Public Choice 91: 27^7, 1997. 27
@ 1991 Kluwer Academic Publishers. Printed in the Netherlands.
Do people care about democracy? An experiment exploring the
value of voting rights *
WERNER GUTH' & HANNELORE WECK-HANNEMANN"
^ Humboldt-University of Berlin, Spandauer Strafie I. D-10178 Berlin, Gennany: ^Institute
for Public Finance. University of Innsbruck, Henog-Friedrich-Strafie 3, A-6020, Innsbruck,
Austria
Abstract. In almost all democratic national elections an individual vote cannot change the
election outcome. The fact that many individuals nevertheless participate voluntarily in such
elections suggests that people do care about democracy as such. This experiment investigates
the value of democratic voting rights by providing participants the chance to sell them. More
specifically, an incentive compatible mechanism is used to elicit the willingness-to-accept
value of the voting right in the election of the German Bundestag on 16 October 1994. A
postexperimental questionnaire makes it possible to assess the relative importance of answers
to the frequently raised question: Why do people vote?
1. Introduction
In democratic national elections an individual vote matters only in highly
unlikely situations. It can be said that it is more likely to win the jackpot of
the German state lottery than to become once the decisive voter in a lifetime
or to be hit in a thunderstorm than to change the voting results. Here we do not
u'ant to engage in a discussion of how one compares two such unlikely events.
In our view, it is an obvious and therefore an accepted fact that individual
voters almost surely cannot alter an election result regardless of whether they
vote or not and how they vote.
In public choice theory it is assumed that individuals as voters act rationally,
i.e., by their voting behavior they try to maximise their own benefits.' Given
the fact that voting means to invest some effort (gathering and evaluating
information, going to the polling booth or requesting an absentee ballot for
postal vote), the extremely low probability of one's own vote being decisive
• We would like to express our thanks to Matthias von Bodecker, Manfred Konigsteln, and
Lutz Hildebrand {all Humboldl-LJniversity of Berlin) fcr their help in carrying out the experiment,
and to Hanno Ernst (Humboldt-University of Berlin), Monika Scheiber and Franziska
Taferner (both University cf Innsbruck) for computing assistance. The helpful comments of
Axel Borsch-Supan, Bruno S. Frey, Hartmut Kliemt, Jan P. Krahnen, Michael O'Brien, Arthur
Schram, Martin Weber and an anonymous referee are gratefully acknowledged.
28
makes voting apparently irrational. Because of this conclusion it is often
asked^ why people vote, and one tries to give reasons.
Our emphasis in this paper is not to discuss these arguments theoretically,
but to address the problem directly by attempting to measure the values
which people assign to their voting rights. More specifically, we report on
an experiment undertaken before the national election of the German Bundestag
in 1994 in which participants had the opportunity to sell their voting
rights for possibly substantial amounts. One usually requires replicability in
experimental research. Although one no longer can sell the voting right for
the election in 1994, analoguous studies could be performed for any election
with given voting rights. Since the experiment was based on an incentive
compatible mechanism (Becker, de Groot and Marshak, 1963), the price bid
attained can be interpreted as the trtie (willingness-to-accept) value for one's
voting right. Usually the willingness-to-pay value is strictly smaller than the
willingness-to-accept value (Romer, 1991). This is mostly attributed to an
endowment effect which, however, would be difficult to detect in our experiment.
Using a follow-up (postexperimental) questionnaire asking for reasons
to vote and not to vote in addition to some personal characteristics, it is possible
to assess the importance of the various reasons for why people vote or
not and compare these to the individual values for being able to vote.
The results are striking: Most of our participants did not want to sell their
voting right even at the top price of DM 200 (German marks) which had been
offered. Only a surprisingly small minority of participants (2.8%) would have
sold their voting right for any positive price. That is, only very few people
view their individual voting power as inessential as suggested by the "why
do people vote?" paradox. Another intermediate group (of about one quarter
of our 174 participants) was willing to sell their voting right for substantial
prices, but refused to do so at very low prices.
In our view, such results can be used to explain the different participation
rates in different elections if one interprets non-willingness to sell one's voting
right as willingness to vote. There is a small group of voters who hardly ever
vote since they consider it to be inessential or worthless. Others only vote
when the opportunity costs of voting are not prohibitive or if the election
results are especially interesting like the outcome of some sports events. The
majority of voters, however, considers it as their (moral) duty to participate
in democratic elections. The answers ofthe postexperimental questionnaire
provide further clues as to why most voters decide to vote and never consider
not to vote.
Voters who care about their voting right although their vote is very unlikely
to matter must view voting in democratic elections as a moral duty and/or
attribute some intrinsic value to democratic voting acts. They do not behave
29
opportunistically! But then the question arises: Can one, like in public choice
theory, assume opportunistic behavior when analysing which party a voter
will select? Since a voter will hardly argue morally when deciding to vote
and opportunistically when choosing a party, our experimental results seem
to question the fundamental approach of public choice theory.
The value of voting rights can also be detected by comparing the market
prices of cotnmon stocks and so-called preferred stocks (without voting rights)
if the latter ones exist. Although preferred stocks have higher and/or more
certain dividend payments, their market prices are more often below those of
common stocks than otherwise.^ Assuming that an individual shareholder is
nearly as insignificant as an individual voter in democratic (national) elections,
the prices of preferred stocks should clearly exceed those of common stocks.
