Frey and Meier: Pro-social behavior in a natural setting

Frey and Meier. 2004. Pro-social behavior in a natural setting. Journal of Economic Behavior and Organization 54:65-88.

In Brief

The economic paradigm of self-interest has been successful within a circumscribed field of activity, but it is increasingly necessary to qualify that view in the face of empirical evidence that people act altruistically in situations that fall outside that narrow zone. Voting, tax paying, and charitable giving are all examples of situations that defy the logic of self-interest. This paper describes an experiment in which charitable giving cannot be explained in terms of self-interest, and weighs alternative explanations before reservedly accepting several.

Experiment

They test whether students of a Swiss university systematically donated to either (or both) of two funds, one intended to provide cheap loans to poor students, one to provide economic assistance to foreign students. Students have no economic motivation to contribute to the fund, since the only benefit they derive is their share of the collective marginal benefit, which is much smaller than their contribution. Moreover, students need fear no social repercussions for not contributing, since the process is anonymous. Nevertheless, 68.6% of the student body did contribute to one or both of the funds, and those who did contribute tended to keep doing so, while those who did not tended not to begin to; the authors discuss free riders and responsible contributors as (somewhat) separate "types." They also mail out a questionnaire afterwards to collect data on respondents' personal characteristics and motivations for contributing to the funds. Their preliminary hypothesis, H1, confirmed, the question becomes: Why do people act this way?

Explanations:

Conditional cooperation

People contribute to a public good if, and to the extent to which, others contribute to it. They test this explanation with a large online survey, in which they ask a random sample of all students what percentage of the student body they thought had contributed to the fund. The responses correlated with the probability that the respondent had herself contributed, to a statistically significant level.

However, the idea that students contribute based on conditional cooperation is contradicted by the observation that only one in five students ever actually talks to anyone else about their decision to contribute, or knows about the decisions of others. So reciprocal considerations cannot be the full story.

Comment: Sure it can. If I expect that others do it, I will do it--I don't need to talk to others to form an expectation that others contribute. Of course, now we move into the realm of norms. Is there a (perceived) norm in favor of giving? Perhaps the school paper publishes information about how many people have donated so far, or even how much each department's students have donated so far. Then I don't need to talk to anybody to engage in conditional cooperation.

Environmental and institutional conditions: two kinds

'Framing effects': being asked to contribute makes a difference in one's likelihood of contributing; so does the manner in which one is asked. In 1998, the University of Zurich changed the way it handled donations to the funds. Prior to that time, students received two separate invoices, one billing them for their tuition plus their donation, one just for their tuition, and they chose which invoice to return (i.e. with or without the donation). Afterwards, students ticked boxes indicating their willingness to contribute, but didn't pay immediately; one month later, they received an invoice billing them for their donation (according to their previous mark on the tuition bill), with no remaining alternative on whether to donate. The percentage of contributors went from 44 to 62 percent after this change. The authors conjecture that people might have "self-control" problems: so deciding to pay today and paying today is harder for us than deciding to pay today and paying in a month. But the authors acknowledge that this has nothing to do with pro-social motivations.

'Group identifications': The authors hypothesize that student contributions to the funds should be greatest when students have their strongest affiliation with the university, i.e., when they are studying there. They find that the probability of a first-semester student contributing money is 2.3 percentage points lower than that of students in subsequent years. This holds until the final semester, when the likelihood decreases by 6.6 percent. Also, foreign students tend to disproportionately contribute to the fund for helping foreigners, further evidence of group identification playing a role in donation.

Heterogeneity in pro-social behavior

People tend to fall into two groups: those who free ride from the beginning and those who do not. They test this by assuming these types can be correlated with choice of major at university. They find that arts students are better people than law students (73% to 64%), accounting for differences in age and gender, and that theology students are bad until they have been at university for a semester.

Comment: Couldn't this be caused by conditional cooperation? Suppose I expect that more arts majors will contribute than law majors. Then I will be more likely to contribute if I am an arts major. This critique also applies to the "group identifications" argument.

Hypotheses:

• H1: A substantial number of people are prepared to act in a pro-social way in an anonymous situation, even after several rounds.
• H2: Expectations about the contributions of other people matter; the more people expect others to cooperate, the more they cooperate themselves.
• H3: The environment under which the donations take place matters. In particular, it is essential that people are asked to contribute in a way they perceive as acceptable.
• H4: People differ in their pro-social attitudes. The type of person (as reflected partly by people's choice of study) influences donating even when standard personal characteristics (gender and age) are controlled for.

Conclusions:

• H1 holds, but conditional cooperation cannot be excluded.
• H2 is not contradicted, but evidence is ambiguous. The coefficient between expected group giving and personal giving is statistically significant, but most students do not have a good idea of what others are giving because the funds are not a topic of much discussion.
• H3 is supported. The way students were asked to donate made a difference in the number that donated. Also, identification with groups seems important as students contribute more when they are in university, and foreign students tend to support the fund for foreigners.
• H4 is supported (?) as the "type" of person (free-rider or responsible citizen), measured by choice of major, plays a role in determining whether we contribute. Certain types of people are more likely than others to donate; the authors assert that their data suggest that the students choose their majors based on differences in pro-social preferences. Comment: Perhaps this connection is tenuous.