This paper makes three contributions to the literature on private provision of public goods. First, we identify limitations of the frequently used specification test that distinguishes between the standard models of pure and impure altruism based on the extent of crowding out. While the literature takes as given the result that crowding out should be less with impure altruism than with pure altruism, we show that it can be either more or less. Second, we propose a new, more general test based on the presence of crowding in or greater than one-for-one crowding out, both of which are consistent with impure altruism only. Third, we provide empirical evidence. Using a unique panel data set on volunteerism in U.S. National Parks, we estimate the causal effect of changes in a park’s public funding on the amount of within-park volunteerism. The overall finding is that each additional dollar of public expenditure crowds in 27 cents worth of volunteerism on average. We show how the estimates of crowding in, along with heterogeneity based on park and volunteer hour types, are theoretically consistent with the mainstay model of impure altruism.
Read full abstract