Subjective time discount rates among teenagers and adults: Evidence from Israel

Please cite this article in press as: Lahav, E., et al., Subjective time discount rates among teenagers and adults: Evidence from Israel. J. Socio-Econ.(2010), doi:10.1016/j.socec.2010.02.012

ARTICLE IN PRESSG Model

SOCECO-962; No. of Pages 8

The Journal of Socio-Economics xxx (2010) xxx–xxx

Contents lists available at ScienceDirect

The Journal of Socio-Economics

journa l homepage: www.e lsev ier .com/ locate /soceco

Subjective time discount rates among teenagers and adults:Evidence from Israel

Eyal Lahava,∗, Uri Benziona, Tal Shavitb

a Department of Economics, Ben-Gurion University, Israelb School of Business Administration, College of Management, Israel

a r t i c l e i n f o

Article history:Received 12 August 2009Received in revised form 17 January 2010Accepted 11 February 2010

JEL classification:D90, D91

Keywords:Time preferenceDiscount ratesTeenagersCushion effect

a b s t r a c t

This study is an empirical examination of the personal discount rates used by teenagers and adults. Theparticipants answered a time-preference questionnaire and were asked about the bank interest paid fordeposits. The results demonstrate that teenagers have a very high personal discount rate in comparisonto adults. In addition, we found that receiving a regular allowance increases the teenagers’ willingnessto wait. Our results indicate that the influence of teenagers’ poor financial knowledge, which ought toincrease the personal discount rate, is stronger than the effect of their safety net that should reduce thepersonal discount rate.

© 2010 Elsevier Inc. All rights reserved.

1. Introduction

The personal discount rate is the rate at which individuals tradecurrent value for future outcome. This rate varies from person toperson, depending on each one’s willingness to wait. In general, anindividual who values the present more than the future will have ahigher personal discount rate in comparison to a person who placesmore value on the future.

Over the last three decades, research in the field of intertem-poral choices has attempted to measure personal discount rateswith studies that estimated the personal discount rates associatedwith choices for various durable items such as energy using homeappliances (Hausman, 1979; Gately, 1980; Ruderman et al., 1987),and for non-durable items such as retirement plans (e.g., Gilman,1976; Black, 1984; Warner and Pleeter, 2001) or sums of money(e.g., Laibson, 1997; Read, 2001; Benzion et al., 2004; Mahajna etal., 2008; Grignon, 2009). Initial research in this field focused pri-marily on monetary questions, finding that personal discount ratesfor adults vary according to time delay, amount of money and direc-tion of trade (see Benzion et al., 1989; Thaler, 1981; Chapman andElstein, 1995; Laibson, 1997; Frederick et al., 2002; Myerson et al.,2003; Prelec, 2004).

∗ Corresponding author at: Department of Economics, Ben-Gurion University, P.O.Box 653, Beer-Sheva, 84105, Israel. Tel.: +972 9 7967938; fax: +972 9 7962821.

E-mail address: [email protected] (E. Lahav).

Earlier studies also investigated the personal discount rates ofchildren and teenagers. In their pioneering work, Mischel et al.(1969) studied the development of self-control among children. Inone of their experiments, children were shown two rewards, onebetter than the other (for example, two marshmallows instead ofone marshmallow). They could receive the smaller reward imme-diately, or the better (bigger) reward if they waited for the personconducting the experiment to leave the room and then return.Mischel and his colleagues found that most very young childrencould not wait, suggesting a very high discount rate, but as they gotolder they developed abstract ways of thinking about the temptingreward so that visceral arousal was reduced. In later studies withthe same children, Mischel and his colleagues found that a child’scapacity to delay gratification correlated significantly with futureacademic achievement as adolescents (Mischel et al., 1988; Ayduket al., 2000). While Mischel and his colleagues’ studies considerednon-monetary issues, Green et al. (1994, 1997, 1999) were the firstto study the discounting of (delayed) monetary rewards amongadolescents. They compared personal discount rates in three agegroups: sixth grade children (n = 12, mean age = 12), young adults(mean age = 20) and older adults (mean age = 67), and found thatthe sixth graders had a higher subjective discount rate than theyoung adults, who in turn had a higher discount rate than the oldestgroup. These findings are consistent with the idea that self-controlincreases with age (Thaler and Shefrin, 1981). Read and Read (2004)tested three theories of intertemporal choice, each one predictinga different pattern of discounting over the lifespan. The first the-ory is in line with Green et al.’s (1994) findings. It predicts that

1053-5357/$ – see front matter © 2010 Elsevier Inc. All rights reserved.doi:10.1016/j.socec.2010.02.012

dx.doi.org/10.1016/j.socec.2010.02.012

http://www.sciencedirect.com/science/journal/10535357

http://www.elsevier.com/locate/soceco

mailto:[email protected]

dx.doi.org/10.1016/j.socec.2010.02.012




2 E. Lahav et al. / The Journal of Socio-Economics xxx (2010) xxx–xxx

discounting (subjective discount rate) will decrease over the lifes-pan, because the ability to enjoy consumption decreases with age,at an increasing rate, so the older we are the more we devaluefuture experiences (Trostel and Taylor, 2001). The second theory,by Rogers (1994), suggests that utility is defined by Darwinian fit-ness. Therefore it predicts that young children and pre-teens shoulddiscount less than young adults, because they cannot yet pro-duce offspring, and so they are better off waiting, by transformingresources to their future selves. However, young adults can get themost benefit from a unit of resource, and so they should discountmore than children and older adults. Older adults are becomingincreasingly less able to directly transform resources into offspring,and so they become more willing to delay, and will even choose totransfer resources to their offspring.1 The third theory, by Sozouand Seymour (2003), predicts a curvilinear relationship betweenage and discounting: middle-aged people will discount least, andyounger and older adults will discount more. For younger peoplethis is due to uncertainty early in life caused by little experienceabout the environment, which makes them act “as if there is notomorrow”, as Sozou and Seymour (2003) put it. For older persons,the uncertainty about the environment is reduced; however con-cern about health and declining capacity late in life arises (whichleads to more spending and less saving).

In order to test the three theories, Read and Read (2004) com-pared the time preferences of 123 respondents in three differentage groups (mean age 25, 44 and 75, respectively). They used bothcomputerized and paper and pencil questionnaires,2 with mone-tary and non-monetary questions. The findings suggest an inversehump relationship between age and time discounting, correspond-ing with the predictions of Sozou and Seymour (2003).

The notion of age-related self-control was reinforced by astudy of monetary time preferences conducted by Bettinger andSlonim (2007), who examined the discount rate of 191 children,divided into three age groups (mean ages: 6, 9 and 12 years). Theresearchers found that the children in the higher age groups weremore patient.

Most of these time preferences studies concentrated on adultsand some on children, but not on teenagers. However, Doss et al.(1995) and Johnson and Lino (2000) demonstrated that childrenand teenagers both spend and save large amounts of money, whichthey receive from various sources. In some places, including Israelwhere this study was conducted, teenagers may now open a bankaccount and even receive a credit card, which makes the study oftheir personal discount rates significant for financial institutions,businesses, parents and the education system.

In the current study, we examine the impact of postponingreceipt of a payment over different time periods (1 week, 2 months,6 months and 1 year) on the discount rate of teenagers and adultparticipants. We suggest three factors that should be consideredwhen comparing the subjective discount rates of teenagers andadults. First, teenagers exhibit a lower willingness to wait thanadults. Babcock (2004) discusses adolescent choices about lifetimeearnings. He focuses on the assumption that adolescents are “impa-tient agents, fully aware of negative consequences of low-effortchoices but unwilling to trade-off present leisure against futuregains” (see also Thaler and Shefrin, 1981; Bettinger and Slonim,2007). As a result of their relatively higher impatience, teenagers

1 Trostel and Taylor (2001) also argue that the utility is defined by Darwinianfitness, but rather than focusing on the direct contribution of consumption to Dar-winian fitness, they focus on the negative correlation between consumption and theenjoyment we get from it.

2 The computerized questions involved choosing between a larger later amountof money and a smaller sooner amount. According to the participant answers, theamounts changed until reaching indifference point. The time spans were from 1 to10 years.

should exhibit higher subjective discount rates than those of adults,corresponding with Trostel and Taylor (2001).

The second factor, the cushion effect, has the opposite impact(see Weber and Milliman, 1997; Mahajna et al., 2008; Rosenboim etal., 2010). According to this effect, when an individual has a safetynet, or a feeling that he or she will be supported (or protected) intimes of need, he will be less risk-averse, which results in a lowersubjective discount rate. For example, in individualistic societieslike the United States, there is no safety net and as a result indi-viduals are more risk-averse (see also Weber and Milliman, 1997).Teenagers, who are still in high school and live with their parents,worry less about their personal needs due to the safety net providedby their parents. Conversely, adults have to work to earn a livingand usually do not have a safety net. The teenagers, as a result oftheir safety net, ought to exhibit a higher willingness to wait andhave correspondingly lower subjective discount rates.

The third factor is financial experience and knowledge. A num-ber of studies have found that a person with more financialknowledge and experience tends to exhibit a greater level of pref-erence for financial risk (e.g., Shelbecker et al., 1990; Cutler, 1995;Grable and Joo, 1997, 2004; Grable, 2000). Many studies that dealwith time preferences consider the bank interest as a base for deter-mining personal discount rate (e.g., Benzion et al., 1989; Fredericket al., 2002). However, when dealing with time preferences amongteenagers, one must take into consideration that they normally donot have access to the capital market. Furthermore, a teenager witha bank account may deposit money in a savings account but may nottake a loan. In most cases, the only way for teenagers to who go intodebt is to borrow from their parents or relatives, who generally donot require them to pay interest. In addition, teenagers do not haveto deal with as many financial transactions as adults do. Consider-ing all of these factors, teenagers should exhibit higher subjectivediscount rates than adults who are used to financial transactions.

In this paper we also examine two possible factors that mightaffect the time preferences of teenagers, and are not relevant for:having a bank account and receiving an allowance from their par-ents. These two factors can be used as tools to educate the teenagersto exhibit a higher willingness to wait.

Dustmann et al. (2004), who studied data from the BritishNational Child Development Study (collected in the spring of1974), found that approximately two-thirds of 16-year-old Britishteenagers had bank accounts (under their name or their parent’sname). In the US, one of the results of the 2008 National Jump-startCoalition Survey of High School Seniors and College Students forPersonal Financial Literacy is that teenagers with bank accountstend to be more financially literate than those without accounts,although they acknowledge that this might also reflect differencesin income. We examine whether teenagers with a bank accountare more financially educated and as a result exhibit lower timediscount rates.

Other researchers discuss the issue of allowances and their effecton economic decisions (e.g., Abramovitch et al., 1991; Doss et al.,1995; Meeks, 1998; Pabilonia, 2001; Markowitz and Tauras, 2006)but none of them discuss the connection to time discounting. How-ever, Abramovitch et al. (1991) do explain that the pattern of resultsin their experiment suggests that children who receive allowancesare more sophisticated about money than those who do not. Weexamine whether teenagers who receive an allowance from theirparents feel the presence of a safety net (cushion) more stronglythan those who do not, and as a result exhibit a lower time discountrate.

Our survey was designed to enlarge the empirical data aboutthe time preferences of teenagers, and while the few studiesconcerning children concentrate on non-monetary questions, wefocused on monetary questions. Furthermore, we investigated thecorrelation of money related time preferences to other monetary

dx.doi.org/10.1016/j.socec.2010.02.012




E. Lahav et al. / The Journal of Socio-Economics xxx (2010) xxx–xxx 3

issues (e.g., allowance and a bank account), which can help inexplaining the discounting of teenagers. Being in the early stagesof their budget constrain understanding, we regarded the cushioneffect, created by parental protection, as a possible explanation forour findings.

The rest of the paper is organized as follows: Section 2 describesthe experimental design. Section 3 presents the results of the timepreference comparison of teenagers and adults. We also present inthis section the effect of bank account ownership and the receivingof an allowance on teenagers’ time preferences. At the end of thissection we discuss the correlation of our data to the Hyperbolic andExponential discounting models. Finally, Section 4 summarizes thepaper and presents its conclusions.

2. Experimental design

2.1. Participants

The teenaged participants were 58 high school students fromtwo Israeli high schools (average age 16.6, age range: 15–18 years,22 males and 36 females). The survey was conducted during regularschool hours and participation was encouraged by the offering of areward (cinema tickets for two participants in each class). The adultparticipants were 71 MBA students (average age 29.04, ranging inage from 24 to 35 years, 31 males and 40 females) from the Collegeof Management, Israel.3

2.2. Procedure

The survey began by informing the participants that they wereabout to take part in a survey in the field of behavioral economics, inwhich they will be asked to answer questions regarding economicsituations that occur in everyday life. They were further instructedthat the questions did not have ‘correct’ answers, so each answershould reflect their personal preferences.

In the main scenario (postponing receipt of income), partici-pants were informed they had earned x for working in a largefashion chain store, and they would receive that amount the nextday.4 The chain’s management offered them the option of postpon-ing receipt of x to a later time (t). The participants were instructed towrite down the amount they would ask to receive at time t ratherthan receiving x tomorrow. The time periods (t) were: 1 week, 2months, 6 months and 1 year. They were also asked to assume thatthe chain is reliable and there is no doubt they would be paid ontheir chosen future date if the management accepted their offer.The idea of the scenario was based on a paper by Benzion et al.(1989) that asks about a financial firm. The point of the statementabout the chain store’s behavior is that it made the scenario clearer.The reason for this is that teenagers in Israel are aware that the Min-istry of Labor has strong control on the employers. They know thatan employer who disobeys the labor laws pays high fines, especiallywhen hiring teenagers (there are many advertisements by the gov-ernment regarding this issue, especially during school vacations).

A sample of the postponed receipt question:

3 The MBA students in the survey had a non-financial bachelor’s degree such aspsychology, social science or law. This means that they did not have an earlier aca-demic background in finance or any professional background in finance. Studentparticipants are used in many behavioral surveys (see, for example, Tan and Johnson,1996; Du et al., 2002).

4 Kirby and Santiesteban (2003) found a one-day FED (front end delay) to be insuf-ficient to overcome the subjective costs and resulting fixed premium. Therefore, webelieve that ‘receiving the income tomorrow’ rather than ‘today’ should not have asignificant impact on the results. We decided not to ask about receiving the incomeon the same day in order to make the scenario more realistic.

The sum of NIS5 200 is going to be deposited in your bankaccount tomorrow. However, we are offering you the option ofpostponing this deposit and receiving another amount in oneyear. What is the minimum amount you are willing to acceptone year from now in order to postpone receipt of the sum?Instead of NIS 200 tomorrow, I am willing to accept NIS inone year.

The questions referred to two different amounts: NIS 200 andNIS 600, so the postponement scenario had eight different ques-tions with four different time periods and two different amounts.All participants went through all eight situations.

Using the same amounts for both groups made the resultscomparable. However, since adults have a higher income thanteenagers, there might have been an income-effect problem, mean-ing that adults might consider NIS 200 a relatively low amount,while it is a relatively large amount for teenagers. In order to avoidthis effect, we also compared the high amount in the adult group(NIS 600) to the low amount in the teenage group (NIS 200).

In another question, participants were instructed to evaluatethe future value of NIS 600 (original sum plus interest) 1 year fromtoday. These answers allowed us to calculate their knowledge ofthe bank interest rate. We chose to use a 1-year interest rate, sinceinterest rates are usually published in annual terms. The questionwas presented as follows:

You have earned NIS 600 that you want to save. If you depositthe money in a bank, what is the sum of money that you willhave in one year:Instead of the NIS 600 that I deposit in the bank today, I willhave NIS in one year.

The main reason we used a question regarding the return on adeposit and not a loan is that for teenagers, a deposit is usually theonly available investment they have (if any). This means that thequestion will mostly measure a degree of knowledge about theironly available option for investment.

The teenaged participants were also asked if they have a bankaccount and if they receive an allowance from their parents.

3. Results

3.1. Comparing teenagers to adults

The average estimated interest rates for 1 year were 27.61%(STDV = 40.23%) and 10.47% (STDV = 13.69%) for teenagers andadults, respectively. The estimated interest rate for a deposit washigher for teenagers than for adults, with a medium size effect (T-value = 3.36, Sig′ = 0.00, d = 0.571). It seems that although teenagerscan open a bank account and even receive a credit card, they usuallydo not become more knowledgeable about interest rates. However,considering that the annual interest rates on deposits in Israel atthe time of the study ranged from 3% to 4%,6 it also seems thatmany adults also did not know the rate of interest paid on depositsor found it difficult to estimate the monetary value of the interest

5 New Israeli Shekel (NIS). The exchange rates at the time of the experi-ment were 4.011NIS to US$ 1, and 6.01N.IS to £1 (source: http://stats.oecd.org/Index.aspx?datasetcode=SNA TABLE4).With 200NIS, a teenager can buy 13 Big Macs or 6 tickets to the cinema.The 2008 GDP per capita in Israel using 2005 PPPs is 26,700 NIS. For comparison,this is about 75%, and 58% of the per capita GDP in the US and the UK, respectively.

