on the latter result. The erratic results with the interaction term do not indicate that the model without it was correctly specified.
The time series analysis by Clotfelter and Steuerle, and Feldstein and Taylor before them, seeks to get around the price-income link by using independent changes in price arising from statutory changes in the tax laws. The approach appears promising, but the standard errors at this stage are awkwardly large. In any case, the estimated price elasticities were generally so low as to give no support to the high constant price elasticities emerging from the cross-section analysis.
The new cross-section work by the authors has been ably carried out, and it has probably pushed this line of inquiry to the limits of its productivity. Their attempts to isolate price and tax effects with the aid of statutory tax rate changes deserve to be followed up. At least two measurement questions should be considered in any further research. The 8 percent contribution-income ratio shown for low incomes suggests that adjusted gross income may be a misleading measure of permanent income and donor capacity at that end of the scale. For example, the income variable might well be revised to exclude capital losses or unusually high deductions in that range.54. Also, the contributions variable itself is suspect according to Treasury estimates indicating substantial overstatement of donations by upper income households.
Finally, the authors' cautious conclusions seem warranted in view of the work to date. They note a "consensus of sorts" showing price elasticities to be significantly negative, but only greater than one in absolute value for higher income classes. They deserve credit for laying this out so well. They also recommend caution in making policy prescriptions on the basis of the findings. Although this warning is certainly in order, an effort to spell out the policy implications of their tentative findings would be very useful. In this context they might put the Treasury file and these models to work simulating revenue effects of various tax policies toward contributions. In this way the estimated effects of tax rate changes, deductibility reductions, and tax credits or above-the-line adjustments to income could advance understanding beyond that achieved by a priori interpretation of elasticities.____________________