Compared to voting in democratic elections, one, furthermore, would not
expect that not caring for one's voting power is morally objectionable. The
fact that preferred stocks are nevertheless usually cheaper than common
stocks thus indicates that people strongly care for their voting right (even if
they do not actively use it, but rely on banks etc.). In our view, at least small
shareholders must either consider a take over as rather likely (in case of a
takeover only common stocks with voting power are strategically relevant) or
they must suffer from a control illusion which typically results if the individual
voter thinks that others decide as he or she does ("if I do this, others will do
the same")In
the following we first describe the experimental design (Section 2) and
comment on the experimental results (Section 3). We then try to explain the
individual bids, i.e., the revealed values of voting rights, by referring to the reasons
(not) to vote which are brought forward in the literature (Section 4), and
by analysing the answers to the postexperimental questionnaire (Section 5).
The conclusion (Section 6) contains an evaluation and some reservations to
our inquiry.
2. Experimental design
The experiment was performed on 11 October, i.e., 5 days before the national
election of the German Parliament (Deutscher Bundestag) on 16 October
1994. It took place during thefirstlecture of a basic course in microeconomics,
i.e., participants were "unspoilt" by economic theory, but biased by their
common inclination to study economics. After some introductory remarks
conceming the intentions and the contents of such a course, students were
asked to participate in an experiment which required strict independence and
therefore no public discussion. Whoever did not want to obey these rules was
asked to leave the lecture room (30 to 40 students left without knowing more
30
about the experiment except for a remark on the course announcement asking
students to bring their voting cards to the first lecture).
The remaining 176 students were told that all questions were to be answered
only privately. We then distributed the instruction sheet and the decision
form.'* Both contained the code number identifying the individual participant.
Only one participant left the lecture room complaining (privately) about the
experiment. Five other participants, who it seemed would have been able to
fulfill our requirements (they did not indicate otherwise), did not participate
properly by returning an empty decision form.
Most questions revealed some difficulties in understanding the incentive
compatible sales mechanism: The price was determined by randomly selecting
one price out of an urn containing 201 integer prices ranging from DM 0,
DM 1,DM2,. ..to the maximal price of DM 200.^ After the price was drawn,
only those participants whose bid did not exceed the actual price were eligible
to sell their voting right. Among these, four were to be selected by chance
to actually be paid in reward for destroying their voting card themselves and
leaving their identity card over the election weekend. Actually this did not
rule out all legal possibilities to vote. No participant, however, mentioned
this. Apparently all our participants thought that voting without a voting card
is impossible or prohibitively troublesome.
It should be clear that it is an undominated strategy to submit a bid at which
one is (due to the integer restriction, nearly) indifferent between selling and
not selling one's voting right. A few students asked whether bidding DM 201
would suffice to exclude any risk of having to sell one's voting right. Our
standard answer was that any bid above DM 200 will do.
After reading the instructions and filling in the decision form, a task which
required about 15 minutes, the decision forms were collected and the postexperimental
questionnaires^ were distributed. While students answered the
questionnaire, the experimenters drew the price (with a student monitor) and
singled out those decision forms with bids which did not exceed the chosen
price of DM 112. Four of these were then selected by chance.
After collecting the postexperimental questionnaire, we publicly announced
the code numbers of the four participants who were actually selected for
selling their voting right. We did not announce the price which they were
to receive. Altogether we needed 40 minutes for the experiment. The actual
payment was made later and not by one of the two authors, in order to avoid
any fear of a bad reputation for the participants. One of the four did not show
up at all (this participant stated a weird, non-integer price between DM 0
and DM 1, claimed to be neither male nor female and should have been
excluded rightaway; "it" is the only participant who has been excluded from
our data set). Another "seller" suddenly reconsidered his decision when being
31
asked to destroy his voting card. He commented that "having to do it makes
me doubtful whether it is okay" and disappeared without the money. The
two remaining sellers collected the DM 112 and destroyed their voting cards
without any comments.
We expected quite a number of foreign students who were not able to
sell their voting right and we wanted to avoid any discriminatory action. An
overhead transparency - visible to all participants - announced that one could
tum in the voting card later (but before the election) and that those who
cannot vote or show up with their voting card in time are able to participate,
but required to indicate by "N" that they could not be elected for actual
sale. Twenty-eight participants stated "N". Actually we cannot exclude the
possibility that some ofthe participants who morally objected to selling their
voting right preferred to state this by "N". Twenty-three of the 28 N = 1participants
could be regular citizens of the Federal Republic of Germany:
according to question 2 of the postexperimental questionnaire, they grew up
in Gennany. It is rather unlikely that all of them have no German citizenship.
3. Experimental results
The individual bids of the 142 N = 0-participants, i.e., those who potentially
could sell their voting right, are graphically illustrated in Figure I. As already
mentioned, five decision forms were retumed empty: they contained neither
a bid nor an "N", indicating that they could not sell a voting right. Of the
142 N = 0-participants only 52 were in principle willing to sell their voting
right. Many of these 52 potential sellers were clearly guided by prominence
considerations (seven chose DM 50, five DM 100, five DM 150, and six
DM 200). Not all the 14 participants who required at least DM 190, but did
not completely rule out a sale, may have understood the sales mechanism. At
least half of them, however, answered the control question (the last question
of the postexperimental questionnaire) in a reasonable way.