6 The questionnaires were collected during 2007–2008, and the intereston saving accounts was about 3–4% depending on the amounts and lengthof the deposit. Data were collected form Heshev Information Systems Ltd.http://www.heshev.co.il/about.htm. Today, for the same deposits, the banks pay1% or less. Banks in Israel pay no interest on chequing accounts, so a deposit thatearns interest is only optional in a saving account.

dx.doi.org/10.1016/j.socec.2010.02.012

http://stats.oecd.org/Index.aspx%3Fdatasetcode=SNA_TABLE4

http://stats.oecd.org/Index.aspx%3Fdatasetcode=SNA_TABLE4

http://www.heshev.co.il/about.htm





Table 1Annual subjective discount rates—postponing receipt of payment (comparing teenagers to adults).

Group Amount 1 week 2 months 6 months 12 months

Teenagers (n = 58) NIS 200a 1275.34% 394.76% 265.14% 260.97%(1043.83%) (294.55%) (184.97%) (185.71%)

NIS 600a 774.77% 219.74% 158.56% 171.81%(724.25%) (204.70%) (156.48%) (186.59%)

Adults (n = 71) NIS 200a 780.04% 198.72% 131.63% 125.18%(927.58%) (197.95%) (140.51%) (134.19%)

NIS 600a 404.53% 102.75% 66.38% 60.06%(340.68%) (81.38%) (48.13%) (52.62%)

T-test (p-value) 3.82 4.41 4.70 4.82Teenagers NIS 600 (0.000) (0.000) (0.000) (0.000)

Adults NIS 600 d = 0.654 d = 0.751 d = 0.796 d = 0.815


Adults NIS 200 d = 0.638 d = 0.907 d = 0.988 d = 0.995


Adults NIS 600 d = 1.122 d = 1.351 d = 1.471 d = 1.472

a Standard deviations are in parentheses.

they would receive for a 1-year deposit.Table 1 shows the annual subjective discount rate for each

period and amount.The discounting rate for delaying payment was calculated by the

following equation:

r =((

P

X

)− 1

)(12t

), (1)

where P is the amount the participant is willing to accept in tmonths for delaying the receiving of the amount of X tomorrow.

We use a one-tailed T-test to examine the null hypothesisthat the subjective discounting is equal for teenagers and adults.We compared not only the subjective discount rate for the sameamounts (NIS 200 and NIS 600), but also compared adult responsesfor the higher amount to teen responses for the lower amount, asa way of neutralizing the income effect.

The subjective discount rates are very high for both teenagersand adults, with a high variance, suggesting wide variationsbetween the participants. This finding is consistent with earlierstudies about the subjective discount rate, especially for short peri-ods (e.g., Thaler, 1981; Frederick et al., 2002). In all cases, we foundthat the average subjective discount rate was significantly higherfor teenagers than for adults, with a high size effect. This was alsotrue when we avoided an income or size effect by comparing thediscount rates for a relatively low amount for teenagers (NIS 200)and a relatively high amount for adults (NIS 600).

In addition we tested the correlation between the estimatedinterest rate for a deposit and the relevant question about sub-jective discount rate (delaying a NIS 600 receipt for 12 months).For teenagers, we found no significant correlation (Rho = −0.006;Sig′ = 0.481), which suggests that their subjective discount rate wasnot related to their estimate of the bank’s interest rate. However, foradults we found a positive and significant correlation (Rho = 0.293;Sig′ = 0.007), which suggests that they based their subjective dis-count rate on their estimated interest rate.

3.2. Analysis of the teenage group

3.2.1. Bank accountThe teenagers were asked if they have a bank account. Of the

58 participants, 19 said they did not have a bank account and the

other 39 teenagers said that they did. We used this information asan indicator of experience in money management.

The average estimated interest rates for 12 months were 27.80%(STDV = 29.16%) and 27.51% (STDV = 44.99%) for teenagers with-out a bank account and teenagers with bank account, respectively.There was no difference between the interest rates estimated byteenagers without a bank account and those estimated by teenagerswith a bank account (T-value = 0.03, Sig′ = 0.49, d = 0.008). Havinga bank account did not influence their estimation of the interestrate for a bank deposit, meaning that having a bank account doesnot improve the financial knowledge or the objective time valueestimation of teenagers.

Table 2 shows the average subjective discount rate for eachgroup, respectively.

Only for the longer time (12 months) and the higher amount(NIS 600) were the average subjective discount rates higher forteenagers who did not have a bank account, with medium size effect(significance <10%). Other than that, it seems that there was no dif-ference in the annual subjective discount rate between teenagerswith and without a bank account. This means that having a bankaccount does not change the time preference or increase the will-ingness of teenagers to wait.

3.2.2. AllowanceThe teenagers were also asked if they receive an allowance from

their parents. Of the 58 teenaged participants, 16 replied that theyreceive an allowance7 from their parents and the other 42 teenagersreplied that they do not. We used this information as an indicatorof a high safety net (cushion effect).

7 A new Internet survey from 2009, by a commercial company, asked 7500 Israelichildren age 7–14 about consumption and other habits, and found that 65% ofthem did not get a regular allowance. Available at http://www.mako.co.il/finances-hitech/media/Article-2754a0cf179de11004.htm. This figure is close to our data,and it corresponds with a survey regarding 6–15 years old children in France,According to this survey, by the Crédit Agricole bank, in 2009, 40% of thechildren receive a parental allowance. http://www.calcalist.co.il/world/articles/0,7340,L-3358103,00.html, These figures are lower than surveys in the USwhich found that about half of the children receive parental allowance (Carol,1998; AMEX Luxury Survey available at: http://home3.americanexpress.com/corp/pc/2006/platlux survey.asp).

dx.doi.org/10.1016/j.socec.2010.02.012

http://www.mako.co.il/finances-hitech/media/Article-2754a0cf179de11004.htm

http://www.mako.co.il/finances-hitech/media/Article-2754a0cf179de11004.htm

http://www.calcalist.co.il/world/articles/0,7340,L-3358103,00.html

http://www.calcalist.co.il/world/articles/0,7340,L-3358103,00.html

http://home3.americanexpress.com/corp/pc/2006/platlux_survey.asp

http://home3.americanexpress.com/corp/pc/2006/platlux_survey.asp





Table 2Annual (1 year) subjective discount rates (comparing teenagers with and without a bank account).

Scenario Group 1 week 2 months 6 months 12 months

Postponing receipt of payment NIS 200 Without a bank accounta (n = 19) 1354.74% 406.11% 253.05% 232.45%(1067.02%) (274.42%) (167.44%) (137.57%)

With a bank accounta (n = 39) 848.42% 206.05% 121.75% 119.78%(921.09%) (173.72%) (95.64%) (94.52%)

T-test (p-value) 0.40 0.20 −0.34 −0.81(0.345) (0.420) (0.366) (0.210)d = 0.135 d = 0.068 d = −0.104 d = −0.236

Postponing receipt of payment NIS 600 Without a bank accounta (n = 19) 1236.67% 389.23% 271.03% 274.87%(617.16%) (220.04%) (177.17%) (214.43%)

With a bank accounta (n = 39) 738.89% 226.41% 176.50% 197.16%(617.16%) (220.04%) (177.17%) (214.43%)

T-test (p-value) 0.54 −0.35 −1.26 −1.50(0.297) (0.363) (0.107) (0.070)d = 0.140 d = −0.103 d = −0.385 d = −0.467


Table 3Annual subjective discount rates (comparing teenagers receiving and not receiving an allowance).

Scenario Group 1 week 2 months 6 months 12 months

Postponing receipt of a payment NIS 200 Not receiving an allowancea (n = 42) 1442.38% 445.86% 296.86% 287.30%(1109.35%) (322.28%) (204.10%) (198.58%)

Receiving an allowancea (n = 16) 880.08% 245.95% 181.27% 199.68%(777.21%) (226.94%) (175.35%) (208.85%)

T-test (p-value) 2.03 2.21 2.18 1.78(0.024) (0.016) (0.017) (0.040)d = 0.709 d = 0.771 d = 0.763 d = 0.639

Postponing receipt of a payment NIS 600 Not receiving an allowancea (n = 42) 836.88% 260.63% 181.88% 191.88%(479.16%) (107.15%) (61.38%) (72.01%)

Receiving an allowancea (n = 16) 498.33% 150.94% 98.96% 98.65%(479.16%) (107.15%) (61.38%) (72.01%)

T-test (p-value) 1.83 1.60 1.83 1.88(0.036) (0.057) (0.037) (0.032)d = 0.591 d = 0.535 d = 0.627 d = 0.647


The average estimated interest rates for 12 months were 22.69%(STDV = 27.23%) and 40.52% (STDV = 62.26%) for teenagers who didnot receive an allowance and for teenagers who did receive anallowance, respectively. The estimated interest rate for teenagerswho received an allowance was higher than for teenagers whodid not, with a medium size effect (T-value = −1.53, Sig′ = 0.066,d = −0.371). It seems that receiving an allowance reduced theteenager’s knowledge and, as a result, his or her ability to estimatethe interest rate. Table 3 shows the respective average discount ratefor these groups.

The average subjective discount rates were higher for teenagerswho did not receive an allowance, compared to teenagers whoreceived an allowance with large size effect. These results indi-cate that receiving an allowance from their parents increases theteenagers’ willingness to wait. It is possible that when teenagersreceive an allowance from their parents, the effect of the safety netis stronger (cushion effect), and this leads to an increase in theirwillingness to wait.

3.2.3. Correlations and multivariate analysesWe found a small positive correlation (Rho = 0.184; Sig′ = 0.083)

between teenagers with a bank account and those who getallowance, which suggests that in our sample, allowance is notthe main source of income for most teenagers with a bankaccount.

No significant correlation was found between the subjectivediscount rate and the inferred interest rate for teenagers with abank account (Rho = 0.003; Sig′ = 0.49), for teenagers without abank account (Rho = −0.075; Sig′ = 0.38), and for teenagers whodid not get allowance (Rho = −0.142; Sig′ = 0.18). However, wefound a high positive correlation between the subjective discountrate and the inferred interest rate for the teenagers who got anallowance (Rho = 0.76; Sig′ = 0.00). For adults we found a positiveand significant correlation between the subjective discount rateand the inferred interest rate (Rho = 0.29; Sig′ = 0.01). This maysuggest that only teenagers who received an allowance based theirsubjective discount rate on their estimated interest rate, similar tothe adult group.

We conducted a multivariate regression analysis for theteenager group, using dummy variable for allowance (noallowance = 0) and a dummy variable for having a bank account(no bank account = 0). The dependent variable was the subjectivediscount rate for a delay of t and an amount of x. In all of the 8combinations of time and amount, getting an allowance reducedthe discount rate. Having a bank account had a significant positiveeffect for questions regarding an amount of 600NIS, for 6 and 12months delays. The estimated interest rate for a deposit had no sig-nificant effect, and the R squares of the models were between 0.056and 0.123 but not significant (Table 4).

dx.doi.org/10.1016/j.socec.2010.02.012





Table 4Regression analysis for the teenagers group+.

Amount 600 NIS 200 NIS

Delay 12 months 6 months 2 months 1 week 12 months 6 months 2 months 1 week

Independent variables(Constant)* 1.32 1.35 2.31 8.94 2.38 2.66 4.22 14.29

(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

Allowance* −1.25 −0.98 −0.96 −3.83 −1.16 −1.28 −1.99 −6.18(0.03) (0.04) (0.13) (0.09) (0.04) (0.02) (0.03) (0.06)

Bank account* 0.99 0.72 0.37 −0.42 0.63 0.40 0.18 −0.09(0.06) (0.10) (0.53) (0.84) (0.23) (0.44) (0.83) (0.97)

Interest on a deposit* 0.25 0.08 −0.36 0.55 0.45 0.27 0.56 0.83(0.68) (0.88) (0.60) (0.82) (0.46) (0.66) (0.57) (0.81)

Model Sig. (F-test) 0.07 0.12 0.37 0.35 0.18 0.16 0.18 0.27R-Square 0.123 0.102 0.056 0.058 0.086 0.092 0.087 0.069

+ The dependent variable is the subjective discount rate for a delay of t months, and an amount of X.* Significance levels are in parentheses.

Table 5Hyperbolic and exponential models—non-linear regression.a A non-linear regression analysis for the following equations, from Johnson and Bickel (2002): (1) Hyperbolicmodel: P = X/(1 + kt) and (2) Exponential model: P = Xe−kt .


Delay Teenagers Adults Teenagers Adults

Hyperbolic Exponential Hyperbolic Exponential Hyperbolic Exponential Hyperbolic Exponential

kb 1.70 1.02 0.62 0.49 2.65 1.32 1.28 0.84(0.12) (0.05) (0.03) (0.02) (0.13) (0.04) (0.08) (0.04)

Residual 242.45 245.04 25.72 26.17 266.37 283.90 172.64 175.66R-Square 0.254 0.246 0.289 0.277 0.405 0.366 0.217 0.203

a In both models, the dependent variable is X/P.b Standard errors are in parentheses.

Table 6Hyperbolic models for sub-groups.a A non-linear regression analysis for the following equation: P = X/(1 + kt).


Allowance Bank account Allowance Bank account

Receive Do not receive Have Do not have Receive Do not receive Have Do not have

kb 1.92 2.93 2.77 2.41 1.00 1.97 1.94 1.23(0.15) (0.16) (0.17) (0.17) (0.09) (0.16) (0.17) (0.11)

Residual 28.16 223.01 210.50 53.72 10.01 218.42 211.04 23.02R-Square 0.500 0.413 0.398 0.446 0.414 2.720 0.266 0.321

a The dependent variable is X/P.b Standard errors are in parentheses.

As mentioned before, the only parameter that affectedteenagers’ ability to wait, and as a result reduce their subjectivediscounting, was getting an allowance from their parents. This wasalso true in a multivariate analysis.

It seems that teenagers who get an allowance from their parentsact more like adults. They have a stronger ability to wait, and basetheir subjective discounting on the estimated interest rate as adults.

3.3. Hyperbolic and exponential discounting models

In order to fit the indifference point reached in the results tothe hyperbolic and exponential models, we conducted a non-linearregression analysis for the following equations, from Johnson andBickel (2002):

Hyperbolic model : P = X

1 + kt(2)

Exponential model : P = Xe−kt (3)

In both equations, P is the present (discounted) value of thereward, X is the objective (undiscounted) value of the reward, tis the delay in years from the choice until the receipt of the reward,and k is a free parameter. In Eq. (2), e is Euler’s number.

Table 5 presents the regression analysis for each model.The results indicate that the time effect8 was higher for the

teenager group compared to the adults. According to the residu-als and R squares, we found a better fit to the hyperbolic model,for both the teenagers and the adults groups. This is consistentwith many earlier studies regarding young adults or children (e.g.,Green et al., 1994; Johnson and Bickel, 2002; Read and Read,2004).

We separated the teenager group into sub-groups according toownership of a bank account, and receiving of an allowance, and

8 The free parameter k, presents the time effect. A higher k, indicates a highertime effect. This implies a higher decrease in the subjective discount rate as thedelay becomes longer.

dx.doi.org/10.1016/j.socec.2010.02.012





tested the hyperbolic model for each sub-group. Table 6 presentsthe non-linear regression results.

We found that for both amounts, teenagers that received regu-lar allowance had the best fit to the hyperbolic model comparedto the other three groups (not receiving allowance, and havingor not having a bank account). Furthermore, according to the R-square, separating the teenagers to sub-groups increased the fit tothe hyperbolic model in each sub-group, compared to the teenagegroup as a whole.

4. Conclusion

We studied teenagers and adults in Israel in order to estimatethe effect of age and allowance on subjective time discountingand interest rate estimation. The subjective discount rates werevery high for both teenagers and adults, with a high variance inboth groups suggesting that there were high individual differ-ences between the participants, consistent with Thaler (1981) andFrederick et al. (2002).

While Read and Read (2004) focus on the time discountingpattern in adulthood, the current paper addresses teenagers, andcompares the results to adults. This allowed us to test the opposingpredictions of Rogers’ (1994) and Sozou and Seymour’s (2003) the-ories, about the time discounting pattern in early years. Althoughthe cushion effect supports Rogers’ (1994) pattern of time dis-counting along the lifespan, we found that when compared toadults, teenagers exhibited a higher interest rate estimation andhigher subjective discount rate, as proposed by Sozou and Seymour(2003).

These results indicate that the effect of lower knowledge isstronger than the effect of the safety net. However, we suggest thatgiving a teenager an allowance can be a tool for increasing theirwillingness to wait. On the other hand, managing a bank accountdoes not affect the teenagers’ willingness to wait.

It may be that since usually teenagers’ financial needs (includingliving allowance, medical care, food and education) are supplied bytheir parents, their safety net (cushion) limits their financial experi-ence and knowledge. This, in turn, makes them more risk-averse forfinancial decision-making and, as a result, they discount the futureheavily. The finding that teenagers who receive an allowance give ahigher estimation for the bank interest rate indicates that a strongersafety net reduces financial experience and knowledge.

The comparison of teenagers and adults is interesting in twoways. First, teenagers (unlike most adults) are supported bytheir parents, and therefore have a stronger safety net (cushioneffect), which should reduce their time discounting. We show thatteenagers who receive an allowance – which increases the impactof their safety net (cushion) – do exhibit lower time discount rates,as expected.

