CRS Analyzes Charitable Contribution Deduction

The 107^th Congress version of (H.R. 7), passed by the House, had eight new tax provisions designed to benefit charities and charitable giving, the most important one being the charitable deduction for non-itemizers. This deduction, which was subject to a phased-in cap of $100 ($200 for joint returns), was in the President's initial tax proposal, although it was not capped. In 2002, S. 1924, providing a temporary deduction with a $400/$800 cap and other provisions was introduced in the Senate: the Senate Finance Committee adopted a version of S. 1924 with a floor and ceiling, also reported out as S. 476 in the 108^th Congress and passed April 9, 2003. A similar non-itemizer deduction was in H.R. 7, reported by the Ways and Means Committee and passed by the House on September 17, 2003. S. 6 in the 109^th Congress includes the same deduction. The President did not include the deduction in the budget for FY 2006. This paper focuses on the economic effect of the deduction for non-itemizers, assessing the incentive such deductions would create for increased charitable giving. It does not attempt to estimate other types of societal impacts.

Economic analysis suggests that the impact of the original proposed deduction on charitable giving is likely to be relatively small, due to the proposed $100/$200 deduction cap. For individuals and couples already contributing more than the cap, there is no additional incentive for further giving. Nonetheless, the deduction would still generate a significant government revenue loss. Effects are also reduced by estimates that project a limited response by taxpayers, particularly lower income taxpayers, to this kind of charitable giving incentive. Analysis suggested that even when the deduction is fully phased in , a dollar of revenue loss is likely to increase charitable giving by three cents, even under relatively optimistic assumptions. The effect would be only half as large if one netted out the percentage of contributions that go to provide sacramental services to church members. The higher caps in the 108^th Congress proposals and in S. 6 in the 109^th, as well as the floor is estimated to increase that amount to 18 cents.

The capped charitable contribution deduction for non-itemizers will add complexity to the tax system and to the simplified tax forms used by millions of taxpayers, since an additional line would have to be added. Charitable deductions are also particularly complicated for record keeping purposes because there is no unified reporting system. Finally, because of the small amounts in question and the absence of requirements for documentation for small gifts, tax evasion may increase; there are likely to be deductions taken even when no contributions were made.

This report will be updated to reflect legislative developments.

 Contents

This report summarizes the provisions affecting charitable contribution deductions of individuals, and then analyzes the incentive such a deduction would create for increased charitable giving beginning with the original proposal for a relatively low cap and then considering other approaches including the current one. It does not attempt to estimate other types of societal impacts. The non-itemizer's charitable deduction was the single most important tax provision in the original version of H.R. 7. In S. 1924, S. 476, and the current version of H.R. 7, the non-itemizer provision was temporary and had a higher cap (and a floor). The provision affecting rollovers from IRAs, which can also function as a deduction for non- itemizers, is also discussed briefly.

The next section describes the proposed changes. The following sections discuss three basic issues: how effective and efficient the proposed non-itemizer deduction might be in increasing charitable contributions, how equitable the change is, and how the revision might affect tax complexity, administration, and compliance.

Under current law a taxpayer can either itemize deductions (with the major deductions being charitable contributions, excess medical expenses, mortgage interest, and state and local income and property taxes) or choose the standard deduction. The latter is advantageous if that amount is larger than total itemized deductions. H.R. 7 would allow someone who takes the standard deduction to deduct charitable deductions in addition, but with caps as shown in Table 1. (The deduction is taken from adjusted gross income, and thus would not affect that measure.) S. 1924 initially would have imposed a cap of $400 ($800 for joint returns). The version passed on April 9 would ,like the President's current proposal, provide a floor and cap of $250/$500 ($500/$1000 for joint returns); a similar provision was adopted on September 9 by the Ways and Means Committee.

       Table 1: Caps on Charitable Contribution Deduction for

IRA Rollover Provision

The second largest provision in the original version of H.R. 7 (and retained in other versions) is one allowing tax free distributions by individuals aged 70 and1/2 and over from individual retirement accounts to charitable organizations. While this treatment may appear no different from simply including the amounts in adjusted gross income and then deducting them as itemized deductions, it can provide several types of benefits even to those who itemize. Apparently an important motivation is to reduce adjusted gross income which can trigger a variety of phase-outs and phase-ins, including the phase-in of taxation of Social Security benefits. (Another potentially important phase-out effect, that for itemized deductions, is now scheduled to be eliminated.) There are also some income limits on charitable contributions.