The support for the public choice argument that individual votes do not
matter and are therefore worthless is surprisingly small: only seven of the
142 participants made a bid below DM 10. Our participants were mainly
unspoilt by public choice theory as revealed by question 1 of the postexperimental
questionnaire which asked whether one had already studied the
analysis of social decision-making. One-hundred-and-thirty-eight out of 174
cases replied negatively; 31 N = 0-cases and five N = 1-cases answered
affirmatively.^
Among the 142 N = 0-participants, 90 (that is, roughly 63% of this group)
did not want to sell their voting right (at least not for DM 200). This majority
of all the 142 N = 0-participants is in line with the even higher share of voters
32
c
a
u
n
1
M
40
30
10
.0
0
JH25.00
50
T.l.OO 11S,00
M 100,00 IJO
[39 1
175.00 115.00 300,00 300.00 700/)0 900,00
M 200,00 150.00 W0,00 COO,DD UO.OO 99V,O0
Bld> (is Cknnu WDU'DM)
IS,00 75,00 111,00 L7I,00 121,00 }O0,00 500.00 700.00 900,00
50,00 100,00 150.00 200,00 ua,oo 400,00 too.oo 100,00 999.00
Bid! lin dmun'MiiIla' DM)
Figure 1.
always participating in national elections. In the Federal Republic of Germany
the participation rate in national elections has always exceeded 77%. For most
of our participants the reasons to vote were clearly outweighed by the reasons
for not voting. In addition, four further N = 0-participants did not submit a
bid at all. Ofthe 28 N = 1-participants 13 also refused to specify a numerical
value for their voting right. Other eight N = 1 -participants bid above DM 200
and seven were in principle willing to sell.
It is an unintended result that so many of the 28 N = 1-participants did not
submit a bid. They only were asked to state an "N" next to their decision. The
instructions to do this were illustrated on the overhead transparency visible
to all participants.
"N DM (only integer DM-amounts, please)"
33
Thirteen of the N = l-participanls returned their forms with an "N" but
no money value. Either they misunderstood the instructions as an order to
leave the decision box empty or they viewed this as an easy way out of the
"dilemma" of having to specify a numerical value for their voting right.
The bids of the 142 N = 0-participants can be used to interpret actual
participation rates in national elections. A majority of voters will participate
whenever the costs of voting are reasonably low. Therefore, the rate of 63%
of those, who refused the idea to sell their voting right, seems quite realistic.
A very small minority of voters will hardly ever vote and can easily be
seduced to abstain from voting. The proportion of about 4% of the N = 0participants
bidding below DM 10 appears very small. It could be that students
of economics are more interested in political elections than usual voters. As
far as we know, however, there is no empirical proof for such a conjecture.
The 45 N = 0-participants who are willing to give up their voting right if they
receive something substantial in retum (e.g., an amount of at least DM 10)
could explain the variation in the voter participation of national elections (the
lowest participation rates in the Federal Republic of Germany were 84.3% in
1987 and 77.8% in 1990, the highest was 91.1% in 1972). Apparently such
voters participate if the opportunity costs of voting are negligible, but they do
not mind (at all) to abstain from voting otherwise, e.g., when voting would
mean to give up some other exciting opportunity.
4. On the reasons (not) to vote
With our incentive compatible procedure of buying the voting rights from our
participants, we leam only the range of prices p (with 0 < p < b,) at which
participants i with bid bj want to keep their voting power as well as the range
of the prices p (with bj < p < DM 200) at which they are willing to sell their
voting right. The first interval is, of course, empty in case of bj = 0, and the
second interval, when b, > DM 200.
By interpreting bi as participant i's tme (willingness-to-accept) value v, of
his or her voting right, one assumes that participant i understood and accepted
the reason why it does not pay to over- or underbid his or her tme value Vj in
the sense of b, ^ v,. Unfortunately, this cannot always be expected. Although
incentive compatibility is a very compelling normative principle, inexperienced
participants do not always understand it, even when they are told why
bidding tmthfully (bi=Vi) is optimal. According to the control question of our
postexperimental questionnaire (question Fl 1 in Appendix A), where the true
value was exogenously predetermined (at DM 52.51), only 34.7% of those
participants actually stating a bid (and 29.1% of all 174 participants) bid
between DM 50 and DM 55. Here we do not want to discuss the differences
34
between normative and behavioral incentive compatibility. The fact that in
total only 41 of our 174 participants hid exactly DM 52.51 as required by
incentive compatibility proves that normative incentive compatibility does
not guarantee truthful bidding.
In the following we want to discuss the essential reasons why a participant's
true value v, of his or her voting right is high or low, and to weigh thenn hy
the relative frequencies of such arguments in answering question 4 of the
postexperimental questionnaire (Appendix A). The typical reason given by
public choice theorists why v, should be close to 0 is the extremely low
probability that one's own vote is decisive.^ Among the 12 participants,
however, who denied question 4 ("are you willing to participate in political
elections?") only one chose the justification that the own vote would not
change the result.
Besides the low probability justification for not voting ("my vote won't
matter"), the costs of voting offer another justification. These costs include
the necessary preparations (one needs the voting card or, as in the USA,
is not even registered automatically by the authorities but has to take the
initiative oneself). The most important costs of voting are, of course, the
opportunity costs in terms of lost working or leisure time. Among those
who denied question 4 of the questionnaire nobody indicated that voting is
uncomfortable, i.e., requires too much effort.
Since in economics one usually assumes given preferences, most studies
neglect the costs of constructing preferences, i.e., in the case at hand the costs
of making up one's mind which party one should vote for. From a behavioral
point of view, preferences over parties or coalitions are not given but the result
of some cognitive and sometimes uncomfortable processes. According to the
psychological theory of cognitive dissonance (Festinger, 1957; Akerlof and
Dickens, 1982), a clear-cut preference for one party is usually the result of
reevaluating, reinterpreting or even neglecting facts. By deciding not to vote,
all these mental efforts can be avoided. Question 4 was denied only once
for the reason that one does not know for which party/which politician one
should vote.
