Substitution Between (and Motivations for) Charitable Contributions: An Experimental Study

I run a series of laboratory experiments to estimate and describe the extent to which an individual's charitable donation to one cause displaces his or her giving to another cause. The experiments also investigate motivations for giving, including the simple warm-glow and public-goods models, and the desire for influence, reputation, and normalizing behavior. In the first wave of experiments I allow 49 subjects to donate or keep any amount of their 10 'endowment' to any of three charities in each of six stages (with one stage randomly chosen for payments). Some of the stages include 'shocks' (to certain charities), such as an expanded choice set, a higher 'match' rate, and a promotional video. The second wave of experiments (48 subjects) has 13 stages, a larger set of treatments, and a 20 'endowment'. All of the treatments have the hypothesized effect on giving, including the 'price' shock; in contrast to previous experiments, the subjects exhibit price-elastic demand. Subjects also give significantly more when their decisions and identity are observed. The results demonstrate that 'expenditure substitution' among charities can be seen in a laboratory setting. I find large own-price elasticities and very large cross-price elasticities - these charities are gross substitutes in the conventional sense. The substitution is stronger where the charities serve similar purposes, such as UNICEF and Care. In the "reduced form" model, using instrumental variables estimation, I estimate an expenditure substitution coefficient of 37% - when gifts to one charity are increased (or decreased) by a certain amount because of a shock, the sum of gifts to other charities decreases by 37% of this amount. The conditional-on-positive estimate of this substitution is 80%: where gifts to both charities remain positive, the 'crowding out' is 0.80 for every 1.