However, unlike the case in the President's proposal, this provision can also benefit those who take the standard deduction by allowing exclusions that would not be otherwise allowed. In effect, a return with the standard deduction would have no cap on charitable deductions up to the limit of aggregate IRA amounts, since the taxpayer could simply channel contributions through an IRA. (This effect did not matter in the President's proposal which had no cap).

Other Provisions

The remaining provisions include a phased in increase in the cap on corporate charitable contributions; a change in the treatment of contributed property by firms, modifications of excise taxes on net investment income (and the current bill introduced in the House includes a provision restricting the counting of some administrative costs as part of a foundation's minimum distribution requirement), changes in the treatment of unrelated business income, changes in the treatment of self constructed property, and modification of basis for certain appreciated assets. The increase in the corporate charitable deduction limit was also included in the President's proposal; this provision and the excise tax revisions are the largest in revenue terms of the remaining provisions.¹ The Senate-passed bill, however, excludes these provisions but contains other ones.

Economic theory also recognizes an efficiency rationale for subsidies to charitable giving. When individuals give to charity, they do not take into account the benefits their contribution provides to others and therefore charitable giving is under supplied in a market economy. Indeed, other potential donors in society can receive the benefits from the charitable activities of others without contributing. This free-rider problem is, in fact, one of the justifications for government activities that use mandated contributions (taxes) to provide a variety of services. Some activities (such as national defense) would probably be impossible to provide privately. Most of the support of the poor is also undertaken by government. Other activities that may have broad benefits to society (e.g. research and development, art, education) are provided both publicly and privately.

There are advantages and disadvantages in using direct funding versus encouraging private charitable donations. For example, government provision may more generally reflect the collective preferences of society, but private provision may provide added diversity.

The under supply of charitable activities is greater in some circumstances than in others. One important factor is the size of the contributing group. In general, the larger the pool of contributors or potential contributors, the greater the free-rider problem. It is also more difficult to obtain a better market outcome if the benefits of the activity are not excludable. For example, contributions that finance health research benefit all members of society; for that reason government is heavily involved in providing funds for such research and in providing a legal framework (e.g. patent laws) that encourage the profit-making sector to invest. Art museums, however, have the option of charging admission to finance their services so that these institutions would still exist (perhaps in smaller numbers) in the absence of both charitable donations and government support.

Another of the problems with stimulating charitable activities through private giving is that it is not always easy to separate pure charitable giving and giving in which the donor receives a direct benefit. A contributor to the opera, for example, who receives a seating preference, is receiving a benefit. Contributors to institutions of higher education may find it more likely that their children are admitted to prestigious schools. And, one of the more obvious examples of small groups that receive benefits are contributions to churches, where contributions are partly used to provide services to members. This provision of sacramental and similar services is particularly important in evaluating the deduction for non-itemizers because of the very large fraction of contributions that go to religious organizations.

There is also some leakage in the tax system in that individuals will claim deductions for expenditures they did not incur. This problem may be particularly likely to arise with small cash donations since there is no requirement that small donations be substantiated and it is not efficient for the tax authorities to make efforts to enforce compliance associated with small dollar amounts.

Another issue is whether a dollar spent stimulating private giving will result in contributions (regardless of their use) that are greater than a dollar or smaller than a dollar. If the induced contribution is larger, private charitable tax incentives are more efficient than direct spending, assuming the objectives of private and public spending are equally desirable. The stimulative effect of a tax subsidy is measured by the price elasticity. As a rule of thumb, without considering other factors (such as the whether the contributions actually go to a charitable purpose and the effects of caps), a dollar of revenue cost will result in more than a dollar of giving if the price elasticity is above one in absolute value. (All further references to price elasticities will refer to absolute values, without negative signs.)

Finally, an important issue with a capped deduction is what portion of the revenue loss is associated with individuals whose deductions, in the absence of the subsidy, are less than the cap. If individuals are already giving more than the cap, there is no economic incentive for further giving, since further giving will not affect tax liability. However, the government still incurs a revenue loss.

Therefore, the amount of induced giving per dollar of revenue cost for the non-itemizer capped deduction is the product of three factors: the absolute value of the price elasticity, the share of the cost going to individuals not subject to the cap, and the share of the induced contributions going to desired charitable activities. The following subsections explore each of these in turn.