Teenagers have less financial experience than adults. As a result,they should exhibit higher subjective time discounting (Shelbeckeret al., 1990; Cutler, 1995; Grable and Joo, 1997, 2004; Grable, 2000).An alternative explanation for the relative high discount rate mightbe that although teenagers are at the beginning of adulthood (theycan work, open a bank account, etc.), they remain children in someways. Regarding the subjective discount rate, they still have diffi-culty delaying satisfaction (e.g., Mischel et al., 1988; Bettinger andSlonim, 2007), which is reflected in the high subjective discountrate.

Teenagers are becoming more and more important to banks andother financial institutions. A 1994 estimate of teenage expendi-tures showed that US teenagers had an aggregate income of US$96 billion and spent 68% of that income (Zollo, 1995). Doss et al.(1995) reported that middle school children, ages 10–15, spend73% of their income, save 18% and use the rest for gifts. Banks andother institutions should take into consideration the high subjec-

tive discount rate of teenagers when offering their teenage clientsa long-term saving program (e.g., giving them a gift with the firstpayment). Moreover, as we have demonstrated here, in some casesthe ‘safety net’ reduces the individual’s knowledge of the financialsystem.

In the current paper, we focus on the effect of age on time dis-counting. However, risk aversion also changes with age and can be aconfounder of our results. It is reasonable to assume that risk aver-sion decreases when one moves from the teen years to adulthood9

(due to the fact that there is more exposure to contracts and rela-tionships involving potential money loss). Since more risk-aversepeople tend to spend now rather than waiting, pure time discount-ing might be spuriously over-estimated in the survey. Similarly,differences in stated preferences might stem from differences inself-control (as suggested by Thaler and Shefrin, 1981), rather thanfrom differences in true perceptions of future versus present.

Our sample compared teenagers to adults, and we found verydifferent time preferences. The adults were students of MBA butwho had no financial background. Choosing a different comparisonsample which is more similar to the ‘average adult’ might decreasethe difference in time discounting between teenagers and adults.However, it will probably not change the main conclusion, becauseas earlier work shows, time discounting is influenced by many fac-tures other than education, such as earnings, risk aversion and ofcourse age (Warner and Pleeter, 2001; Read and Read, 2004), allpredicting higher discount rates for teenagers.

Our findings are consistent with the view that by managingtheir allowance, children learn about independence, decision-making, responsibility, priorities, and of course about delayinggratification.10

Future research that tests the effect of the decision-maker’s age,should also take into consideration the strength of their safety net.Any comparison between two sub-groups with different safety netsor different financial knowledge is problematic, and depends on thestrength of each effect. For example, a comparison between middle-aged adults and older adults would also be problematic becausesome older people have pension funds, social security allowancesor assets that increase the impact of their safety net.

References

Abramovitch, R., Freedman, J.L., Pliner, P., 1991. Children and money: getting anallowance, credit versus cash and knowledge of pricing. Journal of EconomicPsychology 12 (1), 27–45.

Ayduk, O., Mendoza-Denton, R., Mischel, W., Downey, G., Peake, P.K., Rodriguez,M., 2000. Regulating the interpersonal self: strategic self-regulation for copingwith rejection sensitivity. Journal of Personality and Social Psychology 79 (5),776–792.

Babcock, P.S., 2004. The Rational Adolescent: Privilege, Policy, and Myopic HumanCapital Acquisition. Working Paper.

Benzion, U., Rapoport, A., Yagil, J., 1989. Discount rates inferred from decisions: anexperimental study. Management Science 35, 270–284.

Benzion, U., Shachmurove, Y., Yagil, J., 2004. Subjective discount functions—anexperimental approach. Applied Financial Economics 14 (5/1), 299–311.

Bettinger, E., Slonim, R., 2007. Patience among children. Journal of Public Economics91 (1–2), 343–363.

Black, M., 1984. Personal Discount Rates: Estimates for the Military Population.Final Report of the Fifth Quadrennial Review of Military Compensation, vol. 1B,Appendix I. U.S. Department of Defense, Washington, DC.

Carol, B.M., 1998. Factors influencing adolescents’ income and expenditures. Journalof Family and Economic Issues 19 (2), 131–150.

Chapman, G.B., Elstein, A.S., 1995. Valuing the future: temporal discounting of healthand money. Medical Decision Making 15 (4), 373–386.

9 See for example: Halek and Joseph (2001), Sozou and Seymour (2003) and VanPraag and Booij (2003).

10 There are many Internet sites, and parents’ forums that discuss the influ-ence of parental allowance. See for example: http://www.activeallowance.com/kidsallowance/benefits.html http://life.familyeducation.com/allowance/jobs-and-chores/34438.html.

dx.doi.org/10.1016/j.socec.2010.02.012

http://www.activeallowance.com/kidsallowance/benefits.html

http://www.activeallowance.com/kidsallowance/benefits.html

http://life.familyeducation.com/allowance/jobs-and-chores/34438.html





Cutler, N.E., 1995. Three myths of risk-tolerance: what clients are not telling you.Journal of the American Society of CLU & ChFC 49, 33–37.

Doss, V.S., Marlowe, J., Godwin, D.D., 1995. Middle-school children’s sources anduses of money. Journal of Consumer Affairs 29 (1), 219–241.

Du, W., Green, L., Myerson, J., 2002. Cross-cultural comparisons of discountingdelayed and probabilistic rewards. The Psychological Record 52, 479–494.

Dustmann, C., Micklewright, J., Van Soest, A.H.O., 2004. In-school Work Experience,Parental Allowances, and Wages. IZA Discussion Paper No. 1235. Available atSSRN: http://ssrn.com/abstract=574067.

Frederick, S., Loewenstein, G., O’Donoghue, T., 2002. Time discounting and timepreference: a critical review. Journal of Economic Literature 40, 351–401.

Gately, D., 1980. Individual discount rates and the purchase and utilization of energy-using durables: comment. Bell Journal of Economics 11, 373–374.

Gilman, H.J., 1976. Determinants of Implicit Discount Rates: An Empirical Examina-tion of Voluntary Pension Contributions of Employees in Four Firms. Center forNaval Analyses, Arlington, VA.

Grable, J.E., 2000. Financial risk tolerance and additional factors that affect risk takingin everyday money matters. Journal of Business and Psychology 14 (4), 625–630.

Grable, J.E., Joo, S.H., 1997. Determinants of risk preference: implications for familyand consumer science professionals. Family Economics and Resource Manage-ment Biennial 2, 19–24.

Grable, J.E., Joo, S., 2004. Environmental and bio psychosocial factors associated withfinancial risk tolerance. Financial Counseling and Planning 15 (1), 73–82.

Green, L., Fry, A.F., Myerson, J., 1994. Discounting of delayed rewards: a lifespancomparison. Psychological Science 51, 33–36.

Green, L., Myerson, J., McFadden, E., 1997. Rate of temporal discounting decreaseswith amount of reward. Memory and Cognition 25 (5), 715–723.

Green, L., Myerson, J., Ostaszeski, P., 1999. Discounting of delayed rewards acrossthe life span: age differences in individual discounting functions. BehavioralProcesses 46 (1), 89–96.

Grignon, M., 2009. An empirical investigation of heterogeneity in time preferencesand smoking behaviors. Journal of Socio-Economics 38 (5), 739–751.

Halek, M., Joseph, G., 2001. Demography of risk aversion. Journal of Risk and Insur-ance 68 (1), 1–24.

Hausman, J., 1979. Individual discount rates and the purchase and utilization ofenergy-using durables. Bell Journal of Economics 10 (1), 33–54.

Johnson, M.W., Bickel, W.K., 2002. Within-subject comparison of real and hypothet-ical money rewards in delay discounting. Journal of Experimental Analysis ofBehavior 77, 129–146.

Johnson, D.S., Lino, M., 2000. Teenagers: employment and contributions to familyspending. Monthly Labor Review 123 (9), 15–25.

Kirby, K., Santiesteban, M., 2003. Concave utility, transaction costs, and risk in mea-suring discounting of delayed rewards. Journal of Experimental Psychology:Learning, Memory, and Cognition 29 (1), 66–79.

Laibson, D., 1997. Golden eggs and hyperbolic discounting. The Quarterly Journal ofEconomics 112 (2), 443–477.

Mahajna, A., Benziona, U., Bogairea, R., Shavit, T., 2008. Subjective discount ratesamong Israeli Arabs and Israeli Jews. Journal of Socio-Economics 37 (6),2513–2522.

Markowitz, S., Tauras, J.A., 2006. Even for Teenagers, Money Does Not Grow on Trees:Teenage Substance Use and Budget Constraints. NBER Working Paper Series,w12300. Available at SSRN: http://ssrn.com/abstract=908539.

Meeks, C.B., 1998. Factors influencing adolescents’ income and expenditures. Journalof Family and Economic Issues 19 (2), 131–150.

Mischel, W., Grusec, J., Masters, J.C., 1969. Effects of expected delay time on thesubjective value of rewards and punishments. Journal of Personality and SocialPsychology 11 (4), 363–373.

Mischel, W., Shoda, Y., Peake, P., 1988. The nature of adolescent competenciespredicted by preschool delay of gratification. Journal of Personality and SocialPsychology 54 (4), 687–696.

Myerson, J., Green, L., Hanson, J.S., Holt, D., Estle, S.J., 2003. Discounting delayed andprobabilistic rewards: processes and traits. Journal of Economic Psychology 24,619–635.

Pabilonia, S.W., 2001. Evidence on youth employment, earnings and parental trans-fers in the national longitudinal survey of youth 1997. Journal of HumanResources 36 (4), 795–822.

Prelec, D, 2004. Decreasing impatience: a criterion for non-stationary time prefer-ence and hyperbolic discounting. Scandinavian Journal of Economics 106 (3),511–532.

Read, D., 2001. Is time-discounting hyperbolic or subadditive? Journal of Risk andUncertainty 23 (1), 5–32.

Read, D., Read, N.L., 2004. Time discounting over the lifespan. Organization Behaviorand Human Decision Processes 94, 22–32.

Rogers, A.R., 1994. Evolution of time preference by natural selection. American Eco-nomic Review 84, 460–481.

Rosenboim, M., Shavit, T., Shoham, A., 2010. Financial decision making in collectivesociety—a field test on Israeli kibbutz members and city residents. Journal ofSocio-Economics 39 (1), 30–36.

Ruderman, H., Levine, M.D., McMahon, J.E., 1987. The behavior of the market forenergy efficiency in residential appliances including heating and cooling equip-ment. Energy Journal 8 (1), 101–124.

Shelbecker, G.E., Roszkowski, M.J., Culter, N.E., 1990. Investors’ risk tolerance andreturn aspirations, and financial advisor’s interpretations: a conceptual modeland exploratory data. Journal of Behavioral Economics 19, 377–393.

Sozou, P.D., Seymour, R.M., 2003. Augmented discounting: interaction betweenaging and time-preference behavior. Proceedings of the Royal Society of LondonB 270, 1047–1053.

Tan, C.T., Johnson, R.D., 1996. To wait or not to wait: the influence of culture on dis-counting behavior. In: Loke, W.H. (Ed.), Perspectives on Judgment and DecisionMaking. Scarecrow, Lanham, MD, pp. 97–106.

Thaler, R.H., 1981. Some empirical evidence on dynamic inconsistency. EconomicLetters 8, 201–207.

Thaler, R.H., Shefrin, H.M., 1981. An economic theory of self-control. Journal of Polit-ical Economy 89 (2), 392–410.

The 2008 National Jumpstart Coalition Survey of High School Seniors and CollegeStudents for the Jump-Start Coalition® for Personal Financial Literacy. Availableat http://www.jumpstart.org/upload/2009 FinLit-Mandell.pdf.

Trostel, P.A., Taylor, G.A., 2001. A theory of time preference. Economic Inquiry 39(3), 379–395.

Van Praag, B.M.S., Booij, A.S., 2003. Risk Aversion and the Subjective TimeDiscount Rate: A Joint Approach. CESifo Working Paper Series No. 923;Tinbergen Institute Discussion Paper No. 2003-018/3. Available at SSRN:http://ssrn.com/abstract=396563.

Warner, J.T., Pleeter, S., 2001. The personal discount rate: evidence from militarydownsizing programs. American Economic Review 91 (1), 33–53.

Weber, E.U., Milliman, R.A., 1997. Perceived risk attitudes: relating risk perceptionto risky choice. Management Science 43 (2), 123–144.

Zollo, P., 1995. Talking to teens. American Demographics 17 (11), 22–28.

dx.doi.org/10.1016/j.socec.2010.02.012

http://ssrn.com/abstract=574067


http://www.jumpstart.org/upload/2009_FinLit-Mandell.pdf


FACTORS AFFECTING SOLDIERS’ TIME PREFERENCE 1

1

Factors affecting soldiers’ time preference:

A field study in Israel


2

Abstract

The current field study examines how the day of the week, optimism level and other personal

characteristics influences the time preference of soldiers. To do this, we compare the time

discount of soldiers in the Israel Defense Forces at the beginning of the work week (Sunday

in Israel) and just prior to the weekend (on Thursday afternoon in Israel). The soldiers were

asked to answer questionnaires regarding their time preferences, and dispositional optimism.

We found that the soldiers have a higher subjective discount rate on Thursday when they

need money for weekend activities. In addition, we found that optimism, being an older

brother, having parents with higher earnings, and time remaining until discharge are

negatively related to subjective discount rate. Conversely, having a balanced bank account is

positively related to subjective discount rate.

Keywords: Time preference; Discount rates; Soldiers; Optimism; Military; Birth order.


3

Factors affecting soldiers’ time preference:

A field study in Israel

The subjective discount rate is the rate at which individuals trade current value for

future outcome. This rate varies from person to person, depending on each one’s willingness

to wait. In general, an individual who values the present more than the future will have a

higher subjective discount rate than a person who places more value on the future. Initial

research in the field of time preferences focused primarily on monetary questions, and found

that subjective discount rates vary according to age,1 time delay, amount of money, risk and

direction of trade (see Benzion, Rapoport, & Yagil, 1989; Thaler, 1981; Stevenson, 1992;

Chapman & Elstein, 1995; Laibson, 1997; Frederick, Loewenstein, & O’Donoghue, 2002;

Myerson, Green, Hanson, Holt, & Estle, 2003; Prelec, 2004).2 Other important effects that

may influence subjective time discounts are morbidity and mortality. Chao, Szrek, Pereira,

and Pauly (2009) found that subjective time discount is related to both physical health and

subjective expectation of survival. Their research was done in South Africa where, due to

HIV/AIDS, the middle-age mortality rate is much higher than in developed countries. Liu and

Aaker (2007) show that experience with death of a close other by cancer, which bears

information about time, is associated with making decisions that favor the long-term future

over short-term interests. Carstensen and his colleagues (Carstensen, Isaacowitz, & Charles,

1999; Carstensen, Fung, & Charles, 2003) also separated age from mortality in their research

to develop the socio-emotional selectivity theory. According to this theory, when time is

1 The research on the effect of age on subjective discount is not consistent. Some found decreases with age and

others found curvilinear relationships (Green, Fry, & Myerson, 1994; Green, Myerson, & McFadden, 1997;

Green, Myerson, & Ostaszeski, 1999; Harrison, Lau, & Williams, 2002; Lahav, Benzion & Shavit, 2010; Read

and Read, 2004; Sozou and Seymour , 2003; Thaler & Shefrin, 1981).2 There are also many articles regarding non-monetary questions, such as choices for various durable items

(Gately, 1980; Ruderman, Levine, & McMahon, 1987), and even for beers and bags of chips (Tsukayama &

Duckworth, 2010).


4

perceived as open-ended, people are future-oriented and concentrate on knowledge-related

goals. However, when time is perceived as limited, people become present-oriented and start

concentrating on emotional goals. The researchers checked their theory and found that young

people with symptomatic HIV do become present-oriented, and concentrate on emotional

goals.

Lahav, Benzion, and Shavit (2011) compared the time preferences of soldiers,

university students and high school students in Israel. During compulsory service, the soldiers

live in a violent atmosphere where they face great uncertainty about the near future and a

high risk of mortality. University students and teenagers live in much calmer environment,

and are tested for performance only periodically. The soldier-subjects showed relatively high

subjective discount rate when compared to the other two groups. The authors concluded that

the higher subjective discount rate among soldiers could be the result of high perceived risk3

they face in the army or higher mortality risk.

In the current paper, we examine how the day of the week, optimism level and other

personal characteristics influence the time preference of soldiers. To do this, we compare the

time discount of soldiers in the Israel Defense Forces at the beginning of the work week

(Sunday in Israel) and just prior to the weekend (on Thursday afternoon in Israel). The day of

the week is known to have its effect on people’s feelings. For example, the phrase “Monday

blues” is widely used in songs and literature to express how people feel on Monday

mornings, while people are presumed to feel better on the weekend. In Israel, Saturday is the

day of rest and many people do not work on Friday. However, Sunday is an ordinary

3 Perceived risk is the subjective judgment of risk that people attribute to a situation. According to the

sociological approach, perceptions are socially constructed by institutions, cultural values, and ways of life

(Wildavsky & Dake, 1990).


