Spending Time versus Spending Money

Journal of Consumer Research, Inc.

Spending Time versus Spending MoneyAuthor(s): Erica Mina Okada and Stephen J. HochSource: Journal of Consumer Research, Vol. 31, No. 2 (September 2004), pp. 313-323Published by: The University of Chicago PressStable URL: http://www.jstor.org/stable/10.1086/422110 .

Accessed: 20/12/2014 10:40

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

The University of Chicago Press and Journal of Consumer Research, Inc. are collaborating with JSTOR todigitize, preserve and extend access to Journal of Consumer Research.

http://www.jstor.org

This content downloaded from 169.230.243.252 on Sat, 20 Dec 2014 10:40:56 AMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=ucpress

http://www.jstor.org/stable/10.1086/422110?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp


313

� 2004 by JOURNAL OF CONSUMER RESEARCH, Inc.● Vol. 31 ● September 2004All rights reserved. 0093-5301/2004/3102-0007$10.00

Spending Time versus Spending Money

ERICA MINA OKADASTEPHEN J. HOCH*

We find systematic differences in the way that people spend time versus money.Ex post, people are able to more easily accommodate negative outcomes byadjusting the value of their temporal inputs. Ex ante, people are willing to spendmore time for higher risk, higher return options whereas when spending moneythe pattern is reversed and the more standard pattern of increasing risk aversionis observed. Although accurate assessment of the opportunity costs of time is keyto making good decisions, ambiguity in the value of time promotes accommodationand rationalization.

Benjamin Franklin said, “Remember time is money,”implying that time is valuable just like money. A moral

interpretation might be that time should be spent wisely. Aneconomic interpretation might be that the value of one’stime can be expressed in monetary terms as an opportunitycost, often represented by one’s after-tax wage rate (Becker1965). Either way, both time and money are exchange me-diums. People can acquire products by paying hard cash orexpending effort, but typically there is a trade-off betweenthe two currencies. Consumers generally pay a premium forconvenience and incur temporal transaction costs in the pro-cess of information search and uncertainty reduction (Carl-son and Gieseke 1983; Marmorstein, Grewal, and Fishe1992; Stigler 1961) or as an additional cost in the form ofdelays (Taylor 1994).

But there are reasons to suggest that consumers do nottreat time and money in the same fashion, even if norma-tively they should. In economics, an individual’s hourlywage ratew is a commonly used benchmark for one’s op-portunity cost, and accordingly one should be indifferentbetween payingw in cash and spending an hour of timeworking in order to acquire a product. One systematic de-viation from this assumption is the underweighting of op-portunity costs (Thaler 1980, 1999), which leads individualsto work one hour for even though they are only willing≥ wto pay in cash to spare the expenditure of their own! weffort. People fail to calculate the opportunity costs of timewhen unstated (Neumann and Friedman 1980) and under-estimate them when prompted (Hoskin 1983). Underappre-ciation of opportunity costs produces temporal spendthriftsand monetary misers, wasting time on lower-value activities

*Erica Mina Okada is assistant professor of marketing at the Universityof Washington, Box 353200, Seattle, WA 98195 ([email protected]), and Stephen J. Hoch is John J. Pomerantz professor of marketingat the Wharton School, University of Pennsylvania, Philadelphia, PA 19035([email protected]). The authors thank the editor, associate editor,and reviewers for their input.

at the expense of higher-value pursuits, though Larrick, Mor-gan, and Nisbett (1990) have shown that people can betrained to use normative cost-benefit rules. Another devia-tion from the model is that people’s perceived valuation oftime is not as precise asw. This ambiguity in the value oftime is the focus of our research.

THE VALUE OF TIME IS AMBIGUOUSA key difference between temporal and monetary cur-

rencies is that the opportunity cost of money is easy toassess, whereas the opportunity cost of time is more am-biguous. Opportunity cost captures the concept of the nextbest use for a resource. The question is: what is the nextbest thing that one can do with the money or time if onechooses not to spend it on the item at hand? Money has areadily exchangeable market, is highly liquid and fungible,and can be saved. A dollar is a dollar no matter the trans-action type, and so what comes to mind as the next bestuse for money remains fairly constant across situations. Incontrast, the range of the second-best uses for one’s time isvariable. Time is not as readily exchangeable; it is perishableand despite some ability for postponement cannot be in-ventoried easily for later use. People may be more practicedand therefore reliable at spending money than bartering timefor goods. At the same time, people have plenty of oppor-tunities each day to spend (or waste) their time, but thetransactions may be more ad hoc and informal than thoseinvolving money.

Spending time and money also differ because consumersface different set budget constraints. When consumers spendmoney, they face two constraints: one chronic, their totalwealth, and one acute, due to momentary liquidity issuesthat often can be overcome through the borrowing or savingprocess. The current low U.S. savings rate and mountingcredit card debt suggest monetary budget constraints oftenare quite real. Consumers also face temporal budget con-straints, as evidenced by the many media accounts of the



314 JOURNAL OF CONSUMER RESEARCH

harried consumer and complaints by middle-aged parents,teenagers, and even some senior citizens that they neverhave enough time. We believe, however, that temporalbudget constraints are pretty soft or at least elastic (Hsee1995). There are 24 hr. in a day. With 8 hr. devoted to workand 8 for sleep, that still leaves 8 hr. for discretionary ac-tivities. Since time use is discretionary, so is its valuation.

How valuable is this time? Should it be valued at one’swage rate, or are the opportunity costs associated with thistime more valuable due to scarcity or less valuable due toabundance? Clearly there may be important cultural (U.S.,EU, Asia) and life-stage (teen, middle-aged, senior) differ-ences, but our view is that the value of time is flexibleregardless. People can more easily adjust the value of theirtime to the particulars of the situation. Because we facemany decisions each day about whether and how to spendthis discretionary time, we have more degrees of freedomabout how we can spend this resource. We recognize thatpeople in some professions (e.g., doctors, lawyers, con-sultants, commissioned sales people) have had lots of prac-tice at pricing their work hours and have a more precisenotion of the opportunity costs of their discretionary time,though as Scitovsky (1976) cogently pointed out, sometimesthese people act as if work is actually consumption. Wewould argue that everyone ends up spending a significantamount of available time that falls in between the extremeof eating and sleeping and required hours on the job forpay.

