Tax Incentives and Charitable Giving: Evidence from New Survey Data

A new survey data set is used to obtain estimates of the tax price elasticity of personal giving to tax-deductible charitable causes. Like other surveys, the data here yield a large elasticity estimate, roughly two and a half for a representative household when Tobit estimation is used. One hypothesis for the discrepancy between such large estimates and values close to unity found in tax data is that there is an “itemization effect” reflecting nonrandom selection in tax data; the Florida data do not support this hypothesis. Another explanation for the discrepancy between tax-file-based and survey-based estimates is that the standard use of OLS rather than Tobit biases the elasticity more in survey data, where many people report zero gifts. For the Florida data, using OLS increases the estimated elasticity by about 30%; while this effect cannot explain why the Florida data produce such large elasticities, it suggests that OLS estimates in earlier studies should be used with caution.