February), 179–84.
7. Cochran, Winona and Abraham Tesser (1996), “The ‘What the Hell’ Effect: Some Effects of Goal Proximity and Goal Framing on Performance,” in Striving and Feeling: Interactions among Goals, Affect and Self‐Regulation, ed. Leonard L. Martin and Abraham Tesser, Mahwah, NJ: Erlbaum, 99–120.
8. Dhar, Ravi, Joel Huber, and Uzma Khan (2007), “The Shopping Momentum Effect,” Journal of Marketing Research, 44 (3), 370–78.
9. Gourville, John T. (1998), “Pennies‐a‐Day: The Effect of Temporal Reframing on Transaction Evaluation,” Journal of Consumer Research, 24 (March), 395–408.
10. Heath, Chip and Jack Soll (1996), “Mental Budgeting and Consumer Decisions,” Journal of Consumer Research, 23 (June), 40–52.
11. Hoch, Stephen J. and George F. Loewenstein (1991), “Time‐Inconsistent Preferences and Consumer Self‐Control,” Journal of Consumer Research, 17 (March), 492–507.
12. Kahneman, Daniel and Amos Tversky (1979), “Prospect Theory: An Analysis of Decision under Risk,” Econometrica, 47 (March), 263–91.
13. Mishra, Himanshu, Arul Mishra, and Dhananjay Nayakankuppam (2006), “Money: A Bias for the Whole,” Journal of Consumer Research, 32 (March), 541–49.
14. O’Guinn, Thomas C. and Ronald J. Faber (1989), “Compulsive Buying: A Phenomenological Exploration,” Journal of Consumer Research, 16 (September), 147–57.
15. Pham, Michel T., Joel B. Cohen, John W. Pracejus, and G. David Hughes (2001), “Affect Monitoring and Primacy of Feelings in Judgment,” Journal of Consumer Research, 28 (September), 167–88.
16. Prelec, Drazen and George Loewenstein (1998), “The Red and the Black: Mental Accounting of Savings and Debt,” Marketing Science, 17 (Winter), 4–28.
17. Raghubir, Priya and Joydeep Srivastava (2002), “Effect of Face Value on Product Valuation in Foreign Currencies,” Journal of Consumer Research, 29 (December), 335–47.
18. ——— (2008), “Monopoly Money: The Effect of Payment Coupling and Form on Spending Behavior,” Journal of Experimental Psychology: Applied, 14 (3), 213–25.
19. Rick, Scott I., Cynthia E. Cryder, and George Loewenstein (2008), “Tightwads and Spendthrifts,” Journal of Consumer Research, 34 (April), 767–82.
20. Rook, Dennis W. (1987), “The Buying Impulse,” Journal of Consumer Research, 14 (September), 189–99.
21. Schelling, Thomas C. (1978), “Egonomics, or the Art of Self Management,” Proceedings of the American Economic Association, 68 (February), 290–94.
22. Shafir, Eldar, Peter Diamond, and Amos Tversky (1997), “Money Illusion,” Quarterly Journal of Economics, 112 (May), 341–74.
23. Shefrin, Hersh M. and Richard H. Thaler (1988), “The Behavioral Life Cycle Hypothesis,” Economic Inquiry, 26 (October), 609–43.
24. Soman, Dilip and Amar Cheema (2004), “When Goals Are Counterproductive: The Effects of Violation of a Behavior Goal on Subsequent Performance,” Journal of Consumer Research, 31 (June), 52–62.
25. Srivastava, Joydeep and Priya Raghubir (2002), “Debiasing Using Decomposition: The Case of Memory‐Based Credit Card Expense Estimates,” Journal of Consumer Psychology, 12 (3), 253–64.
26. Thaler, Richard H. (1985), “Mental Accounting and Consumer Choice,” Marketing Science, 4 (3), 199–214.
27. Thaler, Richard H. and Hersh M. Shefrin (1981), “An Economic Theory of Self‐Control,” Journal of Political Economy, 89 (2), 392–406.
28. Tversky, Amos, Shmuel Sattath, and Paul Slovic (1988), “Contingent Weighting in Judgment and Choice,” Psychological Review, 95 (July), 371–84.
29. Wertenbroch, Klaus (1998), “Consumption Self‐Control by Rationing Purchase Quantities of Virtue and Vice,” Marketing Science, 17 (4), 317–37.
30. Wertenbroch, Klaus, Amitava Chattopadhyay, and Dilip Soman (2007), “On the Perceived Value of Money: The Reference Dependence of Currency Numerosity Effects,” Journal of Consumer Research, 34 (June), 1–10.