The reasons why one should vote are the familiar ethical and philosophical
justifications.^ If nobody would vote, the "my vote won't matter" justification
can no longer be applied. Thus, the only general recommendation in the sense
of Kant's (1785) ethical principle is to vote.'^ One could describe non-voters
as freeriders who enjoy the benefits of living in a democracy without accepting
their democratic responsibilities. Among the 162 participants who answered
question 4 affirmatively, 92 gave the reason that voting is a civil duty to
preserve democracy.
35
Reasons of more behavioral sense also have to rely on psychological concepts.
Participating in democratic elections certainly awakens political interests
and thereby all the fun and inspirations induced by more or less active
participation in political debates. Forty of the affirmative answers to question
4 were justified by indicating that voting keeps one's political interests
awake. One also may enjoy the act of voting as such (most individuals are
rather proud when they are allowed to vote for the first time) regardless of
the election result. Seven of the affirmative answers reasoned that it is fun to
vote and 83 argued that by voting one reveals one's political views.
The rational voter theory as applied in the literature on political support
functions" somehow neglects the low probability argument for not voting
and claims that people decide in public elections as in economic situations. In
other words: a voter chooses that party or coalition whose program or expected
policy is best for him or herself personally. According to such a view, parties
are nothing more than competing firms trying to satisfy the voters' demands
(given a perfectly competitive political system). That one votes to serve one's
personal interests has been argued by 53 of the 162 affirmative answers of
question 4 in the postexperimental questionnaire.
To justify such an approach one has to argue why the low probability argument
does not apply. One way would be that most voters' subjective probabilities
for being decisive are rather unrelated to the objective probabilities.'^
In the minimax-regret model of Ferejohn and Fiorina (1974, 1975) voting is
interpreted as a decision under uncertainty, i.e., voters are unable to estimate
the probabilities of the altemative outcomes. Thirty-six of the affirmative
answers of question 4 were supported by the argument that one's own vote is
decisive.
Others view voting not exclusively instrumental to determine the winning
party or coalition, but as a private consumption act from which benefits
accrue independent of the outcome of the election.'^ We saw no easy way
to distinguish such reasons for voting in spite ofthe "my vote won't matter"
argument from the familiar ethical and philosophical justifications mentioned
above. If a norm is internalized or socially controlled, norm-conforming
behavior often appears as individually rational.
Since the answers to the postexperimental questionnaire did not influence
the monetary expectations, one may view the weights for the traditional arguments
why people (do not) vote as unreliable. Another objection may be that
out of an ego-defensive attitude participants answered the postexperimental
questionnaire in a way which justified their revealed valuejudgement for their
voting right. None of these objections applies to the actual decisions, i.e., the
revealed values. The bids, however, only tell us how participants evaluate
their voting rights. As has been elaborated above, such an evaluation can be
36
determined by many "pros" and "cons" whose relative importance can be
hardly assessed if one just observes the bids. If, however, the bid is positive,
the reasons to vote seem to dominate those for not voting. If one even does
not sell at all possible prices, this dominance appears to be strikingly robust.
Thus the actual decision data may not allow the assessment of the importance
of the individual "pros" and "cons", but they definitely allow one to weigh
the importance of all "pros" against the one of all "cons".
5. An attempt to explain the value of voting rights
In the following, nevertheless, we will use the answers to the postexperimental
questionnaire as well as the additional information given on the decision form
(age, number of semesters of studying, subject of study, and sex) to try to
explain the bids, i.e., the revealed values of voting rights.
For the 90 N = 0-participants who would not even sell their voting right at
the maximal possible price of DM 200, we cannot explain the exact value, but
only argue why it is higher than those of the 52 remaining N = 0-participants
who submitted a bid. In Table 1 we compare the share of potential sellers
with that of non-sellers for the seven specific reasons for voting as well as for
the six specific reasons for not voting.
Although 52 participants were in principle willing to sell their voting right,
only eight (and other four N = 0-participants) denied question 4, i.e., claimed
not to be interested in actively taking part in political elections. Thus only
eight participants have exposed themselves as habitual non-voters for whom
the experiment provided a nice opportunity to get paid for something they
would have done anyhow (even though, according to question 5, only four
of them stated that they actually intend not to participate in the next national
election). The remaining 44 potential sellers may be described as voters who,
in principle, want to vote, but who can be distracted from doing so by offering a
rather small positive incentive. The eight N = 0-participants of Table I justified
their propensity for non-voting predominately with other reasons, i.e., with
the lack of time (1), the lack of a real alternative (1), and their opposition
to the political system in general (2). It is surprising, however, that the most
frequent reason for non-voting (3) is that one cannot make up one's mind, i.e.,
the cognitive dissonance related to voting decisions. Observe that the public
choice justification "my vote won't matter" is given only once. It appears that
(social) psychologists might be able to better explain the variance in voting
participation than public choice theorists.
Those of the N = 0-participants who (according to Table 1) vote in principle
gave lots of reasons (97 reasons by potential sellers and 241 by non-sellers).
The number of "other reasons" amounts to 14. Many of these other reasons
37
Table I. Reasons to vote and not to vote by the 142 N = 0-participants
F4
1
No. of panicipanis who are decided
- to participate in political elections
- not to padicipale in elections
Rea.wns to vote:
F4,l
F4.2
F4,3
F4,4
F4,5
F4,6
F4.7
F4,8
Total
for revealing my political views
my vote is decisive
il keeps my political interests awake
il is fun to vote
it series my personal interests
I'm supporter/opponent of a party/policy
it is a civil duty to preserve democracy
other reasons
Rea.sons not to vote:
F4,9
F4,IO
F4.11
F4,12
F4.13
F4.14
F4.15
Total
my vote would nol change anything
I don't know for which party/politician to vote
it is too uncomfortable
1 don't care for politicians
1 don't know how to serve my
personal views/interests best
I'm not familiar with the party programs
other reasons
Type of voters
Potential sellers
(n=52)
44
8
18
9
11
2
15
18
21
3
97
1
1
0
0
3
1
6
12
Non-sellers
(n=90)
90
0
51
22
19
3
30
48
57
11
24t
0
0
0
Q
0
Q
0
0
Both
(n=142)
134
8
69
31
30
5
45
66
78
14
338
1
1
0
0
3
1
6
12
are, however, at least partly covered by those which we offered for justifying
an affirmative answer of question 4. Most subjects viewed voting as civil duty
(78) or as a possibility to state one's opinion (69). Frequent reasons are also
that one is a supporter or an opponent of a specific party or policy (66) or that
one is closely affiliated with a certain political party or line of politics (45).
