A clear and present minority: heterogeneity in the source of endowments and the provision of public goods.
Oxoby, Robert J. ; Spraggon, John
I. INTRODUCTION
There is extensive evidence that heterogeneous populations invest
less in social capital. For example, Alesina and Ferrara (2000) find
that participation in social activities is lower in racially and
ethnically diverse areas. This in turn may result in lower levels of
public goods provision (Alesina, Baqir, and Easterly 1999) and lower per
student educational spending (Poterba 1997). Moreover, heterogeneity
(racial or otherwise) may increase the number of social categories and
thereby reduce the extent to which individuals identify with one
another. As discussed by Akerlof and Kranton (2000, 2005), the lack of
identification with one another may result in individuals engaging in
less cooperation and displaying greater self-interest (Eckel and
Grossman 2005; Van Vugt and De Cremer 1999). On the level of policy
implementation, Luttmer (2001) suggests that these group differences
reduce individuals' incentives to support redistribution. This
research suggests that heterogeneity matters in reducing the effects of
social preferences and the incentives to invest in social capital.
But what exactly is the source of this heterogeneity? On the one
hand, Alesina, Glaeser, and Sacerdote (2001) and Lee and Roemer (2004)
propose that the source of this heterogeneity is racial: The presence of
a clearly identifiable racial minority reduces individuals'
incentives to engage in redistribution. Moreover, Alesina and Ferrara
(2000) and Knack and Keefer (1997) find that social capital is lowest in
those U.S. states and countries characterized by greater racially
heterogeneous populations (in which the group of interest is a racial
minority). On the other hand, we propose that one source of this
heterogeneity is the difference in individuals' efforts and hence
the source of (and perceived property rights over) their resources.
While this second dimension of heterogeneity may be strongly correlated
with racial diversity, it has strikingly different policy implications.
It is this latter form of heterogeneity which emphasizes the importance
of understanding and providing proper incentives to individuals. For
example, Oxoby (2004) suggests that individuals living in poverty may
adhere to different sets of norms which are related to the subculture of
the poor or the underclass. These norms may include withdrawal from the
labor market or welfare dependency. These differences in norms provide
an opportunity to reduce this heterogeneity which may be correlated with
racial lines but can be addressed through policies such as welfare
reform and work-fare programs which alter the extent to which these
individuals are viewed as minorities in terms of their deservingness to
public goods. Indeed, if unemployment is a dimension which defines
heterogeneity in terms of legitimacy to assets and access to public
goods, the Phelps (2000) argument for employment subsidies may not only
serve his intended purpose of reducing social exclusion, but also serve
to reduce the type of heterogeneity we identify as a cause of reductions
in the contribution to public goods. Thus, this form of heterogeneity
(i.e., differences in the legitimacy one has to one's assets or
differences in individuals' perceived deservingness to benefitting
from public goods) emphasizes basic policies such as reducing
unemployment as a means of increasing contributions to the public good.
Heterogeneity in the source of individuals' resources and the
presence of a minority among individuals with differing sources of
endowments is the focus of this paper. Specifically, we conduct public
goods games in which groups of participants differ with respect to the
source of their endowments: some individuals within the group earned
their endowments while others had their endowments allocated by the
experimenter. This type of heterogeneity, albeit on a small scale,
mirrors the type of heterogeneity described above, and may underlie the
way individuals invest in social capital and assess their support for
redistribution. Our results suggest that the presence of a clear
minority (in terms of endowment source) significantly reduces
contributions to public goods, leading many to behave in accord with the
Nash prediction based on traditional self-interested preferences. This
demonstrates the importance of heterogeneity in endowment sources in
focusing subjects on self-interested play, suggesting that social
preferences are muted when individuals differ in, say, perceived
property rights or perceived "deservingness."
This points to a second important aspect of our experiment. Just as
Cherry, Frykblom, and Shogren (2002) demonstrated the importance of
property rights to eliminate offers above the Nash equilibrium in the
dictator game, our experiment provides strong evidence of Nash behavior
in a linear public goods environment. Specifically, we observe 70% and
80% of participants choosing the minimal (Nash) strategy in some of our
treatments. Other experiments have argued that the non-Nash behavior is
reduced by heterogeneity in wealth levels (Buckley and Croson 2006; Chan
et al. 1996; Ledyard 1995; Isaac and Walker 1988; Zelmer 2003), the use
of taxation to crowd out contributions (Andreoni 1993; Chan et al.
2002), and providing participants with payoff tables (Charness,
Frechette, and Kagel 2004; Saijo and Nakamura 1995). While these
treatments have well-documented effects on reducing contributions in
public goods games, our results show that the effect of heterogeneity in
the source of endowments (and in particular, the presence of a minority
with a different endowment source than the majority) dwarfs these other
manipulations and dramatically increases Nash behavior. Interestingly,
two earlier studies (Cherry, Shogren, and Kroll 2005; Clark 2002)
address the issue of property rights in public goods games,
demonstrating that when all subjects earn their endowments (or bring
their own resources for use in an experiment) aggregate contributions
are just as high as when endowments are allocated by the experimenter.
We replicate this finding but show that the presence of a minority
significantly reduces contributions in the standard public goods
experiment. (1)
We proceed as follows: in Section II we describe our experiment.
Section III presents our results, focusing on the effects of
heterogeneity of endowment source on contributions. In Section IV, we
discuss the implications in light of recent literature on identity,
reciprocal altruism, and the debate on the relationship between
heterogeneity within an economy and the structure of redistributive
systems. Section V concludes.