The Price Elasticity

Research on the price elasticity of charitable contributions had long suggested that, other things equal, charitable contributions deductions might be efficient since elasticities averaged above one. However, these studies were criticized because they did not control for transitory timing effects. High income individuals would choose to make contributions when their tax rates and income were temporarily high, but this effect would not be a permanent response. Estimates correcting for this effect found elasticities below one. A recent study of the effect of fundamental tax reform used, as a base case, an elasticity of 0.5, corrected for timing effects, but also considered the older elasticity of 1.3 reflecting general findings of earlier studies.³

A recent study (Duquette, 1999) of non-itemizer contributions also suggests that elasticities will be below one in absolute value.⁴ This study examined the non-itemizer deductions allowed in 1985 and 1986 (when ceilings were not present), and also estimated price elasticities for itemizers. Note that this study could not correct for transitory effects, including the temporary nature of the deduction for non-itemizers which would have encouraged individuals to shift deductions, especially to 1986, particularly those at higher income levels. This study found elasticities, for 1985 and 1986, of 1 and 1.24 for itemizers, but elasticities of 0.8 and 0.6 for non-itemizers. (Note that the estimate for non-itemizers in 1985, while statistically significant, was not measured very precisely; the estimate for 1986, while measured somewhat more precisely, may be overstated because individuals would have an incentive to concentrate their contributions in that year.)

These findings of lower elasticities for non-itemizers are also consistent with evidence that suggests that price elasticities are not constant across income levels, but are lower at lower levels. Examining itemized returns that do not suffer as much from transitory effects, Duquette finds price elasticities for the $1 to $40,000 income class to be 0.08 in 1985 and 0.25 in 1986, with neither estimate statistically different from zero. Elasticities rise as income increases (although some of that increase may reflect the greater importance of transitory effects at higher income levels).

As shown in Table 2, where the 1986 data are reported by income class (but restated in 2001 income levels) non-itemizer deductions are concentrated in lower income levels, simply because it is in the lower income levels that the standard deduction is normally taken. Based on Duquette's elasticities by income level, an average elasticity weighted by the number of contributors in each class suggests an elasticity of 0.10 based on 1985 data and 0.29 based on 1986 data. (These estimates are, however, highly imprecise.) The effects of caps, which tend to concentrate the marginal benefit even more among lower income levels, would slightly decrease these estimates.

On the whole, therefore, the evidence suggests price elasticities that are considerably lower than one, and perhaps not very different from zero. In the analysis below, we use the estimate of 0.5, which reflects a typical level from the literature corrected for transitory effects, is somewhat below Duquette's direct estimates (which do not correct for transitory effects) and somewhat above the income weighted estimates. We also discuss the effects of alternative choices.

Effect of the Caps in the Original Version of H.R. 7

A second issue is the effect of the caps. In order to estimate this effect, we combine the data in Table 2, which reports on average contributions by income class with survey data that provides a distribution of non-itemizer contributors by size of contribution.⁵ This survey data indicated that 29.3% of non- itemizing households contributed less than $100 and 44.6% contributed less than $200.

Although these numbers suggest a substantial fraction of non- itemizers might affected by the cap, these numbers are much too large to provide a guide to the projected efficiency of H.R. 7. Indeed, we estimate that the share of tax cuts accruing to those under the cap is only 3% initially and only 6% when the provision is fully phased in.

A more detailed explanation of the corrections that should be made in these data is provided in the Appendix. However, some simple approximations will explain why this share is relatively small. To begin with, we need to estimate an aggregate that reflects the shares of joint versus single returns; since only 20% of non-itemizer returns are joint, a number of about 32% is appropriate. (Our actual estimates are slightly different because shares are adjusted by income class.)

But out of the class with contributions below the ceiling, the average deduction will also be below the ceiling, while for those above the ceiling the average deduction will also be the ceiling amount. If, say, the average were only half the cap, the percentage of deductions below the ceiling would fall to 19% (0.32 X 0.5)/(0.32X0.5 +0.68). However, this number is also too large because the data indicate the cumulative share of the population under a ceiling do not rise proportionally with ceiling increases; that is, the function is not a linear one. For that reason, the average contribution in the bottom class tends to be less than half the ceiling: a functional form that fits the data quite well suggests that the average will be about 32% of the ceiling. This correction would lead to about 13% of the cost associated with the group under the ceiling (0.32 X 0.32)/(0.32X0.32 +0.68).