The shape of the distributions for potential sellers and non-sellers is surprisingly
similar. A binomial test for the various reasons to vote (questions
F4.1 to F4.8) reveals a statistically significant difference only in case of questions
F4.I, F4.6 and F4.7. The non-sellers consider it to be more important
that voting enables them to reveal their political views (including support or
38
opposition to a specific party or policy), and that voting is a civil duty to
preserve democracy.
A similar comparison for the remaining questions of the postexperimental
questionnaire as well as for the information in the decision form (age, number
of semesters, subject of study, sex) is contained in Table 2. In the group of all
142 N = O-participants there is no apparent gender effect nor a statistically
significant difference in average age and length of study of the groups of
potential sellers and non-sellers. According to a binomial test, the subject of
study of the political sellers is, however, different from that one of non-sellers
who have less students of business administration.
According to Table 2, most students did not engage in studies of social
decision-making or public choice. There were slightly more East Germans
in the sample of potential sellers than in the sample of non-sellers. Surprisingly,
the latter effect does not go along with a different income distribution.
Most participants (135) indicated their willingness to vote in the forthcoming
national election, whereby, as expected, the non-voters mainly belong to the
group of potential sellers. Rather few (57 altogether) revealed a close affiliation
to a political party. Question 7 ofthe postexperimental questionnaire
revealed a surprisingly high number of strategic voters (86), whereas only 54
participants stated that they would vote for their favourite party. Few (32) participants
viewed the campaigns by political parties as inspiring participation
in elections.
The remaining answers of Table 2 were related to the experiment: 91
participants considered it as morally objectionable to sell one's voting right
(50 did not); 65 found it difficult to specify a numerical monetary value for
one's voting right (75 did not). Not surprisingly, the non-sellers had more
moral objections than those who actually offered a bid up to DM 200, but
they had less difficulties in specifying a numerical monetary value.'''
For the 52 participants with bids not exceeding DM 200 we tried to explain
the variance of bids by dummy variables representing the various reasons
(not) to vote specified in question 4 of the postexperimental questionnaire
as well as by socio-economic dummies, e.g., growing up in former East
or in former West Germany and income. We ran Tobit-regressions with all
142 N = O-participants'^ and the dependent variable, the individual bid v,,
being modelled as right-censored variable (with DM 201 representing the
right-censored value). In addition, we calculated OLS-estimates for the 52
potential sellers who had stated an accurate bid. Altogether, several aspects
tum out to be statistically significant in determining the individual bids (the
regression results are given in Appendix B): First, the bid is smaller, when
participants have more experience with the analysis of social decision-making.
Second, the intention not to vote in the forthcoming national election or in
39
Table 2. Answers to the postexperimental questionnaire and additional information, given by
the 142 N = 0-participants
Questions
Age (average)
Number of semesler (average)
Subject of study - business economics
- political economy
- otherwise
Sex - female
- male
Fi: already been concemed with the analysis
of social decision-making - yes
-no
F2: grown up ~ in old provinces of Germany
- in new provinces of Gemiany
-other
F3: amount of money at one's disposal
- less than DM 800
-between DM 800and DM 1200
-morethan DM 1200
F5: intention (o vote in the next national election - yes
-no
F6: closely connected to a political party - yes
- n o
F7: vote for favourite party - yes
- n o
F8: inspiration by election campaigns - yes
-no
F9: considering it as immoral to sell ballot-paper -yes
-no
FIO: difficulties to state an accurate DM-amount -yes
-no
Type of voters
Potential sellers
(n=52)
21.4
1.23
34
9
9
19
33
12
40
21
30
1
31
14
7
46
6
17
35
20
31
15
37
25
27
36
16
Non-sellers
(n=90)
22.1
1.30
53
27
10
33
SJ
14
n46
43
1
42
»
1ft
t»
1
40
50
34
55
17
73
66
24
29
61
Both
(n=142)
21.8
1.27
87
36
19
52
26
IM
67
73
2
73
43
3S
mf
SJ
85
54
86
32
no
91
51
65
77
elections generally lowers the individual bid in a statistically significant way.
