7. Predicting Giving

In Chapter VII of Giving Credit Where Credit is Due: A New Approach to Welfare Funding, the Beacon Hill Institute estimates the relationship between the price and quantity of charitable contributions made by individual tax filers who itemize their deductions. We also make extrapolations and inferences concerning the behavior of nonitemizers, estimating taxpayer response to the enactment of a tax credit for charitable contributions to certain charitable organizations. Here we show how to project aggregate giving using this methodology.

It is generally believed that charitable contributions are normal goods subject to the law of demand. This means that if the price of giving increases, individuals will reduce the amount of money they give to charitable organizations, and vice versa. The conventional determinants of demand include the price of the good, income, tastes and preferences, and prices of related goods. Along with price and income, we include demographic variables because they help predict an individual’s personal disposition toward giving. Moreover, since the price of giving is related to an individual’s marginal tax rate, changes in the price of giving have public policy implications.

Empirical tests can show if giving is related to these variables.One form of the model is:

Equation 1 states that giving, G, depends on (1) P, the price of the contribution, as measured by the after-tax cost to the consumer, (2) Y, individual income, measured in different ways by different researchers, and (3) Z, a vector of exogenous variables, mostly demographic, which indicate an individual's propensity to contribute to charity.

Assuming a log-linear demand function in a quasi Cobb-Douglas specification of the model yields,

which can be estimated econometrically in a straightforward manner. While the model is linear in its parameters only, the estimated coefficients b₁ and b₂ are the elasticity estimates of income and price, respectively. In this specification, these elasticities are assumed constant.

where charitable giving, G, is measured in dollars and TAXPRICE is the after-tax cost of a dollar’s worth of contributions, that is, one minus the marginal rate of taxation on income before deductions. The first dollar contribution is used to remove problems of simultaneity that occur between income and price. In general, the price of a contribution is the consumption foregone as a direct result of the contribution. INCOME is measured by taxable income.

The demographic variables are XTOT91, MARRY, and AGE65. The variable XTOT91 is the total number of exemptions claimed by the filer of the return and is a proxy for the number of dependents. MARRY is a dummy variable set equal to one if the taxpayer is married and zero if not. AGE65 is a dummy variable set equal to one if any member of the taxpayer's household is over age 65 and zero otherwise. Other demographic variables that would help explain charitable giving are not specified here, due to data limitations. They are represented by u, the error term.

We expect the sign of the coefficient for price to be negative, and the other signs to be positive. That is, a lower price of giving should increase charitable contributions. We expect married and older taxpayers with more dependents and higher income to give more to charitable organizations. Conversely, we expect younger and single taxpayers with fewer dependents and lower income levels to give less.

To estimate Equation 4, we use data from the University of Michigan/Office of Tax Policy Research, which contains a sample of more than 115,000 individual U.S. tax returns for the year 1991. Selecting only returns for taxpayers who itemize deductions leaves us with 64,422 individual returns to examine. The results of the estimation are shown in Table 1.

Variable	Parameter Estimate	Standard Error	T for H0: Parameter = 0	Prob > \|T\|
INTERCEPT	2.0799	0.0774	26.866	0.0001
TAXPRICE	-1.1175	0.1044	-10.702	0.0001
INCOME	0.3431	0.0093	36.739	0.0001
XTOT91	0.1008	0.0059	17.072	0.0001
MARRY	0.2753	0.0185	14.826	0.0001
AGE65	1.0378	0.0239	43.340	0.0001
F Value	Prob>F	Adjusted R²	Root MSE	Dep Mean
1464.931	0.0001	0.1020	35.3542	6.4483

While many outside variables that could help explain charitable contributions are not included in our regression, we find that the effect of our tax policy is significant. The individual coefficients of our regression estimation all have signs as hypothesized and all are statistically significant at very high levels of confidence, with levels of significance of less than .0001. The negative sign associated with TAXPRICE indicates that as the price of giving decreases (as tax incentives increase), more giving occurs. The estimated parameter -1.12 is the elasticity of giving with respect to the tax price. This value implies that giving is price elastic, which means that a decrease in the price of giving, as occurs with the enactment of a tax credit, will cause giving to rise by more than tax revenues fall.