Consumers are likely to be more adaptable in what theybelieve their time is worth and what constitutes an accept-able implicit wage rate. The greater ambiguity of the valueof time likely supports what Hsee (1995, 1996) has calledelastic justification, allowing individuals to be more oppor-tunistic in their valuation of time. We believe that peoplehave a much easier time living with wasting time than wast-ing money, a view supported by Soman’s (2001) findingthat sunk costs involving expenditures of time are moreeasily ignored than expenditures of money. And if consum-ers have an easier time self-justifying losses of time, thenwe might expect that they are more willing to take riskswhen making temporal investments than when making mon-etary investments.

In a series of five experiments, we provide evidence thatit is the inherent ambiguity in the value of time that supportsand justifies a different spending pattern than that observedwith money. The first two studies examine ex post expen-ditures of time versus money. Experiment 1 shows that sat-isfaction with an acquisition is less sensitive to the con-sumption experience when people pay in time than in moneyas people have an easier time accepting bad outcomes whenthey pay with time. The experiment provides direct evidencethat people flexibly adjust the value of their temporal inputsto be congruent with the realized outcome. Basically youshould get what you pay for, so if the outcome is positive(negative), people infer a higher (lower) value of the timethat they expended in the acquisition. This kind of flexiblevaluation is harder to do with money. Note that these results

cannot be explained by a cognitive dissonance story sincethe need for dissonance reduction should be greater afterspending money (Soman 2001). Experiment 2 shows thatwhen people pay using a fictional currency with a volatileexchange rate (suggesting ambiguity in value) their ex postevaluations of positive and negative outcomes are more likethose made when paying with time, again suggesting thatthe ambiguity in the value of the currency sustains the ob-served pattern.

The remaining studies examine ex ante expenditures oftime and money. Experiments 3 and 4 utilize simple, equal-expected-value lotteries of both monetary and nonmonetaryoutcomes. When people spend money, they display the stan-dard pattern of increasing risk aversion to higher-variance(higher-risk, higher-reward) gambles. When paying withtime, the pattern completely reverses and respondents arewilling to pay more time for higher-variance gambles. Thefinal experiment uses multiattribute products, either averageon all attributes or strong on some and weak on others, andagain demonstrates that people are willing to take more riskswhen paying with time rather than money, presumably be-cause they can more easily adjust the value of their temporalexpenditures to whatever is the realized outcome. We donot wish to argue that the observed differences in spendingtime and money are purely cognitive or purely motivationalas we find that distinction artificial (Tetlock and Levi 1982).We simply provide evidence that the greater ambiguity inthe value of time allows people to be more creative in theirmotivated reasoning (Kunda 1990) about ex post and exante decisions involving time.

EXPERIMENT 1

Experiment 1 tested the prediction that having paid intime attenuates the level of satisfaction with the overallexchange transaction because the opportunity cost of timeis relatively flexible and furthermore reflects the consump-tion outcome and that having paid in money amplifies thelevel of satisfaction with the overall exchange transactionbecause the opportunity cost of money is more immutable.The basic idea is that when the consumption experienceturns out bad, people will have an easier time rationalizingan expenditure of time than money because the value oftime is more labile and easier to write off. We measureddirectly the effect of the realized consumption experienceon people’s valuation of their prior temporal expenditures.

Method

Three hundred sixty undergraduate students participatedin the experiment. The respondents completed a question-naire presenting two scenarios where they had already con-sumed a product that they had previously acquired throughthe expenditure of either time or money. Two product classeswere used: a dinner for two at a downtown restaurant anda pair of athletic shoes, both relevant and familiar to therespondent pool.

There were two levels of currency: time and money. A



SPENDING TIME VERSUS SPENDING MONEY 315

TABLE 1

SATISFACTION RATINGS (1–7) DEPENDENT ON CURRENCYAND OUTCOME TYPE

Valence of outcome

Athletic shoes Restaurant

Money Time Money Time

Positive experience 6.2 5.5 6.0 5.1Negative experience 1.8 2.1 1.6 1.9Difference 4.4 3.4 4.4 3.2

pretest ( ) determined that for $50, respondents weren p 26willing to perform about 4 hr. of data-entry work, a wagerate of about $12.50/hr. In the time condition of the mainexperiment, respondents were told they had spent 4 hr. doingdata entry in order to acquire either a dinner for two or apair of athletic shoes. In the money condition, they weretold that they had spent $50 for the acquisition. The ex-periment implemented a design according to Winer, plan 5(1971). The design was a Latin square crossing two levelsof product with two levels of currency. The respondentswere randomly assigned either to group 1 or group 2 tocontrol for presentation order.

There were two between-subjects variables: consumptionoutcome and purchase decision context. The between-sub-jects consumption outcome, either positive or negative, wascrossed with the Latin square. Respondents in the positive(negative) condition were given a scenario where the con-sumption experience at the restaurant or with the athleticshoes was described in a very favorable (unfavorable) way.The positive and negative outcomes for the restaurant sce-nario appear below:

Positive Outcome: When you arrived at the restaurant fordinner, you right away liked the ambiance. The table whereyou were seated was cozy and out of the way of foot traffic.The service was very prompt and courteous, and the foodwas delicious, the best you’ve had in a long time. Your friendand you had a great time. You would definitely go back thereagain and would recommend the restaurant to family andfriends.

Negative Outcome: When you arrived at the restaurant fordinner, you right away disliked the ambiance. The table whereyou were seated was tight and right next to the noisy servicestation. The service was very slow and rude, and the foodwas horrific, the worst you’ve had in a long time. Your friendand you had an awful time. You would definitely never goback there again and would discourage any family or friendfrom ever going there.

The other between-subjects variable had three levels,which differed according to the context in which the initialpurchase decisions were made. One-third of the respondentswere given a scenario where they were told that they hadmade the initial purchase decisions individually. Anotherthird were told that they had made the initial purchase de-cisions as a part of a group: “You and some of your class-mates decided as a group to . . . ” And the other third weretold that they had made the initial purchase decisions duringa very busy week of final examinations. Previous researchhas demonstrated that the need for self-justification becomesless relevant after group decisions (Whyte 1991), and wewere interested to see whether the relative flexibility in thevaluation of time would support accommodation to the re-alized outcome even when motivation to rationalize waslower. We introduced the constrained time condition to testthe robustness of any observed flexibility in the valuationof time when time scarcity is made more salient and the

opportunity cost (time spent on data entry vs. studying forfinals) more apparent.