II. EXPERIMENTAL DESIGN
Our experiment was designed to test the effect of heterogeneity in
endowment source on contributions in a one-shot, four-person, linear
public goods environment. (2) To do this we use a two-stage game. In the
first stage subjects are randomly assigned to either earn their
endowment or have it allocated randomly. In the second stage,
participants choose how much of their endowment [[omega].sub.i] [member
of] {100, 200, 300} to contribute to a public good (denominated in
laboratory dollars; 1 lab dollar = $0.10). To avoid biases for or
against the lower boundary, we restricted individuals'
contributions ci to a minimum 5% of their endowments. Individuals'
payoffs were given by
(1) [[pi].sub.i] = ([[omega].sub.i] - [c.sub.i]) + 0.4
([4.summation over (j=1)] [c.sub.j]).
where [c.sub.i] [member of] [0.05 x [[omega].sub.i],
[[omega].sub.i]]. As such, the dominant strategy Nash equilibrium in
this game is for each participant to contribute [c.sub.i] = 0.05 x
[[omega].sub.i]. The marginal per capita return here is 40%. This is
consistent with the conditional cooperation paper of Fehr and
Fischbacher (2002) among others.
Prior to individuals playing the above public goods games, we
introduced our treatments wherein we altered the sources of
individuals' endowments in the game. We conducted five treatments.
Of these, we conducted four earnings treatments in which either one,
two, three, or four participants in each group earned their wealth via a
15-minute, 12 question exam culled from the GMAT and GRE. This yields
our one-, two-, three-, and four-earner treatments. Endowments for those
who earned their endowments (earners) were based on the following
scoring scheme: if the participant answered fewer than six questions
correctly, she received an endowment of 100; if she answered between six
and nine questions correctly, she received 200; if she answered ten or
more questions correctly, she received 300. The remaining group members
(non-earners) were randomly assigned wealth levels of 100, 200, or 300.
In our final treatment (our endowed (control) treatment) individuals
were randomly assigned an endowment of either 100, 200, or 300 lab
dollars. Within groups in the endowed treatment, assignments were made
such that the distribution of wealth within a group was either (100,
100, 200, 300) or (100, 200, 200, 300). This treatment mirrored other
experiments in which participants' wealth was determined by the
experimenter (e.g., Buckley and Croson 2006). Our goal here was to
introduce heterogeneity in the source of these endowments (i.e., earned
or allocated by the experimenter). In addition to receiving a
description of the payoff function (Equation 1), individuals were also
given a payoff table indicating their payoff for each level of private
contribution and a level of aggregate contributions by others. (3)
Within this design, our hypotheses concern the effect of
heterogeneity in endowment source on contributions in the public goods
game. Previous experiments on dictator and ultimatum games (e.g., Cherry
2001; Cherry, Frykblom, and Shogren 2002; Hoffman and Spitzer 1985;
Oxoby and Spraggon 2008; Ruffle 1998) have demonstrated that a source of
one's endowment can have profound effects on the way in which
moneys are allocated in ultimatum and dictator games. Perhaps most
striking are the experiments of Cherry, Frykblom, and Shogren (2002) in
which 98% of dictators who had earned their endowments followed the Nash
prediction. Germane to our analysis, one interpretation of this result
is that endowments were allocated not by chance (i.e., random assignment
of roles and wealth levels) but by one participant (the dictator)
exerting effort on an exam, thereby introducing an additional dimension
of heterogeneity between the dictator and the receiver. (4)
In our public goods game, we expect a similar result: interpreting
the linear public goods game as a game of redistribution (toward a
socially optimal outcome), heterogeneity of endowment source alters the
manner in which one views others as deserving of redistribution. Just as
dictators in Cherry, Frykblom, and Shogren (2002) and Oxoby and Spraggon
(2008) who earned their wealth revealed their recognition of their
property rights, a similar recognition should occur among earners in our
public game. This should manifest itself in lower contributions among
earners. Anticipating this, non-earners should also contribute less
based on theories of conditional cooperation (Fischbacher, Gachter, and
Fehr 2001; Keser and van Winden 2000). Heterogeneity of endowment source
should therefore reduce aggregate contributions. (5)
Hypothesis 1: When endowments are heterogeneous in their source
(earned versus allocated), individuals will contribute less to the
public good than when endowments are homogeneous in their source.
In addition, the distribution of this heterogeneity in endowment
source may matter in how individuals behave in this environment. For
example, one interpretation of the aforementioned research on the
broader effects of heterogeneity (Alesina and Ferrara 2000; Alesina,
Glaeser, and Sacerdote 2001; Knack and Keefer 1997; Lee and Roemer 2004)
is that regions in which there are clearly identifiable minorities
exhibit lower levels of redistribution and (global) public goods
provision: when there is a clear minority who differs from the majority
in an identifiable way, attention turns to the exclusion of this group
through reductions in redistributive policies. This attention focuses on
the behaviors (e.g., prevalence of welfare dependency) or the
circumstances (e.g., long-term unemployment) of the minority and how
these relate to the deservingness of individuals for global public goods
and redistribution. Although these reductions in public goods may reduce
the welfare of individuals in the majority, it is the attention to the
minority (e.g., racism as discussed in Alesina, Glaeser, and Sacerdote
2001; Lee and Roemer 2004) which appears to drive majoritarian behavior.