The positive sign on income suggests that individuals give more to charity as their incomes increases, with a 10% increase in income leading to a 3.5% increase in contributions. This indicates that charitable giving is a “normal good,” meaning that consumers give more as income increases. The positive sign on dependents indicates that individuals with more dependents give more, as do married and older taxpayers, as we expected.

Predicting Future Giving

What does this mean for a tax credit? How can we model the change in giving that results from a drop in the price of giving? To do that, we need to know current levels of giving, current tax rates, current income, and how people respond to changes in price and income, (estimated above). Following Clotfelter (1985), current giving (G₀) can be written as Equation 5, while projected giving (G₁) can be written as Equation 6 as follows:

Current giving depends on current price and current income, P₀ and Y₀, respectively, while projected giving depends on projected price and income, P₁ and Y₁. Individual behavior is assumed to remain the same over a range of tax policies implying that b₁ and b₂ remain constant. Individual demographics also remain constant. However, price and income can change when the tax code is altered. Taking the difference between Equation 6 and Equation 5 gives

Because we know the values of current giving, price, and income, as well as projected price and projected income and the estimates of

from the above econometric estimation, we can simulate the impacts of our proposed tax credit by examining several different scenarios. The first problem is that we do not know how nonitemizers react to changes in the tax price of giving, since the regression was run for itemizers only. We assume they are as responsive to price changes as are itemizers.

Giving Under a Tax Credit

To predict future giving using this methodology, we must know the present level of giving for itemizers and nonitemizers. Only itemizers report charitable contributions on their Schedule A forms, which they file along with their tax returns. This value is open to fraud or error if itemizers consistently over or under-estimate their charitable giving. We shall assume that itemizers are, on average, fairly accurate. Nonitemizers report zero contributions, but surely they contribute something even if they do not receive any tax preference for their donation.

To predict future giving, we need to know the price of giving, P₀. So we must calculate the effective federal and state marginal tax rates, t_f and t_s, respectively. In addition, the term d_s indicates whether a state allows deductions for contributions. To give one dollar, the relevant price facing the itemizing taxpayer is one minus the federal marginal tax rate minus the state marginal tax rate plus the interaction term

due to the increase in federal taxes because of lower state deductions on schedule A, as shown in Equation 11.

If the state does not allow contribution deductions, d_s will equal zero and the price of giving one dollar will be one minus the federal marginal tax rate. For nonitemizers, the net price of a contribution is the contribution itself, since the taxpayer receives no deduction; hence the price for nonitemizers is one.

To calculate the new price of giving for itemizers under a tax credit regime, we need to subtract the new state tax credit, t_c.

due to the increase in federal taxes because of the lower state tax deduction on Schedule A. The new price of giving is shown in Equation 12.

The price of giving P₁ for nonitemizers would be one minus the tax credit, as shown in equation 13.

To estimate the change in giving, we use Equations 10 through 12 for itemizers and nonitemizers. Below we provide two examples of how to project the response of itemizers and nonitemizers in the States of Oklahoma and Massachusetts.

Oklahoma

In Oklahoma, itemizers gave approximately $898 million in total cash and check charitable contributions in 1996. According to recent U.S. tax return data, itemizers account for 59% of total giving by individuals. Nonitemizers account for approximately 41% of total giving, implying that nonitemizer contributions amount to approximately 69% of itemizer contributions. However, Clotfelter (1985) suggests that itemizers account for approximately two-thirds of total giving, or approximately twice as much as nonitemizer contributions. Here, we take a conservative approach and suggest that nonitemizers in all states contribute approximately 55% of what itemizers contribute. Thus total contributions by nonitemizers amounted to $497 million in 1996, and all contributions totaled $1.39 billion.

Total eligible current charitable contributions in Oklahoma are estimated to be approximately 5% of total contributions, or $70 million. Thus, approximately $45 million is given by itemizers and $25 million by nonitemizers. Table Q shows the distribution of current eligible giving by form of deduction.

