researc

1. Institute for Empirical Research in Economics
University of Zurich
Working Paper Series
ISSN 1424-0459
Published in: Journal of Economic Behavior and Organization 54. 65-88. (2004)
Title: Pro-Social Behavior in a Natural Setting.
Working Paper No. 107
Pro-Social Behavior, Reciprocity or Both?
Bruno S. Frey and Stephan Meier
February 2002

2. 1
Pro-Social Behavior, Reciprocity or Both?
"RUN& S. FRE,
S-EP/0N MEIER!
-his draft; 21 January 2002
Abstract
Empirical evidence is provided for the importance of non-reciprocal pro-social Jehavior of
individuals in an anonymousK n-person pure puJlic good setting. 0 uniMue panel data set of
1NOK000 oJservations is matched with an eQtensive survey.
Even under anonymous conditionsK a large numJer of individuals are prepared to donate a not
insignificant sum of money. Rooperation conditional on giving Jy specific other persons
SreciprocityT is present Jut the causal relationship is amJiguous. It is crucially importantK
whetherK and in what wayK one is asked to donate. Identification with the organiVation is also
important.
Ke#words; PuJlic WoodsK Pro-social JehaviorK ReciprocityK Xonation
)EL,classification; /41K XO4K Z1N
I. Beyond Self-Interest
0ssuming that the Jehavior of human Jeings is driven Jy self-interest has Jeen a powerful
approach when studying the economy. -he self-interest hypothesis predicts individuals
choices pretty well in most competitive markets. -his also holds for most types of Jehavior
outside the traditional economic markets. [Economic imperialism orK more generallyK
rational choice analysisK which Juilds on the self-interest assumptionK has had great success in
! "runo S. Frey is Professor of EconomicsK and Stephan Meier is Research 0ssistantK at the Institute for
Empirical Economic Research of the University of Zurich. "luemlisalpstrasse 10K R/- ]00O ZürichK
SwitVerland_ Phone; ``41-1-ON4 Na 2]K FaQ; ``41-1-ON4 49 0aK e-mail; Jsfreyciew.uniVh.chK
smeierciew.uniVh.ch.

3. 2
many areas outside of economicsK eQamples Jeing politicsK historyK lawK the artsK or the family
Se.g. "ecker 19aOK 199OK Stigler 19]4K Frey 1999K daVear 2000T.
"ut not all applications of the self-interest hypothesis meet the necessary conditions under
which it works well_ many markets are characteriVed Jy rigiditiesK incomplete contractsK or Jy
a small numJer of traders. Warnings have Jeen issued that the application of the calculus of
self-interest may face decreasing marginal returns Se.g. /irshleifer 19]5K Frey 2001T. 0 recent
Jook on giving and altruismK appearing under the auspices of the 4nternational,Econo5ic
6ssociation, even understands itself as an [oJituary of /omo oeconomicus Sgolm 2000a;
N2T. Studies of important activitiesK such as charitaJle giving Se.g. 0ndreoni 1990K 199]K
WeisJrod 199]TK voting Se.g. Mueller 19]9K 199aTK and taQ paying Se.g. Slemrod 1992K
0ndreoniK Erard and Feinstein 199]T have convincingly argued that such actions cannot Je
eQplained Jy relying on the strict self-interest aQiom. -husK for eQampleK it has Jeen stated
that [h0i purely economic analysis of the evasion gamJle implies that most individuals would
evade if they are [rationalK Jecause it is unlikely that cheaters will Je caught and penalised
S0lmK McRlelland and SchulVe 1992;22_ similarly WraetV and Wilde 19]5;N5]K Skinner and
Slemrod 19]5T. "ut most people actually pay their taQ dues. -aQ payment can therefore Je
considered a [Muasi-voluntary act Sdevi 19]]T. -he self-interest model has Jeen clearly
rejected in a great numJer of laJoratory eQperiments Ssee Fehr and Schmidt 1999K Fehr and
Wkchter 2000T. Most importantlyK eQperiments of the Ultimatum Wame in 15 societies
eQhiJiting a wide variety of economic and cultural conditions reveal that [the canonical
model of the self-interested material pay-off maQimiVing actor is systematically violated
S/enrich et al. 2001; aaT.
0s a result of these findingsK the reciprocity model has recently gained much attention.
Reciprocity occurs when persons act in a more cooperative manner as a response to the
friendly Jehavior of others and act in a hostile way when treated in an unfriendly way Jy
others. It has Jeen claimed that [ Practically all life in society includes and implies
reciprocitiesK and reciprocity has Jeen seen as the Jasic glue that makes people constitute
groups or societies Sgolm 2000J; 115T. RecentlyK a large numJer of laJoratory eQperiments
have Jeen devoted to the study of reciprocity in economics Ssee the surveys in e.g. Fehr and
Wkchter 2000K Fehr and Schmidt 2000K Wkchter and Falk 2001K Falk and FischJacher 2001T. It
supplements evidence that has Jeen provided for decades Jy anthropology Se.g. Mauss 1924K
Sahlins 19a0K Wodelier 2000TK Jy sociology and social psychology Se.g. Wouldner 19O0K
0dam 19ONK 19O5T as well as Jy economics itself Searly contriJutions are "oulding 19aNK
golm 19aNK 19]4K Rollard 19a]T. "ut the emphasis on reciprocity comes at a price. Forms of

4. N
pro-social preferences not implying reciprocity are often neglected in theoretical as well as
empirical analyses. -his paper argues that there is very important pro9social,be;avior,t;at
goas,be#ond,reciprocit#. People are prepared to act in a non-selfishK altruistic and non-strategic
way and l aJove all l such Jehavior is contingent on environmental and institutional
conditions.
-his paper provides empirical evidence for the importance of non-reciprocal pro-social
Jehavior of individuals in an anonymousK n-person pure puJlic good setting. We use a uniMue
panel data set of 1NOK000 oJservations Sroughly NNK000 personsT concerning the decisions of
students at the University of Zurich on whether or not to contriJute to two Social Funds
administered Jy the University. -hese field oJservations are matched with an eQtensive
survey of the same group of people to find out more aJout the conditions and motives for
giving.
We are aJle to show that even under these anonymous conditionsK a large numJer of
individuals are prepared to donate a not insignificant sum of money. -his finding is not
trivial_ it contradicts for instance the statement that [Positive and,stable,contriJutions to the
puJlic good are very unlikely ... free riding will Je pervasive under conditions of anonymous
interactions SFischJacherK Wkchter and Fehr 2000; 9T. We identify evidence of indirect
reciprocity; when students eQpect that others also contriJuteK they themselves tend to donate
moreK Jut of course the direction of causality is not at all clearK i.e. onems own willingness to
donate may lead one to eQpect that others Jehave in the same way.1 MoreoverK due to
complete anonymityK the individuals in our sample do not know what the others are doing and
seem not to Je interested in knowing. While cooperation that is conditional on giving Jy
specific other persons is present Jut amJiguousK the donations are contingent on the
environmental and institutional conditions under which the donations take place. In particularK
it is crucially important whetherK and in what wayK one is asked to donate. -he Jehavior of the
students seems to indicate that not only the way the Muestion is framed is vital Jut also that
identification with the organiVation is an important institutional condition which influences
their pro-social Jehavior.
FinallyK our data suggests that individuals differ in the eQtent of their pro-social preferences.
-o a certain eQtent they self-select according to their own preferences for the different
disciplines SfacultiesT taught at the university. 0 novel result in this conteQt is the impact of
1 For a discussion of the so-called nfalse consensuso effect see Ross et al. 19aa and Xawes et al. 19aa.

5. 4
teachingK i.e. whether giving is influenced Jy what professors teach in class aJout the
rationality of contriJuting to puJlic goods.
Section II discusses the concept of reciprocity in the conteQt of giving Jehavior and confronts
it with pro-social Jehavior. &n that JasisK theoretical hypotheses are derived. -he following
section III presents the actual case studied and the data collected. -he econometric analysis to
test the hypotheses is undertaken in section Ip. -he last section p offers conclusions.
II. Reciprocity and 6ore 7eneral Pro-Social Behavior
-he effect of reciprocity on giving for the voluntary funding of a SlinearT pure puJlic good
Swhich is Joth non-rival and non-eQcludaJleT can Je analyVed in the following way Se.g.
Rroson 199OT; Each potential contriJutor i in the group of n identical persons involved has an
income ,iK which she can either use to donate to a fund F or to consume private goods. If di is
the contriJution to the fundK the individual is privately aJle to consume ,i -di . -he
individualos earning from the fund is a multiple2 m of the sum of donations from all the
participantsK m di . 0 puJlic good proJlem eQists whenever 1qn<m<1. When m<1K it is never
optimal for a self-interested person to contriJute to the puJlic good Jecause the contriJution
costs her one unit Jut earns her only m. When 1qn<mK contriJuting to the puJlic good is
always optimal for the group as a wholeK Jecause donating one unit to the puJlic good costs
the individual one unit Jut earns nrm for the group.
For selfish individualsK there is a uniMue dominant strategy eMuiliJrium in which all persons in
the group,free,ride,Si.e. contriJute nothingTK independent of what the others do. IfK howeverK
an individualos utility function depends on another individualos Sor group of individualsoT
utility S"ecker 19a4TK a higher contriJution Jy the other memJers to the fund induces a
reduction in the individualos own contriJution SSugden 19]2T. If an individualos utility
moreover positively depends on the amount she contriJutes S0ndreonios 19]9K 1990 [warm
glow of givingTK her donation and the contriJutions Jy the others are imperfect suJstitutes.
-he reduction in the individualos contriJution is smaller than in the case of "eckeros pure
altruism. In the conteQt of the puJlic goods modelK Sugden S19]4T formaliVes reciprocity Jy
assuming that an individual contriJutes the minimum of SiT the smallest donation of all other
group memJers and SiiT the donation level she would most prefer other group memJers to
2 -he multiple m is the marginal return for each individual when he or she contriJutes one unit to the fund.