In support of these ideas, Deffuant, Huet, and Amblard (2005) and others
find that the presence of a small minority yields the greatest
polarization effects while the presence of larger minorities creates the
need and incentives for moderation and incorporation. (6) As such, we
have the following hypothesis:
Hypothesis 2: When endowments are heterogeneous in their source
(earned versus allocated), individuals will contribute less when there
is a clear minority whose endowment source differs from that of the
majority.
Following Hypothesis II, we expect to observe lower contributions
in our one- and three-earner groups as these groups have the clearest
minorities (one individual who earned her endowment and one individual
who was allocated an endowment). On the other hand, while there is
heterogeneity in the two-earner groups, these groups have no strict
minority. As such, the polarization effects of heterogeneity are
softened in these groups (Deffuant, Huet, and Amblard 2005). Viewed in
the context of identity, the one- and three-earner groups are noteworthy
as in each there is at least one type of individual (an earner or
non-earner) for who there is no other individual with whom she can
identify. This individual may therefore opt not to contribute, reducing
the contributions of all individuals through their attention to norms of
conditional cooperation. Such an effect is muted and non-existent in the
two-earner, four-earner, and endowed groups.
III. ANALYSIS OF RESULTS
Three-hundred and sixteen participants (79 four-person groups) were
recruited from the undergraduate student body at the University of
Calgary. We have 16 groups in each of the control (in which all
participants received their endowments), one-, two- and three-earner
treatments (where either one, two, or three participants in the group
earned their endowments). We have 15 groups in the four-earner treatment
(where all four participants were randomly selected to earn their
endowment). The experiments were conducted in the university's
experimental economics laboratory and programmed in z-Tree (Fischbacher
2007).
In what follows we discuss the results of these sessions. Recall
that all of the sessions were "one-shot" games and so we have
316 independent observations. Fundamentally, we are interested in the
effect of heterogeneity in property rights: if the distribution of
wealth (by source) is inequitable (in the sense that some subjects
earned their stake while others were simply allocated resources) does
the cooperation that we observe in the standard public goods game
disappear?
Because wealth levels differ both between individuals and groups we
focus on percent aggregate contribution (the total contribution of the
group divided by the group's wealth ([[summation].sup.4.sub.j=1]]
[c.sub.j]/[[summation].sup.4.sub.j=1] [[omega].sub.j])) at the group
level and percent contribution [c.sub.j]/[[omega].sub.j] at the
individual level.
Table 1 presents percentage aggregate contributions, percentage
contributions, and the percentage of contributions which are equal to
the Nash prediction (within 1%) by treatment. (7) The results in the
endowment and four-earner treatments (30% contributions and 30% of
subjects choosing the self-interested Nash decision) are remarkably
consistent with the literature on public goods experiments (Clark 2002;
Cherry, Shogren, and Kroll 2005; Ledyard 1995; Zelmer 2003), while the
results for the one- and three-earner treatments are consistent with
self-interested Nash payoff maximization. Although these results are not
entirely consistent with Hypothesis II, they are consistent with our
Hypothesis II. We observe much lower contributions and more Nash play
among the groups with either one earner or three earners than we do in
the other groups. Specifically, notice that the groups in which either a
single member either earned or was allocated her endowment (one- or
three-earner treatments) have the lowest (aggregate and percentage)
contributions and the highest percentage of Nash play.
Figure 1 presents the percentage contribution and 95% confidence
intervals for each treatment. The obvious "W" shape of this
graph highlights the importance of there being a clear minority in
reducing the contributions of all players. Note that while we observe a
reduction in contributions with heterogeneity (i.e., from the
homogeneous groups to the two-earner groups), there is yet a greater
reduction in groups with a clear minority. This is also observable in
Figure 2, the cumulative and density of percent contributions. Notice
the difference in the amount of Nash play (percentage of subjects
contributing 5% of their endowments) between the one- and three-earner
treatments relative to the other treatments.
In analyzing these results note that, as in Cherry, Shogren, and
Kroll (2005) and Clark (2002), we identify no difference in
contributions between the control treatment and the four-earner groups.
(8) In these groups, the homogeneity of endowment source among
participants resulted in play similar to that observed in other linear
public goods experiments. On the other hand, we observe marked
contribution differences in the one- and three-earner treatments in
which we observe 73% and 84% support for the Nash prediction (i.e., 5%
contribution). We attribute this to the existence of a clear minority
within these groups. Regression results for aggregate decision making
are presented in the second column of Table 3. (9) The positive and
significant constant suggests that contributions are in the 27% range
for our baseline treatment where all subjects were provided with an
endowment. That the coefficient for the four-earner treatment is
insignificant is again consistent with the Cherry, Shogren, and Kroll
(2005) result--when everyone in the group earns their endowment the
results in public goods experiments are no different than when they are
all given their endowment. This regression also does not indicate a
significant difference in contributions between the earner and
two-earner treatments. It does, however, support our contention that
contributions are significantly less among groups with a clear minority:
the one- and three-earner groups.
[FIGURE 1 OMITTED]
[FIGURE 2 OMITTED]
In Table 2, we break contributions down by role (earner or
non-earner). Notice that in all cases earners contribute more and are
less likely to choose the minimal (Nash) contribution than non-earners.