  Table 2: Use of Above the Line Deduction in 1986, at 2001 Income

Two other factors tend to lower the share, although they interact with the correction described above in ways that cause only a small additional reduction. First, the survey data cover the entire population while tax benefits go only to the taxpaying population. About 20% of returns have no tax liability before credits.

Since low levels of donations are associated with lower levels of income, the share of the taxpaying population with small donations should be considerably smaller than the share of the total population. Adjustments for these effects suggest that about 24% of the taxpaying population would fall below the caps.

Finally the caps were phased in and began at $25/$50 rather than $100/$200. Moreover, even as caps increase in nominal value, they are also declining in value relative to income, so the shares will fall below 10%. An adjustment is also made for these effects, which vary by year, but even in the year that benefits are fully phased in, the caps are cut approximately in half. (Further details of the methodology used to make these adjustments are provided in Appendix A.)

Table 3 presents the estimated share of revenue loss that falls under the cap as a result of these calculations. After 2011 the share would continue to decline as nominal values fall relative to incomes.

  Table 3: Share of Revenue Cost Accruing to Taxpayers Subject to

Target of Induced Contributions

A final issue that would affect the target effectiveness of the non-itemizers deduction is what types of contributions are induced. This issue is particularly important because of the amount of giving to religious organizations. To the extent that those contributions provide for member benefits, they may be less likely to efficiently address the market failures in charitable contributions identified by economic theory or the objectives stated directly in the legislation (to provide for private provision of services that might otherwise be provided by the government).

The most important recipient of all charitable giving is religion, which accounted for 36.5% of the total $203.45 billion of giving in 2000.⁷ It is almost three times as large as the next largest recipient area (education) which accounts for 13.8%. The remaining categories (health, human services, arts, culture, public/society, environment and international affairs) account for less than 10% each. Religious giving accounts for an even larger share, 43.4%, of giving, excluding foundations. (Foundations account for 12% of giving, individuals for 75%, bequests for 7.8% and corporate giving for 5.3%.) However, giving to religious organizations is even more pronounced among individuals at lower and moderate incomes, as shown in Table 4, which reproduces data from 1992, restated at 2001 levels of income.

When the shares in Table 4 are matched with the data in Table 2, the results indicate that 73% of contributions for non-itemizers in general, and 74% for non-itemizers who fall below the cap, are made to religious organizations.

Of course, religious organizations engage in activities that provide benefits beyond the local congregation and even outside of strictly religious functions. A study of the disposition of these funds, however, indicated that about 70% of the receipts of religious organizations go to provide sacramental services and similar services for members.⁸ Thus, assuming that those under the caps have similar characteristics in the choice of giving to those over the caps, one could make a case for excluding 52% of induced contributions (0.7 times 0.74) in measuring the efficiency of the charitable deduction provision.

   Table 4: Recipients of Individual giving by Income Class, 2001

In this section we combine the three estimating parameters to estimate the induced giving in cents per dollar of revenue cost. These estimates are provided in Table 5, and they indicate that initially, excluding religious member services, each dollar of revenue lost will result in one cent of induced contributions. This amount will gradually rise to two cents before beginning to fall. If all giving is included, the induced amount begins at slightly over two cents, gradually rising to almost four cents.

    Table 5: Estimated Cents of Charitable Giving Per Dollar of

By any of the measures, very little induced giving occurs as a result of the capped itemizer deduction in H.R. 7 or even in S. 1924. The most important limiting factor is the cap on deductions which causes most of the revenue loss to accrue as a benefit (or windfall) to individuals who are already giving in excess of the cap. However, the effectiveness is also limited by the relatively low price elasticity (which alone would cause only 50 cents of induced giving per dollar of revenue loss, even without the caps or exclusion of religious member services). A deduction with no caps would result in only 24 cents of induced giving out side of religious member services. Thus, all three factors act to limit the effectiveness of the deduction for non-itemizers.

These numbers are, of course, sensitive to the estimates of the share affected and the estimates of price elasticity and exclusion of member religious services. Evidence suggests, however, that the price elasticity, if anything, is probably smaller. Using Duquette's elasticity estimate weighted by income would result in a weighted elasticity of 0.29 without the caps and 0.26 with the caps, even using the higher 1986 estimates. With the lower 1985 estimates, the elasticity would be 0.12 without a cap and 0.10 with a cap. Estimating the share attributable to caps is also subject to a number of potential uncertainties because of limitations of the data.