5

workday, equivalent to Monday in other western countries, while Thursday evening is similar

to the western Friday evening.

The soldiers in Israel are a unique group because they are drafted at young age, prior to

beginning their tertiary education, and all earn similar, very low salaries. By law, soldiers

serve full-time for two (females) or three (males) years, and then continue serving in the

reserves. Those who continue in the military career earn a full salary, according to their rank

and position. Their “employer” is a harsh and total institution,4 with very clear values. They

are intensively trained and tested from the first day of basic training. During compulsory

service, most soldiers are stationed at bases that are at least a few hours’ drive from home,

and spend their weeks on the base. On Thursday afternoon, many return home where they

“hang out” with their friends during the weekend.

In order to prevent a selection bias, we distributed the questionnaires to the soldiers

when they were travelling on the train. The benefits of this method are discussed in greater

detail below in the experimental design section. We found that the soldiers have a higher

subjective discount rate on Thursday when they need the money for weekend activities. In

addition, we found that optimism is negatively related to subjective discount rate.

Furthermore, subjective discount rate is negatively related to gender (female), being an older

brother, risk aversion, higher parents earnings, time until army discharge and higher

education level. Conversely, having a balanced bank account is positively related to

subjective discount rate.

The rest of the paper is organized as follows: in the next section, we discuss time

preferences, day of the week affects, and optimism, and relate them to our research. Then we

4 A total institution is an institution where all parts of life of individuals under its authority are subordinate to

and dependent upon the hierarchy of the organization.


6

present the hypotheses and describe the experimental design, followed by the results,

summary and conclusions.

Time Preferences, Optimism, and Day of the Week.

Personal characteristics such as risk aversion, impulsivity and compulsivity are argued

to influence time preferences (Andersen, Harrison, Lau, & Rutström, 2008; Frederick et al.,

2002; Frederick, 2003; Loewenstein, Weber, Flory, Manuck, & Muldoon, 2001). For

example, a more impulsive person may have a harder time resisting present consumption, and

therefore have difficulty saving for the future. Such behavior is related to a stronger present

preference, and a higher subjective discount rate. Optimism, a positive view of events in life,

is another personal characteristic that has been found to influence decision making. Scheier

and Carver (1985) suggested that people tend to be either generally optimistic or generally

pessimistic. “Dispositional optimism” is the generalized expectancy that good things will

happen in the future and bad things will be minimal (Scheier & Carver, 1992). An individual

who is more optimistic will face life events, whether good or bad, with an overall positive

outlook. An individual who is more pessimistic will view these events in the opposite way,

believing that their outcome will be negative (Hayes and Weathington, 2007). Optimism,

when viewed as a relatively stable personality characteristic, may account for the method by

which individuals regulate their actions (Peterson, 2000; Scheier & Carver, 1985). If

individuals take an optimistic approach to difficult situations, they may be better able to

adjust and overcome challenges. For example, Scheier and Carver (1992) found that college

students scoring higher in levels of optimism reported less depression, loneliness, and stress.

Taylor and Brown (1988) found that people with high optimism level exhibited higher levels

of motivation, determination, and performance. Optimistic athletes exhibited a higher ability

to cope and stronger mental health (Nicholls, Polman, Andrew, & Backhouse, 2008). Haack


7

and Mullington (2005) restricted the sleep hours of 20 adult subjects to 50% of their habitual

amount for 12 consecutive days. They found that optimism-sociability declined

progressively, by up to 15%, and bodily discomfort increased, in comparison to a control

group. Their finding is consistent with “The 2002 Sleep in America Poll” (National Sleep

Foundation, 2002) which reported that people who get more than six hours of sleep on a

weekday are more likely to be optimistic and/or satisfied with life; and that those sleeping

seven to eight hours have the highest ratings for optimism and life satisfaction. Regarding

financial markets, Heaton (2002) argued that corporate managers are considered too

optimistic, leading them to overvalue their own corporate projects, and undervalue the

possibility of bad performance.

Emotions and optimism have also been related to the day of the week. Early research

on this issue, using student-subjects, found that Monday is associated with more negative

emotions, (Farber, 1953; Pecjak, 1970; Rossi & Rossi, 1977). Christie and Venables (1973)

examined a wider age range of male subjects, aged 23 to 53, and found their mood to be

lower on Monday and higher on Friday. This day of the week also has a well-known effect on

equity markets. In several studies, Monday correlated with decreasing securities prices in US

equity markets, while securities prices tended to go up on Fridays. (Cross, 1973; French,

1980; Keim and Stambaugh, 1984). Pettengill (2003) reviewed more recent articles and found

that the tendency of securities prices to go down on Monday (“Monday effect”) is also valid

in many developing and developed countries. Kenourgios and Samitas (2008) found that the

tendency of securities prices to go down on Monday and up on Friday (“day of the week

effect”) is also valid in the Greek equity market. Rystrom and Benson (1989) argued that

these effects are, among other things, the result of behavioral influences. They explained that

on Monday people tend to feel “blue,” making them more pessimistic about the outlook for

the securities they hold. This attitude may lead them to sell for less, while more benign


8

attitudes later in the week could lead to relatively greater demand for securities. To the best

of our knowledge, no prior research has examined the effect of optimism on time preference

or dealt specifically with soldiers.

Hypotheses

Day of the Week Affect

During mandatory service, soldiers leave in a violent atmosphere, and face great

uncertainty about the near future. Beside the mortality and morbidity risk, during service they

face uncertainty about daily activities: their commanders control their schedule, and can

change it instantly if the soldier, or even other soldiers, misbehaves in any way. For example,

it is quite common for soldiers to have weekend leave canceled for misbehavior or for not

achieving a training goal. In some cases, they can be grounded on base for weeks as a

punishment.5 Lahav et al. (2011) suggest that this uncertainty about daily activities and the

violent atmosphere of military service can lead to higher perceived risk about the future

(Slovic, Fischhoff, & Lichtenstein, 1982; Weber & Milliman, 1997), which produces a high

preference for the present. The authors found that the soldier-subjects showed a relatively

high subjective discount rate when compared to university and high school students. This is

also consistent with Carstensen et al. (1999, 2003) whose socio-emotional selectivity theory

suggests that when time is perceived as limited, people become present-oriented.

At the beginning of the week (Sunday morning), soldiers have no immediate need for

money, as the army takes care of their personal needs, such as food, clothing and a place to

sleep. Compulsory service soldiers earn an extremely low salary,6 but have no immediate

5 In Israel, most forces are posted within a few hours drive of their homes and soldiers usually receive a fewdays leave at least once every three weeks, and often every week.6 The salary of a soldier during mandatory service ranges from about NIS (New Israeli Shekel) 350/month fornon-combat positions to about NIS 750/month for combat positions. The average monthly salary in Israel at thetime of the experiment was NIS 8,477 (source: http://www.cbs.gov.il/hodaot2011n/26_11_100t2.pdf ) . Theexchange rates at the time of the experiment were NIS 3.623 to USD 1 and NIS 5.895 to GBP 1 (source:http://www.bankisrael.gov.il/heb.shearim/index.php?day=1&month=3&year=2011 ).


9

need for money until the end of the week. On Thursday afternoon, just after they leave their

base on their way home, their need for money for the upcoming weekend leisure activities

increases. The soldiers usually go out nightly during the weekend, to clubs, pubs and similar

places, to unwind the stress of a week in the army. Financing these activities can be quite

difficult for them, because that their monthly salaries are less than 10% of the average

monthly salary in Israel.

As a result of the difference in need for money we expect to find that:

H1: The subjective discount rate of the soldiers will be higher on Thursday, in comparison to

Sunday.

Optimism

Optimism level is an important personal characteristic that was found to influence

decision making (Carver et al., 1979; Heaton, 2002; Nicholls et al., 2008; Taylor and Brown,

1988). For the best of our knowledge, no previous research has correlated time preferences

and optimism, however, a person’s subjective discount rate is determined, among other thing,

by his belief of an actual future payment. Therefore, increased optimism (meaning a lower

expectation of negative events in the future) should result in a reduced risk premium for the

future, which could results in a reduced subjective discount rate. As a result, we expect that:

H2: Subjective discount rate will decrease with an increase in optimism level.

Demographic and Financial Characteristics

Birth order.

Birth order has been a standard variable in psychological research for many decades

(see Healey and Ellis, 2007; Rink 2010; Sulloway, 1996; Sulloway and Zweigenhhalf, 2010).

Firstborns are usually described in the literature as being conformists and having a high need

for achievement (Forer, 1977). Sulloway (1996) used a Darwinian approach, and argues that

children compete for parental resources by creating distinctive niches. Firstborn children,


10

whose position is already established, tend to be responsible, competitive, and conventional;

later-born children must distinguish themselves by being playful, cooperative, and especially

rebellious. Although presumably strongest in early childhood, Sulloway (2001) points out

that the influence of birth order on personality may continue to operate in adulthood. We

argue that higher responsibility and conformity of firstborns make them focus more on the

future. Therefore, they will be more future oriented in comparison to non-firstborns, and so

we expect to find that:

H3: Firstborn children will exhibit lower subjective discount rates than non-firstborns.

Parental earning.

A person’s earnings can explain his financial situation, and were found to be negatively

correlated with subjective discount rate (Harrison, 2002; Warner and Pleeter, 2001).

However, salaries during compulsory service are less than 10% of the average salary, making

solders dependent on their parents’ income. Therefore, for young soldiers we expect that

parental income correlates with discounting in the same way that own-earnings have been

found to correlate with discounting for older subjects:

H4: Subjective discount rate will increase with a decrease in parental earnings.

Financial situation.

An unbalanced bank account is another estimator of a person’s financial situation. For

compulsory service soldiers, this might be related to their expenses, and to their parents’

financial situation. Overdrawing one’s bank account is an accepted and widely-used way of

taking a small loan in Israel, as a possible alternative for credit card debt. Each account has a

defined limit for its overdraft in his bank account, and the overdraft interest rate, which is

very high, is determined in advanced. Therefore, being overdrawn might suggest a stronger or

more immediate need for money, in order to stop paying the high interest, and could result in

a higher subjective discount rate. Consequently, we expect that:


11

H5: Soldiers with an overdrawn bank account will show a higher subjective discount rate

than those who are not overdrawn.

Army Service Characteristics

Time left until discharge.

During mandatory service, the army takes care of the soldiers’ needs, and controls their

schedule. Such an institution affects uncertainty regarding the next few days, but on the other

hand creates a cushion effect (Lahav et al., 2010; Mahajna et al., 2008; Rosenboim, Shavit, &

Shoham, 2009, Weber and Milliman, 1997) regarding the long run. According to this effect,

when an individual has a safety net or feeling that he or she will be supported (or protected)

in times of need, this leads to a reduction in risk premium regarding the future, which can

result in a reduced subjective discount rate. As the time remaining in a soldiers’ military

service lessens, this effect weakens and the soldiers start concentrating on the future with all

its uncertainties, including planning for a post-army trip,7 academic direction and

professional direction. Plans for higher education and a trip also increase the need for money

to finance them. As a result we expect that:

H6: Subjective discount rate is negatively related to time left until army discharge.

Prolonging army service.

A compulsory service soldier can decide to prolong the army service, for one or more

years (at a full salary) in the standing army. By delaying discharge, a soldier who considers

prolonging his or her service, prolongs the influence of the cushion effect. We expect to find

that:

7 In Israel, many young adults finish their army service, and start looking for a job in order to save money for along trip abroad. The term “post-army trip” is commonly used in Israel to describe this trip. The most populardestinations are the Far East and South America, followed by Europe and the USA. When they return from thetrip, they move on to academic studies. In our sample, 89.9% claimed they are planning a "post-army trip." Ofthese, 36% planned to travel in South America; 15.2%, in the Far East, and 25%, in the USA or Europe.


12

H7: Soldiers who intend to prolong their army service will show lower subjective discount

rate than soldiers who do not intent to prolong their army service .

Experimental Design

Subjects

The subjects consisted of 178 Israeli soldiers during their compulsory service period

(average age 19.53, age range: 18-22,8 92 males and 86 females). The Sunday group

consisted of 92 Soldiers (average age 19.41, age range: 18-22, 47 males and 45 females) and

the Thursday group consisted of 86 Soldiers (average age 19.65, age range: 18-22, 45 males

and 41 females). The questionnaires were distributed on Sunday mornings: just after the

weekend was over, but before the soldiers start the routine schedule inside the base, and on

Thursday afternoon: just after they have left the harsh environment of the army, and can

concentrate on the upcoming weekend.

Procedure

The questionnaires were distributed on trains travelling between two large cities in

Israel (Tel Aviv and Be’er Sheva). This train line is used extensively by soldiers, who are

exempt from paying fares on public transportation, and need to get to and from bases in the

south of Israel. The subjects were given approximately 40 minutes to fill the questionnaire,

and were paid NIS 20 for their time. They were asked not to speak with each other while

completing the questionnaires.

Similar to Lahav et al. (2011), we chose to distribute the questionnaires on trains to

facilitate access to soldiers, since the IDF usually does not allow civilian researchers an

access to soldiers on bases. Furthermore, even if we had been able to enter one or two

military bases to conduct the current survey, it would create selection bias, since soldiers in

8 In exceptional circumstances, such as being a new immigrant, a person can start his mandatory service a fewyears after the age of 18. This is the case for the oldest solders in our sample.


13

each service (Air Force, Navy, or Army) have certain shared traits. However, the train is a

comfortable means of transportation used by young soldiers from all services. This helped us

to avoid selection bias, and also enabled us to access the soldiers when the “day of the week

effect” ought to be very strong.

The questionnaires (see Appendix A for a translated questionnaire) began by informing

the subjects that they are about to complete questionnaires in the field of decision making.

They were further instructed that the questions did not have “correct” answers, so each

answer should reflect their personal preferences.

Time preferences.

In the first part of the questionnaires (postponing receipt), participants were asked to

chose between a sum (x) to receive today, and a delayed sum (p) at a future time (t). They

were asked to fill in the minimum future sum that they are willing to accept (WTA) at time t

(see Appendix A, Part 1).

The questions referred to two different amounts: NIS 600 and NIS 3,000, and three

different delays: one week, two months, and 12 months, so this part of the questionnaire had

six different questions with three different time periods and two different amounts. All

subjects answered all six scenarios.

Optimism.

In order to assess participants’ level of dispositional optimism, the second part of the

questionnaire included 10 items that evaluate generalized expectancies for either positive or

negative outcomes (see Appendix A, Part 2). These items included the six items of the Life

Orientation Test–Revised (LOT-R) questionnaire, developed by Scheier et al. (1994), and

four additional items from the Extended Life Orientation Test (E-LOT), developed by Chang

et al. (1997). The E-LOT Questionnaire combined items from the LOT-R questionnaire with

items from the longer Optimism/Pessimism Scale (OPS) question developed by Dember,


14

Martin, Hummer, Howe, & Melton (1989). The four additional items are the optimism items

that are included in the E-LOT questionnaire, but did not originate in the LOT-R

questionnaire. The subjects were asked to circle their answer on a 5-point Likert-type scale

from 1 (strongly disagree) to 5 (strongly agree). Participants who score higher on the

questionnaire are considered to have higher levels of dispositional optimism.

Additional explanatory variables.

In order to further understand possible influences on time preferences, the

questionnaire included short questions regarding socio-demographics characteristics, and

army service characteristics. Specifically, we collected information on age, gender, parental

allowance, birth order, educational level, bank balance, time left until discharge, tiredness,

weekend plans, and plans for a post-army trip (see Appendix A, General Data section).

Results

Comparing Time Discount by Day of the Week

Table 1 shows the annual subjective discount rate for each period and amount. The

annual discount rate for delaying payment was calculated as follows:

r = ( 1)( XP )* )12(

t, (1)

where P is the amount the subject is willing to accept in t months for delaying the receiving

of the amount of X today.

We used a one-tailed T-test to examine the null hypothesis that the subjective

discounting of soldiers is equal on Sunday and Thursday, and Cohen’s d to measure the

strength of the difference.


15

Table 1

Average (STDV) Annual Subjective Discount Rates (Comparing Sunday to Thursday)

Day Amount 1 week 2 months 12 months

Sunday

(n = 92)

NIS 600 * 1021.41% 332.45% 216.34%

(3136.07%) (298.52%) (244.02%)

NIS 3,000 * 602.96% 241.48% 138.83%

(877.20%) (265.83%) (146.41%)

Thursday

(n = 86)

NIS 600 * 2395.40% 673.49% 380.31%

(3136.07%) (712.76%) (326.29%)

NIS 3,000 * 1339.40% 419.88% 254.38%

(1809.82%) (459.78%) (246.51%)

T-test (d.f),

p-value

NIS 600

t(176) = 3.94 t(176) = 4.21 t(176) = 3.81

p = .000 p = .000 p = .004

d = 0.438 d = 0.624 d = 0.569

T-test (d.f),

p-value

NIS 3,000

t(176) = 3.49 t(176) = 3.19 t(176) = 3.83

p = .000 p = .001 p = .000

d = 0.518 d = 0.475 d = 0.570

* Standard deviations are in parentheses.