Also notice that in the one-earner treatment the earners still
contribute 20% of their endowment while in the three-earner treatment
they contribute much less (13%). We use the regression results in Table
3 to determine whether or not the differences in contributions between
the treatments and roles are significant. As we are dealing with percent
contribution, our data is truncated at 5% and 100% of wealth
contributed. To address this issue, we follow Greene (2003, 773-80) and
conduct ordinary least squares and Tobit analysis (allowing for
censoring at both 5% and 100%) controlling for the different earner
treatments, whether the subject had earned her endowment crossed with
the earning treatment, and the subjects percentage of total wealth. We
also employ regression analysis to determine whether there are
significant differences in the percent of Nash play. To do this, we use
a probit model (Greene 2003, 812-18).
At the individual level (for both percent contribution and
percentage Nash), the regressions include the treatment dummy variables,
the one-, two-, and three-earner treatment dummies interacted with a
dummy variable for whether or not the individual was an earner, as well
as the individual's wealth as a percentage of the group's
total wealth (percent wealth). (10) Notice in Table 2 that earners (in
the four-earners treatment) contribute more (35% as opposed to 29%) but
are also more likely (30% as opposed to 20%) to make the minimum
contribution when compared to the non-earner baseline. The regression
results, in columns 3 and 4 of Table 3, suggest that neither of these
differences is statistically significant. Thus we conclude (again
consistently with Cherry, Shogren, and Kroll 2005) that endowment source
or property rights do not affect the results of public goods experiments
when they are homogeneous. Next let's turn to our other treatment
which seems consistent with this conclusion--the two-earner treatment.
Here we observe (Table 2), that non-earners contribute much less (16%)
compared to the baseline and are much more likely (50%) to make their
minimum contribution. Earners on the other hand contribute about the
same (30%) and seem just a little more likely to make their minimum
contribution (35%). The regression analysis, Table 3, suggests that the
contribution differences are significant for non-earners but not for
earners, (11) and that the likelihood of non-earners making Nash
contributions is significantly higher. The likelihood of earners making
Nash contributions is not significantly different from nonearners in
this treatment. For the one-earner treatment we again observe a
significant reduction in contributions for the non-earners and a
significant increase in their likelihood to make the minimal
contribution. For earners in this treatment we observe no significant
differences in behavior from the non-earners in this treatment. Finally
consider the three-earner treatment. Again, contributions for
non-earners are significantly lower, and the likelihood of making
minimal contributions is significantly higher. For earners, the story is
different, they contribute significantly more than non-earners (but less
than those in the baseline (p = .000)), and there is no significant
difference in the effect on their likelihood of making the minimal
contribution as compared to non-earners.
In summary, we observe that heterogeneity in endowment source
always results in significant reductions in contributions and increasing
likelihood of making minimal contributions among those who are given
their endowment. For those who earn their endowment we observe an
identical result in terms of the likelihood of making Nash
contributions. However, earners' contributions are higher (and
significantly higher in the two- and three-earner treatments) than the
non-earners they are matched with. This can be seen in Figure 3 which
shows percentage contribution by role (earner or non-earner) and
treatment. Notice that the contributions of both earners and non-earners
are consistent with the "W" shape observed in Figure
1--contributions are closest to Nash for the one- and three-earner
treatments.
IV. DISCUSSION
An interpretation of our environment is as a general case public
goods experiment of which the standard public goods game is a special
case (i.e., a case in which property rights are homogeneous).
Interpreted in this way, our results help delineate cases in which one
would expect to observe Nash contributions. If one conjectures that
non-Nash behavior in standard public goods games results from
other-regarding behavior based on outcomes, our results suggest that
heterogeneity in property rights trumps these concerns. (12) Moreover,
if one thinks of reciprocity (manifest as conditional cooperation) as
the driving force in non-Nash contributions, our results suggest that
heterogeneity in property rights significantly alters the ways in which
individuals behave. Strikingly, it is the presence of a minority which
appears to drive down individuals' contributions, regardless of how
this minority's endowment was determined.
One reason for these changes in behavior may be changes in the
perceptions of identity between participants. As discussed by Akerlof
and Kranton (2000, 2005), identity within a group of individuals can
serve as a motivator for behavior, facilitating contracting and
cooperation. Similarly, the absence of identity or threats against
one's identity may provide incentives to "act out,"
display an "out-group bias," or engage in otherwise
unacceptable behaviors. Within our environment, identity (or the absence
thereof) may encourage (or inhibit) contributions.
While research in social psychology has demonstrated that it is
seemingly easy to experimentally "create" identity among
individuals (Rabbie and Horowitz 1969; Turner 1982; Van Vugt and De
Cremer 1999; Wilder and Shapiro 1984), the effects of such identity
appear to be small. In economic experiments (i.e., experiments with
salient rewards), it appears to be more difficult to create identity
which motivates significant increases in cooperation and contributions
to public goods. For example, Solow and Kirkwood (2002) and Wit and
Wilke (1992) find that individuals in stranger treatments contributed
less than those in treatments in which identity was fostered. However,
these increases were always relatively small (less than 10% increases in
all cases). In a more comprehensive study, Eckel and Grossman (2005)
find that artificially motivated team identity (e.g., the simple
assignment of a participant to a team) does not affect cooperation.
However (as suggested by Akerlof and Kranton 2005) cooperation does
increase (by about 10%) when team identification is motivated by having
team members work together on a prior goal.