Satisfaction Ratings and Opportunity CostMeasures

Respondents provided a total of three responses to the setof two scenarios that they read. First, for each of the twoscenarios, respondents were asked how satisfied they wouldbe overall after consuming the product. This measured themerit of the transaction, and was on a scale of 1 to 7, with1 being “I would be extremely dissatisfied” and 7 being “Iwould be extremely satisfied.” High (low) merit indicatesthat respondents were happy (unhappy) with the overallexchange transaction. In the time condition only, they werethen asked what they thought the monetary equivalent oftheir time expenditure would be. This provided a direct mea-sure of the opportunity cost of time in monetary terms. Ifthe stated monetary equivalent of time expenditure is higher(lower) when the consumption outcome is positive (nega-tive), it would support the idea that flexible temporal wagerates allow consumers to more effectively accommodate toreceived outcomes by adjusting the value of their timeinputs.

Results and Discussion

Our attempt to manipulate the initial decision context (in-dividual choice, group choice, busy time) had no impact onthe results. The grouping variable and order also were notsignificant. The results after collapsing across these variablesappear in table 1. There is an obvious main effect of outcometype, where positive outcomes are rated as more satisfyingthan negative ones, , . Our mainF(1, 356)p 2,742 p ! .0001prediction, based on the inherent ambiguity in the value oftime, was that the difference in the overall satisfaction levelsbetween the positive and negative consumption outcomeswould be greater when paying with money than time. Thesignificant currency by outcome interaction supports thisprediction, , . This significant in-F(1, 356)p 32.3 p ! .0001teraction indicates that, as predicted, the difference betweenpositive and negative outcomes is greater when paying withmoney ( ) than when paying with time ( ).M p 4.4 M p 3.3For both product scenarios, when people paid $50 in cashand had a positive experience, they were significantly more




FIGURE 1

RELATIONSHIPS AMONG VALENCE, $ EQUIVALENTS, ANDSATISFACTION

satisfied than when they paid with 4 hr. of time spent com-pleting a market research study (both ). In contrast,p ! .01when people paid $50 in cash and had a negative experience,they were significantly less satisfied than when they paidwith 4 hr. of time spent completing a market research study(both ). Paying with time attenuates the ex post sat-p ! .01isfaction that respondents express for both positive and neg-ative outcomes.

To get a better handle on the underlying process, weexamined the monetary equivalents of 4 hr. of work providedby subjects in the time condition. Ex post, subjects valuetheir time at over $20/hr. when their 4-hr. time expenditurewas spent in service of a positive outcome, but when theoutcome is less satisfying, that same 4-hr. investment isworth less than $15/hr. This 44% wage premium accom-panying positive over negative outcomes is statistically sig-nificant, , . It is worth pointing out thatF p 43.2 p ! .0001all of these money equivalents are greater than the 4 hr. thatsubjects were willing to spend for $50 indicated in the pre-test. One possible reason for this is that the question posedhere was the willingness-to-accept/sell rather than the will-ingness-to-pay/spend question.

$ equivalent of 4 hr. spentValence of outcome Athletic shoes RestaurantPositive Experience $82.67 $81.80Negative Experience $57.91 $56.18

Of course we did not predict that the level of satisfactionwith the positive outcome would be higher when payingwith money than time, only that the difference in satisfactionbetween positive and negative outcomes would be attenu-ated with time payments. We conducted a simple path anal-ysis to examine the interrelationships between the valenceof the outcome, the 4-hr. dollar equivalents, and satisfaction.The results appear in figure 1. The numbers are standardizedcoefficients and all of them are significant at . Thep ! .02coefficients in parentheses come from the multiple regres-sion of satisfaction onto valence and dollar equivalents. Va-lence has a positive influence on both satisfaction (.81) anddollar equivalents (.32). Also the simple relationship be-tween dollar equivalents and satisfaction is positive (.19),as previously shown in table 2. After controlling for valence,however, we see a small negative impact of dollar equiva-lents on satisfaction (�.08), suggesting that the simple pos-itive relation is spurious. This makes sense, since satisfactionshould decrease as one pays more holding constant the con-sumption experience. It also provides insight into the ob-served attenuation in satisfaction observed when payingwith time that can be more flexibly valued than when payingwith a more fixed-value currency like money.

The data show that time attenuates or money amplifiesthe happiness or unhappiness that consumers feel after con-sumption. These effects occurred irrespective of the need tojustify a decision (lack of any effect due to group vs. in-dividual decision) or whether subjects were alerted to thecosts of time (busy week of tests). We acknowledge thatthese two manipulations might have been weak, but this

does suggest that the results are somewhat robust. Afterspending time and ending up with a negative outcome, sub-jects are flexible in valuing their time and can increase theirlevel of satisfaction by lowering their implicit wages ratesby over 40%. Verbal protocols support this reasoning. Ex-amples of protocols from subjects who paid in time for anegative outcome generally contained rationalizations suchas:

• “It cost me no money, just time.”• “At least I didn’t pay money.”• “Even though the shoes (were bad), they were free in

monetary terms, and I took only four hours.”

A positive outcome leads people to infer a higher value fortheir time to match their greater satisfaction with the out-come. People who have paid money for the positive outcomeend up being the most satisfied, probably due to self-per-ception and ego bolstering for making a high return on theirinitial $50 investment. Protocols from people who paid inmoney and received a positive outcome support this self-perception process:

• “If the shoes are awesome, $50 is hardly a cost.”• “The money is a small factor when the service and food

are superior.”• “I love food. $50 for a great dinner is totally worth it.”

In making most judgments and decisions, people haveonly partial information and must construe or construct themissing pieces (Griffin and Ross 1991). Because of the am-biguous nature of time and people’s relative inexperiencein exchanging it, the set of information available for valuingtime is not as complete as for valuing money. And so thereis greater construal in valuing time, which leads to greatervariance in its valuation.