The subjective discount rates are very high, with high variance, suggesting wide

variations between the subjects. This finding is consistent with earlier studies about the

subjective discount rate, especially for short periods (e.g. Frederick et al., 2002, Lahav et al.,

2010, 2011; Thaler, 1981). In all cases, we found that the average subjective discount rates

were significantly higher for the soldiers on Thursday, with relatively high size effect,

consistent with hypothesis 1.


16

We also asked the participants if they already had weekend plans. On Sunday, 52% had

plans, while 79% had weekend plans on Thursday.9 The preferred recreations were going to a

pub (47.8%), and clubbing (25.8%); both are expensive and can cost a substantial percentage

of the soldiers’ salaries. The fact that soldiers on active duty have almost no need for money

during the week, and that most have weekend plans on Thursday supports our hypothesis

about the influence of a specific need for the money on subjective discount rate.

In addition, we found that the subjective discount rates were significantly lower for the

higher amount (NIS 3,000) than for the lower amount (NIS 600), for all the periods

(p < 0.001), consistent with the results of Thaler (1981), which pointed to lower subjective

discount rates for higher amounts.

Regression Analysis for Subjective Discount Rate

In order to further understand possible influences on time preferences, we conducted a

multivariate regression analysis of subjective discount rate on a dummy variable for the

higher amount (NIS 600=0; NIS 3,000=1), time variable (1 week, 2 months or 1 year ), and a

dummy variable for day of the week (Sunday = 1; Thursday = 0). The regression analyses

also included additional independent variables which might influence the soldiers’ time

preferences, according to the hypotheses listed above. These variables and the dummy

variables created to represent some of them are:

Optimism –To check for differences in time preferences resulting from optimism

levels, we first created an index for the ten questions in the dispositional optimism

questionnaire (Optimism Index, henceforth, “OI Index”). The Cronbach’s α of the OI Index

was 0.743. The index was calculated as an average of the answers to the ten optimism items,

9 We used a two-tailed Z-test to examine the null hypothesis that the same proportions of soldiers have weekendplans on Sunday and Thursday. The test show a significant difference (z(176)= 3.77, Sig.= .000) meaning thatmore soldiers have weekend plans on Thursday in comparison to Sunday.


17

calculated separately for each subject. The index values ranged from 1 (less optimistic) to 5

(more optimistic).

Risk preferences – In order to control for risk aversion, we used a lottery question

similar to a question used in Booij & van Praag (2009) (see Appendix A, Part 1). The risk

aversion measure is the willingness to pay (WTP) for the lottery. Willingness to pay a lower

amount to participate in the lottery indicates a higher risk aversion.

Birth order – To understand whether the greater responsibility and conformity of

firstborn children affects time preferences, the questionnaire included a question regarding

birth order. 73 out of the 178 subjects were firstborn, and we added a dummy variable for

being firstborn (firstborn = 1, other = 0).

Gender – In order to control for gender differences, the questionnaire included a

question regarding gender. Of the 178 subjects, 86 were females, and we added a dummy

variable for gender (female = 1, male = 0).

Education Level –The questionnaire included two questions regarding subject’s high

school education. The first one regarded their math level. In Israeli high-schools, students are

sorted into three tracks for mathematics, by ability. The highest level is “5 points” and the

lowest is “3 points.” In response to our question regarding which mathematic matriculation

exam they had taken, 73 subjects answered “3 points;” 57, “4 points”, and 48, the highest

level or “5 points.” We created a dummy variable for the highest level (5 points of math = 1,

other = 0). The second variable regarded extended studies. A high school student can take the

basic matriculation exams, or choose to expand his studies and study extra subjects. In this

case, the students are again divided into three groups, but not only by ability, but also

according to their desire to expand their knowledge on a specific subject. Choosing several “5

point” subjects means more intensive learning, and extra practice at home. We asked the

participants how many “5 points” matriculation exams they had taken, in addition to math, if


18

any. We then created the scale variable “Other 5 points in matriculation” as a proxy of higher

level of education.10

Parental earnings – In response to question regarding parental earnings, 86 subjects

reported parents with above average earnings, and the parents of the remaining 92 reported

average or below average earnings. The dummy variable 1 was assigned to above average

parental earnings and 0 for average or below average earnings.

Financial situation – To evaluate subject’s financial situation, the questionnaire

included a question the state of their bank accounts. In response, 116 subjects claimed they

are never overdrawn, 38 claimed that their account is overdrawn at times and the remaining

22 said that always is in overdraft. The answers were grouped to a dummy variable with

never having an overdraft = 1, other = 0.

Time left until discharge – The questionnaire included two questions relating to time

remaining in the army. The first was a straight forward question about the time left until the

subject’s discharge from compulsory service (in months). The second question asked, “Are

you considering signing on for service in the regular army?” This is a common way of

referring to prolonging the army service for one or more years (at a much higher salary), in

the career military. Of our subjects, 119 said they were considering this option and 59 were

not. The dummy variable was set as yes = 1, other = 0.

Combat Soldier11 – We asked the soldiers if they are combat soldiers or not. Of the

178 subjects, 103 were and 75 were not. The relevant dummy variable is combat soldier = 1,

non-combat soldier = 0).

10 The answers ranged from 1 (20.2%) to 5 (2.2%): 33.1% took 2 “5 point” exams, 29.8% took 3 “5 point”exams, and the remaining 11.2% took 4.11 Combat soldiers are soldiers who train for and perform field missions. Their training is longer and harder, andif they actually do field missions, the mortality and morbidity chances become higher. Lahav et al. (2011) foundthat combat and non-combat soldiers had no significant differences in subjective discount rates.


19

Table 2 presents descriptive statistics for these variables. In addition, it presents the

results of a Z-test/T-test for the null hypothesis that the specified variable is equal on Sunday

and on Thursday. The purpose of these tests is to examine whether both samples are equal in

characteristics.

Table 2

Descriptive Statistics for Variables, by Day of Week

The variable* Day N Mean STDV 2-tailed Z/T-test (d.f.) ; Sig.

OI index Thursday 86 3.72 0.58 t (176) = -3.049; p= 0.003

Sunday 92 3.96 0.48

WTP for lottery Thursday 86 46.08 34.10 t (176) = -1.22 ; p= 0.224

Sunday 92 52.14 32.19

Firstborn Thursday 86 0.42 0.50 z (176)= 0.271;p= 0.786

Sunday 92 0.40 0.49

Female Thursday 86 0.48 0.50 z (176)= -0.133; p= 0.894

Sunday 92 0.49 0.50

Math 5 points Thursday 86 0.29 0.46 t (176)= 0.609; p= 0.544

Sunday 92 0.25 0.44

Other 5 points Thursday 86 2.34 1.20 t (176)= 0.200; p= 0.842

Sunday 92 2.30 0.98

Parents earning more

than average

Thursday 86 0.47 0.50 z (176)= -0.400; p= 0.689

Sunday 92 0.50 0.50

Bank account never

overdrawn

Thursday 86 0.60 0.49 z (176)= -1.399; p= 0.161

Sunday 92 0.70 0.46


20

Time until discharge Thursday 86 19.70 13.15 t (176)= 0.149; p= 0.882

Sunday 92 19.45 9.17

Considering signing Thursday 86 0.33 0.47 z (176)= -0.141; p= 0.887

Sunday 92 0.34 0.48

Combat soldier Thursday 86 0.66 0.48 z (176)= 2.159; p= 0.031

Sunday 92 0.50 0.50

* The results of the independent sample T-tests and Z-tests (for percentages) are presented in the last column of the table.

As Table 2 shows, except for combat soldiers and optimism, all variables had no

significant differences, which suggest that the Sunday and the Thursday samples are mostly

equal in characteristics. The OI index was significantly higher on Sunday. A possible

explanation for the difference could be the effect of extra sleep (Haack and Mullington,

2005), and the weekend recreation, which might have a stronger influence on optimism than

the “Monday blues.” During the week many soldiers guard at night, and train for long periods

of 24 to 48 hours; sometimes operational activity also continues for more than a day.

Although the IDF issued a directive mandating an average of 6 sleeping hours per night,12

these activities disrupt the sleeping habits, which results in increased tiredness by the end of

the week.13 Optimism is supposedly influenced by many factors, and a “re-energizing”

weekend after a hard week in the army might also effects a soldier’s optimism level.

We also asked the participants to rate how tired they are, on a scale of 1 (not tired at

all) to 5 (very tired). The mean answer for the Sunday group (mean= 3.08) was lower than

that of the Thursday group (mean=3.39).14 This means that, on average, the subjects were less

12 For the directive (in Hebrew) see,http://www.aka.idf.il/rights/asp/info.asp?moduleId=2&catId=22703&docId=2272113 According to the National Sleep Foundation, young adults need 7 to 9 hours of sleep/night.14 We used a one-tailed T-test to examine the null hypothesis that the tiredness level of soldiers is equal onSunday and Thursday. The test show a significant difference (Sig.= .04, df.=176).


21

tired on Sunday, which supports our suggestion of the effect of tiredness and consistent with

Haack and Mullington (2005) findings of decreasing optimism with restricted sleep.

The regression analysis for time preferences is presented in Table 3.

Table 3

Regression Analysis for Subjective Discount Rate

Independent Variable*

Coefficient

(Sig’)

(Constant) 29.293

(0.00)

Sunday -4.315

(0.000)

Time -7.924

(0.00)

3000 NIS -3.323

(0.00)

OI index -1.665

(0.02)

WTP for lottery -.041

(0.00)

Firstborn -3.348

(0.00)

Female -1.717

(0.03)

Math 5 points -.540

(0.56)


22

Other 5 points -1.092

(0.00)

Parents Earning More Than Average -1.950

(0.01)

NO Minus In Bank Account 2.082

(0.01)

Time until discharge -.111

(0.00)

Considering signing -2.817

(0.00)

Combat Soldier .552

(0.49)

R- Square .189

Model Sig. 0.000

+ The dependent variable is subjective discount rate.

*Significance levels are in parentheses.

The results in table 3 show that subjective discount rate decreases with time and with

the increase in the amount, as table 1 already showed.

The subjective discount rate decreases with the increase in dispositional optimism (OI

index). In line with hypothesis 2, this indicates that an increase in optimism results in a

decrease in the subjective discount rate. We also see that subjective discount rate decreases

with the increase in WTP for the lottery. This means that a decrease in risk aversion

(increasing WTP) results in a decreased subjective time discount, consistent with earlier


23

research (Anderhub, Güth, Gneezy, & Sonsino, 2001; Andersen et al., 2008; Lahav et al.,

2011).

For the dummy variable “firstborn,” we found a significant negative coefficient, which

indicates that being the oldest sibling results in a reduced subjective discount rate. This

finding is in line with hypothesis 3 which states that firstborns will exhibit comparably lower

subjective discount rates. We also find significant negative effect for females, suggesting that

they present lower subjective discount rates. Earlier research on time preferences found

conflicting results of gender influences.15 Most articles provide no hypothesis about the

influence of gender on discounting, and usually use gender as a control variable.

For the dummy variable regarding math level we find no significant effect. However,

for the second education level variable (“Other 5 point matriculation exams”), we did find a

significant negative effect, meaning that higher educational level in high school is related to

lower subjective discount rate. This finding corresponds to earlier findings of negative

correlation between education and subjective discount rate (Fersterer & Winter-Ember, 2003;

Harrison, 2002; Warner and Pleeter, 2001).

Parents’ earnings had a significant negative effect on discounting. Higher parental

earnings correlate with lower subjective discount rate, consistent with hypothesis 4. Bank

account balance had a significant effect as well. Subjects who never have an unbalanced

(overdrawn) bank account showed a significantly higher subjective discount rate. This

finding is inconsistent with hypothesis 5. One would expect an opposing effect in which

those with an overdraft, who pay high interest rates, have a stronger need for money and

therefore a higher subjective discount rate. A possible explanation for this finding is that

15 Van Praag and Booij (2003) found higher subjective discount rates for women, while Harrison (2002) andLahav et al. (2011), found no influence of gender on time preferences. Warner and Pleeter (2001), Bütler andTeppa (2005), and Bauer and Chytilová (2009) found a lower subjective discount rate for women.


24

people who do not overdraw their account are fearful of having an unbalanced account and

therefore prefer not to postpone a receipt. However, this is only speculation.

“Time to discharge” had a significant negative effect. This is in line with hypothesis 6

which states that subjective discount rate is negatively related to time remaining in service.

“Considering prolonged army service” also had a significant negative effect on subjective

discount rate, consistent with hypothesis 7 which states that subjective discount rate is

negatively related to a soldiers intentions to prolong the army service.

The last army-related variable was “combat soldier,” for which we found no significant

differences between combat and non-combat soldiers. This finding is consistent with Lahav

et al. (2011).

Conclusion

In the current field study, we examine how the day of the week, optimism level and

other personal characteristics influences the time preference of soldiers. To do this, we

compared the time discount of soldiers in the IDF on Sunday morning and on Thursday

afternoon, while they were travelling on the train. This method enabled us to study a group

with many similar characteristics (age, earnings, education, work place act.) and examine the

effect of natural manipulation on that group.

In our view, compulsory service soldiers in Israel could serve as a good example for the

correlation between time preference and the need for money, since their need for money

changes dramatically during the week: On Sunday mornings, they face a full week during

which the army will take care of their schedule and most of their personal needs. During the

weekend they are responsible for their schedule, and need money to finance their activities.

Therefore, their subjective discount rate increases.

We find it interesting that subjective discount rate were higher on Thursday, not only

for delays of one week in which the need for money actually changes, rather also for long


25

delays of 2 months and a year.16 If we take the example of a year delay, this finding implies

that a soldiers discount rate for one year have a dramatic decrease at the beginning of the

week, and a dramatic increase at the end of the same week, and so on. Apparently the need

for money in the short run biases the time preferences of the soldiers even for the long run.

We do, however, also see an influence of future plans and events on time preference.

As time until discharge shortens and the uncertainty of the after army arise, subjective

discount rate increases. Interestingly, soldiers who are planning on prolonging the army

service are also prolonging the influence of the “cushion effect” of the army, and therefore

exhibit lower subjective discount rates.

In addition, we found that dispositional optimism is negatively related to subjective

discount rate. We are unaware of any previous research that correlates time preferences and

optimism, and so we can only speculate if this is due to a reduced risk premium regarding the

future.

Furthermore, we did not find any earlier research that correlates time preferences and

birth order. In the current experiment, firstborn subjects had lower subjective discount rates,

which might be the results of higher responsibility in comparison to non-firstborn subjects.

This finding is in line with Sulloways’ (2001) argument regarding the influence of birth order

in early adulthood.

Finally, we find support for earlier findings about the influence of earnings and

education on adults’ time preferences (Fersterer & Winter-Ember, 2003; Harrison, 2002;

Warner and Pleeter, 2001). In the current study, parents’ earnings influence time preferences

the same way own-earnings influence time preferences for adult subjects. In addition,

education usually correlates with earnings, and it is hard to separate the effects of education

16 We find similar results when comparing the increase in the WTA amount from one week delay to 2-monthand 12-month delays. The subjective discount rate of the increase in WTA was significantly higher on Thursday(p < .01).


26

and earnings on discounting. However, the earnings of all soldiers in this group are very low,

and we still find a negative effect of education on discounting. However, this effect was not

due to difference in years of education, as earlier research found, rather to education level. As

Van-Praag and Booij (2003) explained, education is an investment in the long-run that is

related to long-term vision and better self-control. In high school, this self-control enables a

student to pursue more advanced studies that require longer hours of studying. Therefore, a

higher level of education may be a mediator between self-control and discounting.

To conclude, the results of the current study add to the existing research, which

indicates that personal and demographic characteristics influence time preferences.

Nevertheless, the current study shows that a specific institution or situation can have a

stronger effect than other characteristics, and can cause a person to repeatedly change his

time preferences depending on a specific date or event. More field studies are needed in order

to find the relationship between time preference and personal characteristics, including but

not limited to optimism and birth order, or the effect of the day of the week on time

preferences.


27

References

Anderhub, V., Güth, W., Gneezy, U., & Sonsino, D. (2001). On the interaction of risk and

time preferences: An experimental study. German Economic Review, 2, 239-253.

Andersen S., Harrison, G. W., Lau, M. I., & Rutström E. E. (2008). Eliciting risk and time

preferences. Econometrica, 76, 583-618.

Bauer M. and Chytilová J. (2009) Do children make women more patient? Experimental

evidence from Indian villages. IES Working Paper. retrieved from

http://ies.fsv.cuni.cz/default/file/download/id/10197.

Benzion, U., Rapoport, A., & Yagil, J. (1989). Discount rates inferred from decisions: an

experimental study. Management Science, 35, 270-287.

Booij, A. S., & van Praag, B. M. S. (2009). A simultaneous approach to the estimation of risk

aversion and the subjective time discount rate. Journal of Economic Behavior &

Organization, 70, 374-388.

Bütler, M., & Teppa, F. (2005). Should You take a lump-sum or annuitize? Results from

Swiss pension funds. CEPR Discussion Papers 5316, C.E.P.R. Discussion Papers.

Carver, C. S., Blaney, P. H., & Scheier, M. F. (1979). Focus of attention, chronic expectancy,

and responses to a feared stimulus. Journal of Personality and Social Psychology, 37,

1186-1195.