[FIGURE 3 OMITTED]
In our experiment, we observe the presence of a minority in
endowment source yielding reduced cooperation (i.e., contributions). In
some sense, we can consider the homogeneous groups as having more
consistent (albeit perhaps weak) identification than the heterogeneous
groups. Under this interpretation, heterogeneity in perceived property
rights appears to fracture group identity, resulting in significantly
less cooperative behavior. This is particularly strong in the one- and
three-earner groups in which there are no other individuals with whom a
member of the minority can identify with in terms of endowment source.
Thus, our results suggest that in economic environments with salient
incentives, the fracturing of identity appears to be easier and yield
larger behavioral changes than does the construction of identity.
Broadly speaking, the endowment heterogeneity in our environment is
observed generally in non-experimental economies (i.e., resources
created through an individual's efforts vs. resources received due
to luck or other reasons). As such, our research dovetails with previous
research on the relation between "heterogeneity" and
preferences regarding community participation and redistributive
policies. For example, Alesina and Ferrara (2000) use survey data on
participation in social activities to discern the effect of community
heterogeneity (in terms of income and ethnicity). Results indicate that
participation is lower in more heterogeneous communities. (13) Our
results provide further evidence of this tendency: as the degree of
heterogeneity in endowment source increased (being highest in our one-
and three-earner treatments) we observed lower "participation"
as manifest in contribution decisions.
Ethnic heterogeneity appears to also affect individuals'
tastes for redistribution. For example, Alesina, Glaeser, and Sacerdote
(2001) find a strong relationship between the degree of redistribution
and the size of an economy's poor ethnic minority. Relatedly,
Luttmer (2001) find that individuals' support for redistribution
decreases as the share of welfare recipients of their own racial group
declines. (14) One explanation for this evidence may be in the manner
ethnic heterogeneity weakens identity structures (i.e., creates more
social categories) and thereby reduces reciprocal altruism (Becket
1974). It is worth noting that in these studies, the racial group of
interest are always in minority status relative to the population under
study. The results from our experiment emphasize the importance of
understanding these effects not necessarily in terms of the racial
component per se, but rather the position of the racial component in the
broader society (i.e., its minority status).
With respect to redistribution, Bowles and Park (2005) and Gilens
(1999) argue that support for redistribution is conditional on norms of
reciprocity and the obligations of others. As such, the heterogeneity of
income and wealth sources across an economy reduces individuals'
support for redistributive policies (which may tax those who exert
effort to earn income and subsidize those who do not). In our
experiments, we identify a similar pattern in which heterogeneity of
endowment source yields markedly lower contributions, demonstrating a
reduction in altruism (larger than Nash contributions) and a
corresponding response based on reciprocity. It is interesting to note
that all participants reduced their contributions in the presence of a
clear minority (i.e., our one- and three-earner groups). Thus, as
discussed in Luttmer (2001) and Lee and Roemer (2004), greater
heterogeneity and the presence of a minority reduced redistribution by
reducing the extent to which individuals positively account for the
payoffs and the deservingness (via intentions and ex ante behaviors) of
others.
V. CONCLUDING REMARKS
The extent to which governments exist to provide public goods
versus protect property rights is a fundamental question. Indeed this
Pigouvian-Coasian debate underlies much economic research and arises in
many policy debates. However, the importance of property rights is often
ignored in experimental investigations into the provision of public
goods. As a result, these studies often invoke altruism (or other
outcome-based motives) or conditional cooperation to explain the
observed deviations from the predictions of the self-interested model.
In this paper, we compare voluntary contributions across
environments with varying degrees of heterogeneity in endowment source
(and potentially perceived property rights). As in previous studies, we
observed non-Nash contributions in environments where there is ambiguity
regarding the allocation and legitimacy of resources (our non-earner
treatment). However, as we move to an environment with a clear minority
we observe a sharp reduction in contributions, suggesting that this
heterogeneity supplants altruistic motives and positive reciprocity,
thereby emphasizing self-interest. Most importantly, it appears that the
presence of a clear minority (particularly on differing in property
rights) trumps alternative preferences based on others' outcomes.
One should think of our experimental environment as one in which
the presence of a minority (in terms of property rights) confounds
individuals' social preferences. In essence, the presence of a
minority with a differing degree of asset legitimacy substantially
complicates individuals' determination of what is an
"equitable" outcome. These same factors are at play in regards
to more fundamental economic questions regarding the provision of local
public goods (e.g., debates over school choice) and policies of
redistribution (e.g., welfare reform). To this end, our experiment sheds
light on how heterogeneous property rights and the presence of a
minority affect other regarding preferences.
REFERENCES
Akerlof, G., and R. Kranton. "Identity and the Economics of
Organizations." Journal of Economic Perspectives, 19(1), 2005,
9-32.
Akerlof, G. A., and R. E. Kranton. "Economics and
Identity." Quarterly Journal of Economics 15(3), 2000, 715-58.
Alesina, A., R. Baqir, and W. Easterly. "Public Goods and
Ethnic Divisions." Quarterly Journal of Economies, 114(4), 1999,
1243-84.
Alesina, A., and E. Ferrara. "Participation in Heterogeneous
Communities." Quarterly Journal of Economics, 115(3), 2000,
847-904.
Alesina, A., E. Glaeser, and B. Sacerdote. "Why Doesn't
the U.S. Have a European Style Welfare State?" Brookings Papers on
Economic Activity, 2, 2001, 187-278.
Andreoni, J. "An Experimental Test of the Public Goods
Crowding-Out Hypothesis." American Economic Review, 83(5), 1993,
1317-27.
Becker, G., "A Theory of Social Interactions." Journal of
Political Economy, 82, 1974, 1063-93.
