Altruism with Social Roots: An Emerging Literature
Pablo Branas-Garza y Maria Paz Espinosa
MRQ- with the only difference that they do not know with whom;
dictators know only that they will play with a friend in Fi (which im-
plies pi(f) = 1 and 0 < pi(j) ≤ 1, see Table 2). Then they are informed
that they will play with any randomly chosen close friend, that is any
element of Fi. Observe that although pi(f) = 1 the probability of being
matched with a specific friend j is less than 1, in particular pi(j) = 1/fi,
with fi being the total number of friends11. In contrast, the baseline
treatment comprises dictators playing with strangers.
Note that pi(j) ≤ 1 implies on the one hand that reciprocity has less
importance since the dictator does not know exactly who the recipi-
ent is (and therefore obtaining ex-post benefits is made more diffi-
cult), and on the other hand it eliminates the effect that knowledge of
the recipient’s identity may have on the dictator. Interestingly, even
though direct reciprocity and knowledge of identity are removed, there
is still a friendship effect in BDE, in a similar direction to that re-
ported by MRQ:
R1’: In the absence of direct reciprocity, social proximity increases
giving.
Goeree, McConnell, Mitchell, Tromp and Yariv (2006) (GMMTY) con-
ducted a survey among 10-12 year old girls in Pasadena, eliciting
friendship relations among these fifth and sixth graders (see Table 1)
as well as a large number of personal attributes. An interesting feature
is that subjects play 10 different dictator game decisions with recipi-
ents at different social distances -from the dictators- although only
one of the decisions is implemented (randomly).
As in MRQ the matching mechanism explores different relations be-
tween players: friends, friends of friends and greater social distances.
This procedure allows proximity to be graduated more accurately and
enables the effect of spatial measures of intensity of relations on gen-
erosity to be analyzed. Their study of giving as a function of distance
between players yields a clear result: there are large and significant
effects of (social) distance on giving. They also control for the dicta-
11 p( j) = 1 only for those subjects with fi = 1.
252