It is possible that small contributors are less likely to be making regular contributions to churches and the estimates in column two may be somewhat understated. At the same time the estimates do not take into the account the possibility that a lot of small deductions may simply be claimed without making a contribution, given the fact that such small deductions are likely to be unchallenged by the tax authorities.

Note that the provision allowing IRA rollovers will have a greater impact per dollar of cost than the deduction for non- itemizers, because the effective caps are likely to be larger.

Effects of a Floor

A capped deduction is less costly than an uncapped deduction but also less efficient. It is possible to increase the efficiency level by use of a floor and the Senate adopted a proposal that has a cap as well as a floor.⁹ As the Appendix shows, this same data can be used to estimate a floor; in the case of a $500 ($1000 for joint returns) floor, we estimate that each dollar of revenue loss will induce 64 cents of giving (and 32 cents if religious services were excluded). The floor increases the power of the effect (which would otherwise be 50 cents total on the dollar with a deduction with neither cap nor floor), and an important constraint is the elasticity. A floor, however, tends to shift the distributional effect to higher income individuals compared to a ceiling.

One approach that is likely to be even more efficient and which does not shift the distribution as much is to have a floor as a percentage of income. It is more efficient because the floor can be moved up to reflect the higher giving levels of higher income individuals. While our data do not permit the full projection of large floors (and thus effects can be calculated only at lower incomes, initial results suggest that this approach would be considerably more effective per dollar of revenue. For the $12,000 to $24,000 income class, a floor that was 2% of income was estimated to induce 81 cents of contributions per dollar. In the $24,000 to $36,000 class, each dollar of loss is estimated to induce $1.04 of contributions. In the $33,000 to $48,000 income level, each dollar of loss is estimated to induce $1.31 of contributions. These effects would, of course, vary with the percentage of income, falling as the percentage falls and rising as it rises, and about half would go to religious services.¹⁰

It may be argued that individuals should be able to deduct charitable contributions separately because they do not benefit from them. However, economic theory does not provide a particular justification for deducting charitable transfers on equity grounds since individuals freely make such contributions and derive some benefit from them (even if the benefit is only a good feeling). Moreover, individuals who take the standard deduction generally elect to take it because it is larger than their itemized deductions, including charitable contributions.

Another claim that might be made is that charitable giving directly benefits the poor as recipients of funds. Of course, as indicated above, very little of the revenue loss translates into additional giving with caps. In any case, only a small fraction of money that goes to charity is focused on poor people. For example, only about 6% of giving to religious organizations is estimated to go to benefit poor individuals. And while the share benefitting the poor is somewhat higher in some other categories, we estimate that no more than 10% of all charitable giving directly benefits the poor.¹¹ Moreover, very little of this amount is in the form of cash transfers.

In addition, because of the small sums involved with a cap and the lack of a requirement for record-keeping for these small sums, the allowance of an deduction for charitable contributions may induce more than average amounts of tax evasion. Many individuals will probably figure they can safely report a small amount of fictitious contributions without any consequences, although there were many individuals who made no contributions at all when the non-itemizers deduction was allowed in the early 1980s.

An approach that would have an even more pronounced effect on targeting the marginal contribution would be to institute a floor, rather than a ceiling, as is the case for medical deductions. Such a floor increased the bang for the buck somewhat for the version of S. 1924 reported by the Senate Finance Committee, but its effect was still constrained by the cap. A flat dollar floor would simplify administration and reduce false claims of deductions. A floor based on income would be most effective, although more complicated to compute. A floor might also direct even more of the benefits to religious activities that provide direct services to the contributors.

The third option is to use the revenue that would have been directed at the tax cut to fund spending programs aimed at the same objective, either through a direct government program or a grant that might be administered by a private entity.

We cannot directly use these latter numbers for several reasons. First, these survey data on distribution of contributions by size of contribution cover individuals who would have no tax liability, and thus would not have a benefit from the deduction. Since there is a correlation between small contribution size, small income and lack of tax liability, the percentage would be smaller. Secondly, the dollar limits depend on filing status (joint or single) so that the numbers must be weighted to reflect ceilings averaged over filing status.