Carstensen L. L., Fung H. F., & Charles S. T. (2003). Socioemotional selectivity theory and

the regulation of emotion in the second half of life. Motivation and Emotion, 27, 103-

123.

Carstensen L. L., Isaacowitz D. M., & Charles S. T. (1999). Taking time seriously. A theory

of socioemotional selectivity. American Psychology, 54, 165-181.


28

Chang, E. C., Maydeu-Olivares, A., & D’Zurilla, T. J. (1997). Optimism and pessimism as

partially independent constructs: Relationship to positive and negative affectivity and

psychological well-being. Personality and Individual Differences, 23, 433-440.

Chao L. W., Szrek H., Pereira N. S., & Pauly M. V. (2009). Time preference and its

relationship with age, health, and survival probability. Judgment and Decision Making,

4, 1-19.

Chapman, G. B., & Elstein, A. S. (1995). Valuing the future: temporal discounting of health

and money. Medical Decision Making, 15, 373-386.

Christie J., & Venables, P. (1973). Mood changes in relation to age, EPI scores, time and day,

British Journal of Social and Clinical Psychology, 12, 61-72.

Cross, F. (1973). The behavior of stock prices on Fridays and Mondays, Financial Analysts

Journal, 29, 67-69.

Dember, W. N., Martin, S. H., Hummer, M. K., Howe, S. R., & Melton, R. S. (1989). The

measurement of optimism and pessimism. Current Psychological Research and

Reviews, 8, 109-119.

Farber, M. (1953). Time-Perspective and feeling-tone: A study in the perception of the days.

Journal of Psychology, 36, 253-257.

Fersterer, J., & Winter-Ebmer, R. (2003). Smoking, discount rates, and returns to education,

Economics of Education Review, 22(6), 561-566.

Forer, L. (1977). The birth order factor. New York: Pocket Books.

Frederick S. (2003) Time preference and personal identity. In G. Loewenstein, D. Read, & R.

Baumeister (Eds.) Time and decision: Psychological perspectives in intertemporal

choice. New York: Russell Sage Foundation Press.

Frederick, S., Loewenstein, G., & O’Donoghue, T. (2002). Time discounting and time

preference: A critical review. Journal of Economic Literature, 40, 351-401.


29

French, K. (1980). Stock Returns and the Weekend Effect. Journal of Financial Economics,

8, 55-69.

Gately, D. (1980). Individual discount rates and the purchase and utilization of energy-using

durables: Comment. Bell Journal of Economics, 11, 373-374.

Green, L., Fry, A. F., & Myerson, J. (1994). Discounting of delayed rewards: A lifespan

comparison. Psychological Science, 51, 33-36.

Green, L., Myerson, J., & McFadden, E. (1997). Rate of temporal discounting decreases with

amount of reward. Memory and Cognition, 25, 715-723.

Green, L., Myerson J., & Ostaszeski, P. (1999). Discounting of delayed rewards across the

life span: Age differences in individual discounting functions. Behavioral Processes,

46, 89-96.

Haack M., & Mullington J. M. (2005). Sustained sleep restriction reduces emotional and

physical well-being, PAIN, 119(1) , pp 56-64.

Harrison, G. W., Lau, M. I., & Williams, M. B. (2002). Estimating individual discount rates

in Denmark: A field experiment. The American Economic Review, 92, 1606-1617.

Hayes C.T. and Weathington B.L. 2007. Optimism, stress, life satisfaction, and job burnout

in restaurant managers. The Journal of Psychology, 141(6), 565–579.

Healey, M. and B. Ellis (2007). Birth order, conscientiousness, and openness to experience,

Evolution and Human Behavior, 28, 55-59.

Heaton J. B. (2002). Managerial optimism and corporate finance. Financial Management, 31,

33-45.

Keim, D. Stambaugh, R. (1984). A Further investigation of the weekend effect in stock

returns. Journal of Finance, 39, 819-835.


30

Kenourgios, D., & Samitas A. (2008). the day of the week effect patterns on stock market

return and volatility: evidence for the Athens stock exchange. International Research

Journal of Finance and Economics, 15, 78-89.

Lahav, E., Benzion U., & Shavit, T. (2010). Subjective time discount rates among teenagers

and adults: Evidence from Israel. Journal of Socio-Economics, 39, 458-465.

Lahav, E., Benzion U., & Shavit, T. (2011). The effect of military service on soldiers’ time

preferences – Evidence from Israel. Judgment and Decision Making, 6(2), 130-138.

Laibson, D. (1997). Golden eggs and hyperbolic discounting. The Quarterly Journal of

Economics, 112, 443-477.

Liu, W., & Aaker, J. (2007). Do you look to the future or focus on today? The impact of life

experience on intertemporal decisions. Organization Behavior and Human Decision

Processes, 102 (2), 212-225.

Loewenstein, G., Weber, R., Flory, J., Manuck, S., & Muldoon M. (2001, November)

Dimensions of time discounting. Paper presented at the Conference on Survey Research

on Household Expectations and Preferences, Ann Arbor, Michigan.

Mahajna, A., Benzion, U., Bogaire, R., & Shavit, T. (2008). Subjective discount rates among

Israeli Arabs and Israeli Jews. Journal of Socio Economics, 37(6), 2513-2522.

Myerson, J., Green, L., Hanson, J. S., Holt, D., & Estle, S. J. (2003). Discounting delayed and

probabilistic rewards: processes and traits. Journal of Economic Psychology, 24, 619-

635.

National Sleep Foundation. (2002) “Sleep in America” poll: Executive summary. Retrieved

from http://www.sleepfoundation.org/sites/default/files/2002SleepInAmericaPoll.pdf.

National Sleep Foundation. (n.d.). How Much Sleep Do We Really Need?. Retrieved from

http://www.sleepfoundation.org/article/how-sleep-works/how-much-sleep-do-we-

really-need


31

Nicholls R., Polman C. J., Andrew R. L., & Backhouse S. H. (2008). Mental toughness,

Optimism, and coping among athletes. Personality and Individual Differences, 44,

1182-1192.

Pecjak, V. (1970). Verbal synesthesiae of colors, emotions, and days of the week. Journal of

Verbal Learning and Verbal Behavior, 9, 623-626.

Peterson, C. (2000). The future of optimism. American Psychologist, 55, 44–55.

Pettengill, G. (2003). A survey of the Monday effect literature. Journal of Business and

Economics, 42, 3-27.

Prelec, D. (2004). Decreasing impatience: a criterion for non-stationary time preference and

hyperbolic discounting. Scandinavian Journal of Economics, 106, 511-532.

Read, D., & Read, N. L. (2004). Time discounting over the lifespan. Organization Behavior

and Human Decision Processes, 94, 22-32.

Rink, D,R. 2010, The impact of birth order upon consumers’ decision-making, buying, and

post-purchase processes: a conceptualization. Innovative Marketing, 6(4) 71-79.

Rosenboim, M., Shavit, T. and Shoam, A. 2010. Financial decision making in collective

society - A field test on Israeli kibbutz members and city residents. Journal of Socio

Economics, 39(1), 30-36.

Rossi, A. and Rossi, P. (1977). Body Time and Social Time: Mood Patterns by Menstrual

Cycle Phase and Day of the Week. Social Science Research, 273-308.

Ruderman, H., Levine, M. D., & McMahon, J. E. (1987). The behavior of the market for

energy efficiency in residential appliances including heating and cooling equipment.

Energy Journal, 8, 101-124.

Rystrom, D.S., & Benson E.D. (1989). Investor psychology and the day-of-the-week effect.

Financial Analysts Journal, 45, 75-78.


32

Scheier, M. F., & Carver, C. S. (1985). Optimism, coping, and health: Assessment

andimplications of generalized outcome expectancies. Health Psychology, 4, 219–247.

Scheier, M. F., & Carver, C. S. (1992). Effects of optimism on psychological and physical

well-being: Theoretical overview and empirical update. Cognitive Therapy and

Research, 16, 201–228.

Scheier, M. F., Carver, C. S., and Bridges, M. W. (1994). Distinguishing optimism from

neuroticism: A reevaluation of the life orientation test. Journal of Personality and

Social Psychology, 5, 1063-1078.

Slovic, P., Fischhoff, B., & Lichtenstein S. (1982). Why Study Risk Perception? Risk

Analysis. 2, 83-93.

Sozou, P. D., & Seymour, R. M. (2003). Augmented discounting: Interaction between aging

and time-preference behavior. Proceedings of the Royal Society of London B 270,

1047-1053.

Stevenson, M.K. (1992). The impact of temporal context and risk on the judged value of

future outcomes. Organization Behavior and Human Decision Processes, 52(3), 455-

491.

Sulloway, F. (1996). Born to rebel: Birth order, family dynamics, and creative lives. New

York, Pantheon.

Sulloway, F. (2001); Birth order, sibling competition, and human behavior. In H. Holcomb,

III (Ed.) Conceptual challenges in evolutionary psychology: innovative research

strategies. Dordrecht, Netherlands: Kluwer Academic Publishers.

Sulloway, F. J., & Zweigenhaft, R. L. (2010). Birth order and risk taking in athletics: A Meta-

Analysis and study of major league baseball. Personality and Social Psychology

Review. 14(4), 402-416


33

Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: A social psychological

perspective on mental health. Psychological Bulletin, 103, 193-210.

Thaler, R. H. (1981). Some empirical evidence on dynamic inconsistency. Economic Letters,

8, 201-207.

Thaler, R. H., & Shefrin, H. M. (1981). An economic theory of self-control. Journal of

Political Economy, 89, 392-410.

Tsukayama E., & Duckworth A. L. (2010). Domain-specific temporal discounting and

temptation. Judgment and Decision Making, 5, 72-82.

Van Praag, B. M. S., & Booij, A. S. (2003). Risk Aversion and the subjective time discount

rate: a joint approach. CESifo Working Paper Series No. 923. Tinbergen Institute

Discussion Paper No. 2003-018/3. Retrieved from http://ssrn.com/abstract=396563.

Warner, J. T., & Pleeter, S. (2001). The personal discount rate: Evidence from military

downsizing programs. American Economic Review, 91, 33-53.

Weber, E. U., & Milliman, R. A. (1997). Perceived risk attitudes: Relating risk perception to

risky choice. Management Science, 43, 123–144.

Wildavsky, A., & Dake, K. (1990). Theories of Risk Perception: Who fears what and why?

American Academy of Arts and Sciences, 119, 41-60.


34

Appendix A – Questionnaire Sample

Dear Subject,

The following experiment is part of a research discipline called “Decision-Making” that

tries to understand people’s behavior and their decision-making process on economic and

financial issues.

In order to receive a cash prize of NIS 20, you need to complete the questionnaire. The

data gathered will only be used for research purposes.

The questions in the questionnaire do not have a correct answer or an incorrect answer.

Therefore we ask that the answers represent your personal preference, for each and every

question. Please do not use a calculator, as this will cause a deviation in the research results.

The questionnaire is divided into 3 parts. Please take your time to answer each part.


35

Part 1:

In the following questions, please write your answer at the blank that appears towards

the end of each question.

1. The sum of NIS 600 is going to be deposited in your bank account today. However, we

are offering you the option of postponing this deposit and receiving another amount

in one year. What is the minimum amount you are willing to accept one year from

now in order to postpone receipt of the sum?

Instead of NIS 600 today, I am willing to accept NIS____ in one year.

2. The sum of NIS 3000 is going to be deposited in your bank account today. However,

we are offering you the option of postponing this deposit and receiving another

amount in one week. What is the minimum amount you are willing to accept one year

from now in order to postpone receipt of the sum?

Instead of NIS 3000 today, I am willing to accept NIS ____ in one week.



in two months. What is the minimum amount you are willing to accept two months


Instead of NIS 600 today, I am willing to accept NIS____ in two months.


36



amount in two months. What is the minimum amount you are willing to accept one

year from now in order to postpone receipt of the sum?

Instead of NIS 3000 today, I am willing to accept NIS____ in two months.



amount in one year. What is the minimum amount you are willing to accept one year

from now in order to postpone receipt of the sum?

Instead of NIS 3000 today, I am willing to accept NIS____ in one year.



in one week. What is the minimum amount you are willing to accept one year from


Instead of NIS 600 today, I am willing to accept NIS____ in one week.

7. Suppose you were offered a lottery ticket in a lottery with ten participants (so you

have a 1:10 chance of winning). The prize is NIS 1,000 in cash. What is the maximum

amount you are willing to pay for the ticket?

I am willing to pay no more than NIS____ in order to purchase the lottery ticket.


37

Part 2:

For the following statements, please circle to how much you agree with each statement, on a

scale from 1 (Strongly disagree) to 5 (Strongly agree)*.

1. In uncertain times, I usually expect the best.

5

Strongly agree

4321

Strongly disagree

2. I always look on the bright side of things.

5

Strongly agree

4321

Strongly disagree

3. I hardly ever expect things to go my way. **

5

Strongly agree

4321

Strongly disagree

4. I’m always optimistic about my future.

5

Strongly agree

4321

Strongly disagree

5. When I undertake something new, I expect to succeed.

5

Strongly agree

4321

Strongly disagree

6. If something can go wrong for me, it will.**

5

Strongly agree

4321

Strongly disagree

7. When there’s a will, there’s a way.


38

5

Strongly agree

4321

Strongly disagree

8. In general, things turn out all right in the end.

5

Strongly agree

4321

Strongly disagree

9. I rarely count on good things happening to me. **

5

Strongly agree

4321

Strongly disagree

10. Overall, I expect more good things to happen to me than bad.

5

Strongly agree

4321

Strongly disagree

* The questionnaire also included 6 additional statements, which are not relevant to this

study, that serve as fillers.

** These items were reverse scored before scoring and analyses.


39

General Data

Age: _____ Gender: Male / Female

Do you serve at a combat unit? A. Yes B. No.

Birth Order: A. Oldest B. Youngest C. Middle child D. Only child.

The average family income in Israel is about 13,000 NIS per month.

Your Parents’ income (together) is:

A. Much less than average. B. Less than average. C. About average. D. More than

average. E. Much more than average.

Are there times when you have an overdraft in your bank account?

A. Never. B. Sometimes. C. Always.

Are you considering signing on for the regular army? A. Yes. B. No.

Time left until discharge from compulsory service (in months)_______

Math level in matriculation exam: A. 3 Points. B. 4 Points. C. 5 Points.

In how many subjects in addition to math (if any), did you take a “5 point”

matriculation exam ? __________

Do you have plans for the upcoming weekend? A. Yes. B. No.

Preferred recreation for the weekend:

A. Club. B. Pub. C. Cinema. D. Coffee Shop. E. Staying Home.

Are you planning an post-army trip abroad? A. Yes. B. No.

If so, to where do you plan to travel?

A. Far East. B. South America. C. Europe/USA. D. Australia/New-Zealand. E. Other.

How tired are you right now?

5

Very tired

4321

Not tired at all

Judgment and Decision Making, Vol. 6, No. 2, February 2011, pp. 130–138

The effect of military service on soldiers’ time preferences —Evidence from Israel

Eyal Lahav∗ Uri Benzion∗ Tal Shavit†

Abstract

The current field study compares the time preferences of young adults of similar ages but in two very differentenvironments, one more dangerous and uncertain than the other. Soldiers, college students and a control group ofteenagers answered questionnaires about their time preferences. During mandatory service, soldiers live in a violentatmosphere where they face great uncertainty about the near future and high risk of mortality (measured by probabilityof survival). University students and teenagers live in much calmer environment and are tested for performance onlyperiodically. The soldier-subjects show relatively high subjective discount rates when compared to the other two groups.We suggest that the higher subjective discount rate among soldiers can be the result of high perceived risk in the armyas an institution, or higher mortality risk.

Keywords: time preference; discount rates; soldiers; perceived risk; military.

1 Introduction

The subjective discount rate is the rate at which individ-uals trade current value for future outcome. This ratevaries from person to person, depending on each one’swillingness to wait. In general, an individual who valuesthe present more than the future will have a higher sub-jective discount rate than a person who places more valueon the future.

In the current paper, we examine how military serviceinfluences the time preference of soldiers. We comparethe time discount of soldiers in the Israel Defense Forces(IDF) to the time discount of university and high schoolstudents. Since its establishment in 1948, the State ofIsrael has had universal conscription for all youths (maleand female) upon completing high school at the age of 18.By law, soldiers serve full-time for two (females) or three(males) years, and then continue serving in the reserves.1

The IDF plays an important part in Israeli society, and issupported by a national consensus of the Israel population(Gal, 1986).

The soldiers are a unique group because they aredrafted at young age, prior to beginning their tertiaryeducation, and earn similar, very low salaries. Their“employer” is a harsh and total institution—where all

We wish to thank the Research Fund of the School of Business Ad-ministration at The College of Management, for providing the researchfunding for this study.∗Department of Economics, Ben-Gurion University, Israel.†Corresponding author: Finance Department, School of Business

Administration, The College of Management, 7 Rabin Ave., RishonLezion. Israel. Email: [email protected].