Bowles, S., and Y. Park. "Emulation, Inequality, and Work
Hours: Was Tborsten Veblen Right?" The Economic Journal 115(507),
2005, 397-412.
Buckley, E., and R. Croson. "Income and Wealth Heterogeneity
in the Voluntary Provision of Linear Public Goods." Journal of
Public Economics, 90(4-5), 2006, 935-55.
Chan, K. S., R. Godby, S. Mestelman, and R. A. Muller. "Spite,
Guilt and the Voluntary Provision of Public Goods When Income Is Not
Distributed Equally." Canadian Journal of Economics, 29(2), 1996,
S605-9.
--. "Crowding-Out Voluntary Contributions to Public
Goods." Journal of Economic Behavior and Organization, 48, 2002,
305-17.
Charness, G., G. R. Frechette, and J. H. Kagel. 2004. "How
Robust Is Laboratory Gift-Exchange?" Experimental Economics, 7(2),
189-205.
Cherry, T.L. "Mental Accounting and Other-Regarding Behavior:
Evidence from the Lab." Journal of Economic Psychology, 22, 2001,
605-15.
Cherry, T.L., P. Frykblom, and J.F. Shogren. "Hard-nose the
Dictator." American Economic Review. 92(4), 2002, 1218-21.
Cherry, T. L., J.F. Shogren, and S. Kroll. "The Impact of
Endowment Heterogeneity and Origin on Public Good Contributions:
Evidence from the Lab." Journal of Economic Behavior and
Organization, 57(3), 2005, 357-65.
Clark, J. "House Money Effects in Public Goods
Experiments." Experimental Economics, 5(3), 2002, 223-32.
Deffuant, G., S. Huet, and F. Amblard. "An Individual-Based
Model of Innovation Diffusion Mixing Social Value and Individual
Benefit." American Journal of Sociology, 110(4), 2005, 1041-69.
Eckel, C.C., and P.J. Grossman. "Managing Diversity by
Creating Team Identity." Journal of Economic Behavior and
Organization, 58(3), 2005, 371-92.
Fehr, E., and U. Fischbacher. "Why Social Preferences
Matter--The Impact of Non-Selfish Motives on Competition, Cooperation,
and Incentives." The Economic Journal, 112, 2002, C1-C33.
Fershtman, C., and U. Gneezy. "Discrimination in a Segmented
Society." Quarterly Journal of Economics, 116(1), 2001, 351-77.
Fischbacher, U. "z-tree: Zurich Toolbox for Ready-Made
Economic Experiments." Experimental Economies, 10(2), 2007, 171-78.
Fischbacher, U., S. Gachter, and E. Fehr. "Are People
Conditionally Cooperative? Evidence from a Public Goods
Experiment." Economics Letters, 71, 2001, 397-404.
Gilens, M. Why Americans Hate Welfare: Race, Media, and the
Politics of Antipoverty Policy. Chicago: University of Chicago Press,
1999.
Greene, W. H. Econometric Analysis. 5th ed. Upper Saddle River, NJ:
Prentice Hall, 2003.
Harrison, G.W. "House Money Effects in Public Good
Experiments: Comment." Experimental Economics, 10, 2007, 429-37.
Hoffman, E., K. McCabe, and V.L. Smith. "Preferences, Property
Rights, and Anonymity in Bargaining Games." Games and Economic
Behavior, 7, 1994, 346-80.
Hoffman, E., and M. Spitzer. "Entitlements, Rights, and
Fairness: An Experimental Examination of Subjects' Concepts of
Distributive Justice." Journal of Legal Studies, 14, 1985, 259-97.
Isaac, R., and J. Walker. "Group Size Effects in Public Goods
Provision: The Voluntary Contributions Mechanism." Quarterly
Journal of Economics, 103(1), 1988, 179-99.
Keser, C., and F. van Winden. "Conditional Cooperation and
Voluntary Contributions to Public Goods." Scandinavian Journal of
Economics. 102(1), 2000, 23-39.
Knack, S., and P. Keefer. "Does Social Capital Have an
Economic Payoff? A Cross-Country Investigation." Quarterly Journal
of Economics. 112(4). 1997, 1251-88.
Konow, J. "Fair Shares: Accountability and Cognitive
Dissonance in Allocation Decisions." American Economic Review,
90(4), 2000, 1072-91.
--. "Which Is the Fairest One of All? A Positive Analysis of
Justice Theories." Journal of Economic Literature. 41 (4), 2003,
1186-237.
Ledyard, J. O. "Public Goods: A Survey of Experimental
Research." in Handbook of Experimental Economics, edited by J. H.
Kagel, and A. E. Roth. Princeton, NJ: Princeton University Press, 1995.
Lee, W., and J. E. Roemer. "Racism and Redistribution in the
United States: A Solution to the Problem of American
Exceptionalism." Cowles Foundation Discussion Paper No. 1462, 2004.
Luttmer, E. "Group Loyalty and the Taste for
Redistribution." Journal of Political Economy, 109(3), 2001,
500-28.
Oxoby, R. J. "Cognitive Dissonance, Status, and Growth of the
Underclass." The Economic Journal. 114, 2004, 729-49.
Oxoby, R.J., and J. Spraggon. "Mine and Yours: Property Rights
in Dictator Games." Journal of Economic Behavior and Organization,
65, 2008, 703-13.
Phelps, E. S. "The Importance of Inclusion and the Power of
Job Subsidies to Increase It." O.E.C.D. Economic Studies, 0(31),
2000, 85-113.