6. 5
make. It follows that there may Je a positive correlation Jetween the donation of a particular
person and the contriJutions Jy the other memJers. -his result is consistent with reciprocity
in a one-shot eMuiliJrium conteQt. 0 more general theory of reciprocity has Jeen advanced Jy
RaJin S199NT. Reciprocity is defined as the desire to Je kind to those who are perceived to
Jehave kindly towards youK and to punish those who are perceived to act in a hostile way
towards you. -his theoryK which Jases reciprocity on intentionsNK again predicts that a person
donates more if the others contriJute too Ssee also XufwenJerg and girchsteiger 1999K Falk
and FischJacher 2001T.
Xepending on the issue to Je analyVedK there are many reciprocity concepts4. In the conteQt of
our studyK the idea of conditionalit#,in the reciprocity concept is crucial. Individuals are
defined to Je conditional cooperators when the positive correlation discussed aJove oJtainsK
i.e. when people contriJute the more to a puJlic goodK the more the others contriJute5. In a
recent standard puJlic good eQperimentK for eQampleK it was identified thatK according to this
definitionK roughly 50 percent of the suJjects are conditional cooperators while a third of the
suJjects act as free riders SFischJacherK Wkchter and Fehr 2001T.
-his study argues that there are important cases in which this kind of conditional reciprocity
cannot solely eQplain the suJstantial share of individuals willing to contriJute to the puJlic
good. -his non-reciprocal giving will Je called [pro9social,be;avior. While it is not
conditional on the contriJution of others Stype I conditionalityTK it strongly depends on
another kind of conditionality Stype II conditionalityTK namely the environmental and
institutional conditions under which the contriJutions take place. Several studies support type
II conditionality in a specific sense; "eing asked is an important factor in eQplaining why
people contriJute to a puJlic good and offer voluntary work Sparese and ,aish 2000_ &pp
2001; N-5_ Freeman 199a_ Foster et al. 2001T. In addition to Jeing askedK pro-social Jehavior
also depends on the way one is asked. 0lthough the decision remains the sameK the
institutional change has significant Jehavioral conseMuences. -his conteQt dependence has
Jeen laJeled ninstitutional framingo Jy Isaac et al. S1991T. -his is consistent with findings in
Ultimatum Wame eQperiments that have Jeen conducted in 15 cultures; [... the preferences
N "ecause intentions are crucialK this reciprocity model differs from fairness modelsK where the Jehavioral
responses are solely caused Jy ineMuity aversion S"olton and &ckenfels 2000K Fehr and Schmidt 1999T.
4 For eQampleK direct and indirectK strong and weakK chain and general reciprocity Ssee e.g. "owles and Wintis
2001K golm 2000JT.
5 See e.g. the eQperiments Jy geser and van Winden 2000K SonnemansK Schram and &fferman 1999K Rroson
199].

7. O
over economic choices ... are shaped Jy the economic and social interactions of everyday life
S/enrichs et al. 2001;aaT. -his indicates that outside the laJK conditionality of type II is
crucially important.
Pro-social Jehavior varies consideraJly Jetween individuals. While some persons act
according to the economic assumption and therefore free-ride in social dilemma situationsK
others reveal suJstantial pro-social preferences. 0ccording to these differencesK individuals
select into specific groups. 0s has Jeen shown Jy prior studiesK economists seem to constitute
one such group Se.g. Rarter and Irons 1991 and Frank et al. 199NT. It is thus crucial to analyVe
such differences l Jetween economists and non-economists l outside the laJ. Especially if
one is interested in whether economics training destroys pro-social JehaviorK one should look
at Jehavior in a natural setting.
&n the Jasis of this discussionK we advance the following hypotheses for the case of
contriJuting to a social fund;
/1; 0 suJstantial numJer of people are prepared to act in a pro-social way in an anonymous
situation l even after several rounds.
/2; EQpectations aJout the contriJutions of other people matter. -he more people eQpect
others to cooperateK the more they cooperate themselves SRonditionality of type IT.
/N; -he environment under which the donations take place matters SRonditionality of type
IIT. In particularK it is essential that people are asked to contriJute in a way they perceive
as acceptaJle.
/4; People differ in their pro-social attitudes. -he type of person Sas reflected in part Jy their
choice of studyT influences donating even when standard personal characteristics SgenderK
and ageT are controlled for.
/5; -eachers have no influence on who donates and who does not donate.
In the following sectionsK we test these hypotheses empirically.
III. ;he Empirical >ase and the Data
Each semesterK all the students at the University of Zurich have to decide whether or not they
want to contriJute to two official Social Funds l in addition to the compulsory tuition fee. &n
the official letter for renewing their registrationK the students are asked whether they want to

8. a
voluntarily give a specific amount of money SR/F a.-K aJout USs 4.20T to a fund which offers
cheap loans to students in financial need andqor a specific amount of money SR/F 5.-K aJout
USs NT to a second fund supporting foreigners who study at the University of Zurich. Without
their eQplicit consent SJy marking a JoQTK students do not contriJute to any fund at all. &ur
data refers to the decisions made in the seven semesters from the winter semester 199]q99 up
to and including the winter semester 2001q2002. -he fact that every student at the University
of Zurich has to decide anew each semester whether he or she is willing to contriJute to one
or Joth of the Social Funds generates a large numJer of oJservations. -aJle 1 presents
summary statistics of the data set. -he taJle also shows the amount of students who contriJute
to at least one of the funds for each variaJle. We oJserve the decisions of N2K9O1 students
who decide on an average 5.4 timesK depending on how many semesters they are there. -he
decisions from the seven semesters are pooledK generating 1NOK]O2 oJservations. -he panel
structure allows us to analyVe the effect of repetition on the decision to contriJute. 0s well as
thatK we have aggregated data since 199NK which allows us to test the effects of environmental
and institutional conditions on giving Jehavior.
In additionK an anonymous on-line survey among the same student group of the University of
Zurich was undertaken.O -he response rate was 1] percent. From this sampleK we were aJle to
use No25O answersK containing responses to all the Muestions relevant for our conteQt. -his
sample is not totally representative Snot surprisinglyK a larger numJer of economics students
responded to the Muestionnaire sent out Jy two economistsTK Jut with respect to gender and
ageK the sample corresponds to the distriJution of students at the University of Zurich. -he
survey again asked whether the person contriJuted money to one or Joth of the funds. aN
percent responded that they didK compared to the O] percent who actually contriJuted. -his
difference Jetween survey answers and actual Jehavior is found in a lot of survey-Jased
studies. While the differences can Je the result of people lying Ssee EichenJerger and
&JerholVer-Wee S199]T for differences Jetween hypothetical and real decisionsTK a more
convincing eQplanation is that people who contriJuted to the Funds are more likely to
respond. -he differences should Je kept in mind while interpreting the survey data.
-he donations Jy the students to the Social Funds have three important characteristics;
S1T 0 large numJer of people are involved Smore than NNK000 peopleT. -he condition 1qn<m
is certainly fulfilled and rational selfish individuals would not contriJute to the fund.
O -he on-line Muestionnaire is reproduced at http;qqwww.iew.uniVh.chqgrpqfreyqfrageJogen.htm.