An especially pronounced impact among the various reasons for voting is that
"it is a civil duty to preserve democracy" (F4.7). Stating a bid in line with the
true value (between DM 50 and DM 55) in the control question (FI 1) in the
postexperimental questionnaire has a positive, but only for the 52 potential
sellers statistically significatit impact on the willingness-to-accept value. The
Tobit-estimates suggest that those among the 142 N = O-participants who
40
took the view that it is immoral to sell one's ballot-paper (ceteris paribus)
are more inclined to state a higher bid. In addition, these estimates reveal that
those who admitted that they had difficulties to state an accurate DM-amount
not surprisingly are those with a lower bid, i.e., they seem to be the potential
sellers who actually had to state an accurate bid.'^
6. Conclusion
Although we refer to our empirical study (how people evaluate their voting
rights) as an experiment, this is somewhat debatable. After all, participants
were asked to state a monetary value for their actual voting right which they
(seemingly) gave up by tearing their voting card into pieces. Furthermore,
we do not know what political consequences the individuals expected when
they abstained from voting. There wer3 very likely none, although voters
might rely on an illusion of control. Because of these aspects our study
differs fundamentally from other experimental studies of voting behavior
(e.g., Schram and Sonnemans, 1994, 1995). These may have a better control
of most relevant variables (like voting costs and voting benefits) but cannot
rely on actual democratic voting rights.'^
With our experiment we mainly wanted to investigate the prices at which
voters are indifferent between selling and non-selling their voting right in a
specific and seemingly interesting election. As cleariy revealed by the control
question (FI 1) of our postexperimental questionnaire, one cannot be sure that
all our participants completely understood that bidding truthfully is optimal.
However, we leamed from later discussions with our participants, as well as
with those who assisted in the experiment, that it was seemingly more difficult
to understand and answer the control question than the rules conceming the
actual decision.
The bids which from a normative point of view should reveal the true
values for one's voting right, show a strikingly strong non-willingness to sell
one's voting right: A strong majority would not even sell at the maximal
price of DM 200 which had been offered. Furthermore, the group of voters
who hardly assign any positive value to their voting right is ciose to being
negligible. Thus the famous public choice question "why do people vote?"
can be answered in a roundabout way by stating: "The individual right to vote
appears to be something precious". Some decision forms contained additional
comments like "one cannot/should not sell one's voting right", other bids were
outrageously large numbers showing that for some participants the voting
right is a personal value which they would not easily give up for something
less personal, e.g., money, in exchange.
41
It is, of course, a more demanding task to explore the specific reasons why
people vote (or not) and apparently care for their voting right. Whereas the
bids only tell us whether the reasons to vote dominate those for non-voting,
the answers of the postexperimental questionnaire, especially question 4,
indicate the relative importance ofthe main arguments as discussed in Section
4 above. Here we do not want to repeat these results, but briefly discuss how to
weigh the "soft" data ofthe postexperimental questionnaire versus the "hard"
decision data - where "soft", respectively "hard", means that no or substantial
monetary incentives are involved.
In our view, the questionnaire data are less reliable than the decision data and
were answered with less care. This was clearly revealed by the answers to the
control question (Fl 1), where fewer participants required further explanation
than for the actual decision. Thus the questionnaire results can only provide
a preliminary idea whether or not the various arguments to vote or not to vote
matter. There seems to be, however, no easy experiment to elicit the relative
importance of these arguments and whose observations are all based on - due
to monetary incentives - relevant choices. At least for us the combination of
"hard" decision data and "soft" questionnaire results seemed to be the most
fruitful combination.
One may object to our results and say that we relied on a highly special
sample by relying on beginning students enrolled in a basic course of microeconomics.
Although we do not expect dramatically different results, one
may perform a similar experiment with a different sample, but hopefully an
equally interesting election. Our sample had at least two advantages: Students
were not yet spoilt by economic theory (it was their first lecture in economics)
and they were not very experienced with the details of national elections (only
for those who were 22 years old or older it was the second national election in
which they had the opportunity to actively take part; for citizens of Berlin and
the eastern part of Germany this was not the case in former national elections,
anyway).
Actually, the experimenters themselves when designing the experiment
and most students when participating thought that without a voting card it
is impossible to vote. Other voters might have been more familiar with the
legal rules - as we are now. To rule out a voting act it does not suffice to
destroy one's voting card. In addition, one must keep the personal identity
card and the passport over the election weekend. Even then one may not be
sure whether the vote had already been submitted by mail (here, of course,
the participants would have lied, e.g., by falsely claiming that (s)he has lost
the voting card). This illustrates that a completely cheatproof experiment is
hard to perform and that our experiment seems to be an easy and reasonable
compromise.
42
Often when discussing this experiment with colleagues, we confronted
comments like: "you cannot run such an experiment", "this is illegal", or "I
would not do it". In our view, legal violations have been avoided by asking
the sellers themselves to destroy their voting card. As actually occurred, a
potential seller had many opportunities to give up his or her plans, e.g., by
leaving the lecture room before the experiment, by indicating an "N" on the
decision form, by stating a bid greater than DM 200 or no bid at all, by not
collecting the money. Furthermore, we could have never enforced such a sales
contract legally.
When planning the experiment we were more worried that some participants
would make public statements, probably by morally condemning the
experiment, and thereby question the independence of individual bids. To
take care of that we made the announcement that we need independent observations
and have therefore to rule out any public comments. By allowing
those who did not accept such an obligation to leave (30 to 40 students left),
we were surprisingly successful in performing a "clean" experiment in spite
of the many (altogether 175) active participants. Apparently all remaining
students by their voluntary decision to participate, felt extremely committed
to obeying the rules.
The experimental study reported above is unusual since the usual reservation
against experimental results - that they do not reflect real life behavior
- does not apply. We have explored real life behavior (our participants were
mainly actual voters, they had to destroy their "real" voting card, etc.). Thus
there remains only one reservation, namely that we have relied on a biased
sample of student beginners. This, however, is no general restriction of our
experimental procedure. Our experiment could be repeated with any group
of participants where one only should take care to explain the optimality of
truthful bidding in an appropriate way.
Notes
1. Theraiional voter theory was first developed by Downs (1957), TuUock (1967), and Riker
and Ordeshook (1968), and it is summarized by Mueller (1989), Stmthers and Young
(1989), Schram (1991) and Kirchgassner (1992).
