the fact that contributors fail to take into account the fact that any dollar contributed
raises the well-being of other individuals as well (via increased provision of the public
good). On the other hand, there is a re-distributive motive that calls for taxing
contributions. To see this, note that in the presence of charitable contributions we
have essentially two consumption goods: c and z. Furthermore, z is a normal good (as
shown above). Thus, a tax on contributions accompanied by an upward adjustment in
the lump-sum transfer (to maintain the government revenue constraint) is progressive
and therefore enhances re-distribution. As shown by Deaton (1979), when the utility
function is both separable (between leisure and the set of consumption goods) and
homothetic (with respect to the set of consumption goods), commodity taxation is
redundant in the presence of an optimal linear labor income tax. Thus, there are no re-
distributive gains from taxing contributions in this case, and we are left with the
Pigouvian motive, suggesting that charitable contributions should be subsidized. This
establishes the following proposition (the proof is relegated to Appendix A):
Proposition 1: Suppose that the contribution motive is purely altruistic (β= 0), if the
function H(c,z) ≡ u(c) + v(z) is homothetic, the optimal tax on charitable
contributions is negative.
This proposition justifies tax deductibility of or tax credits to charitable
contributions when the latter are motivated purely by altruism. Naturally, this result
extends also to the case where the altruistic motive is sufficiently strong relative to the
status seeking motive (namely, for β sufficiently small). Note, however, that in the
absence of separability and homotheticity, the redistributive motive that may call for
taxing charitable contributions (especially when charitable giving is a sort of a luxury
good) could dominate the Pigouvian motive that calls for subsidizing such
contributions, and a tax on charitable contributions may be called for.
14