9. ]
S2T It is not generally known who receives actual support from the funds.
SNT Whether a student donates or not to the Social Funds remains completely anonymous.
Social pressure can therefore Je eQcluded as a motive.
0s will Je eQplained in detail in the neQt sectionK these characteristics e?clude,direct
reciprocit#,as defined aJove. Xonations to the Social Funds at the University of Zurich are
therefore more a case of pro9social,be;avior.
IV. Analysis and Results
@.,Bro9social,be;avior
-he raw data suggest that the students in our sample do not act like the traditional economic
model of selfish individuals that it predicts. 0 large proportion of the students are prepared to
contriJute to the funds. "etween the years 199] l 2001K on average more than O]t of the
individuals contriJuted to at least one of the funds Ssee taJle 2T. More than O1t contriJuted to
Joth funds.a
-0"dE 2 0"&U- /ERE.
0s the decision is not a laJoratory dictator or puJlic good gameK we can eQclude that pro-social
Jehavior is due to an eQperimenter effect or some other sort of direct reciprocal
reaction mentioned Jy /offman et al. S199OT. -hey Jelieve that the fact that anonymity is not
completely guaranteed can indeed eQplain the remaining level of donation in their dictator
game. Johannesson and Persson S2000TK on the other handK Jy increasing social distance
Jetween dictator and recipients even moreK find evidence of non-reciprocal altruism. In our
caseK the students decide at home and under totally anonymous conditions. SoK direct
reciprocity either SiT Jy the students andqor SiiT Jy the University can Je ruled out as an
eQplanation of the contriJutionK as will Je eQplained in detail Jelow.
SiT Students who do not contriJute to the funds cannot Je punished or accused of Jeing selfish
Jy their colleagues. -he mechanism of punishment for sustaining cooperationK discussed e.g.
Jy Fehr and Wkchter S2000TK cannot eQplain the high numJer of people acting pro-socially. In
a In dictator gamesK the contriJution of the suJjects is much lower. -hese differences can Je eQplained Jy the
fact that recipients differ. Eckel and Wrossman S199OT show in an anonymous dictator game that contriJutions
are much Jigger if the suJjects can give money to an estaJlished charity rather than another student.

10. 9
the aJsence of any form of punishmentK one would eQpect that repetition strongly decreases
cooperationK as shown Jy puJlic goods eQperiments SXawes and -haler 19]]K dedyard 1995K
Fehr and Wkchter 2000T. Figure 1 shows the willingness to give money to the Social Funds to
Je dependent on the numJer of semesters the students study at the University. -he repetition
of the decision only weakly decreases the level of contriJution. -his oJservation supports the
idea that in the situation analyVedK direct reciprocity activated through a punishment
mechanism is not necessary in order to sustain the pro-social Jehavior of the students.
/oweverK the sustainaJility of contriJutions under anonymity cannot eQclude indirect
reciprocity Snconditional cooperationoT. We will test this notion in the neQt section.
SiiT Xirect reciprocity on Jehalf of the University is also eQcluded. -he contriJution of the
students does not influence any possiJle future support from the funds in case of needK nor
does it in any way affect the relationship Jetween the students and the University.
FIWURE 1 0"&U- /ERE.
-he results of the descriptive statistics are consistent with ;#pot;esis,@; even after several
roundsK a large numJer of students act pro-socially in an anonymous decision setting. "ecause
no mechanism of punishment and therefore direct reciprocity is at handK this result supports
the eQistence of non-reciprocal pro-social preferences. Ronditional cooperation cannot Je
eQcluded from the evidence presented. In the following sections we analyVe whether this
indirect type of reciprocity can eQplain the contriJution of the students to the funds.
C.,4ndications,of,indirect,reciprocit#
-heories of reciprocity suggest that people contriJute to a puJlic good dependent on the
Jehavior of others. Individuals dislike Jeing a so-called msuckermK i.e. Jeing the only one who
contriJutes to a puJlic good while the others free-ride. -he more one thinks that others
cooperateK the greater is the proJaJility that this person contriJutes him- or herself. -o test
this notionK the students were asked via a large-scale online surveyK how many other students
they eQpect to Je contriJuting. -he results of our survey show that eQpectations aJout others
correlate with the individual decision to contriJute to the Social Funds. -he coefficient of the
correlation Jetween the eQpressed eQpectation and the contriJution to at least one fund is 0.N4.
-his correlation is Muite large and statistically significant at a 99t-level SF1KN1O]u415.4aK
pv0.01Tand. 0s can Je seen from Figure 2K the marginal effect is suJstantial.] 0n increase of
] -he marginal effect vector in a proJit analysis eMuals 0.00O2 Ss.e. 0.000N5T.

11. 10
the perceived cooperation of others Jy 10 percentage points increases the individual
proJaJility of contriJuting Jy O percentage points.
FIWURE 2
"ut the causalit# is not at all clear. While for the notion of conditional cooperation it is
important that individuals cooperate conditional on the Jehavior of othersK the causality can
Je the other way around; people who Jehave in a cooperative way also eQpect others to
cooperate. -hey deduce from their own Jehavior how they think other people will Jehave.
-he perceived eQpectation aJout the cooperation of others is therefore a good indicator of
peopleos own pro-social Jehavior and not evidence for conditionality of type I. SimilarlyK
Wlaeser et al. S2000T found evidence of such an effect in their study aJout trust and conclude;
n... the Jest way to determine whether or not a person is trustworthy is to ask him whether or
not he trusts otherso. -his proJlem of causality gets even more oJvious if one looks at how
little students really know aJout the real contriJution rate.
Students do not know for certain what others do and also do not seem to Je interested in the
Jehavior of the othersK as they donmt talk with their colleagues aJout the two Social Funds.
-0"dE N 0"&U- /ERE.
-aJle N shows the answers to two Muestions designed to find out whether the students are
aware of the Jehavior of others and whether they actually talk with each other aJout the
Funds. -he results indicate that more than three Muarters of the students do not tell their
friends whether they contriJuted or not. -hree Muarters of the students never talk with their
colleagues aJout the funds. -hese results throw even more douJt on the notion that the
causality of the correlation goes from the eQpectations aJout the share of contriJutions to
oneos own contriJution.
-he results of the empirical analysis are consistent with ;#pot;esis 1 and also partly apply for
;#pot;esis,C. 0 large numJer of students Jehave pro-socially. -he anonymous situation does
not allow for direct reciprocity. "ut there is empirical support that students do compare with
others and that their actions depend on the Jehavior of others. -he evidence for this indirect
reciprocity in the form of conditional cooperation SRonditionality of type IT is amJiguous.
While the correlation Jetween the eQpected cooperation rate and the actual contriJution of the
students is Muite JigK the causality is more than unclear. &nly approQimately every fifth
student knows the Jehavior of her colleagues or talks with others to find out aJout the
appropriateness of their own Jehavior. We conclude that students Jehave pro-socially not

12. 11
eQclusively conditionally on the Jehavior of others. Pro-social Jehavior depends more on the
environmental and institutional conditionsK as will Je argued in the neQt section.
D.,Bro9social,be;avior,depends,on,Environ5ental,and,4nstitutional,Conditions
We have adduced empirical evidence that contriJuting to the two Social Funds is not only due
to reciprocal considerations Jut to other pro-social attitudes. -his pro-social Jehavior depends
on institutional conditions. 0 crucial institutional feature supporting pro-social Jehavior is
Jeing asked to do so. MoreoverK it very much depends on how one is asked. Xifferent ways of
framing the same Muestion institutionally can change the prevalence of pro-social Jehavior
dramatically Sfor framing effects see e.g. wuattrone and -versky 19]]K dindenJerg 1992K
Sonnemans et al. 199]K 0ndreoni 1992 and Elliott and /ayward 199]T.
0t the University of ZurichK an eQogenous variation of the institutional conditions allows us to
test the effect on pro-social Jehavior. Xue to a restructuring of the administrationK the
University of Zurich changed the official letter for renewing studentso registration for the
winter term in 199]. 0fter this semester termK the administration was aJle to handle studentso
decisions electronically. -he students are now asked to contriJute in the following way; they
have to tick JoQes to decide if they want to donate money to one or the otherK to Joth or to
none of the Funds. 0fter a monthK they receive an invoice with the compulsory tuition fee
plus the chosen amount for the Social Funds. "efore the winter term 199]K students received
two invoices and had to chose Jetween the two_ one with the amount of the compulsory
tuition fee on itK and the other with the amount of the tuition fee plus the amount due for
contriJutions to Joth Funds.
Figure N shows the effect of the eQogenous change in the institutional setting on pro-social
Jehavior. 0fter the change effected in the summer term 199]K the percentage of people
contriJuting to the two Social Funds increased from an average of 44t to O2t. -he
difference is statistically significant St-test of differences of distriJution; t-valueu-11.1T.
MoreoverK according to the new systemK the students can also opt for only one of the fundsK so
the percentage of people who contriJute to at least one of the funds saw an even Jigger
increase. -his result is consistent with hypothesis N that pro-social Jehavior is sensitive to
changes in the institutional conditions.
FIWURE N 0"&U- /ERE.