2. See, for instance, Schram and van Windcn( 1994) and Mueller {1989: Ch. 18).
3. See, for instance, the dala summarized by Weber, Berg and Kruse (1992), and Knise, Berg
and Weber (1993).
4. The detailed formulation in German language (or in English translation) is available from
the authors upon request.
5. If one uses discrete instead of continuous prices, the incentive compatibility is weaker: If
the true value Vj happens to be one of the discrete prices, i.e., Vi = k DM with 0 < k < 2(K),
both bids bi, i.e., the truthful bid bj = k DM and the next higher bid bi = (k+I) DM,
are optimal since they imply different implications only for the price p = Vi which yields
0-profit.
43
6. The questions are given in English translation in Appendix A. Moreover, ihe detailed
(German) formulation of the postexperimental questionnaire is available from the authors
upon request.
7. It seems possible that these participants did not study social decision-making as such, but
thought thai they did so since they are members of some political party.
8. See, for instance, Mueller (1989: Ch. 18), and Struthers and Young (1989).
9. See, for instance, Riker and Ordeshook (1968), or Barzel and Silberberg (1973).
10. Harsanyi (1979), for instance, distinguishes norm from act utilitarism, and Downs (1957)
views voting as an insurance against a breakdown of democracies.
11. See, for a recent survey, Nannestad and Paldam (1994).
12. For a discussion, see Struthers and Young (1989), and a related argument referring to
self-deception and diagnostic voting is given by Quattrone and Tversky (1985).
13. See Stigler (1972) and also, e.g., Guttman, Hilger and Shachmurove (1994) for the distinction
between voting as consumption and voting as investment. Fiorina (1976) refers to
the benefits as "expressive" and "instrumental", respectively. See also the arguments given
in the theory of low-cost decisions by Kliemt (1986) and Kirchgas.sner(1992).
14. In effect, 39 indicated a very high monetary amount, even up to infinity, and all these
values are illustrated in Figure I by including the value 999.
15. In fact, the data basis was reduced to only 133 observations due to missing values in the
individual answers, i.e., the exogenous (dummy) variables.
16. The likelihood to sell or not sell one's voting right was explained also in a Logit-regression
for all 142 N = O-participants. Overall, these estimates support the results of the Tobit-
regressions.
17. One especially interesting aspect ofthe studies by Schram and Sonnemans (1994, 1995),
which is largely neglected in other discussions, is the focas on the fact that most democratic
elections can be viewed as a combination of intragroup (among party members and/or
voters favoring a party) and intergroup (party) competition. It seems that voters are more
likely to vote and therefore to care for their democratic voting right when parties are in a
very close and tough race for winning the election.
References
Akerlof, G. A. and Dickens, W.T. (1982). The economic consequences of cognitive dissonanee.
American Economic Review 12(iune): 307-319.
Barzel, V. and Sitberberg, E. (1973). Is the act of voting rational? Public Choice 16(Fall):
51-58.
Becker, G., de Groot, M.H. and Marshak, J. (1963). An experimental study of some stochastic
models for wagers. Behavioral Science 8: 199-202.
Downs, A. (1957). An economic theory of democracy. New York: Harper & Row.
Ferejohn, J.A. and Fiorina, M.P. (1974). The paradox of not voting: A decision-theoretic
analysis. American Political Science Review 68: 525-536.
Ferejohn, J.A. and Fiorina, M.P. (1975). Closeness counts only in horseshoes and dancing.
American Political Science Review 69: 920-925.
Festinger, L. (1957). A theory of cognitive dissonance. Evanslon, IL: Roar.
Fiorina, M.P. (1976). The voting decision: Instrumental and expressive aspects. Journal of
Politics 3S: 39a~4]3.
Guttman, J.M., Hilger, N. and Shachmurove, Y. (1994). Voting as investment vs. voting as
consumption: New evidence. KyklosAl: 197-207.
Harsanyi, J.C. (1979). Bayesian decision theory, rule utilitarism, and Arrow's impossibility
theorem. Theory and Decision 11: 289-317.
44
Kant, I. (1785). Groundwork of the metaphysics of morals. Translated and analysed by P.J.
Paton. New York: Harper & Row, 1964.
Kirchgassner, G. (1992). Towards a theory of low-cost decisions. European Journal ofPolitical
Economy 8: 305-320.
Kliemt, H. (1986). The veil of insignificance. European Journal of Political Economy 2/3:
333-344.
Kruse, H., Berg, E. and Weber, M. (1993). Erklaren untemehmensspezifische Faktoren
den Kursuntcrschied von Stamm- und Vorzugsaktien? Zeitschrift fur Bankrecht und
Bankwirtschaft 5: 23-31.
Mueller, D.C. (1989). Public choice II. 2nd ed. Cambridge, MA: Cambridge University Press.
Nannestad, P and Paldam, M. (1994). The VP-function: A survey ofthe literature on vote and
popularity after 25 years. Public Choice 79: 213-245.
Quattrone, G.A. and Tversky, A. (1985), Self-deception and the voter's illusion. In J. Elster
(Ed.), The multiple self, 35-58. Cambridge, MA: Cambridge University Press.
Riker, W.H. and Ordeshook, P.C. (1968). A theory ofthe calculus of voting. American Political
Science Review 62: 25-42.
Romer, A.U. (1991). Der kontingente Bewertungsansatz: Einegeeignete Methode zurBewertung
umweltverbessernder MaBnahmen? ZeitschriftfUr Uniwelipolitik und Umweltrechi 4:
41
Schram, A.J.H.C. (1991). Voter behavior in economic perspective. Berlin: Springer-Verlag.
Schram, A. and Sonnemans, J. (1996a). Voter tumout as a participation game: An experimental
investigation. International Journal of Game Theory 25: 385^06.