EXPERIMENT 2

The first study provides support for our prediction thatpeople have an easier time accommodating positive and neg-ative purchase outcomes when paying in a more flexibly




TABLE 2

SATISFACTION RATINGS (1–7) DEPENDENT ON CURRENCY AND OUTCOME TYPE

Valence of outcome

Restaurant Athletic shoes

Fixed currency Volatile currency Time Fixed currency Volatile currency Time

Positive experience 6.3 6.1 5.4 6.2 5.6 5.5Negative experience 1.6 1.9 2.3 1.8 2.1 2.3Difference 4.7 4.2 3.1 4.4 3.5 3.2

valued currency like time than when paying with a morefixed value currency like money. The purpose of this studyis to determine if it is indeed the ambiguity of the value ofa currency that supports the observed effect. To do this, wemanipulated the ambiguity of the value of money. We in-troduced a hypothetical foreign currency, , and used it asEthe monetary unit instead of $US. In the ambiguous con-dition, the exchange rate between $US and was volatile.ERespondents were told that they had exchanged $US for

twice in the past year, once at , and once atˇ ˇE E p $0.80. The current exchange rate was , withˇ ˇE p $1.20 Ep $1.00

some analysts predicting an appreciation to inE p $1.50next month or two, and others predicting a devaluation to

. In the fixed condition, the exchange rate wasE p $0.50stable at . Because the value of money is moreE p $1.00ambiguous in the volatile than fixed condition, our pre-diction is that having paid in a higher-variance currencyshould attenuate the overall merit of the transaction in amanner similar to time. Our manipulation of the ambiguityof money is similar in spirit to that of Soman (2001), whoshowed that people treat sunk costs after paying with timethe same as when paying money when their wage rate ismade salient.

Method

One hundred eighty undergraduate students participatedin this study. There were three levels of currency: 4 hr. oftime, volatile , and fixed . As in experiment 1, we crossedˇ ˇE Ea Latin square with a between-subjects factor, but in thedesign of experiment 2, the Latin square crossed the sametwo products with consumption experience, and the be-tween-subjects factor was type of currency.


The data appear in table 2. As before we found a maineffect of outcome type, where positive outcomes are ratedas more satisfying than negative ones, ,F(1, 168)p 546.0

. Now with three levels of currency, we duplicatedp ! .0001the significant currency by outcome interaction,

, . Across the two product sce-F(2, 168)p 6.07 p ! .003narios, the difference between positive and negative out-comes was greatest for the fixed currency ( ), nextM p 4.55largest for the volatile currency ( ), and smallestM p 3.85when paying with time ( ). These three differencesM p 3.15all are significantly different from each other, . Thep ! .03

data clearly show that the ambiguity of the currency, inincreasing order from fixed foreign currency to volatile for-eign currency to time, attenuates the happiness or unhap-piness that consumers feel after consumption.

Experiment 1 sheds light on how individuals value moneyand time as currencies of exchange. Even when the usageexperience is described in the same positive (negative) way,how satisfied (dissatisfied) individuals feel about the overallexchange transaction depends on whether they paid for theproduct in time or money. Furthermore, experiment 2 sup-ports our assumption that the ambiguity of the value of timefacilitates this difference in satisfaction. When the ambiguityof the value of money is increased, individuals start ac-counting for expenditures in a volatile currency more likethey would when paying with time. An alternative approachto increasing the variance of the opportunity cost of moneymight be to present concrete examples of alternative waysin which money can be spent. Participants could be askedto think about the second best thing that can be bought for$50, as well as to recall occasions where they had spent $50on a frivolous purchase that they ended up never using orwhere they had simply lost $50. Making a diverse set ofpotential outcomes salient and available for recall (Schwarzand Vaughn 2002) might likewise increase ambiguity in thevaluation of money.

Our results demonstrate that the valuation of time ex-penditures is more flexible than the valuation of monetaryexpenditures, and the value of expenditures in ambiguouscurrencies is derived at least in part from the usage expe-rience. If the consumption experience is positive, individualsassign a greater value to the time spent in the exchangetransaction, and if the experience is negative, they infer alesser value to the same amount of time. The same goes formonetary expenditures of ambiguous value, to a lesser ex-tent. Furthermore, this effect does not change when the needfor self-justification is reduced or when the urgency of timeis made more salient.

EXPERIMENT 3

The first two studies illustrated the differential ex postvaluation of time and money as units of expenditure. Payingin time instead of money ends up working like an insurancepolicy against a negative consumption outcome. In exper-iments 3 and 4 we go on to examine whether or not thisdistinction between time and money influences behavior atthe time of purchase.




TABLE 3

WILLINGNESS-TO-PAY IN MONEY AND TIME FOREXPERIMENT 3

Variance inoutcomes

EV p $50 EV p $100

$ Minutes $ Minutes

Low 18.13 78.1 34.27 110.3Medium 18.42 89.5 33.67 128.2High 16.68 98.1 28.68 154.4

NOTE.—EV p expected value.

Using simple lotteries where we hold constant expectedvalue, we predict that people will display a greater prefer-ence for higher-variance and therefore higher-risk gambleswhen paying with time rather than when paying with money.If people are more flexible in their thinking about the valueof time, then they should be willing to pay more for higher-variance gambles that offer a lower probability of winninga larger amount. With time, if they win, then great, and ifthey lose then it is only a limited waste of time, not hardcash.

Rottenstreich and Hsee (2001) compared people’s riskpreferences for receiving affectively rich stimuli (kisses, va-cations, electric shocks) to those for money. Respondentswere less sensitive to probabilities when setting willingness-to-pay and willingness-to-avoid prices for noncash as tocompared to monetary outcomes. They argue that peopleprefer affect rich stimuli at low probabilities, whereas athigh probabilities they prefer money due to the strength ofimagery-induced emotions. Similarly Rettinger and Hastie(2001) found risk aversion with gambles but risk seekingwhen the same problem was imbedded in a different contentdomain (e.g., traffic tickets). Although there are differencesin the task we used in experiments 3 and 4, in a similarvein, we expected people to be less sensitive to probabilitieswhen paying with time rather than money.