13. 12
Standard economic reasoning would consider the two decisions identicalK Jecause the
underlying decision to Je taken is the same; does one want to contriJute money to the two
Funds or notx 0nd the prediction is also straightforward; no homo oeconomicus will donate
any money in either of the two anonymous decision settings. "ut even for non-traditional
eQplanations of cooperative Jehavior Se.g. reciprocity and pure altruismTK the different settings
should not affect the Jehavior of the suJjects. IfK for eQampleK cooperation is only conditional
on the Jehavior of othersK no Jehavioral difference should Je oJserved in the two settings.
-husK it is oJvious that our theory must go be#ond an assumption of reciprocity or pure
altruism Jecause they are unaJle to eQplain the results presented here. /oweverK the concept
of pro-social JehaviorK as presented aJoveK depends on the institutional conditions or the
conteQt of the decision.9
-he conteQt dependent pro-social Jehavior has Jeen laJeled ninstitutional framingo Jy Isaac et
al S1991T. Frey and "ohnet S1995T and "ohnet and Frey S1999T further developed the idea and
present evidence that institutions affect fairness consideration in eQperiments. -hey allow for
one-way identification in a dictator game and oJserve that the amount shared increases
suJstantially. -heir analysis indicates the importance of identification with the nvictimo10. -he
increase seems to support the concept of conteQt dependent pro-social JehaviorK which goes
Jeyond reciprocity and pure altruism. -he same holds for identification with an organiVation.
0s has Jeen shown in other studiesK especially in studies concerning alumni giving to
universitiesK attachment to an organiVation is an important factor in eQplaining pro-social
Jehavior SRlotfelter 2001K Mael and 0shforth 1992T. In the case of the contriJution to the two
social funds of the University of ZurichK changes in the institutional conditions which affect
the identification with the UniversityK should eQplain some of the variation in giving Jehavior.
&ne such change in the environmental and institutional conditions takes place at the
Jeginning and the end of a studentms University life. For Joth periodsK studentso physical
attendance at the University is lower than in the periods in Jetween. "efore taking up their
studiesK i.e. at the very JeginningK students oJviously have not attended the University at all_
at the end of their studiesK students no longer attend classes Jut prepare for their eQams over
an eQtended period Smore than half a year in the Swiss University systemT and therefore
9 0ndreoni S1992T presents evidence that positive framing leads to more cooperation in a puJlic good eQperiment
than negative framing of the same decision. /e eQplains this difference in the light of the nwarm glowo effect.
nS...T it must Je that people enjoy doing a good deed more than they enjoy not doing a Jad deedoSp. 11T.
10 -he idea of the so-called nidentifiaJle victim effecto goes Jack to Schelling S19O]T and has recently Jeen
analyVed Jy Jenni and doewenstein S199aT.

14. 1N
attend the University only sporadically. -he strongest identification with oneos University
should eQist when students regularly attend courses and feel themselves to Je a part of the
student Jody and of their al5a,5ater. 0s a conseMuenceK students are eQpected to contriJute
significantly less to the Social Funds at the Jeginning and end of their studies.
-he first decision of whether to contriJute to the Funds is taken Jefore the students are
actually attending the University. -hey are registeredK Jut the actual decision concerning their
contriJution takes place Jefore the start of the freshman semester. 0gainK the traditional
economic prediction would not change. "ut as can Jeen seen from a proJit modelK estimated
for the period 199]-2001 and reproduced in taJle 4K this change in environmental condition
systematically affects Jehavior. -he proJaJility that a first semester student contriJutes
money is 2.N percentage points lower compared to those in the following semesters Sthe
reference period is the basic studyT. -his effect is statistically significant at the 99t-level and
persists in an estimation with personal fiQed-effects Ssee the conditional logit model in taJle
4T.
-0"dE 4 0"&U- /ERE.
-he effect on contriJuting while Jeing in the last semester is also shown in taJle 4. -he
variaJle for the last semester takes the value 1 if a student is in her last semester and 0
otherwise. "oth models show a significant effect for Jeing in the last semester. -he
proJaJility of contriJuting to at least one fund decreases O.O percentage points compared to
the preceding periods.
-he two Jehavioral regularities oJserved l that students tend to contriJute less Jefore they
start their studiesK and at the very end of their studies l is consistent with a changing
identification with the University as an organiVation.
-he control variaJles show the eQpected signs; the different stages in a studentos studies
SFain,stage and B;.G.T do not have a significant effect on his or her pro-social Jehavior in
the fiQed-effects estimation which controls for individual heterogeneity. 0s could already Je
seen in the descriptive statisticsK the Hu5ber,of,se5esters decreases the proJaJility of a
contriJution to the Funds. Gender does not have an effect on giving Jehavior. In contrastK
Farried students are more generous than their single colleagues. /oweverK the effect is not
statistically significant in a fiQed-effects model. Marriage itself does not make one more
generousK Jut married students are a special selection. &ver timeK the willingness to contriJute
increases as indicated Jy the period dummies. InterestinglyK the proJaJility that foreign

15. 14
students contriJute to the Social Funds is smaller than for Swiss students. -his Jehavior of
foreign students is of interest Jecause one of the two Funds is eQclusively designed to support
foreigners. It could Je thatK if foreigners contriJute they tend to prefer to support other
foreigners.
-aJle 5 shows the descriptive statistics for the contriJutions of foreigners to the Social Funds.
Foreign studentsK to the eQtent they contriJute at allK mainly have a tendency to support other
foreigners. -he result is supported in a multinominal regression model which includes control
variaJles. -his pattern of pro-social Jehavior can Je interpreted as further support for the
importance of identification for giving. Foreigners feel more identified with other foreigners.
-his evidence is consistent with various studiesK which find that group identity eQplains a lot
aJout pro-social Jehavior SSimon 199NK Xawes and -haler 19]]K -urner and Wiles 19]1K
0kerlof and granton 2000T. In contrast to these eQperimentsK where group identity is achieved
through discussionK our data suggest that even anonymous group attachment can evolve.
-0"dE 5 0"&U- /ERE
-he empirical results in this section show that pro-social Jehavior depends on environmental
and institutional conditions S;#pot;esis,DT. Most of allK the way one is asked to contriJute to a
puJlic good is of great importanceK even in the aJsence of personal contact. MoreoverK our
results support the crucial effect of identification and identity for giving Jehavior.
J.,Effect,of,Education,on,Bro9Kocial,Le;avior
People seem to differ in their pro-social preferencesK which leads to different Jehavior asK for
eQampleK is reflected in eQperimental settings Se.g. Fehr and Wkchter 2000K Weimann 1994T.
Some of them free-ride right from the Jeginning of the game and thus Jehave according to the
standard economic predictionsK while others deviate from this prediction suJstantially and act
in a pro-social way. We test this notion aJout different types of persons in our data set
Shypothesis 4T Jy looking at potential selection effects. People with similar preferences select
themselves into similar suJjects at the University. If this is the caseK we should oJserve that
the distriJution of selfish types is not randomK Jut systematic.11 -o test this hypothesisK we
look at the very first decision to contriJute to the two Funds at the University. 0t this date
11 SimilarlyK &ckenfels and Weinmann S1999T compare the preferences of East- and West-Wermans in laJoratory
eQperiments and find differences in their cooperative Jehavior. RadsJy and Maynes S199]T compared the
Jehavior of nurses with economics students in an eQperimental puJlic good game.

16. 15
time students have not yet attended any lectures at the UniversityK so we can eQclude any
effects resulting from the influence of University training.
While aNt of arts students contriJute to the FundsK only O4t of law students do so.12 -he
chosen discipline of study partially reflects the type of students and their JehaviorK which is
consistent with ;#pot;esis,J. "ecause the students of the faculties could differ systematically
with regard to other characteristicsK such as seQ or ageK which correlate with giving JehaviorK
we estimate a multivariate regression model. -he results in taJle O support our hypothesis that
students differ in their social preferences and select according to these preferences into
different suJjects. -he control variaJles in this estimation for students in their first semester
show the same effects as in taJle 4. 0n eQception is the Jehavior of women. Rounting all the
semestersK women do not Jehave significantly differently from menK Jut the proJaJility that
women contriJute to the funds is almost 4 percentage points lower in the first semester than it
is for men. -his result contradicts previous results of Jehavioral differences Jetween women
and men SEckel and Wrossman 199]K &rtmann and -ichy 2000T. Further research should
investigate which conditions in the data set used influence the Jehavior of the two seQes Sfor
similar eQperimentsK see 0ndreoni and pesterlund 2001T.
-o test whether pro-social preferences are influenced Jy educationK we consider students who
are eQtremely eQposed to nindoctrinationo effects; economics students. 0 lot of critics of
economic imperialism Jelieve that studying economics changes cooperative Jehavior S&strom
199]K gelman 19]aT. -hereforeK we test whether economics students adapt to the predictions
of learned SconventionalT theoryK and whether it is taken as normative advice with respect to
pro-social Jehavior. Prior research has looked at the differences Jetween economists and non-economists
in their cooperative Jehavior Se.g. Marwell and 0mes 19]1K Rarter and Iron 1991K
Frey et al. 199N and Frank and SchulVe 2000T. Especially Frank et al. S199NK 199OT identified
Jehavioral differences in Ultimatum games Jetween economists and non-economists.
MoreoverK they found that training in economic theory changes the willingness to Jehave pro-socially
for the economics students. We call this the indoctrination hypothesis and test it with
our data of the University of Zurich.1N
0s can Je seen in taJle OK the Jehavior of students of different disciplines is mostly due to
selectivity effects. Even Jefore attending the first lectureK differences in pro-social Jehavior
12 -he respective results for students of other faculties are; aat of students of theologyK O5t of medical
studentsK 5]t of veterinary medicine students and Oat of natural science students contriJute to at least one
fund.