Schram, A. and Sonnemans, J. (1996b). Why people vote: Experimental evidence. Journal of
Economic Psychology 17: 417-442.
Schram, A. and van Winden, F. (1994). Why people vote: The role of inter- and intragroup
interaction in the tumout decision. In H. Brandstatter and W. Guth (Eds.), Essays on
Economic Psychology. 213-250. Berlin: Springer-Verlag.
Stigler, G. (1972). Economic competition and political competition. Public Choice 13(Fall):
91-106.
Struthers, J. and Young, A. (1989), Economics of voting: Theories and evidence. Journal of
Economic Studies 16/5: 1-42.
Tultock. G. (1967). Toward matliematics ofpolitics. Ann Arbor: University of Michigan Press.
Weber, M., Berg, E. and Kruse, H. (1992). Kurs- und Renditevergleich von Stamm- und
Vorzugsaktien - Eine empidsche Analyse. ZeitschriftJur betriebswirtschaftliche Forschung
44/6: 548-565.
45
Appendix A: Questions (Fi) of the postexpertmental questionnaire
Fl; "Have you already been concemed in your studies with the analysis of social decisionmaking?"
(yes / no)
F2: "Where have you grown up, in Ihe old or new provinces of Germany?" (old (FRG) / new
(GDR) / other)
F3; "Which ainouni of money you have at your disposal on average monthly?" (less than
DM 800/between DM 800and DM 1200/more than DM 1200)
F4: "Are you decided on participating in political elections actively?" (yes / no) i
Reasonsfor participating in political elections:
F4,1: - "for revealing my political views"
F4.2: - "my vote is decisive"
F4.3: - "it keeps my political interests awake"
F4.4: -"it is fun to vote"
F4.5: - "serving my personal interests"
F4.6: - "I'm supporter/opponent of a specific party/policy"
F4.7: - "it is a civil duty to preserve democracy"
F4,8: - other reasons.
Rea.wnsfor not participating in political elections:
F4.9: - "my vote would not change anything"
F4.10: - "1 don't know for which party/politician 1 should vote"
F4.11: - "it is too uncomfortable"
F4.12: -"1 don't care for politicians"
F4.13: - "1 don't know how to serve my personal views/interesLs best"
F4.14: - "I'm not familiar with the programs of the political parties"
F4.15: -other reasons
F5: "Do you ititend to vote in the next national election (provided you will not be one of the
four participants who will sell their voting right)?" (yes / no)
F6: "Are you closely connected to a political party?" (yes / no)
F7: "Will you ever vote for the party with which you are closely connected?" (yes / no)
P8: "Do you believe that you are inspired to participate in the vote by election campaigns of
political parties?" (yes / no)
F9: "Do you consider it as immoral to sell your ballot-paper?" (yes / no)
FIO "Did you have difficulties to state an accurate DM-amount as your minimal selling price?"
(yes / no)
Fll: individual bid (in DM) to the control question."
"Control question: "Gehen Sie davon aus, daB ein Gut besitzen, das fur Sie keinen eigetien
Wert besitzt. Sie konnen das Gut auf zwei Arten an uns verauBem. Einmal konnen Sie analog
zu den Regeln des vorherigen Experiments das Gut an tins verkaufen, d.h. der Preis wird per
Zufall aus einer Ume tnit alien Preisen von DM 0 bis DM 200 ausgewahlt und Sie mussen
den Betrag festlegen, ab dem Sie zum Verkauf bereit sind. Falls Sie das Gut nicht wie oben
beschrieben verkaufen, konnen Sie das Gut an uns abgeben zum Betrag von DM 52,51. Wir
sind daran inleressiert zu erfahren, ab welchem Belrag Sie bereit sind, das Gut an uns zu
verkaufen. - Ihre Antwort: Ich bin bereit, das Gut ab dem Betrag von DM . . . zu verkaufen".
46
a
>-§
o VI
> A • %
I
a
^ 9
rri ^.-' —
m
l/^ ^—^ V^ - - ^ ^ * . .1-1. O j . ^ ^ ^ . o
o c r o o c ~ - t ^ r ^ o i o o
IT •
^ o , . - , O i ^ ' p . , - , o . - ^ o o , - v ^ / > - - ^
oo r j WT. O. wi — r\ o t*i -^ ^ ^_
CTvoOOaomcsorSfS'iOoo
*^^ J ^^ ^ ""^^ *'.-' CS *—' ^ —< I
o\ o ^ ' o eT •** vi* o «> •*'
— ' I — - K — ' i w J ^ w | I
^ O ; --V oo ^
in f-j —^ K v\
o' oC o ^ ' <3
S. -a
E .a
; :5 s. K
a E •§ I .1
i ^
47
•o VI
— ^-, 00 ^-. 00 --. "n
<fl T^ r-; r-i^ Ml (N_ (S
CT> O oo — m (N Q\
•*-• cs — V —' I
o s
— I T,
" " fn V^ QO QO "^
£ d B S C 5
3 ^-. O ' - . — ^-. ^
O OS -" CTv (*1 t^
O r- o o — l / l > 0 - H O v C I 0 O ^
-.w." ^ -.w* op ^-^ ^^^ CT*
r i O cn* —w m —•
O <S O. (« t-.
3
a
"c
o
U
CQ
ii « o
o 2
t = I -si
" ^ ra u o ^ x!
^ 11^ o f^ n
•J T3
O <
H -J a:
the
o
c
>.
aratel
a.
u
o
P
aoi)
3
o-i «
:ant
y i :
callv
^ 1 - •
U- •« ^
i-s
•s ^
u
ue
•o
11
o5
O\
> o