Method

Three hundred sixty-three undergraduate students partic-ipated in the experiment, which was conducted as a penciland paper questionnaire. The respondents were presentedwith six lotteries and were asked to indicate their willing-ness-to-pay in time (number of hours and minutes) and thenin money (dollars) for each, a total of twelve measures fromeach respondent. For the time-based willingness-to-paymeasure, respondents were told that they would be com-pleting research assistant work (data coding, data entry)related to a market research project. A short filler taskseparated the two sets of judgments. A subset of 90 re-spondents provided their initial set of twelve willingness-to-pay judgments, completed about a half hour of unrelatedpencil and paper experiments, and then made the sametwelve willingness-to-pay judgments a second time. Thisallowed us to investigate test-retest reliability for both timeand money judgments.

The six lotteries varied in expected value and risk. Threeof the lotteries had an expected value of $50, and the otherthree had an expected value of $100. At each level ofexpected value, there were three ranges of outcomes. Thehighest-variance lotteries offered a 20% chance of winningsomething (and the winning payoff was the highest) butan 80% chance of winning nothing at all. The medium-risk lotteries had a 50% chance of winning something, anda 50% chance of winning nothing at all. The lowest-risklotteries had an 80% chance of winning something (butthe winning payoff was low) and only a 20% chance ofwinning nothing. The presentation order between time and

money and the order among the six lotteries were random-ized.

Range ofoutcomes EV p $50 EV p $100Low 20% # 0 + 80% # $62.50 20% # 0 + 80% # $125Medium 50% # 0 + 50% # $100 50% # 0 + 50% # $200High 80% # 0 + 20% # $250 80% # 0 + 20% # $500


We analyzed the willingness-to-pay data using a repeated-measures analysis of variance. Willingness to pay in bothtime and money currencies served as the dependent variablewith expected value ($50 or $100) and variance in outcomes(low, medium, high) as the other design variables. All thedata were within subjects. In order to deal with the differentscale properties (mean and variance) of the money and timemeasures, we separately transformed the time and moneydata intoz-scores using the mean and standard deviationcalculated across lotteries and respondents. This of courseeliminates the main effect of currency since the grand meanof both is rescaled to zero, but it also resulted in data thatdid not violate the homogeneity of variance assumption. Theresults were very similar when analyzing the raw data. Theraw data are displayed in table 3.

The key effect is the interaction of currency and variancein outcomes, which is quite significant, ,F(2, 362)p 30.3

. The key driver of this interaction a significantp ! .0001linear contrast across the three levels of variance in out-comes, , . Follow-up simple-F(1, 362)p 36.3 p ! .0001effects tests showed that the interaction occurred becauserespondents became relatively more risk averse as the var-iance of the gambles increased when paying with money,

, , whereas they became more riskF(2, 362)p 3.9 p ! .01seeking with higher-variance gambles when paying withtime, , . There also was a sig-F(2, 362)p 15.1 p ! .0001nificant three-way interaction between currency type byexpected value by variance in outcomes,F(2, 362)p

, . This interaction was driven by the fact16.1 p p .0001that risk aversion increased more for higher-variance gam-bles with higher expected value when paying with money,whereas when paying with time the exact opposite oc-




curred—respondents became more risk seeking withhigher-variance outcomes with greater expected value.

Examination of the willingness-to-pay judgments withmoney shows that respondents display the standard riskaversion found with these types of gambles. On average,respondents were willing to pay only about one-third of theactual expected value of each of the gambles. Moreover, asthe variance in outcomes increased, respondents becameeven more risk averse. For lotteries with an expected valueof $50, the willingness to pay in money decreased 8% whencomparing the high-variance to the low-variance gambles.Likewise, for those lotteries with an expected value of $100,the willingness to pay in time decreased 16% when com-paring the high-variance with the low-variance gambles. Asimple two-way interaction of expected value with variancesuggests that the 16% decrease in willingness to pay for$100 expected value gambles is greater than the 8% declinefor $50 expected value gambles, ,F(2, 362)p 4.2 p p

..04The willingness to pay in time results are almost the mir-

ror image of those for willingness to pay in money. As thevariance in outcomes increased, respondents displayed lessrisk aversion. For lotteries with an expected value of $50,the willingness to pay in time increased 26% when com-paring the high-variance to the low-variance gambles. Forlotteries with an expected value of $100, the willingness topay in time increased 40% when comparing the high-vari-ance to the low-variance gambles. As with the willingnessto pay money in data, a simple two-way interaction of ex-pected value with variance suggests that the 40% increasein willingness to pay for $100 expected value gambles isgreater than the 26% increase for $50 expected value gam-bles, , .F(2, 362)p 13.6 p ! .001

Although these respondents displayed increased risk aver-sion to higher-variance gambles when asked to pay withmoney, a standard result with the magnitude of the lotteriesthat utilized here, they displayed decreased risk aversion forhigher-variance gambles when paying with time. It was al-most as if paying with time were like playing with someoneelse’s money. If they win the high-variance gamble, theycan treat it as a windfall, and if they lose, it only costs thema little time. When paying with money, they ex ante antic-ipate the regret that they will feel if they lose the higher-variance gamble, which is pretty likely (80% chance).

For the subset of 90 people who provided two sets oftime and money willingness-to-pay judgments, we calcu-lated two individual-level correlation coefficients, one fortime and one for money. Respondents displayed significantlygreater reliability expressing their willingness to pay inmoney (average ) than in time (average ).r p .69 r p .55A paired t-test ( ) on the correlations (after Fisher’st p 2.5r to z transformation) revealed that the difference was sta-tistically significant, though both sets of judgments displaya reasonable level of reliability. We actually were a bit sur-prised that people were so proficient using the time scale,since the direct mapping between the outcomes (expressedin dollars) and time is less transparent. Possibly we would

have seen more degradation in judgment consistency of thetime-based willingness to pay if there had been a long delaybetween the two sets of ratings.

EXPERIMENT 4

The findings of experiment 3 suggest that individuals tendto be less risk averse when paying for monetary lotterieswith time rather than money. The purpose of experiment 4was to rule out the possibility that this somehow could havebeen an artifact due to the congruence or compatibility ofthe reward currency (gambles with dollar outcomes) withthe measurement of willingness to pay in money but notwith time (e.g., Fischer and Hawkins 1993; Tversky, Sattath,and Slovic 1988). Therefore we conducted this study toreplicate our findings, using lotteries with nonmonetaryrewards.