17. 1O
eQist. Not only economicsK Jut also suJjects like law tend to attract selfish students. If
economic theory influences the cooperative Jehavior of economics students negativelyK we
should eQpect the differences Jetween economists and non-economists to widen over time. -o
test whether these Jehavioral differences do indeed increase over timeK we estimate a
conditional logit model with personal fiQed effects. With this modelK we control for individual
heterogeneity. We use interaction terms Ssuch as Econo5istsMHu5ber,of,Ke5estersT as
proQies for economic indoctrination. -aJle a presents the respective results. -he results
support ;#pot;esis,N; teaching economic theory does not influence giving and cooperative
Jehavior. For economics studentsK who are eQposed to a rather eQtreme theory of human
JehaviorK the willingness to contriJute does not decrease in a statistically significant way.
Rompared with other studentsK only students of lawK medicine and veterinary medicine
negatively change their Jehavior during their study. InterestinglyK for students of theology the
willingness to contriJute increases during their studies. Whether this is the effect of religion is
MuestionaJle and lies outside the scope of this paper.14
-o summariVeK our results show that students select into different disciplines of study
according to differences in their pro-social preferences. FurthermoreK economics students are
not influenced Jy teaching where pro-social Jehavior is concerned. Economics students do
not change their pro-social attitudes as a result of their economics training. -he results are
consistent with ;#pot;eses 4 and 5.
>onclusions
-his paper has provided empirical evidence for the importance of non-reciprocal pro-social
Jehavior of individuals in an anonymousK n-person pure puJlic good setting. We use a uniMue
panel data set of 1NOK000 oJservations Sroughly NNK000 personsT concerning the decisions of
students at the University of Zurich to contriJute to two Social Funds administered Jy the
University. -hese field oJservations are matched with an eQtensive survey of the same sample
group of students to find out more aJout the conditions and motives for giving.
Four hypotheses are tested with these data;
1N For an eQtensive test of the differences Jetween economists and non-economistsK see Frey and Meier S2001T.
14 -he influence of religion on the contriJution to puJlic goods is discussed in dipfordK McRormick and -ollison
S199NT and /ull and "old S1995T.

18. 1a
1; 0 suJstantial numJer of people are prepared to act in a pro-social way in an anonymous
situation in which direct reciprocity is aJsent.
-he results of the statistics are consistent with ;#pot;esis,@; Even after several roundsK a large
numJer of students act pro-socially in an anonymous decision setting. "ecause no mechanism
of punishment and therefore direct reciprocity is at handK this result supports the eQistence of
non-reciprocal pro-social preferences. "ut conditional cooperation cannot Je eQcluded Jy the
evidence presented.
2; EQpectations aJout the contriJutions of other people matter. -he more people eQpect others
to cooperateK the more they cooperate themselves Sconditionality of type IT.
-he results of the empirical analysis are consistent with this ;#pot;esis. While the anonymous
situation does not allow for direct reciprocityK students compare themselves with others and
make their actions dependent on their eQpected Jehavior of others. -he evidence for this
indirect reciprocityK in the form of conditional cooperationK is amJiguous. While the
correlation Jetween the eQpected cooperation rate and the actual contriJution of the students
is Muite JigK the causality is unclear. &nly approQimately every fifth student knows the
Jehavior of his or her colleagues or talks with others to find out aJout the appropriateness of
their own Jehavior. Students thus seem to Jehave pro-socially Jut not eQclusively conditional
on the Jehavior of others.
N; -he environment under which the donations take place matter Sconditionality of type IIT.
In particularK it is essential that people are asked to contriJute in a way they conceive to Je
acceptaJle.
-he empirical results suggest that pro-social Jehavior depends on environmental and
institutional conditions. -he way one is asked to contriJute to a puJlic good is of great
importanceK even in the aJsence of any personal contact. MoreoverK our results support the
crucial effect of identification and identity for giving Jehavior.
4; People differ in their pro-social attitudes. -he type of person Sas partially reflected Jy the
choice of studyT influences donating even when standard personal characteristics Sgender and
ageT are controlled for.
&ur data suggest that students indeed select into different disciplines according to differences
in their pro-social preferences.
5; -eachers have no influence on who does or does not donate.

19. 1]
-he empirical analysis indicates that economics students do not change their pro-social
attitudes due to their economics training.
-he results derived are Jased on the Jehavior of the students and a survey carried out at the
University of Zurich. Future research must estaJlish whether the giving Jehavior identified
applies to other persons and to other settings. /oweverK we are confident that our findings are
not peculiar to these students Jut apply more generally. -he students at the University of
Zurich are Muite unlike a student population in many other countriesK especially the United
StatesK Jecause they are consideraJly older and a large numJer of them hold a part-time joJ
while studying. -hey are thus more like the population at large thanK for instanceK 0merican
college students.

20. 19
References
0damK J.S. S19ONT. Wage IneMualitiesK Productivity and Work wuality. 4ndustrial,Relations N.
9-1O.
0damK J.S. S19O5T. IneMuity in Social EQchange. In; d. "erkowitV Sed.T. 6dvances,in
E?peri5ental,Kocial,Bs#c;olog#K vol. 2. New ,ork; 0cademic Press.
0kerlofK Weorge 0. and Rachel E. granton S2000T. Economics and Identity. Puarterl#
)ournal,of,Econo5ics,115SNT. a15-5N.
0lmK JamesK Wary McRlelland and William X. SchulVe S1992T. Why Xo People Pay -aQesx.
)ournal,of,Bublic,Econo5ics 4]; 21-N].
0ndreoniK James S19]9T. Wiving with Impure 0ltruism; 0pplications to Rharity and Ricardian
EMuivalence. )ournal,of,Bolitical,Econo5# 9a SNT. 114a-115].
0ndreoniK James S1990T. Impure 0ltruism and Xonations to PuJlic Woods; 0 -heory of
Warm-Wlow Wiving. Econo5ic,)ournal 100. 4O4-4aa.
0ndreoniK James S1992T. Warm-Wlow persus Rold-Prickle; -he Effects of Positive and
Negative Framing on Rooperation in EQperiments. Puarterl#,)ournal,of,Econo5ics O0
S1T.1-21.
0ndreoniK James S199]T. -oward a -heory of RharitaJle Fund-Raising. )ournal,of,Bolitical
Econo5# 10OSOT. 11]O-121N.
0ndreoniK James S2001T. -he Economics of Philanthropy. -o appear in; N.J. Smelser and P.".
"altes Seds.T. T;e,4nternational,Enc#clopedia,of,t;e,Kocial,and,Le;avioral,Kciences.
dondon; Elsevier.
0ndreoniK JamesK "rian Erard and Jonathan Feinstein S199]T. -aQ Rompliance. )ournal,of
Econo5ic,Literature NOS2T. ]1]-O0.
0ndreoniK James and dise pesterlund S2001T. Which Is the Fair SeQx Wender Xifferences in
0ltruism. Puarterl#,)ournal,of,Econo5ics 11OS1T. 29N-N12.
"eckerK Wary S. S19a4T. 0 -heory of Social Interactions. )ournal,of,Bolitical,Econo5# ]2.
10ON-109N.
"eckerK Wary S. S19aOT. T;e,Econo5ic,6pproac;,to,Ru5an,Le;avior. Rhicagoqdondon;
University of Rhicago Press.
"eckerK Wary S. S199OT. T;e,Econo5ic,Wa#,of,Looking,at,Le;aviorU,t;e,Hobel,Lecture.
Stanford University; /oover Institution on WarK Revolution and Peace.
"ohnetK Iris and "runo S. Frey S1999T. Social Xistance an &ther-Regarding "ehavior in
Xictator Wames; Romment. 65erican,Econo5ic,Review ]9S1T. NN5-NN9.
"oltonK Wary and 0Qel &ckenfels S2000T. ERR l 0 -heory of EMuityK Reciprocity and
Rompetition. 65erican,Econo5ic,Review 90. 1OO-19N.
"ouldingK genneth S19aNT. T;e,Econo5#,of,Love,and,Fear. "elmontK R0; Wadsworth.
"owlesK Samuel and /erJert Wintis S2001T. Social Rapital and Rommunity Wovernance.
Fi5eo. Santa Fe Institute.