Poterba, J. M. "Demographic Structure and the Political
Economy of Public Education." Journal of Public Policy and
Management, 16. 1997, 48-66.
Rabbie, J.M., and M. Horowitz. "Arousal of Ingroup-Outgroup
Bias by a Chance Win or Loss." Journal of Personality and Social
Psychology, 13(3). 1969, 269-77.
Ruffle, B. J. "More Is Better, but Fair Is Fair: Tipping in
Dictator and Ultimatum Games." Games and Economic Behavior, 23(2),
1998, 247-65.
Saijo, T., and H. Nakamura. "The "Spite" Dilemma in
Voluntary Contribution Mechanism Experiments." Journal of Conflict
Resolution, 39(3), 1995, 535.
Selten, R. "The Equity Principle in Economic Behavior,"
in Decision Theory and Social Ethics, edited by H. Gottinger and W.
Leinfellner. Dordrecht, Holland: D. Reidel, 1978, 289-301.
Solow, J. L., and N. Kirkwood. "Group Identity and Gender in
Public Goods Experiments." Journal of Economic Behavior and
Organization, 48, 2002, 403-12.
Turner, J.C. "Towards a Cognitive Redefinition of the Social
Group," in Social Identity and Intergroup Relations, edited by H.
Tajfel. Cambridge: Cambridge University Press, 1982.
Van Vugt, M., and D. De Cremer. "Leadership in Social
Dilemmas: The Effect of Group Identification on Collective Actions to
Provide Public Goods." Journal of Personality and Social
Psychology, 76(4), 1999, 587-99.
Walster, E., G. Walster, and E. Berscheid. Equity: Theory and
Research. Boston: Allyn and Bacon, 1978.
Wilder, D. A., and P. N. Shapiro. "Role of Out-Group Cues in
Determining Social Identity." Journal of Personality and Social
Psychology, 47(2), 1984, 342-48.
Wit, A. P., and H. A. M. Wilke. "The Effect of Social
Categorization on Cooperation in Three Types of Social Dilemmas."
Journal of Economic Psychology, 13, 1992, 135-51.
Zelmer, J. "Linear Public Goods Experiments: A
Meta-Analysis." Experimental Economics, 5, 2003, 299-310.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the online
version of this article:
APPENDIX S1. Participant Instructions.
(1.) Our approach is also related to experimental work focusing on
the effects of property rights and sunk costs on social preferences
(Cherry, Frykblom, and Shogren 2002; Hoffman, McCabe, and Smith 1994;
Hoffman and Spitzer 1985; Konow 2000, 2003; Ruffle 1998). However our
emphasis is on heterogeneity in these sunk costs and the presence of a
minority of individuals with different sunk costs.
(2.) A one-shot game was used for consistency with the methodology
used by Cherry, Shogren, and Kroll (2005) with subjects earning money
and then making a contribution decision. We are not concerned with the
critique that subjects do not have an opportunity for learning as our
results are remarkably consistent with other linear public goods
experiments (Ledyard 1995; Zelmer 2003), and replicate the results of
Cherry, Shogren, and Kroll (2005) and Clark (2002).
(3.) Instructions are available on request from the author's
website and in Appendix S1 (Supporting Information). The payoff tables
and the exam used in the earnings treatments are available upon request.
Each participant received three payoff tables, one for each potential
level of her endowment.
(4.) This is consistent with the idea of equity theory (Hoffman and
Spitzer 1985; Konow 2000: Selten 1978; Walster, Walster, and Berscheid
1978) in which individuals seek to match inputs with outputs in the
determination of what is considered a fair allocation. See Konow (2003)
for an in-depth discussion of these ideas.
(5.) Of course earners may expect non-earners to contribute more
and so contribute more themselves. Neither of these hypotheses are
consistent with the significant contributions observed by ourselves and
Cherry, Shogren, and Kroll (2005) among groups where all subjects earned
their endowments.
(6.) In terms of the experimental literature on found money effects
and asset legitimacy, such an interpretation explains the results of
Cherry, Shogren, and Kroll (2005) and Clark (2002) who find no effects
of property rights on contributions in public goods games when these
property rights are consistent across subjects.
(7.) That percentage aggregate contributions are lower than
percentage contribution suggests that subjects with higher wealth levels
contribute a lower percentage. This is consistent with the literature on
endowment heterogeneity (e.g., Buckley and Croson 2006).
(8.) As in Harrison (2007), we observe more Nash play among the
earners although as in Cherry, Shogren, and Kroll (2005) this difference
is not statistically significant (see the Percent Nash Regression in
Table 3).
(9.) We model percent aggregate contribution simply using treatment
dummy variables (column 2 of Table 3). Variables to control for the
group's wealth, inequality, or the distribution of the wealth
effect were insignificant and so have been excluded from the
presentation. This regression suggests that as suggested by Hypothesis
2, percent aggregate contributions are significantly lower in the one-
and three-earner treatments than in the baseline. These results are also
supported by standard nonparametric tests (Wilcoxon rank-sum).
(10.) Regressions were conducted which included total wealth,
percent wealth, the Gini coefficient, and the correlation coefficient,
and these variables interacted with a variable for whether the
individual was an earner or not. The Gini and correlation coefficient
variables were never statistically significant and when all of the
variables are included only the coefficients on the one- and
three-earner treatment variables are significantly significant. As a
result, we have chosen the regression with percent wealth to present.