Method

In a pretest of 42 respondents, we first approximated theperceived value of two types of products: airline tickets andtelevision sets. The purpose was to create sets of lotteriesof comparable expected value, and varying risk, using non-monetary payoffs. We asked the respondents to estimate thevalue of three alternatives for each product category. Forthe airline tickets, the three alternatives (and their respectiveaverage values) in order of increasing value were a round-trip economy-class ticket to any West Coast destination fromanother West Coast city where the pretest was conducted($225), a round-trip economy-class ticket to any destinationin the lower 48 states ($360), and a round-trip economy-class ticket to any destination in North America, includingHawaii and the Caribbean ($505). For the televisions, thethree alternatives (and their respective average values) inincreasing order of value were a 19-inch-screen set ($181),a 24-inch-screen set ($261), and a 36-inch-screen set ($438).

From these pretest data, we created high-, medium-, andlow-risk lotteries of comparable expected value, using air-line tickets and television sets as payoffs. The strict equalityof the expected values across the three lottery types withina product class is not necessary, so we used round numbersfor winning percentages of all six lotteries. Eighty-one un-dergraduates participated. For each of the six lotteries ofcomparable expected value and varying risk, we measuredthe respondents’ willingness to pay first in time and thenmoney, or in the reverse order. Additionally we measuredthe perceived scarcity of time by asking respondents to whatextent they agreed (1–7 scale) with the statement that theyhad a lot of time to spare.

Range of outcomes Airline ticket Television setLow 80% # West Coast

+ 20% # nothing70% # 9′′ screen

+ 30% # nothingMedium 50% # lower 48


+ 50% # nothingHigh 35% # North America


+ 70% # nothing




TABLE 4

WILLINGNESS-TO-PAY IN MONEY AND TIME FOREXPERIMENT 4

Variance in outcomes

Airline ticket Television set

$ Minutes $ Minutes

Low 69.70 179 40.70 168Medium 51.90 227 36.90 172High 41.50 333 35.10 180


The data appear in table 4 with willingness to pay as thedependent variable and currency, risk, product type, andtheir interactions as the independent variables. Again, thekey effect is the two-way interaction between currency andrisk, which is statistically significant, ,F(2, 160)p 36.3

. Polynomial contrasts indicate that the interactionp ! .0001is driven by opposite linear trends when paying with timeor money, , . As the varianceF(1, 160)p 27.2 p ! .0001(risk) of an outcome increases, individuals are willing topay less with money but more with time. This suggests thatrelative risk aversion in money and risk seeking in time isgeneralizable to nonmonetary lotteries as well. Follow-upsimple-effect tests of variance in outcomes within currencyindicate that the decrease in willingness to pay with moneyis significant, , , as is the increase inF(2, 160)p 5.4 p ! .01willingness to pay with time, ,F(2, 160)p 45.0 p p

. The significant three-way interaction among cur-.0001rency, risk, and product, , ), sug-F(2, 160)p 11.5 p ! .0001gests that the magnitude of both the relative risk aversionin money and risk seeking in time were greater for airlinetickets than for television sets. Inclusion of each respon-dent’s perceived scarcity of time had no influence on theabove pattern of results.

The results of experiments 3 and 4 suggest that individualsdistinguish between the fixed valuation of money and theadjustable valuation of time a priori and choose between thetwo currencies accordingly when making purchase deci-sions. If individuals expect a high variance in the outcome,they prefer to pay in time, anticipating that in the case ofa negative outcome, paying in time would mitigate the dis-satisfaction that they would feel with the overall exchangetransaction. In doing so, individuals effectively buy an in-surance policy against negative consumption outcomes, sim-ilar to Thaler and Johnson’s (1990) demonstration of howplaying with house money affects risky choice. When theoutcome is more certain, insurance against a bad outcomeis not as necessary and paying in money makes sense. Thenext experiment looks at another case of how individualsmatch the flexibility of currency valuation to different ex-pected outcomes.

EXPERIMENT 5

Experiment 5 measures how individuals pay for enrichedversus impoverished alternatives with time and money. Anenriched alternative has more positive, as well as negative,attributes than an impoverished alternative (Shafir 1993).There are more attribute trade-offs associated with an en-riched alternative, and thus the consumer may perceive theoutcome to be more uncertain. Building on the previousargument regarding certain versus uncertain outcomes, theprediction is that individuals would have a relative prefer-ence to pay in time rather than in money for enriched al-ternatives and a relative preference to pay in money ratherthan in time for impoverished alternatives.

Method

Three hundred sixty respondents completed the experi-ment in fulfillment of a class requirement. Two differentproduct categories were selected: personal digital assistantsand digital cameras. For each of the product categories, weconstructed alternatives described on six attributes. Usingthe procedure described in Shafir (1993), we constructedalternatives with either low or high variance across the levelsof each of the attributes. Shafir referred to these as eitherimpoverished (low-variance) or enriched (high-variance) al-ternatives. Brand A is average on all attributes whereasBrand B is outstanding on some and weak on other attrib-utes. Brand B offers some upside potential on some attrib-utes, but there is also some downside risk due to weak ratingson other attributes. The design was completely between sub-jects. In the context of an experimental session with otherunrelated tasks, each respondent saw one of the four possiblemultiattribute product descriptions (two product categories,either impoverished or enriched) and indicated their will-ingness to pay in either money or time (hours spent doingthe data entry task described earlier).

Attribute Brand A Brand BImage quality 5 10Compactness 5 2Features 5 9Ease of use 5 1Warranty 5 3Durability 5 8


The results are displayed in table 5. As can be seen, whenpeople indicate their willingness to pay with money, theyplace higher value on low-variance than on high-varianceoptions. Average on all dimensions looks more attractive torespondents than being strong on some and weak on otherattributes. On average, higher-variance options are valuedat about 80% of the low-variance options. In contrast, whenpaying with time, respondents are willing to pay 30% morefor the higher-variance option. These observations are con-firmed by the significant variance in attribute rating by will-ingness to pay in time versus money interaction,

, . The simple effects of attributeF(1, 352)p 20.0 p ! .0001rating variance are significant for both time and money (both




TABLE 5

WILLINGNESS-TO-PAY FOR HIGH- AND LOW-VARIANCE ALTERNATIVES WITH TIME VERSUS MONEY

IN EXPERIMENT 5

Attributeratings

Personal digitalassistant Camera

$ Hours $ Hours

Low variance 175 4.9 161 2.9High variance 118 5.2 148 4.5

), though obviously in the opposite direction fromp ! .05each other.