To allocate these amounts over incomes, we used the average contribution in each income class from Table 1 as a guide to the differential share of individuals in each class below the cap, by measuring the share that would have fallen below the contribution limit based on a uniform distribution of contributions. This approach produces a total number that is too small (because there are likely to be some individuals in each class with very large contributions that affect the average in a way that a uniform distribution would not). The formula for this ratio is the dollar cap divided by twice the average contribution. This method produces shares that begin at 24% in the lowest class and decline to 12% in the next bracket, 10% in the next and so forth, with an average of 10.4% for all of the income classes.

The $200 and $100 amounts were computed separately and then weighted. In the case of the $200 amount, we begin with an overall average of 44.6%. However, some of that amount reflects individuals who might have not claimed a deduction because of lack of tax liability. We made an adjustment only in the lowest bracket by assuming that the differential between the share claiming the deduction in that bracket and the next bracket was part of the reported share. This amount of 6.3 % reduces the overall average for taxable returns under $200 to 38.3%. A similar procedure reduces the share for the $100 and over contribution class to 23.0%. This adjustment is imperfect, since it assumes all of that amount is due to contributors (overstating the adjustment) but makes no adjustment for the next bracket, where some returns might not be taxable as well (understating the adjustment).

To obtain the totals, each ratio was multiplied by the share derived from the uniform distribution calculation, so that for the $200 category the share in the lowest class was 90% (38.3/10.4 times 0.24), the next class was 46% and so forth. A similar procedure was calculated for the $100 class. In calculating the totals, however, an exclusion of returns expected to have no tax liability under current tax law was made. For the joint return ($200) number, the first bracket was excluded entirely. For the single or head of household return ($100) number, two thirds of the first bracket was excluded based on the exempt levels of singles and heads of household, and the share of returns that are heads of household in that category. The resulting effect of excluding this total was to reduce the under $200 share from 38.3% to 36.9% and the under $100 share from 23% to 22.5%.

Each income class was then weighted according to the share of joint returns filed in that class (based on 1997 data inflated to 2001 income levels). This procedure yielded a total of 24.1% of returns with contributions under the cap.

This 24.1% number is, however, too large to use in estimating contributions induced per dollar of revenue loss. First, those over the cap have a full $200/100 deduction, while those under the cap have only a partial deductions. Thus, it is necessary to estimate the average size of contributions for those below the cap. Secondly, the returns above the cap will have higher marginal tax rates than those under the cap, because the share under the cap falls as income rises. To correct for this effect, we weighted the tax rates for the cap and the tax rates for all returns, and found those with a cap had tax rates approximately 70% of those over the cap. Finally, the caps begin at a smaller level and are phased in nominal terms. In 2002 and 2003, the ceiling is only a quarter of the size of the long run amounts: $25 and $50, rather than $100 and $200. And, the dollar amounts do not rise with income; even the data on the shares with $200 and $100 ceilings from 1999 must be corrected for growth over that time. That is, to maintain the shares of individuals under the caps, the $100 and $200 amounts must have been increased with income growth. Adjusting by growth rates that have already occurred in output and by the projections of the Congressional Budget Office, a $100 deduction in 1999 must have growth by 76% to maintain its relative value in 2010, which is the same as saying, for purposes of applying the data from 1999, that the 2010 deduction is only 54% of its value in 1999.

In adjusting the shares for these lower real amounts, however, another problem is projecting from the data (44.6% giving less than $200 and 29.3% giving less than $100). This function is not linear; that is, if we want to examine the effect of a deduction 25% as big, we cannot simply use a share a quarter as large.

The data suggest some sort of exponential function relating cumulative shares of the population to dollar amounts. We considered two functional forms. The first is a function derived from an exponential probability distribution of the form:

(1) S = 1-exp(-xp),

where S is the cumulative share of the population falling below the dollar value of x, with x ranging from zero to infinity.

The second functional form was:

(2) S = Ax^p

To project these relationships, we used all of the observations of shares up to $1000, assumed an exponential function and fitted the curve. Both curves performed well statistically; however, the first functional form did not match shares very well at the lower end of the distribution. The second function matched these shares very closely, but had the disadvantage that probabilities do not sum to one. We chose the function that fit most closely at the low end of the scale, and which yielded an exponential curve with a power of 0.4476.¹² In general, a limit 25% as large would lead to a share 52% as large (0.25^.4476). It would also indicate that the average contribution within a cumulative interval is p/(1+p) times the ceiling of that cumulative interval. (This amount is obtained by integrating the density function pAx^(p-1) times x, and then dividing by the cumulative probability). This ratio amounts to about 32% (0.4476/(1+0.4476)), indicating that the average contribution in the group of contributors that gave less than $100 is $32. This amount is less than the average of $50 that would occur with a linear relations. Thus the functional form causes a larger share of individuals to remain under the cap compared to a linear form, but a lower average contribution for those individuals.

