ECONOMETRIC ISSUES IN THE ANALYSIS OF CHARITABLE GIVING

In this article, several econometric problems arising from estimating models of charitable contributions are examined An instrumental variable technique proposed by Clotfelter is used to reduce the bias of the estimated price effect and compare it with other methods. Linear dependence between price and income are reduced by including state and local tax rates and show that the degree of collinearity is low. By using demographic characteristics, current and permanent income and contributions to different organizations in the Consumer Expenditure Survey, the results are compared with other models, and bias caused by neglecting government expenditures is examined. These comparisons show that the dummy-intercept and the first-difference model omit important permanent effects and that omitting government welfare expenditures biases price elasticities, although the degree of bias may be small.