21. 20
RadsJyK Rharles "ram and EliVaJeth Maynes S199]T. Rhoosing "etween a Socially Efficient
and Free-Riding EMuiliJrium; Nurses persus Economics and "usiness Students. )ournal,of
Econo5ic,Le;avior,and,OrganiXation Na. O0N-O20.
RarterK John R. and Michael X. Irons S1991TK 0re Economists XifferentK and If SoK Whyx
)ournal,of,Econo5ic,Berspectives 5K 1a1-1aa.
RlotfelterK Rharles -. S199aT. -he Economics of Wiving. In; "arryK John W. und "runo p.
Manno SEds.T. Giving,Letter,,Giving,K5arter. WashingtonK X.R.; National Rommission on
Philanthropy and Rivic Renewal. N1-55.
RlotfelterK Rharles -. S2001T. 0lumni Wiving to Elite Private Rolleges and Universities.
Working,Baper.
RollardK Xavid S19a]T. 6ltruis5,and,t;e,Econo5#U,6,Ktud#,of,Hon9Kelfis;,Econo5ics.
dondon; Martin RoJertson.
XawesK RoJyn M.K J. Mc-avish and /. Shaklee S19aaTK "ehaviorK RommunicationK and
0ssumptions 0Jout &ther Peoplems "ehavior in a Rommons Xilemma SituationK )ournal,of
Bersonalit#,and,Kocial,Bs#c;olog# N5K 1-11.
XawesK RoJyn M. and Richard /. -haler S19]]T. Rooperation. )ournal,of,Econo5ic
Berspectives 2SNT. 1]a-19a.
XufwenJergK Martin and Weorg girchsteiger S199]T. 0 -heory of SeMuential Reciprocity.
Fi5eo. RentER for Economic Research. -ilJurg.
EckelK Ratherine R. and Philip J. Wrossman S199]T. 0re Women dess Selfish -han Menx
Evidence From Xictator EQperiments. Econo5ic,)ournal 10]. a2O-aN5.
EckelK Ratherine R. and Philip J. Wrossman S199OT. 0ltruism in 0nomymous Xictator Wames.
Ga5es,and,Econo5ic,Le;avior 1OS2T. 1]1-191.
EichenJergerK Reiner and FeliQ &JerholVer-Wee S199]TK Rational Moralists; -he Role of
Fairness in Xemocratic Economic PoliticsK Bublic,C;oice 94K 191-210.
ElliottK Ratherine S. and Xonald M. /ayward S199]T. -he EQpanding Xefinition of Framing
and Its Particular Impact on Economic EQperimentation. )ournal,of,Kocio9Econo5ics
2aS2T. 229-24N.
FalkK 0rmin and Urs FischJacher S2001T. 0 -heory of Reciprocity. CEKifo,Working,Baper
No. 45a.
FehrK Ernst and glaus M. Schmidt S1999T. 0 -heory of FairnessK RompetitionK and
Rooperation. Puarterl#,)ournal,of,Econo5ics 114; ]1a-]O].
FehrK Ernst and glaus M. Schmidt S2000T. FairnessK IncentivesK and Rontractual Rhoices.
European,Econo5ic,Review 44 S4-OT; 105a-O].
FehrK Ernst and Simon Wkchter S2000T. Rooperation and Punishment in PuJlic Woods
EQperiments. 65erican,Econo5ic,Review 20; 9]0-994.
FischJacherK UrsK Simon Wkchter and Ernst Fehr S2001T. 0re People Ronditionally
Rooperativex Evidence from a PuJlic Woods EQperiment. Econo5ics,Letters_ a1SNT; N9a-
404.

22. 21
FosterK pivienK Susana MouratoK Xavid Pearce and Ece yVdemiroglu S2001T. T;e,Brice,of
YirtueU,T;e,Econo5ic,Yalue,of,t;e,C;aritable,Kector. Rheltenham U.g.; Edward Elgar
PuJlishing.
FrankK "jörn und Wünther W. SchulVe S2000T. Xoes Economics Make RitiVens Rorruptx
)ournal,of,Econo5ic,Le;avior,and,OrganiXation 4N S1T. 101-11N.
FrankK RoJert /.K -homas Wilovich and Xennis -. Regan S199NT. Xoes Studying Economics
InhiJit Rooperationx )ournal,of,Econo5ic,Berspectives aK 159-1a1.
FrankK RoJert /.K -homas Wilovich and Xennis -. Regan S199OT. Xo Economists Make "ad
RitiVensx )ournal,of,Econo5ic,Berspectives 10K 1]a-192.
FreemanK Richard ". S199aT. Working for Nothing; the Supply of polunteer daJour. )ournal
of,Labour,Econo5ics 15; 140-1OO.
FreyK "runo S. S1999T. Econo5ics,as,a,Kcience,of,Ru5an,Le;aviourU,Towards,a,Hew,Kocial
Kcience,Baradig5. "oston et al.; gluwer 0cademic PuJlishers.
FreyK "runo S. S2001T. 4nspiring,Econo5icsU,Ru5an,Fotivation,in,Bolitical,Econo5#.
Rheltenham; U.g. Edward Elgar PuJlishing.
FreyK "runo S. and Iris "ohnet S1995T. Institutions 0ffect Fairness; EQperimental
Investigations. Journal of Institutional and -heoretical Economics SJI-ET 151S2T. 2]O-N0N.
FreyK "runo S. and Stephan Meier S2001T. 0re Political Economists Selfish and
Indoctrinatedx Institute of Empirical Economic ResearchK University of ZurichK Working
Paper.
FreyK "runo S.K Werner W. Pommerehne and "eat Wygi S199NT. Economics Indoctrination or
Selectionx Some Empirical ResultsK )ournal,of,Econo5ic,Education 24K 2a1-2]1.
WkchterK Simon and 0rmin Falk S2001T. Reputation and Reciprocity - RonseMuences for the
daJour Relation. Kcandinavian,)ournal,of,Econo5ics. Forthcoming.
WlaeserK Edward d.K Xavid I. daiJsonK Jos{ 0. Scheinkman and Rhristine d. Soutter S2000T.
Measuring -rust. T;e,Puarterl#,)ournal,of,Econo5ics 115SNT. ]11-]4O.
WodelierK Maurice S2000T. -hings ,ou Xonot Wive or Sell Jut Which ,ou geep; paluaJle
and Social &Jjects. In; W{rard-paretK douis 0ndr{K Serge-Rhristophe golm and Jean
Mercier ,thier Seds.T. T;e,Econo5ics,of,Reciprocit#,,Giving,and,6ltruis5. /oundmills et
al.; MacMillan Press dtd. 1]2-195.
WouldnerK 0lvin Ward S19O0T. -he Norm of Reciprocity; 0 Preliminary Statement. 65erican
Kociological,Review 25; 1O1-1a].
WraetVK Michael J. and douis d. Wilde S19]5T. -he Economics of -aQ Rompliance; Facts and
Fantasy. Hational,Ta?,)ournal N] SSeptemJerT; N55-NON.
/enrichK Joseph et al. S2001T. In Search of /omo Economicus; "ehavioral EQperiments in 15
Small-Scale Societies. 65erican,Econo5ic,Review 91; aN-a].
/irshleiferK Jack S19]5T. -he EQpanding Xomain of Economics. 65erican,Econo5ic,Review
a5 SMayT; 5N-O].
/offmanK EliVaJethK gevin McRaJe and pernon d. Smith S199OT. Social Xistance and &ther-
Regarding "ehavior in Xictator Wames. 65erican,Econo5ic,Review ]O. O5N-OO0.