The results will differ somewhat, however, depending on what assumptions are made about where the contributors below the limit that have no tax liability fall relative to the contributors below the limit. If these two groups are scattered randomly across the entire population share subject to the caps one can just multiply the population ratios by the original share (of 24.1%) and the constant share by the new dollar ceiling.

This assumption is not entirely reasonable, and an alternative one is that nontaxable contributors take up the very bottom of the distribution and taxable contributors are stacked on top of these individuals. This approach understates somewhat the share of the taxpaying population that is covered, especially in the early years, but overstates the average contribution within that population. In this case, the ratio of individuals included would be multiplied by the share of all individuals (which we calculate at about 35% based on the ratios of joint and single returns) and then subtract the amounts associated with the non-taxable share (which we estimate at about 10.9%). The taxable population share goes down by much more than in the approach used above. However, the average contribution using this approach is much higher: (p/(1+p)X(1+1.09/(NTPS))) [where NTPS is new taxable population share] and also multiplied by the difference between the new ceiling and the ceiling associated with the distribution running up to 10.9%, divided by the new ceilings).

Based on these relationships we predict the shares of individual revenues that would be covered under the cap as shown in Table 3. The formula for dollars spent affecting taxpayers at the margin is (rX0.7/(r*0.7*s+(1-s))) where r is the ratio of the average contribution to the ceiling, and s is the share of the taxable population falling under the cap.

For the $400/$800 cap in S. 1924, the share is estimated at 24%. Data presented that report for 1999 indicated that a significant fraction of contributions were made by individuals contributing less than $800 and $400, 81.3% and 65.4%. As noted above, these numbers are much too high because they must be adjusted for the distribution across taxable returns, and weighted for the share of returns filed by single individuals, who tend to be more important at the lower income levels. These adjustments led to an estimate that about 56.6% of individuals would have contributions under these amounts. When adjusted for the lower value of deductions granted in 2002 compared with 1999 income levels, this adjustment resulted in a share of 50.4%. That is, 50.4% of the individuals who make contributions are likely to fall under the ceiling and thus are likely to have a marginal inducement to contribute. We estimate that the average amount given would be approximately 39% of the cap, while the share of those above the cap would be the cap it self. This factor would cause the share to be 28%. In addition, the tax rate of individuals under the cap will be lower than the rate of those over the cap, and we estimate the ratio of tax rates to be about 70%. This effect would cause the share of revenue to fall to 23.8%. That is, only about 24% of the revenue cost would accrue to individuals where an incentive to give would still exist.

To estimate induced giving, we used an elasticity of 0.5, which resulted in an additional giving of 12 cents (0.5 X 23.8) per dollar of revenue loss. These estimates are, if anything, likely to be a little high. Since about 70% of giving by non-itemizers is to religious organizations and about 70% of that amount is for sacramental services and similar services for members, about half, or 6 cents per dollar, would go to these member services.

In the analysis of the effects of a floor, the analysis uses the estimates of the number of individuals who fall below the floor and their average deduction, along with overall average deductions to determine the average deduction for those above the floor. That is:

AD = sAD_b+(1-s)AD_a

where AD refers to average deduction, s is the share of individuals giving below the is the floor, ADb is the average deduction for individuals below the floor and ADa average deduction for those above the floor. We consider the case of a floor of $500 for a single return and $1000 for a joint return.

With these floors, each dollar of revenue loss is estimated to lead to about 64 cents of induced contributions, with about half of that amount going to provide sacramental or other religious services. According the data referred to above, a significant fraction of contributions were made by individuals contributing less than $1000 and $500, 85.1% and 72.0%. Again, these raw numbers are too high because they must be adjusted for the distribution across taxable returns and weighted across taxpayers, resulting in a share of 61.9%. After adjusting for income levels, the share is estimated at 56.2%. That is, 56.2% of the individuals who make contributions are likely to fall under the floor.