1Those who continue in the career military earn a full salary, accord-ing to their rank and position.

parts of life of the individuals under its authority aresubordinate to and dependent upon the hierarchy of theorganization—with very clear values. The soldiers areintensively trained and tested from the first day of basictraining. Conversely, university students in Israel usuallywork and have higher earnings. They live in much calmerenvironment and are tested for performance only period-ically.

A few studies have tested soldiers’ decision makingand compare them to civilians. Warner and Pleeter (2001)studied the choices regarding early retirement plans ofolder soldiers in the US army. These soldiers were hetero-genic in age, education, and earnings, and choose to jointhe Army as a profession. In a recent article, Haerem etal. (2010), compared educated, military decision makersto business students, and found that the soldiers exhib-ited high levels of self-efficacy that correlated with risk-seeking behavior. Unlike other studies, our study is thefirst to test time preferences of young soldiers who did notjoin the army as a career choice but rather were draftedfor mandatory service.

In order to ensure similar test conditions for the sol-diers and other young adults, we distributed the question-naires to both groups when they were travelling on thetrain. We found that the soldiers have a higher subjectivediscount rate when compared to similar age groups.

The rest of the paper is organized as follows: In sec-tion 2, we discuss risk, morbidity, and mortality and theirrelation to our research. Section 3 presents the hypothe-sis. Section 4 describes the experimental design. Section5 presents the results regarding the time preferences ofsoldiers and students. Finally, Section 6 summarizes thepaper and presents its conclusions.

130

Judgment and Decision Making, Vol. 6, No. 2, February 2011 Time preference in the military 131

2 Risk, morbidity, mortality andtime preferences

Initial research in the field of time preferences focusedprimarily on monetary questions, and found that subjec-tive discount rates vary with age2, time delay, amount ofmoney and direction of trade (see Benzion, et al, 1989;Thaler, 1981; Chapman & Elstein, 1995; Laibson, 1997;Frederick et al., 2002; Myerson, et al., 2003; Prelec,2004).3

An important individual characteristic that may influ-ence time preference is risk preference, the tendency ofan individual to refrain from taking risks or to take them(Brockhaus, 1980; Rabin, 2000; Riley & Chow, 1992;Tochkov, 2009). A more risk-averse person can tolerateless uncertainty about future income, which may lead to ahigher elicited subjective discount rate and a demand forhigher compensation for delaying consumption or a pay-ment (Stevenson, 1986). In an experiment that combinedlotteries with delayed payments Anderhub et al. (2001)found a positive correlation between the degree of riskaversion and the subjective discount rate. In a more re-cent study that included both lottery questions and timepreference questions, Andersen et al. (2008) found a lowpositive correlation between risk aversion and subjectivediscount rate. Ida and Goto (2009), used lottery questionsthat included a time delay for one of the alternatives intheir research regarding smoking behaviors. They foundthat smokers in Japan are more impatient and risk-pronethan nonsmokers, pointing to a negative correlation be-tween risk aversion and subjective discount rate.

Other important effects that may influence subjectivetime discounts and related risks are morbidity and mortal-ity. Chao et al. (2009) found that both physical health andsubjective expectation of survival are related to subjectivetime discount. Their research was done in South Africawhere, due to HIV/AIDS, the middle-age mortality rateis much higher than in developed countries. The sub-jects were asked about their mental and physical health,and about their subjective probabilities of survival in thenext years. Mortality risk (measured by probabilities ofsurvival) is important because it reduces the time avail-able for reproduction and enjoinment. Similarly, mor-bidity risk (measured by health questions) reduces theability to reproduce, consume and enjoy. In addition tothe mortality and morbidity questions, the researchers

2The research on the effect of age on subjective discount is not con-sistent. Some find decreases with age and some curvilinear relationships(Thaler & Shefrin, 1981; Green et al., 1994; Green et al., 1997; Greenet al., 1999, Harrison et al., 2002; Sozou and Seymour, 2003; Read andRead, 2004).

3There are also many articles regarding non-monetary questions,such as choices for various durable items (Gately, 1980; Ruderman etal., 1987), and even for beers and bags of chips (Tsukayama & Duck-worth, 2010).

elicited the subjective discount rate, using a time pref-erence questioners regarding monetary questions. Aftercontrolling for mortality and morbidity, they found thatage is not a significant predictor of time preference. (Theresearchers note, however, that their data may not havecontained enough older people to show the age effect thatothers found.) They also found a U-shaped relationshipbetween physical health and the subjective discount rate(very healthy or very sick people have higher discountrates than people with average health). A similar relation-ship was found between survival probabilities and subjec-tive discount rate (people with a low or high probabilityof survival have a higher discount rate than people withan average probability of survival). The researchers con-clude that “because causes of morbidity and mortality inSouth Africa are not necessarily related to aging, age isno longer a strong predictor of health and expected sur-vival and, hence, of subjective discount rates” (Chao etal., p. 14).

Carstensen et al. (1999, 2003) also separated age frommortality in their research and development of the so-cioemotional selectivity theory. According to this theorywhen time is perceived as open-ended, people are future-oriented and concentrate on knowledge-related goals.However, when time is perceived as limited, people be-come present-oriented, and start concentrating on emo-tional goals. The researchers found that young peoplewith symptomatic HIV concentrated on emotional goals.

3 The hypothesisDuring mandatory service, soldiers leave in a violent at-mosphere, and face great uncertainty about the near fu-ture and greater mortality and morbidity risk. Even ontheir way home and back to base, terrorist organizationsconsider them preferred targets. Beside the mortality andmorbidity risk, soldiers face uncertainty about daily ac-tivities: their commanders control their schedule, and canchange it instantly if the soldier, or even other soldiers,misbehaves in any way. For example, it is quite commonfor soldiers to have weekend leave canceled for misbe-havior or for not achieving a training goal. In some cases,they can be grounded on base for weeks as a punishment.4

The military is regarded as a total institution (Wintre andBen-Knaz, 2000), unlike a high school or university en-vironment.

We suggest that such uncertainty about daily activitiesand violent atmosphere of military service can lead tohigher perceived risk5 in the future (Slovic et al., 1982;

4In Israel most forces are posted within a few hours drive of theirhomes and soldiers usually receive a few days leave at least once everythree weeks.

5Perceived risk is the subjective judgment of risk that people at-tribute to a situation. According to the sociological approach, percep-


Weber & Milliman, 1997), and therefore produces a highpreference to the present. This is also consistent withCarstensen et al. (1999, 2003) whose socioemotional se-lectivity theory suggests that, when time is perceived aslimited, people become present-oriented. Note that ourexplanation and hypothesis are speculative since we donot measure directly the degree of totality of the army in-stitution or uncertainty inherent in military service. In or-der to control for income and age more thoroughly whentesting the hypotheses, we also use earlier data on highschool students. This date were collected and used in La-hav et al. (2010). Soldiers are older than the high schoolstudents, but younger than the university students.6 Notehowever, that all of our ages are very close, and age ef-fects on discounting are relatively small within the rangeincluded in this study.

As a result of the uncertainty inherent in military ser-vice we expect to find that the subjective discount rateof the soldiers will be higher than that of students andteenagers, due to higher perceived risk during mandatoryarmy service.

4 Experimental design

4.1 SubjectsThe subjects in the group of soldiers consisted of 90 Is-raeli soldiers during their mandatory service period (av-erage age 19.93, age range: 18–23, 60 males and 30 fe-males).7 The young adult subjects were 69 students (av-erage age 23.57, ranging in age from 19 to 25 years, 34males and 35 females).8 The questionnaires were dis-tributed on trains traveling between three large cities inIsrael (Haifa, Tel-Aviv and Beer-Sheva) where there aremajor universities. Furthermore, these train lines are alsoused extensively by soldiers, who are exempt from payingfares on public transportation. The subjects were givenapproximately 40 minutes to fill the questionnaire, andwere paid 20 New Israeli Shekels (NIS, about $5 U.S.)for their time. They were asked not to speak with eachother while completing the questionnaires.

We chose to distribute the questionnaires on trains tofacilitate access to soldiers, since the IDF usually does

tions are socially constructed by institutions, cultural values, and waysof life (Wildavsky & Dake, 1990).

6Because of the mandatory service, young adults in Israel usuallystart university only at age 21 or older, and most work while studying.

7In exceptional circumstances, such as being a new immigrant, aperson can start his mandatory service a few years after the age of 18.This is the case for the oldest solders in our sample.

8Since the mandatory draft applies to all 18 year-old Israeli, the fewthat are not drafted are usually considered problematic. Other smallgroups that are not drafted are ultra-orthodox Jews and Arabs. Theseyoung adults are not representative of the general population in Israeland using them as a comparison group would cause a selection bias.Therefore, the soldiers and comparison group are not identical in age.

not allow civilian researchers an access to soldiers onbases. Furthermore, even if we had been able to enterone or two military bases to conduct the current survey, itwould create selection bias, since soldiers in each service(Air Force, Navy, or ground forces) have certain sharedtraits. However, the train is a comfortable means of trans-portation used by young soldiers from all services, as wellas many students. This helped us to avoid selection bias,and created the same testing environment for both sol-diers and students.

The control group included 58 teenagers (average age16.6, age range: 15–18 years, 36 males and 22 females),who we studied in an earlier paper (Lahav et al., 2010).The questionnaires for this group did not include all ofthe questions that were presented to the young adults (seesection 5.1.2).

4.2 Procedure4.2.1 Time preferences

The questionnaires began by informing the subjects thatthey about to complete questionnaires in the field of be-havioral economics. They were further instructed that thequestions did not have “correct” answers, so each answershould reflect their personal preferences.

In the main scenario (postponing receipt of income),subjects were informed they had earned x for working ina large fashion chain store, and they could receive thatamount today. The chain’s management offered them theoption of postponing receipt of x to a later time (t). Thesubjects were instructed to write down the amount theywould ask to receive at time t rather than receiving x to-day. The time periods (t) were: one week, two monthsand one year. They were also asked to assume that thechain is reliable, and there is no doubt they would be paidon their chosen future date if the management acceptedtheir offer. (See the Appendix for the translated instruc-tions.) The idea of the scenario was based on a paper byBenzion et al. (1989) that asked about a financial firm.The statement about the chain store’s behavior was madeto clarify the scenario.

A sample of the postponed receipt question:

The sum of NIS 600 is going to be de-posited in your bank account today. However,we are offering you the option of postponingthis deposit and receiving another amount inone year. What is the minimum amount you arewilling to accept one year from now in order topostpone receipt of the sum?

Instead of NIS 600 today, I am willing toaccept NIS _____ in one year.

The questions referred to two different amounts: NIS600 and NIS 3,000, so the scenarios had six different


questions: three different time periods and two differentamounts. All subjects answered all six scenarios.

Using the same amounts for both groups made the re-sults comparable. However, since students have a higherincome than soldiers, there might have been an income-effect problem, meaning that students might consider NIS600 a relatively low amount, while it is a relatively largeamount for soldiers. For many soldiers, this amount ismore than their monthly salary.9 In order to avoid thiseffect, we also compared the high amount in the studentsgroup (NIS 3,000) to the low amount in the soldiers group(NIS 600).

4.2.2 Risk preferences

In order to control for differences in risk preferencesbetween the two groups the questionnaire included riskpreference questions. The first question was a lotteryquestion similar to a question used by Booij & van Praag(2009):

Suppose you were offered a lottery ticketin a lottery with ten participants (so you havea 1:10 chance of winning). The prize is NIS1,000 in cash. What is the maximum amountyou are willing to pay for the ticket?

I am willing to pay no more than NIS ____in order to purchase the lottery ticket.

In addition, we included 6 questions regarding riskaversion, similar to the ones used in Singer et al. (2005).The subjects were asked to circle their answer on a scalefrom 1 (Strongly disagree) to 6 (Strongly agree) for thefollowing statements:

• I enjoy taking risks.

• I avoid situations of uncertainty.

• I am not troubled by taking risks if my actions mightyield substantial gains.

• I consider the possibility of not taking risks as amain factor in my life.

• People say that I enjoy taking risks.

• I will take risks only very occasionally, if at all, ifthere is another alternative.

9Israeli soldiers earn between NIS 350 to NIS 700 per month duringthe mandatory service period.

5 Results

5.1 Time discount5.1.1 Comparing time discount of soldiers and stu-

dents

Table 1 shows the annual (yearly) subjective discount ratefor each period and amount10. The annual discount ratefor delaying payment was calculated as follows:

r =(

P

X− 1

)· 12

t, (1)

where P is the amount the subject is willing to accept int months for delaying the receiving of the amount of Xtoday.

We used a one-tailed t-test to examine the null hypoth-esis that the subjective discounting is equal for soldiersand students, and Cohen’s d to measure the strength ofthe difference. We compared not only the subjective dis-count rate for the same amounts (NIS 3,000 and NIS 600)but also students’ responses for the higher amount to sol-diers’ responses for the lower amount, as a way of neu-tralizing the income effect.

The subjective discount rates are very high for both sol-diers and students, with high variance, suggesting widevariations between the subjects. This finding is consistentwith earlier studies about the subjective discount rate, es-pecially for short periods (e.g., Thaler, 1981; Frederick etal., 2002). In all cases, we found that the average sub-jective discount rate was substantially and significantlyhigher for soldiers than for students, which is consistentwith our hypothesis. This was also true when we avoidedan income or size effect by comparing the discount ratesfor a relatively low amount for soldiers (NIS 600) and arelatively high amount for students (NIS 3,000).

In addition, we found that the discount rate was sig-nificantly lower for the higher amount (NIS 3,000) thanfor the lower amount (NIS 600), both for soldiers (p <0.001), and students (p < 0.001) consistent with the re-sults of Thaler (1981), which pointed to lower subjectivediscount rates for higher amounts.

Our sample included an equal number of males andfemales in the students’ group, but an unequal numberin the soldiers’ group. To test for the possibility of agender effect, we ran multivariate regression analysis ofsubjective discount rate on amount variable (NIS 600 orNIS 3000), time variable (1 week, 2 months or 1 year),dummy variables for group (Soldier = 1; Students = 0)and for gender (male = 0, female = 1). The results arepresented in Table 2.

We found no significant effect for gender, while sol-diers show higher subjective discount rate consistent with

10The discounting behavior was better captured by the hyperbolicmodel


Table 1: Average (STDV) annual subjective discount rates (comparing soldiers to students). Standard deviations arein parentheses.

Group Amount 1 week 2 months 12 months

Students NIS 600 667.78% 259.01% 117.04%(n = 69) (617.20%) (201.55%) (89.33%)

NIS 3,000 393.20% 140.35% 78.23%(397.35%) (114.14%) (49.86%)

Soldiers NIS 600 1390.31% 472.82% 183.00%(n = 90) (2097.67%) (679.96%) (189.35%)

NIS 3,000 637.92% 237.44% 102.73%(805.10%) (208.74%) (67.79%)

T-test t(157) = 2.77 t(157) = 2.53 t(157) = 2.67Soldiers NIS 600 vs. p = .003 p = .006 p = .004Students NIS 600 d = 0.467 d = 0.426 d = 0.446

T-test t(157) = 2.32 t(157) = 3.48 t(157) = 2.52Soldiers NIS 3,000 vs. p = .011 p = .000 p = .006Students NIS 3,000 d = 0.385 d = 0.577 d = 0.412

T-test t(157) = 3.89 t(157) = 4.02 t(157) = 4.48Soldiers NIS 600 vs. p = .000 p = .000 p = .000Students NIS 3,000 d = 0.660 d = 0.682 d = 0.757

Table 2: Regression analysis with gender. The dependentvariable is subjective discount rate. Significance levelsare in parentheses.

Independent Variable Coefficient (significance)

(Constant) 6.54 (0.00)Amount −0.001 (0.00)Time −5.25 (0.00)Gender 0.55 (0.30)Soldier 2.38 (0.00)R- Square 0.117 (0.00)

the results in Table 1. The amount and time have nega-tive effect on the annual subjective discount rates consis-tent with Thaler (1981) and Benzion et al (1989). We alsocompare female soldiers to female students, and male sol-diers to male soldiers. Using an ANOVA with repeated-measure, we find significant positive difference in the lev-els of annual subjective discount rate between soldiersand students for females (F(1,29) = 6.37, p = .02) andmales (F(1,33) = 9.53, p = <0.01) separately.

We found no difference between combat soldiers (28

subjects) and non-combat soldiers (62 subjects), based ona repeated-measures ANOVA (F(1,27) = 1.89, p = .18).Combat soldiers are soldiers which train for battle, andengage in combat when necessary. Non-combat soldiersengage in assisting the combat soldiers during the fight.These soldiers are involved with tasks involving ammuni-tion, information gathering, arms repairs, etc. All newly-drafted soldiers are trained for combat fighting in thefirst few months of their army service, and combat sol-diers have longer combat training depending on the corpswhere they will serve. Because non-combat soldiers as-sist the combatants, they frequently meet each other. Onweekend soldiers usually meet with their old friends fromhigh school, again giving combat and non-combat sol-diers an opportunity to socialize with each other and pos-sibly communicate attitudes. Although this might be apossible explanation for the lack of difference in the timepreferences of the two groups, another potential expla-nation relates to the uncertainty about a soldier’s dailyactivities in the army. Combat and non-combat soldiershave the same uncertainty about where they will be inthe near future (even the next weekend, as we mentionedearlier). This uncertainty may be the main influence onthe subjective discount rate, while the effect of uncertainmortality is marginal.


