Expectations and Gender Differences in
Altruism (w/ Adam Levine)
A central question in the study of altruism has been
whether there is a systematic gender difference in giving
behavior. In particular, are females more altruistic than
males? There are at least two variables that may influence
rates of giving: the relative price of giving and the
\emph{expectation of receiving}. For example, if giving 1
unit aids a recipient by 3 units then this may influence
how much you are willing to donate. And if you happen to
believe that others in your local population are facing the
same decision of whether and how much to donate, and that
you are a potential recipient, what you expect to receive
may well influence what you decide to give.
As with prior research on gender differences and altrusitic
behavior, we use the dictator game (DG) played between
Dictator and Recipient to model the basic environment (See
Forsythe, et. al, 1994; Dictator is endowed with a pie of
value M and must decide on some distribution of M between
the two players. The Recipient cannot respond; the
distribution of the pie chosen by Dictator is the final
allocation.). Experimental research has consistently found
that female dictators are more generous than male dictators
when the price of giving is equal to 1-- that is, when 1
unit donated aids the recipient by 1 unit (Eckel and
Grossman, 1998; Selten and Ockenfels, 1998; Dickenson and
Tiefenthaler, 2002; Rigdon, et al., 2008). However,
Andreoni and Vesterlund (2001) find that male dictators
give more than female dictators when the price of giving is
less than 1 whereas female dictators give more than male
dictators when the price of giving is greater than 1. That
is, male giving is more price elastic, while female giving
is more price inelastic. The standard and widely reported
result is thus that females and males are differentially
price sensitive in their giving behavior: males are more
generous than females if giving is cheap, and vice versa if
giving is expensive (Andreoni and Vesterlund, 2001;
Dickenson and Tiefenthaler, 2002; Croson and Gneezy, 2009;
Eckel and Grossman, 2008; Andreoni, et al., 2008).
But all of these results ignore how expectations of
receiving might influence giving behavior. This is for one
of two reasons. One reason is because they implement the
standard DG under one-shot conditions (where this is common
knowledge) as in the experiments reported by Eckel and
Grossman, 1998 and Rigdon, et al., 2008. In this case,
expectations of receiving are (trivially) 0 and so cannot
influence giving decisions. The other reason is because, as
in Andreoni and Vesterlund (2001), the modified DG
implemented in which agents have non-trivial expectations
of receiving fails to control for those expectations. But
those expectations may matter to giving behavior --
assuming otherwise simply assumes that any differences in
expectations among people are uniform across genders. Thus,
if there is gender variance in those expectations, that
variance may well influence giving behavior. Rather than
difference in price sensitivity across genders, it may be
that there is instead a difference in the priors about
other people's giving that males and females bring to the
laboratory. That is an empirical question worthy of an
answer.
Our experiment aims to address exactly this issue: is the
apparent different price sensitivity of giving across
genders instead a reflection of differences in
expectations? We manipulate the price of giving across
budget sets in a version of the DG in a laboratory setting,
just as Andreoni and Vesterlund (2001) do. In one
treatment, we let subjects appeal to their homegrown
expectations of receiving. In our other treatment, we
exogenously control those expectations in a non-trivial but
uniform way. When expectations are homegrown, we replicate
the Andreoni and Vesterlund's results; when they are
exogenously determined, all gender
differences disappear. Which is the fairer sex? It depends
on what they believe.
Status: presented research at ESA Tucson 11.2008