For these individuals, we estimate that the average amount given would be approximately 39% of the floor. Using the formula above, and weighting across returns and filing status, we estimate that the average individual contributing over the floor amount would contribute 4.575 times the floor. The floor makes the contribution an estimated 1.28 times as effective for these individuals as a deduction allowed without a floor (4.575/(4.575-1)). This amount must be multiplied in turn by the price elasticity, which is estimated at 0.5, to yield an induced contribution per dollar of 64 cents (0.5 X 1.28). As with other estimates, about half of this amount would go to fund religious services and the remaining half would go to other purposes.

For a moving floor, the shares are estimated for each floor in each income class, the get data on shares and giving ratios. Unfortunately, not all of these effects can be estimated with the data available, since these data do not extend above $1000. Thus, it is impossible to extrapolate above that level.

The results for both a floor and a ceiling as estimated for the Senate proposal indicate that 40.2% of contributors fall below the floor and about 56.2% fall below the ceiling. On average, the tax rates of individuals below the ceiling are about 70% of those above the ceiling (0.177 compared to 0.254). Individuals below the floor have contributions that average about 41% of the floor while those below the ceiling have contributions that average 39% of the ceiling.

The formula for calculating this effect is:

(0.562*.39*C-0.402*0.41F)/(0.562*0.39*C- 0.402*0.41*F+.438*C*(0.254/0.177)-(0.562-0.402)* F-0.438*F*(0.254/0.177)

where C refers to ceiling and F refers to floor. The numerator is the share falling under the ceiling times their average contribution, minus the share falling under the floor times their average contribution. The denominator adds to this term the share with a contribution above the ceiling weighted by their higher tax rate, and subtracts the amounts falling below the ceiling for each of the two groups (above the ceiling, and between floor and ceiling).

This ratio is approximately 37%, indicating that 37% of the revenue loss goes to individuals who would have a marginal incentive. Multiplying by an elasticity of 0.5 indicates that about 18 cents of induced contributions occur per dollar of revenue loss. About half of these would reflect contributions supporting religious services.

² Ibid.

³ See Charles Clotfelter and Richard L. Schmalbeck, "The Impact of Fundamental Tax Reform on Nonprofit Organizations," in Economic Effects of Fundamental Tax Reform, ed. Henry J. Aaron and William G. Gale, Washington, D.C., The Brookings Institute, 1996 for a general discussion. The study correcting for transitory effects is Randolph, William C., "Dynamic Income, progressive Taxes, and the Timing of Charitable Contributions," Journal of Political Economy. Vol. 103 (August 1995), pp. 709-38.

⁴ Christopher M. Duquette. "Is Charitable Giving by Non-Itemizers Responsive to Tax Incentives? New Evidence." National Tax Journal Vol. 52, No. 2, June 1989, pp. 195-206.

⁵ Compiled by Independent Sector, 1999.

⁶ The Congressional Budget Office (CBO) has also done a study of effects of ceilings and floors. See Congressional Budget Office Paper, Effects of Allowing Nonitemizers ro[sic] Deduct Charitable Contributions, December 2002.

⁷ Giving USA 2001. American Association of Fund Raising Counsel.

⁸ See Jeff E. Biddle, "Religious Organizations" In Who Benefits from the Nonprofit Sector, edited by Charles T. Clotfelter, Chicago, University of Chicago Press, 1992. According to the article, 70% of these transfers go to provide for sacramental services and similar services for the members.

⁹ The effect of a floor is also estimated in "Extending the Charitable Deduction to Nonitemizers: Policy Issues and Options," by Joseph Cordes, John O'Hare, and Eugene Steuerle, In Charting Civil Society, No. &, May 2000, the Urban Institute.

¹⁰ The CBO study, Effects of Allowing Nonitemizers to Deduct Charitable Contributions, op cit., finds most proposals with a floor to have small effects in excess of a deduction without a floor, however.

¹¹ These estimates are based on the articles in Who Benefits from the Nonprofit Sector, edited by Charles T. Clotfelter (Chicago: University of Chicago Press, 1992). Further details are available from the author.

¹² For the first seven increments of $100, the shares are 0.293, 0.153, 0.122, 0.086, 0.066, 0.049, and 0.025. The next $300, from $700 to $1000, account for 0.057. Despite the small number of observations, the constant term and exponent were statistically significant and precisely estimated.