Table 3: Subjective discount rates—postponing receipt of income (comparing soldiers to teenagers). Standard devia-tions are in parentheses.

Group Amount 1 week 2 months 12 months

Teenagers NIS 600 774.77% 219.74% 171.81%(n = 58) (724.25%) (204.70%) (186.59%)

Soldiers NIS 600 1390.31% 472.82% 183.00%(n = 90) (2097.67%) (679.96%) (189.35%)

T-test t(146) = 2.15 t(146) = 2.75 t(146) = 0.35Soldiers NIS 600 vs. p = .017 p = .003 p = .362Teenagers NIS 600 d = 0.392 d = 0.504 d = 0.060

5.1.2 Control for age and earnings

This paper focuses on the time preferences of youngadults in two very different institutional frameworks (mil-itary and university). The average ages of these twogroups are close but not identical (Soldiers: 19.93 years;Students: 23.57 years). We used a control group of 58teenagers (average age 16.6) from our earlier paper (La-hav et al., 2010). If there is an age effect even for a dif-ference of three years, we expect the subjective discountrates of the teenage subjects to be higher than that of thesoldiers.

Many teenagers do not work, while teenagers who dowork mostly do so on weekends or vacations, and sotheir own earnings often do not exceed those of manda-tory service soldiers. Earnings were previously foundto be negatively correlated with subjective discount rates(Warner & Pleeter, 2001).11 Therefore, the control groupof teenagers can be used as a control for age and earn-ings.12 The teenagers were not asked about NIS 3,000,and so we compared the questions regarding delaying anincome of NIS 600.13 Table 3 presents the annual subjec-tive discount rate of the income postponement scenariofor each group. We use a one-tailed t-test to examine thenull hypothesis that the subjective discount rate is equalfor soldiers and teenagers.

Table 3 shows that the subjective discount rates are also

11A survey of the Ministry of Industry, Trade and Labor in Israelfound that 5% of 15 to 17 years old teenagers worked while in highschool, and about 1.3% worked and did not attend school, in 2004. In asurvey from 2007, the ministry found that about 75% of 14 to 15 yearolds only work during vacations. The surveys (in Hebrew) can be foundat http://www.moital.gov.il/NR/exeres/1C46576A-890C-4096-BECF-7563F6FB0577.htm

12We do not have an estimator for the soldier or teenager parent’searnings. This could cause an estimation bias because parents financemost of a teenager’s expenses, and in many cases, a part of a soldier’sexpenses as well. For students, this is not always the case, because moststudents in Israel work.

13It is not realistic for a teenager to receive salary of NIS 3,000.

very high for the teenage subjects, with a high variance,suggesting wide variations between the subjects. Theteenagers’ average subjective discount rate was signifi-cantly lower for both one week delay and two month de-lays. These results contradict the predictions of an ageeffect or income effect. For a one year delay, there wasno difference.

The results suggest that the differences we found inthe current questionnaire regarding the time preferencesof soldiers and students were, for the most part, not dueto an age or income effect, but rather the result of othercharacteristics, such as institutional differences.

5.2 Control for risk preferenceIn order to compare the risk aversion of the two groups,we first created a risk aversion index variable for thesix non-numeric questions. The Cronbach’s α of the sixquestions was 0.734. The index was calculated as an av-erage of the answers to the six risk aversion questions,calculated separately for each subject. The index valuesranged from 1 (low risk aversion) to 6 (high risk aver-sion).

The second risk aversion measure is the willingnessto pay (WTP) for a lottery similar to a question used byBooij & van Praag (2009). For this question, willingnessto pay a lower amount to participate in the lottery indi-cates a higher risk aversion.

We used a two-tailed t-test to examine the null hypoth-esis that the index variable and the WTP are equal forsoldiers and students. The average index value is signifi-cantly lower (t(155) = 2.451 , p = 0.015) for the soldiers(mean = 3.6, STDV = 0.70), than for the students (mean= 3.93, STDV = 0.93). The average WTP is significantlyhigher (t(157) = 1.978 , p = 0.05) for soldiers (Mean =NIS 83.79, STDV = 144.4) than for students (Mean = NIS48.76, STDV = 30.87). Both measures indicate that thesoldiers are less risk averse than the students subjects.


Table 4: Regression analysis with lottery for risk-aversion measure. The dependent variable is subjective discountrate. Significance levels are in parentheses.

Independent variable All subjects Students only Soldiers only

Constant 6.32 (0.00) 5.42 (0.19) 9.47 (0.00)Amount −0.001 (0.00) −0.001 (0.00) −0.001 (0.00)Time −5.25 (0.00) −3.33 (0.00) −6.73 (0.00)WTP for lottery (Risk aversion) 0.01 (0.00) −0.006 (0.31) 0.01 (0.00)Soldier 1.93 (0.00)R- Square 0.134 (0.00) 0.186 (0.00) 0.125 (0.00)

The finding of lower risk aversion for soldiers is con-sistent with Haeren et al. (2010). They explain that thesoldiers’ lower risk aversion is due to the military setting,with its specific type of organizational culture. They sug-gest that “operating in such culture can create differentbehaviors than those found in civil organizations.” Ourfindings indicate that the risky and uncertain environmentof the army and the institutional nature in the army are asource for soldiers’ lower risk aversion and higher timediscount.

In order to test whether differences in risk aversioncould mediate the differences we found in discount rates,we used simple regression of risk aversion measures (in-dex and lottery) on dummy variable for soldiers (soldier =1; student = 0). We found a significant negative effect ofthe dummy variable for the index (p = 0.015, R-square =0.037) and significant positive effect of the dummy vari-able for the lottery’s WTP (p = 0.050, R-square = 0.024).Next we conducted a multiple regression analysis of sub-jective discount rate on risk measure (the index or thelottery), amount variable (NIS 600 or NIS 3000), timevariable (1 week, 2 months or 1 year = 0) and a dummyvariable for group (Soldier = 1; Students = 0). We con-ducted a different analysis for each of the two estimationsof risk aversion (the lottery question and the risk aversionindex). When using the index as measure of risk aversion,we found that the risk aversion coefficient is not signifi-cant (p = 0.88), meaning that differences in risk aversionbased on the index could not mediate the differences wefind in discount rates.

When using the lottery for risk aversion measure, wefind that the risk aversion coefficient is significant. Table4 presents the regression analysis when lottery is usedfor risk aversion measure for all the subjects and for stu-dents and soldiers separately. We thus used two methodsfor eliciting risk aversion and found conflicting results.A possible explanation is that the lottery question wasmore suitable than the non-monetary questions for elicit-ing risk aversion, when considering time preferences formonetary issues.

Table 4 shows that subjective discount rate increaseswith the increase in WTP for the lottery, meaning thata decrease in risk aversion (increase in the willingnessto pay) results in an increased subjective time discount.This means that risk preference based on the lottery couldmediate the differences we found in discount rates. Whenwe separate the groups, we found that risk aversion basedon the lottery correlated with time discount rate only forsoldiers. It is possible that soldiers who are more stronglyinfluenced by the risky and uncertain environment of thearmy show lower risk aversion and higher time discountrate than soldiers who are less influenced by the risky anduncertain environment of the army. Our results are alsoin line with Ida and Goto (2009) who found that one highrisk group (heavy smokers) had the highest time discount(most impatient) and the lowest risk aversion, while onthe other hand a lower-risk group (ex-smokers) had thelowest time discount (least impatient) and the highest riskaversion.

6 Conclusion

In the current study we compare the time preferences ofsoldiers and students in Israel, in order to estimate theeffect of the uncertainty and mortality risk due to mili-tary service and the military environment. The subjectivediscount rates were significantly higher for the soldiers’group.

A possible explanation for the high discount rates ofsoldiers is a higher perceived risk (Slovic et al., 1982;Weber & Milliman, 1997), during mandatory service.Soldiers live in a violent atmosphere, face as risk of mor-tality and great uncertainty about the near future. Theircommanders control their schedule, and can instantlychange it if the soldier or, even other soldiers, misbehavein any way. We suggest that such uncertainty regardingeven the near future produces high present preference,and as a result they present higher subjective discountrate. As Dar and Kimhi (2001, p. 3) explain: “Israeli


youth must direct most of their personal resources towardadjusting to a rigid and demanding service and must com-ply with commands and assume roles that they are notinvolved in shaping. They must live for an extended pe-riod in a total institution, which ostensibly provides for alltheir needs but limits their privacy and freedom of choiceand threatens their individuality.”

The results are even more interesting when taking intoaccount that the risk aversion of the soldiers group wassignificantly lower than that of the students group consis-tent with Haerem et al. (2010). We suggest that the riskyand uncertain environment of the army and the institu-tional differences are sources of a different risk attitude.

In the current study, we claim that institutional risksand uncertainty may change the way people discount thefuture. This claim reinforces the findings of Chao etal. (2009) who found that the uncertainty about the fu-ture caused by the HIV/AIDS epidemic in South Africaeffects people’s time preferences. Therefore, future re-searchers who examine decision-maker uncertainty mightalso take into account the perceived risk or mortality riskfor various situations in life, such as the differences be-tween institutional contexts or work places. Collectingdata in different situations or for situations where the sub-jects are face a mortality risk is not easy and almost im-possible in the lab. Only field studies, like the ones con-ducted by Chao et al. (2009) or the present authors cangive us insight regarding the effect of mortality risk orreal future uncertainty on judgment and decision making.More field studies are needed to strengthen the relation-ship between real uncertainty regarding the future or mor-tality risk and decision makers’ judgment and behavior.

References

Anderhub, V., Güth, W., Gneezy, U., & Sonsino, D.(2001). On the interaction of risk and time preferences:An experimental study. German Economic Review, 2,239–253.

Andersen S., Harrison, G. W., Lau, M. I., & Rutström E.E. (2008). Eliciting risk and time preferences. Econo-metrica, 76, 583–618.

Benzion, U., Rapoport, A., & Yagil, J. (1989). Discountrates inferred from decisions: an experimental study.Management Science, 35, 270–287.

Booij, A. S., & van Praag, B. M. S. (2009). A simulta-neous approach to the estimation of risk aversion andthe subjective time discount rate. Journal of EconomicBehavior & Organization, 70, 374–388.

Brockhaus, R. H., Sr. (1980). Risk taking propensity ofentrepreneurs. The Academy of Management Journal,23, 509–520.

Carstensen L. L., Fung H. F., & Charles S. T. (2003). So-cioemotional selectivity theory and the regulation ofemotion in the second half of life. Motivation andEmotion, 27, 103–123.

Carstensen L. L., Isaacowitz D. M., & Charles S. T.(1999). Taking time seriously. A theory of socioe-motional selectivity. American Psychologist, 54, 165–181.

Chao L. W., Szrek H., Pereira N. S., & Pauly M. V.(2009). Time preference and its relationship with age,health, and survival probability. Judgment and Deci-sion Making, 4, 1–19.

Chapman, G. B., & Elstein, A. S. (1995). Valuing the fu-ture: temporal discounting of health and money. Med-ical Decision Making, 15, 373–386.

Dar Y., & Kimhi, S. (2001). Military service and self-perceived maturation among Israeli youth. Journal ofYouth and Adolescence, 30. 427–448.

Frederick, S., Loewenstein, G., & O’Donoghue, T.(2002). Time discounting and time preference: A crit-ical review. Journal of Economic Literature, 40, 351–401.

Gal, R. (1986). A portrait of the Israeli soldier. Westport:Greenwood Publishing Group.

Gately, D. (1980). Individual discount rates and the pur-chase and utilization of energy-using durables: Com-ment. Bell Journal of Economics, 11, 373–374.

Green, L., Fry, A. F., & Myerson, J. (1994). Discountingof delayed rewards: A lifespan comparison. Psycho-logical Science, 51, 33–36.

Green, L., Myerson, J., & McFadden, E. (1997). Rateof temporal discounting decreases with amount of re-ward. Memory and Cognition, 25, 715–723.

Green, L., Myerson J., & Ostaszeski, P. (1999). Discount-ing of delayed rewards across the life span: Age differ-ences in individual discounting functions. BehavioralProcesses, 46, 89–96.

Haerem T., Kuvaas, B., Bakken, B. T. & Karlsen, T.(2010). Do Military decision makers behave as pre-dicted by prospect theory? Journal of Behavioral De-cision Making, in press.

Harrison, G. W., Lau, M. I., & Williams, M. B. (2002).Estimating individual discount rates in Denmark: Afield experiment. The American Economic Review, 92,1606–1617.

Ida T., & Goto R. (2009). simultaneous measurementof time and risk preferences: stated preference dis-crete choice modeling analysis depending on smokingbehavior. International Economic Review, 50, 1169–1182.

Lahav, E., Benzion U., & Shavit, T. (2010). Subjectivetime discount rates among teenagers and adults: Ev-idence from Israel. Journal of Socio-Economics, 39,458–465.


Laibson, D. (1997). Golden eggs and hyperbolic dis-counting. The Quarterly Journal of Economics, 112,443–477.

Myerson, J., Green, L., Hanson, J. S., Holt, D., & Es-tle, S. J. (2003). Discounting delayed and probabilis-tic rewards: processes and traits. Journal of EconomicPsychology, 24, 619–635.

Prelec, D. (2004). Decreasing impatience: a criterionfor non-stationary time preference and hyperbolic dis-counting. Scandinavian Journal of Economics, 106,511–532.

Rabin, M. (2000). Risk aversion and expected-utility the-ory: A calibration theorem. Econometrica, 68, 1281–1292.

Read, D., & Read, N. L. (2004). Time discounting overthe lifespan. Organization Behavior and Human Deci-sion Processes, 94, 22–32.

Riley W. B. Jr., & Chow K. V. (1992). Asset allocationand individual risk aversion. Financial Analysts Jour-nal, 48, 32–37.

Ruderman, H., Levine, M. D., & McMahon, J. E. (1987).The behavior of the market for energy efficiency inresidential appliances including heating and coolingequipment. Energy Journal, 8, 101–124.

Singer, L. G., Gould, M. K., Tomlinson, G., & Theodore,J. (2005). Determinants of health utility in lung andheart-lung transplant recipients. American Journal ofTransplantation, 5, 103–109.

Slovic P., Fischhoff B., & Lichtenstein S. (1982). WhyStudy Risk Perception? Risk Analysis. 2, 83–93.

Sozou, P. D., & Seymour, R. M. (2003). Augmenteddiscounting: Interaction between aging and time-preference behavior. Proceedings of the Royal Societyof London B 270, 1047–1053.

Stevenson, M. K. (1986). A discounting model for de-cisions with delayed positive or negative outcomes.Journal of Experimental Psychology: General, 115,131–154.

Thaler, R. H. (1981). Some empirical evidence on dy-namic inconsistency. Economic Letters, 8, 201–207.

Thaler, R. H., & Shefrin, H. M. (1981). An economictheory of self-control. Journal of Political Economy,89, 392–410.

Tochkov, K. (2009). The effects of anticipated regret onrisk preferences of social and problem gamblers. Judg-ment and Decision Making, 4, 227–234.

Tsukayama E., & Duckworth A. L. (2010). Domain-specific temporal discounting and temptation. Judg-ment and Decision Making, 5, 72–82.

Warner, J. T., & Pleeter, S. (2001). The personal dis-count rate: Evidence from military downsizing pro-grams. American Economic Review, 91, 33–53.

Weber, E. U., & Milliman, R. A. (1997). Perceivedrisk attitudes: Relating risk perception to risky choice.

Management Science, 43, 123–144.Wildavsky, A. & Dake, K. (1990). Theories of Risk Per-

ception: Who fears what and why? American Academyof Arts and Sciences, 119, 41–60.

Wintre M. G., & Ben-Knaz R. (2000). It’s not academic,you’re in the army now: Adjustment to the army as acomparative context for adjustment to university. Jour-nal of Adolescent Research, 15, 145–172.

Appendix — Questionnaire instruc-tionsDear Subject,

The following experiment is part of a research disci-pline called Behavioral Economics that tries to under-stand people’s behavior and their decision-making pro-cess on economic and financial issues.

The goal of the experiment is to gather data on the sub-jects’ preferences for receiving an amount of money at thepresent time, as opposed to receiving a different amountof money at a later time.

In order to receive a cash prize of NIS 20 you shouldcomplete the questionnaire. The data gathered will onlybe used for research purposes.

The questions in the questionnaire do not have a cor-rect or an incorrect answer, therefore we ask that the an-swers represent your personal preference, for each andevery question. Please do not use a calculator, as this willcause a deviation in the research results.

The questionnaire is divided into 3 parts. Please takeyour time to answer each part.

Instructions for the income postponement scenario:Assume that you have worked on weekends in a large

fashion chain store, and you have earned a sum of moneyfor your work, which you are to receive today. Thechain’s management offers all the employees who aresupposed to receive wages today, an opportunity to post-pone receiving their wages until a later point in time. Inreturn, they will receive a larger amount of money (likeinterest on savings). Assume that the chain is reliable andthere is no doubt you will be paid on their chosen futuredate, if the management accepts your offer.

Subjective time discount rates among teenagers and adults: Evidence from Israel

Documents