Similar to our findings in experiments 3 and 4, this studyshows that people systematically spend money differentlythan they spend time when faced with alternatives that differin their level of risk or certainty of the outcome. Peopleplace a higher value on riskier options when paying withtime; the opposite occurs when they pay with money,whether the alternatives are traditional two outcome gamblesor multiattribute products. These results suggest that thelarger ex ante valuations that consumers place on riskieralternatives may occur because respondents realize that theywill have an easier time rationalizing a potentially bad out-come after paying with time than hard earned cash. In thissense, paying with time is similar to purchasing an insurancepolicy.

GENERAL DISCUSSION

Money is the most ubiquitous form of currency used inexchange transactions. It is not, however, a requisite inexchange, and, long before money was institutionalized,goods were exchanged through barter. In certain parts of theworld moneyless exchanges exist as the dominant form oftrade even today. In modern economies, however, consumersare accustomed to transacting primarily in money and havebeen trained in its valuation through cumulative exchangeexperiences. As such, academic research on exchange hasfocused almost exclusively on money-based transactions.This research examined another form of currency—time.Nonmonetary expenditures have been incorporated into mi-croeconomic models, but the basic approach there has beento assign monetary values to nonmonetary currencies andto analyze exchange transactions based on the assumptionthat rational agents consider all currencies equivalently.Though this monetary conversion serves as a useful nor-mative and prescriptive tool, it does not fully capture dif-ferences in the underlying psychology.

The intrinsic difference between spending time andspending money lies in the estimation of their respectiveopportunity costs. Money is highly liquid and storable forfuture use, which makes the estimation of opportunity costrelatively straightforward and consistent over time and con-text. Time, on the other hand, is not as liquid and furthermorehighly perishable, which makes the opportunity cost more

difficult to estimate and more dependent on context. Thevaluation of money remains relatively fixed, while the val-uation of time is flexible. When expenditures are made intime, we found that individuals infer time’s ambiguous valueby referencing other less ambiguous cues, such as usageexperience. Individuals also appear to appreciate this dif-ference in the valuation of the two currencies a priori, andchoose their currency of payment accordingly upon productacquisition. When the outcome of the usage experience isuncertain, individuals prefer to pay in time, which wouldpartially protect them from the downside.

When consumers spend their own time rather than theirown money, they benefit from a built-in insurance policyin the sense that they can fairly easily rationalize both lowerand higher implicit wage rates. This flexibility in valuingone’s time influences: (a) ex ante decisions because con-sumers anticipate that they will have an easier time ration-alizing and writing off bad outcomes if they occur; and (b)ex post decisions because consumers accommodate to what-ever they get by adjusting the value of their temporal inputs.

An interesting question is whether people spend moneyor time more wisely. The answer at this point is not so clear.Spending time confers option value because we can spendcheap time or valuable time depending on the circumstances.Sometimes this can make us happier after the fact and mo-tivate us to take some risks that we would not if moneywere the required currency. Whether the built-in flexibilityand ambiguity that characterizes the expenditure of time isnormatively correct largely comes down to the actual op-portunity costs of time. For most people, one’s wage rateis only a rough indicator of the value of any specific periodof time. Weekend time does not have the same value asnormal working hours or overtime or time spent resting oron vacation. The reason that the value of time is flexible isbecause in many situations we have a large measure ofdiscretion. Whenever we are a distance away from the tem-poral budget constraint, we have plenty of degrees of free-dom in how to go about spending, investing, or wasting ourspare time. Flexibility in spending time may serve a role inmaintaining our psychological immune system (Gilbert etal. 2000; Gilbert and Ebert 2002).

The limitations of the current research suggest directionsfor future work. Since all of our experiments involve un-dergraduates at U.S. universities who presumably have lowopportunity costs (though they surely complain a lot abouthow busy they are) and are not likely to be as experiencedat valuing their time as certain professionals, it would beinteresting to study other populations, both here in Americaand in other cultures. For example, people in different eco-nomic circumstances who are much more practiced at sellingtheir time for services rendered, such as attorneys, doctors,and consultants, may be more precise in their valuation oftime than college students. With less ambiguity in the valueof time, our findings may be attenuated or not replicated.Europeans, as witnessed by their long summer vacations,may value time more dearly; alternatively, people broughtup in certain eastern religions (e.g., Buddhism) may think




about time and money very differently. Travels in India sug-gest that opportunity costs for many people there are verylow, and this may alter how people spend time and money.

In a related fashion, it would be interesting to study sit-uations where people normally spend time rather than money(waiting in line, letting wine age in the cellar, watching atree grow) and see whether the current findings would gen-eralize or a different pattern would obtain. In none of ourfive studies did people have to make a direct side-by-sidechoice between spending time and money. Rather, we stud-ied choice patterns among those who were spending moneyand compared them to those spending time. Future studiesmight provide a direct choice between the two currenciesthat would more closely resemble situations where the con-sumer has the option of spending either time or money. Weconceptualized time as a medium of exchange in our study,where people acquire something in consideration for per-forming some activity, rather than the simple passing ofminutes and hours in exchange for which one receives noth-ing, such as unexpected delays at airports (Leclerc, Schmitt,and Dube 1995). Another issue to consider is the nature ofthe activities engaged in while spending time. Is the timespent completing pleasant or unpleasant work? Is the timeexciting, boring, intense, or relaxed?

This research suggests that it is ambiguity in the valueof time that allows people to be more creative in their mo-tivated reasoning about decisions (Kunda 1990). But itwould be beneficial to develop a deeper understanding ofthe underlying psychological processes that guide people tointerpret the value of time as dear in some cases and cheapin others. Is it a biased retrieval process where people areinfluenced by contextual variables and do not bother to en-gage is an exhaustive consideration of competing uses oftime or is it that people recode their level of effort duringthe time expended?