23. 22
/ullK "rooks ". and Frederick "old S1995T. Preaching Matters; Replication and EQtension.
)ournal,of,Econo5ic,Le;avior,and,OrganiXation 2a S1T. 14N-149.
IsaacK R.M.K X. Mathieu and E.E. Zajac S1991T. Institutional Framing and Perceptions of
Fairness. Constitutional,Bolitical,Econo5# 2. N29-Na0.
JenniK garen E. and Weorge doewenstein S199aT. EQplaining the nIdentifiaJle pictim Effecto.
)ournal,of,Risk,and,Uncertaint# 14 SNT. 2N5-25a.
JohannessonK Magnus and "jörn Persson S2000T. Non-Reciprocal 0ltruism in Xictator
Wames. Econo5ics,Letters O9. 1Na-142.
gelmanK Steven S19]aT. |PuJlic Rhoice| and PuJlic Spirit. T;e,Bublic,4nterest ]a SSpringT.
]0-94.
geserK Rlaudia and Frans van Winden S2000T. Ronditional Rooperation and poluntary
RontriJutions to PuJlic Woods. Kcandinavian,)ournal,of,Econo5ics 102S1T. 2N-N9.
golmK Serge-Rhristophe S19aNT. La,t;[orie,de,la,r[ciprocit[. Paris; REPREM0P.
golmK Serge-Rhristophe S19]4T. La,Lonne,Econo5ieU,La,R[ciprocit[,G[n[rale. Paris; Presses
Universitaires de France.
golmK Serge-Rhristophe S2000aT. Introduction; -he Economics of ReciprocityK Wiving and
0ltruism. In; W{rard-paretK douis 0ndr{K Serge-Rhristophe golm and Jean Mercier ,thier
Seds.T. T;e,Econo5ics,of,Reciprocit#,,Giving,and,6ltruis5. /oundmills et al.; MacMillan
Press dtd. 1-44.
golmK Serge-Rhristophe S2000JT. -he -heory of Reciprocity. In; W{rard-paretK douis 0ndr{K
Serge-Rhristophe golm and Jean Mercier ,thier Seds.T. T;e,Econo5ics,of,Reciprocit#,
Giving,and,6ltruis5. /oundmills et al.; MacMillan Press dtd. 115-141.
daVearK Edward P. S2000T. Economic Imperialism. Puarterl#,)ournal,of,Econo5ics 115 S1T.
99-14O.
dedyardK John &. S1995TK PuJlic Woods; 0 Survey of EQperimental Research. In; John /.
gagel und 0lvin E. Roth SEds.T. Randbook,of,E?peri5ental,Econo5ics.,Princeton;
Princeton University PressK 111-194.
deviK Margret S19]]T. Of,Rule,and,Revenue. "erkeley; University of Ralifornia Press.
dindenJergK S. S1992T. 0n EQtended -heory of Institutions and Rontractual Xiscipline.
)ournal,of,4nstitutional,and,T;eoretical,Econo5ics 14]. 125-154.
dipfordK JodyK RoJert E. McRormick and RoJert X. -ollison S199NT. Preaching Matters.
)ournal,of,Econo5ic,Le;avior,and,OrganiXation 21 SNT. 2N5-250.
MaelK Fred and "lake E. 0shforth S1992T. 0lumni and -heir 0lma Mater; 0 Partial -est of
the Reformulated Model of &rganiVational Identification. )ournal,of,OrganiXational
Le;avior 1N. 10N-12N.
MarchK James W. and J.P. &lsen S1995T. Ge5ocratic,Governance. New ,ork; Free Press.
Marwell Werald and Ruth E. 0mes S19]1T. Economists Free RideK Xoes 0nyone Elsex
EQperiments on the Provision of PuJlic Woods Ip. )ournal,of,Bublic,Econo5ics 15. 295-
N10.
MaussK Marcel S1924T. Essai sur le Xon. Forme et Raison de doEchange dans les Soci{t{s
0rcha}Mues.,L6nn[e,Kociologi]ue 1. N0-1]O.

24. 2N
MuellerK Xennis R. S19]9T. Bublic,C;oice,44. RamJridge Ug; RamJridge University PressK
2nd ed.
MuellerK Xennis R. Sed.T S199aT. Berspectives,on,Bublic,C;oiceU,6,Randbook. RamJridgeK
New ,ork and MelJourne; RamJridge University Press.
&ckenfelsK 0Qel and Joachim Weimann S1999T. -ypes and Patterns; 0n EQperimental East-
West-Werman Romparison of Rooperation and Solidarity. )ournal,of,Bublic,Econo5ics a1.
2a5-2]a.
&ppK garl-Xieter S2001T. Rollective Political 0ction. 6nal#se,^,Kritik 2NS1T. 1-20.
&rtmannK 0ndreas and disa g. -ichy S1999T. Wender Xifferences in the daJoratory; Evidence
from Prisoneros Xilemma Wames. )ournal,of,Econo5ic,Le;avior,and,OrganiXation N9. N2-
NN9.
&stromK Elinor S199]T. 0 "ehavioral 0pproach to the Rational Rhoice -heory of Rollective
0ction. Presidential 0ddressK 0merican Political Science 0ssociation 199a. 65erican
Bolitical,Kcience,Review 92 S1T. 1-22.
wuattroneK W.0. and 0. -versky S19]]T. Rontrasting Rational and Psychological 0nalysis of
Political Rhoice. 65erican,Bolitical,Kcience,Review ]2 SNT. a19-aNO.
RaJinK Matthew S199NT. Incorporating Fairness into Wame -heory and Economics. 65erican
Econo5ic,Review ]1.12]1-1N02.
RossK deeK Xavid Wreene and Pamela /ouse S19aaT. -he mFalse Ronsensus Effectm; 0n
Egocentric "ias in Social Perception and 0ttriJution Processes. )ournal,of,E?peri5ental
Kocial,Bs#c;olog#,1NK 2a9-N01.
SahlinsK Marshall S19a0T. -he Spirit of the Wift; Une EQplication de -eQte. In; Ec;anges,et
Co55unications,,F[langes,offerts,_,Claude,L[vi9Ktrauss,_,loccasion,de,son,`ae
anniversaire. deiden; Mouton.
SchellingK -homas R. S19O]T. -he dife ,ou Save May "e ,our &wn. In; S. Rhase Sed.T.
Broble5s,in,Bublic,E?penditure,6nal#sis. WashingtonK XR; "rookings Institution. 12a-
1O2.
SimonK /erJert S199NT. 0ltruism and Economics. 65erican,Econo5ic,Review ]N S2T. 15O-
1O1.
SkinnerK Jonathan and Joel Slemrod S19]5T. 0n Economic Perspective on -aQ Evasion.
Hational,Ta?,)ournal N]; N45-N5N.
SlemrodK Joel Sed.T S1992T. W;#,Beople,Ba#,Ta?es.,Ta?,Co5pliance,and,Enforce5ent. 0nn
0rJor; University of Michigan Press.
SonnemansK JoepK 0rthur Schram and -eo &fferman S199]T. PuJlic Wood Provision and
PuJlic "ad Prevention; -he Effect of Framing. )ournal,of,Econo5ic,Le;avior,and
OrganiXation N4. 14N-1O1.
SonnemansK JoepK 0rthur Schram and -heo &fferman S1999T. Strategic "ehavior in PuJlic
Wood Wames; When Partners Xrift 0part. Econo5ics,Letters. O2S1T; N5-41.
StiglerK Weorge J. S19]4T. Economics - -he Imperial Sciencex Kcandinavian,)ournal,of
Econo5ics ]O; N01-N1N.
SugdenK RoJert S19]2T. &n the Economics of Philanthropy. Econo5ic,)ournal 92. N41-N50.

25. 24
SugdenK RoJert S19]4T. Reciprocity; -he Supply of PuJlic Woods -hrough poluntary
RontriJutions. Econo5ic,)ournal 94. aa2-a]a.
-urnerK John R. and /oward Wiles S19]1T. 4ntergroup,Le;avior. Rhicago; Rhicago University
Press.
pareseK Federico and Meir ,aish S2000T. -he Importance of "eing 0sked; -he Rescue of
Jews in NaVi Europe. Rationalit#,and,Kociet# 12 SNT; N0a-NN4.
WeimannK Joachim S1994T. Individual "ehavior in a Free Riding EQperiment. )ournal,of
Bublic,Econo5ics 54. 1]5-200.
WeisJrodK "urton S199]T. To,Brofit,or,Hot,to,Brofit.,T;e,Co55ercial,Transfor5ation,of,t;e
Honprofit,Kector. RamJrige; RamJridge University Press.

26. 25
Table 1: Summary Statistics
Variables Number of
Observations
Percentage of
student body
Percentage who contribute
to at least one Fund
Economists 13932 10.18 O2.]2
Non-Economists 122930 89.82 O9.2O
Theology 1NOa 1.00 aa.1]
Law 21aaa 15.91 O4.01
Medicine 15211 11.11 O5.]1
Veterinary Medicine NaO0 2.a5 5].4O
Arts Faculty O20]O 45.NO aN.1N
Natural Science 14524 10.O1 OO.99
Computer Science 4205 N.0a OO.]a
Pre-university economic knowledge 2N944 1a.49 O5.]N
Age, Mean (s.d.) 27.78 (7.97)
0ged Jelow 2O O55ON 4a.90 a1.09
0ge 2O-N0 NaN99 2a.NN ON.O2
0ge N1-N5 1aOaN 12.91 O5.a0
0ge NO-40 ]NOO O.11 O9.aa
0ged over 40 a]O1 5.a4 aO.92
Gender
Women O]4O] 50.0N O9.1O
Men O]N94 49.9a O].05
Nationality
Foreigner 15a]2 11.5N O2.91
Swiss 1210]0 ]].4a O9.N5
Number of semesters, Mean (s.d.) 10.4a S].21T
Period 1 Swinter semester 199]q99T 1950a 14.25 O4.15
Period 2 Ssummer semester 1999T 1]2N1 1N.N2 Oa.0a
Period N Swinter semester 1999q00T 200O0 14.OO O9.0O
Period 4 Ssummer semester 2000T 1]O50 1N.ON O9.10
Period 5 Swinter semester 2000q01T 20NN5 14.]O a0.24
Period O Ssummer semester 2001T 190a5 1N.94 O9.a]
Period a Swinter semester 2001q02T 21004 15.N5 a0.5O
Xata source; Rompiled from data provided Jy the accounting department of the University of Zurich.