To estimate the change in price of giving and a concomitant change in contributions under a tax credit, we use Equations 10 through 12 for itemizers and nonitemizers. First, we use Equation 11 to investigate the current price of giving. The average federal marginal tax rate for itemizers is 23.0%, the average state marginal tax rate is 6.3% and d_s equals one since Oklahoma allows itemizers a deduction for contributions that lower the overall cost of giving. Thus, the current price of giving one dollar, P₀, is approximately 72.1 cents, or .721 .

Next, we use equation 12 to estimate the cost of giving under a tax credit regime. Assume that Oklahoma allows taxpayers to take both a deduction and a credit. The price of giving, P₁, for itemizers under a 50% tax credit approaches 33.6 cents, or .336. In other words, the credit and deduction would cost taxpayers who itemize only 33.6 cents on the dollar.

The price of giving, P₁, for nonitemizers under a 50% tax credit falls to .50 cents.

Once the price of giving under a tax credit is known, we will want to know how much in contributions will be generated. Equation 10 estimates total amount of giving under a tax credit.

Under a tax credit regime, Oklahoma could expect to receive an extra $90 million each year in new charitable contributions for nonprofit organizations that service the poor. This is calculated by subtracting total estimated eligible individual giving from 1996 estimated eligible giving.

If Oklahoma instituted a tax credit for charitable contributions, it could expect to lose $79.8 million of tax revenues. To maintain revenue neutrality Oklahoma would have to cut a concomitant amount from its welfare budget. Private nonprofit organizations would then replace the inefficient state welfare bureaucracy.

Massachusetts

Total charitable giving by itemizers in Massachusetts is estimated at $1.84 billion for 1996. Itemizers accounted for approximately 64% of total giving, while nonitemizers accounted for 36%. Conservatively, we assume that nonitemizers in all states contribute approximately 55% of itemizers’ giving. Taxpayers who itemize their federal tax returns report $1.84 billion, nonitemizers $1.02 billion.

In Massachusetts total eligible charitable giving (i.e. to the poor) by individuals is estimated to be 5% of total giving or $142 million. Of this amount, giving by taxpayers who itemize is $92 million, nonitemizers is $51 million.

To estimate giving under a deduction, P₀, and under a tax credit, P₁, we use equations 10 through 12 for itemizers and nonitemizers. Since Massachusetts does not allow a deduction for charitable contributions, the cost of current giving is one minus the federal marginal tax rate. The federal marginal tax rate for itemizers is 25.7%. Thus the current price of giving one dollar, P₀, is approximately 74.3 cents

Again we use Equation 12 to estimate the cost of giving under a tax credit regime. We assume that Massachusetts allows taxpayers a 50% tax credit for charitable contributions, but not a deduction. The price of giving for itemizers, P₁, under a 50% tax credit is approximately 37.2 cents, or .372. In other words the credit and deduction would cost taxpayers who itemizers only 37.2 cents on the dollar.

Once the price of giving under a tax credit is known, we will want to know how much in contributions will be generated. To predict giving under a tax credit we use Equation 10. For itemizing taxpayers, current giving to eligible charities, G₀, is $92 million. Estimating via Equation 10 suggests the tax credit would generate $199 million in total giving by itemizers.

For nonitemizing taxpayers, current giving to eligible charities, G₀, is $51 million. Under a 50% tax credit, the price of giving falls from 1 to .5. Thus, for nonitemizers, we predict eligible giving will rise to $110 million.

Under a tax credit regime Massachusetts could expect to receive an extra $167 million each year in new charitable contributions for nonprofit organizations that service the poor. This amount is calculated by subtracting total estimated eligible individual giving from 1996 estimated eligible giving.

By implementing a tax credit regime, Massachusetts would lose $247.8 million in tax revenues. Since state welfare services that were provided by the state would be provided by nonprofit charitable organizations, Massachusetts could cut a concomitant amount from its welfare budget.

The Next Step Toward Welfare Reform:

A Manual for Enacting Tax Credits

for Charitable Contributions

March 1998