(11.) The p value for the sum of the Two-Earners and Earner in
Two-Earners variables being equal to zero is 0.9957.
(12.) Similar arguments are made in Oxoby and Spraggon (2008) and
Cherry, Frykblom, and Shogren (2002) with respect to property rights in
dictator games.
(13.) These results were strongest among individuals who expressed
views against racial mixing. Experimentally, Fershtman and Gneezy (2001)
find evidence of discrimination against individuals of different
ethnicity.
(14.) Lee and Roemer (2004) compute the effect of this behavior on
redistribution in the United States, estimating that such "voter
racism" reduced income tax rate by 11%-18%.
ROBERT J. OXOBY and JOHN SPRAGGON *
* We thank Todd Cherry, Jeremy Clark, Rachel Croson, Doug Davis,
David Dickinson, Stuart Mestelman, and participants at the 2005
international meeting of the Economic Science Association (Montreal,
Canada) for helpful comments. The authors thank Kendra McLeish and
Alexander Smith for research assistance. Financial support from the
Canadian Institute for Advanced Research, the Social Science and
Humanities Research Council, and the Institute for Advanced Policy
Research is acknowledged.
Oxoby: Department of Economics, University of Calgary, 2500
University Drive NW, Calgary, Alberta, Canada T2N IN4. Phone (403)
220-2586, Fax (403) 282-5262, E-mail oxoby@ucalgary.ca
Spraggon: Department of Resource Economics, Stockbridge Hall,
University of Massachusetts, 80 Campus Center Way, Amherst, MA
01003-9246. Phone (413) 545-6651, Fax (413) 545-5853, E-mail
jmspragg@resecon.umass. edu
doi: 10.1111/j.1465-7295.2012.00493.x
TABLE 1
Mean Contributions (Standard Errors) by
Treatment
Percent
Aggregate Percent Percent
Contribution Contribution Nash-play
Endowed 26.52% 29.32% 20.3%
(2.87%) (3.34%) (5.06%)
n=16 n=64 n=64
One-earner 12.82% 13.03% 73.44%
(2.88%) (2.70%) (5.56%)
n=16 n=64 n=64
Two-earners 21.37% 23.14% 42.19%
(2.11%) (3.30%) (6.22%)
n=16 n=64 n=64
Three-earners 9.81% 11.45%n 84.38%
(1.75%) (2.7%) (4.57%)
n=16 n=64 n=64
Four-earners 31.99% 34.69% 30.00%
(4.17%) (3.97%n) (5.97%)
n=15 n=60 n=60
Note: Percent Aggregate Contribution is ([[summation].sup.4.sub.j=1]
[c.sub.j] /[[summation].sup.4.sub.j=1] [[omega].sub.j] * 100), Percent
Contribution [c.sub.j]/[w.sub.j]. Percent Nash is calculated by
generating a variable equal to one if percent contribution is less
than 5.1% of one's endowment. Standard errors in parentheses and n is
number of observations.
TABLE 2
Mean Contributions (Standard Errors) by
Treatment: Earners Versus Endowed
Percent Percent
Contribution Nash-play
Endowed
Non-earners 29.32% 20.3%
(3.34%) (5.06%)
n=64 n=64
One-earner
Non-earners 9.56% 75%
(1.53%) (6.32%)
n=48 n=48
Earners 23.44% 68.8%
(9.51%) (11.97%)
n=16 n=16
Two-earner
Non-earners 15.70% 50%
(3.45%) (8.98%)
n=32 n=32
Earners 30.57% 34.38%
(5.38%) (8.53%)
n = 32 n = 32
Three-earner
Non-earners 5.31% 93.75%
(0.31%) (6.13%)
n = 16 n = 16
Earners 13.48% 81.25%
(3.53%) (5.69%)
n = 48 n = 48
Four-earner
Earners 34.69% 30.00%
(3.97%) (5.97%)
n=60 n=60
Note: Percent contribution is [c.sub.j]/[w.sub.j]. Percent Nash is
calculated by generating a variable equal to one if percent
contribution is less than 5.1 % of one's endowment. Standard
errors in parentheses and n is number of observations.
TABLE 3
Statistical Results: Tobit Regression on Percent
Contribution and Probit Regression on Percent
Nash
Percent
Aggregate Percent Percent
Model Contribution Contribution Nash
Constant 0.265 0.496 -1.18
(0.031) (0.098) (0.281)
[0.000] [0.000] [0.000]
One-earner -0.169 -0.521 1.49
(0.046) (0.106) (0.266
[0.000] [0.000] [0.000]
Two-earners -0.051 -0.311 0.879
(0.044) (0.108) (0.287)
[0.246] [0.004] [0.002]
Three-earners -0.217 -0.955 2.41
(0.047) (0.243) (0.529)
[0.000] [0.000] [0.000]
Four-earners 0.055 0.031 0.310
(0.045) (0.085) (0.247)
[0.225] [0.715] [0.209]
Earner in 0.232 -0.099
one-earner (0.159) (0.385)
[0.146] [0.796]
Earner in 0.311 -0.489
two-earners (0.125) (0.324)
[0.013] [0.132]
Earner in 0.452 -0.693
three-earners (0.248) (0.540)
[0.068] [0.200]
Percent wealth -1.01 1.37
(0.319) (0.834)
[0.002] [0.102]
Num. obs. 79 316 316
Note: Standard errors in parentheses and p value in
square brackets.