Time is a resource, as is money. And people can expendtime or money in exchange for value. Whether it be nothaving enough money or time, constrained resources preventpeople from getting and doing whatever they want. If timewere not a resource, the concept of being busy would notexist. People do not always spend money wisely, becausethey cannot always accurately assess the opportunity costsassociated with deferring a purchase or buying an alternativeoption. Our research shows that the inherent ambiguity onthe value of time can make it even more difficult to accu-rately assess the wisdom or folly of expenditures of time.

[Dawn Iacobucci and David Glen Mick served as editorsand Frank Kardes served as associate editor for this

article.]

REFERENCESBecker, Gary S. (1965), “A Theory of the Allocation of Time,”

Economic Journal, 75 (September), 493–517.Carlson, John A. and Robert J. Gieseke (1983), “Price Search in

a Product Market,”Journal of Consumer Research, 9 (March),357–65.

Fischer, Gregory and Scott Hawkins (1993), “Strategy Compati-bility, Scale Compatibility, and the Prominence Effect,”Jour-nal of Experimental Psychology: Human Perception and Per-formance, 19 (June), 580–97.

Gilbert, Daniel T., Ryan P. Brown, Elizabeth C. Pinel, and TimothyD. Wilson (2000), “The Illusion of External Agency,”Journalof Personality and Social Psychology, 79 (November),690–700.

Gilbert, Daniel T. and Jane E. J. Ebert (2002), “Decisions andRevisions: The Affective Forecasting of Changeable Out-comes,”Journal of Personality and Social Psychology, 82(April), 503–14.

Griffin, Dale W. and Lee Ross (1991), “Subjective Construal, So-cial Inference, and Human Misunderstanding,” inAdvancesin Experimental Social Psychology, Vol. 24, ed. Mark P.Zanna, San Diego, CA: Academic Press, 319–59.

Hoskin, R. E. (1983), “Opportunity Cost and Behavior,”Journalof Accounting Research, 21 (Spring), 78–95.

Hsee, Christopher K. (1995), “Elastic Justification: How Temptingbut Task-Irrelevant Factors Influence Decisions,”Organiza-tional Behavior and Human Decision Processes, 62 (June),330–37.

——— (1996), “Elastic Justification: How Unjustifiable FactorsInfluence Judgments,”Organizational Behavior and HumanDecision Processes, 66 (April), 122–29.

Kunda, Ziva (1990), “The Case for Motivated Reasoning,”Psy-chological Bulletin, 108 (November), 480–98.

Larrick, Richard P., James N. Morgan, and Richard E. Nisbett(1990), “Teaching the Use of Cost-Benefit Reasoning in Eve-ryday Life,” Psychological Science, 1 (November), 362–70.

Leclerc, France, Bernt H. Schmitt, and Laurette Dube (1995),“Waiting Time and Decision Making: Is Time Like Money?”Journal of Consumer Research, 22 (June), 110–19.

Marmorstein, Howard, Dhruv Grewal, and Raymond P. H. Fishe(1992), “The Value of Time Spent in Price-Comparison Shop-ping: Survey and Experimental Evidence,”Journal of Con-sumer Research, 19 (June), 52–61.

Neumann, B. R. and L. A. Friedman (1980), “The Effects of Op-portunity Costs on Project Investment Decision: A Replicationand Extension,”Journal of Accounting Research, 18 (Au-tumn), 407–19.

Rettinger, David A. and Reid Hastie (2001), “Content Effects onDecision Making,”Organizational Behavior and Human De-cision Processes, 85 (July), 336–59.

Rottenstreich, Yuval and Christopher K. Hsee (2001), “Money,Kisses, and Electric Shocks: On the Affective Psychology ofRisk,” Psychological Science, 12 (May), 185–90.

Schwarz, Norbert and Leigh Ann Vaughn (2002), “The AvailabilityHeuristic Revisited: Ease of Recall and Content as DistinctSources of Information,” inHeuristics and Biases: The Psy-chology of Intuitive Judgment, ed. Thomas Gilovich, DaleGriffin and Daniel Kahneman, Cambridge: Cambridge Uni-versity Press, 103–19.

Scitovsky, Tibor (1976),The Joyless Economy, New York: OxfordUniversity Press.

Shafir, Eldar (1993), “Choosing versus Rejecting: Why Some Op-tions Are Both Better and Worse than Others,”Memory &Cognition, 21 (June), 546–55.

Soman, Dilip (2001), “The Mental Accounting of Sunk TimeCosts: Why Time Is Not Like Money,”Journal of BehavioralDecision Making, 14 (July), 169–85.

Stigler, George J. (1961), “The Economics of Information,”Jour-nal of Political Economy, 59 (June), 213–25.




Taylor, Shirley (1994), “Waiting for Service: The Relationship be-tween Delays and Evaluations of Service,”Journal of Mar-keting, 58 (April), 56–69.

Tetlock, Phillip E. and Ariel Levi (1982), “Attribution Bias: Onthe Inconclusiveness of the Cognition-Motivation Debate,”Journal of Experimental Social Psychology, 18 (Januray),68–88.

Thaler, Richard H. (1980), “Toward a Positive Theory of ConsumerChoice,”Journal of Economic Behavior and Organization, 1(March), 39–60.

——— (1999), “Mental Accounting Matters,”Journal of Behav-ioral Decision Making, 12 (September), 183–206.

Thaler, Richard H. and Eric J. Johnson (1990), “Gambling withthe House Money and Trying to Break Even: The Effects ofPrior Outcomes on Risky Choices,”Management Science, 36(June), 643–60.

Tversky, Amos, S. Sattath, and Paul Slovic (1988), “ContingentWeighting in Judgment and Choice,”Psychological Review,95 (July), 371–84.

Whyte, Glen (1991), “Diffusion of Responsibility: Effects on theEscalation Tendency,”Journal of Applied Psychology, 76(June), 408–15.

Winer, Benjamin J. (1971),Statistical Principles in ExperimentalDesign, New York: McGraw.



Documents

Spending Time versus Spending Money