Timing vs. Long-run Charitable Giving Behavior: Reconciling Divergent Approaches and Estimates

In this paper we examine the effect of the income tax on charitable giving. An important challenge in this literature has been to estimate the long-run response of giving to a persistent change in tax-price, which can be difficult to distinguish from intertemporal substitution arising from differences between current and expected future tax prices, arising for example due to transitory fluctuations in incomes, life-cycle factors, or preannounced tax reforms. Several papers that have attempted to distinguish these effects have found that the elasticity of charitable giving with respect to a persistent price change is small, while the elasticity with respect to a transitory difference between current and expected future prices is large. Auten, Sieg, and Clotfelter (2002) advance this literature by developing an estimation procedure that incorporates a more sophisticated model of the stochastic process for income. In contrast to previous research on the topic, they find the counterintuitive result that the immediate response to a persistent price change is much larger than the immediate response to a one-period transitory price change. In this paper, we present a new estimation procedure that allows us to implement their assumptions about the stochastic process of income in a more conventional regression framework, and then adapt the procedure to take into account the pre-announced and phased-in nature of tax reforms that occurred during the sample period. In preliminary analysis based on a public-use panel of individual tax return data, we are able to replicate their counterintuitive pattern of price elasticities, and find that incorporating information about pre-announced and phased-in tax law changes reverses their result -- the persistent price elasticity is reduced substantially, and the transitory price elasticity is now the larger of the two. We also try an instrumental variables strategy that relies exclusively on federal and state tax reforms for identification, and this yields similar results. Finally, we incorporate a dynamic adjustment process into the empirical specification, and find evidence that the long-run response to persistent price and income changes is larger than the immediate response.