27. 2O
Table 2: Contibution to two Social Funds, University of Zurich 1998-2001
0Jsolute Percent
RontriJution to Joth Funds Ssa.2T ]4KaO5 O1.9t
RontriJution to Foreigner Fund only SsNT 5949 4.N5t
RontriJution to doan Fund only Ss4.2T N1]4 2.NNt
No contriJution to either Funds 42K9O4 N1.N9t
-otal 1NOK]O2 100.00
Gata,source; Rompiled from data provided Jy the accounting department of the
University of Zurich.
Figure 1F >ontributions depending on number of semesters
)!
(!
'!
&!
%!
$!
#!
"!
!
" # $ % & ' ( ) * "! "" "# "$ "% "&
Number of semesters
+o Contribution Contribution to 5oreigner 5und Contribution to 9oan 5und Contribution to Both 5unds
Hotes; Students are shown up until their 15th semester. Eight semestersK including the eQamsK is the
norm. "ut 22t of the students study longer.
Gata,source; Rompiled from data provided Jy the accounting department of the University of Zurich.

28. 2a
! "!! D
>? pect a tion abou t o tehrs B ebhaCior
Figure 2F EJp ectat ions and Actual B ehavior
"!! D
!

29. 2]
Table 3: Knowledge about the Contribution of Others
• nXo your friends know aJout your contriJutionxo
0Jsolute Percent
NoK we do not know 25O] a].]a
,esK we do know O]] 21.1N
-otal N25O 100.00
• nXo you ever talk aJout the two Social Funds to your friendsxo
0Jsolute Percent
NoK we do not talk 24]] aO.N4
,esK we do talk aa1 2N.OO
-otal N259 100.00
Xata source; &wn survey 2000.
Figure 3F Effect of being asked
in different Mays
)!
(!
'!
&!
%!
$!
#!
"!
!
PeriodsNOear
Contibution to both
Contribution to at least
one

30. 29
Table 4: Contribution to the Social Funds
University of Zurich 199]-2001
Xichotomous dependent variaJle; mRontriJution to at least one fundm u 1
Model I
ProJit estimate
Model II
Ronditional fiQed effect logit
pariaJle Roefficient Z-value Marginal
effect
Roefficient Z-value
Freshmen -0.0O5rr -N.50N -2.Nt -0.2OOrr -4.]5]
Main stage 0.114rr 10.4N] 4.1t -0.101 -1.]2N
Ph.X. 0.01a 1.105 0.Ot -0.010 -0.0]2
dast semester -0.1]4rr -14.O92 -O.Ot -0.1]9rr -4.N2a
NumJer of semesters -0.04Nrr -24.N22 -1.5t -0.2a4rr -N.945
SNumJer of semestersT2 0.001rr 14.202 0.02t -0.001 -1.211
0ge 2O-N0 0.15a 1.N5a 0.Ot -0.094 -1.O15
0ge N1-N5 0.1]9rr 11.]]9 O.]t 0.019 0.1]O
0ge NO-40 0.N4arr 1O.45a 12.4t 0.0]a 0.O0]
0ge over 40 0.541rr 2N.050 19.4t 0.1a4 0.O42
Wender Sfemaleu1T -0.012 -1.49a -0.4t
Nationality Sforeigneru1T -0.10Nrr -].544 -N.at
Married Su1T 0.052rr N.a]0 1.9t 0.210 1.404
Period 2 Ssummer semester 1999T 0.0]2rr O.0N0 2.9t 0.4a9rr O.24a
Period N Swinter semester 1999q00T 0.142rr 10.]10 5.1t 0.]1Orr 5.]a4
Period 4 Ssummer semester 2000T 0.141rr 10.440 5.1t 0.9aarr 4.a49
Period 5 Swinter semester 2000q01T 0.1a9rr 1N.5Oa O.4t 1.229rr 4.529
Period O Ssummer semester 2001T 0.1O]rr 12.45O O.0t 1.NO0rr 4.00]
Ronstant 0.O0Orr 41.2]2
N 115o]5] N9K5]N
dog dikelihood -a1041.a]N -14]11.]5]
dR chi2 S1OT u 1]0.09
Hotes; Reference group consists of mnon-economistsmK mJasic studymK mwithout pre-university economic knowledgemK maged
Jelow 2OmK mmalemK mSwissmK msemester 199]q99m.
Level,of,significance; r 0.01~p~0.05K rr p~0.01
Gata,source; Rompiled from data provided Jy the accounting department of the University of Zurich.

31. N0
Table 5: Contribution of Swiss and Foreigners to the Two Social Funds
University of Zurich 199]-2001
Swiss Foreigners -otal
RontriJution to Joth Funds Ssa.2T ON.2]t 51.O0t O1.9Nt
SNuaOO22T S]14NT S]4aO5T
&nly to Foreigner Fund SsNT N.O1t 9.9at 4.N5t
S4NaOT S15aNT S5949T
&nly to doan Fund Ss4.2T 2.45t 1.N5t 2.NNt
S29a1T S21NT SN1]4T
No contriJution to the Funds N0.O5t Na.09t N1.N9t
SNa111T S5]5NT S429O4T
-otal 100t 100t 100t
S1210]0T S15a]2T S1NO]O2T
HotesU,Pearson Rhi2 SNTu 1]O1.O411
Gata,source; Rompiled from data provided Jy the accounting department of the University of
Zurich.

32. N1
Table 6: Contribution of Students of Different Faculties in the First Semester
University of Zurich 199]-2001
Xichotomous dependent variaJle; mRontriJution to at least one fundm u 1
ProJit estimates
pariaJle Roefficient Z-value Marginal effect
Economics -0.24Orr -5.ON4 -].0t
Romputer Science -0.145r -2.1O9 -4.at
-heology -0.NNar -2.0a0 -11.0t
daw -0.229rr -O.092 -a.4t
Medicine 0.004 0.0a9 0.1t
peterinary medicine -0.100 -1.241 -N.2t
Natural science -0.19arr -4.2]1 -O.4t
Control,variables
0ge 2O-N0 -0.0NN -0.5]5 -1.1t
0ge N1-N5 0.0ON 0.O50 2.1t
0ge NO-40 0.2aO 1.]a4 9.0t
0ged over 40 0.2O1 1.a02 ].5t
Wender Sfemaleu1T -0.11Orr -4.15N -N.]t
Nationality Sforeigneru1T -0.02O -0.510 -0.]t
Married Su1T 0.01a 0.1OO 0.Ot
Period 2 Ssummer semester 1999T 0.1a1 1.512 5.Ot
Period N Swinter semester 1999q00T 0.N]9rr 10.2]9 12.Ot
Period 4 Ssummer semester 2000T 0.15] 1.N]N 5.1t
Period 5 Swinter semester 2000q01T 0.N9Orr 10.NN9 12.9t
Period O Ssummer semester 2001T 0.N4Orr 2.O92 11.Nt
Period a Swinter semester 2001q02T 0.NN2rr ].9aa 10.]t
Ronstant -0.50Nrr -14.1a]
N 10K5]4
dog dikelihood -O0O2.4Na9
Hotes; Reference group consists of m0rts facultymK maged Jelow 2OmK mmalemK nunmarriedoK
mSwissmK msemester 199]q99m.
Level,of,significance; r 0.01~p~0.05K rr p~0.01
Gata,source; Rompiled from data provided Jy the accounting department of the University of
Zurich.

33. N2
Table 7: Indoctrination Effect of Different Faculties
University of Zurich 199]-2001
Xichotomous dependent variaJle; mRontriJution to at least one fundm u 1
Ronditional fiQed effects logit
pariaJle Roefficient Z-value
Semesters in Economics -0.01a -0.95]
Semesters in Romputer Science -0.00a -0.201
Semesters in -heology 0.12Or 2.0]]
Semesters in daw -0.05Orr -N.OO2
Semesters in Medicine -0.0a]rr -4.a1]
Semesters in peterinary Medicine -0.0]5r -2.4a9
Semesters in Natural Science 0.01O 0.992
Freshman -0.2OOrr -5.412
Main stage 0.11ar 2.N99
Ph.X. stage 0.05a 0.4a5
0ge 0.014 0.NO9
NumJer of Semesters -0.1aNrr -N.]OO
Married 0.0]4 0.OOO
Period 2 Ssummer semester 1999T 0.NaNrr 5.4N5
Period N Swinter semester 1999q00T 0.ON1rr O.2N0
Period 4 Ssummer semester 2000T 0.O92rr 4.424
Period 5 Swinter semester 2000q01T 0.]a2rr 4.501
Period O Ssummer semester 2001T 0.909rr N.ONO
Period a Swinter semester 2001q02T 1.040rr N.O11
N 49505
dog dikelihood -1]KO11.O9]
dR chi2S19T 229.14
Hotes; Reference group consists of m0rts facultymK maged Jelow 2OmK nunmarriedoK msemester
199]q99m.
Level,of,significance; r 0.01~p~0.05K rr p~0.01
Gata,source; Rompiled from data provided Jy the accounting department of the University of Zurich.