On Estate Tax Repeal and Charitable Bequests

On Estate Tax Repeal and Charitable Bequests

Article posted in Transfer Taxes on 30 June 2009| comments
audience: National Publication | last updated: 18 May 2011


Legislation enacted in 2001 calls for the repeal of the federal estate tax in 2010, and its full reintroduction in 2011. That law has since generated considerable interest in how changes in the estate tax and its ultimate repeal may affect charitable bequests and the fortunes of philanthropies. In this article, David Joulfaian, from the Department of the Treasury's Office of Tax Analysis, offers his personal analysis and opinion on this issue.

By David Joulfaian
Office of Tax Analysis
Department of the Treasury
Washington, DC 20220

This paper reviews trends in charitable bequests. It also presents a critical review of the recent literature on the effects of the estate tax.  The views presented are those of the author and do not necessarily reflect those of the Department of the Treasury.

Table of Contents

  I.  Introduction
 II.  Pattern of Giving
         A. Trend in Bequests
         B. Profile of Donors
         C. Control for changes in the Tax Base
II.   Econometric evidence
         A. Tax Structure
         B. Model Estimation
         C. Simulation and Predictions
III.  Concluding Comment  


I. Introduction

Legislation enacted in 2001 calls for the repeal of the estate tax in 2010, and its full reintroduction in 2011. That law has since generated considerable interest in how changes in the estate tax and its ultimate repeal may affect charitable bequests and the fortunes of philanthropies. With changes in the nation’s fiscal condition as well as in the legislative environment, the estate tax is back in the news. With reported $20 billion in annual charitable bequests1,  the tax treatment of bequests promises to remain topical.

This paper summarizes the pattern of bequests by the wealthiest estates. It examines the trend in giving over the past two decades, the profile of donors, and bequests during tax regimes that control for spousal bequests. More importantly, it also provides a critical review of the economic literature and the empirical evidence on the effects of estate taxation.

Because the estate tax favors charities over heirs by exempting donations, lower estate tax rates or outright repeal may discourage giving. On the other hand, lower estate tax rates, by increasing disposable wealth may encourage giving. The net effect of the two is an empirical question and there is little agreement on the magnitude of this effect. However, different findings in the literature often reflect differences in estimation methods and data sources, as well as differences in the tax parameters used, which makes it rather difficult to evaluate the findings.

II. The Pattern of Giving

A. Trend in Bequests

Estates valued in excess of a statutory threshold are required to file tax returns. These returns provide information on the size and composition of wealth as well as its disposition, including bequest to charities.  Table 1 captures the trend in such reported bequests over some 20 years, along with information on the relative frequency of giving as well as the share of wealth bequeathed.2

Charitable bequests grew from an estimated $3 billion in 1982 to $17 billion in 2004, peaking at $20 billion for decedents in 2001. Some of this expansion can be explained by growth in the economy and the appreciation of assets in the portfolios of estates. This trend is also influenced by the structure and administration of the estate tax. Raising the filing threshold, for instance, reduces the number of estates required to file, thus reducing the value of reported bequests. Alternatively, not indexing it for inflation increases the pool of estates required to file.

The trend also reflects changes in the sampling scheme for selecting estate tax returns.  Sample estimates of figures reported on tax returns are representative of the filing in which precision is greater in some years than others. The figures reported in the shaded rows of Table 1, for instance, are more reliable as they reflect years where the sampling rate was high, and those years form part of the stratification scheme.3

Because the value of the underlying assets also changes over time, bequests measured as a fraction of the reported wealth may be a better gauge of generosity. Using this measure, in which wealth is defined as net worth less estate expenses, bequests increased from 6.5 percent of wealth in 1982 to about 10 percent in 2004, but fluctuated considerably in the intervening years. A similar pattern is observed for the fraction of estates giving where 16 percent of the estates provided for bequests in 1982 compared to 20 percent in 2004, but again with varying relative frequencies of giving between the two years.

One is tempted to argue that changes in estate tax rates have had little effect on the pattern of giving over the two decades. But, again given the changes in the filing threshold, this is a difficult call to make. As an alternative, Table 2 replicates Table 1 but restricting the figures to estates with wealth in excess of $10 million, stated in $2004. These estates are sampled with certainty in most years.4  There are fewer of them, less than 5 percent of all estate tax filers, but they account for over half of the bequests when compared to the figures in Table 1. The share of wealth transferred was 18.7 percent in 1982, and fluctuated considerably over the years. Similarly, the fraction giving fluctuated over the years between 44 percent in 1982 and 37 percent in 2004. The maximum estate tax rate declined from 65 percent in 1982 to 55 percent for the years 1984 through 2001, and to the temporary lower rates introduced by EGTRRA. As with Table 1, there is little that the aggregate data on bequests can tell us on the effects of the estate tax over giving, notwithstanding all the changes that took place over the years.

B. Profile of Donors

There is a considerable amount of information available on the profile of estate tax decedents and donors, and no need to reproduce this here.5  However, little is know about the extent of donor’s generosity. For example, are the bulk of bequests made by few donors who transfer much of their terminal wealth, or do such transfers represent a small fraction of the wealth of many donors? Depending on which trend dominates estate taxation may elicit a different behavioral response. For instance, estate tax changes are less likely to impact those who leave their entire estate to charity.

Table 3 tabulates bequests reported by the estates of decedents in 2001, disaggregated by the degree of their generosity as measured by the share of wealth bequeathed to charity. Less than 20 percent of estates provide for charitable bequests. Those estates have a net value of $60 billion of which $10.7 billion is paid in estate tax and $20 billion is transferred to heirs net of tax6; the remainder is spousal bequests.

These donor estates exhibit considerable diversity in their pattern of generosity.  The bulk of bequests, some 83 percent, are reported by estates that transfer over one half of wealth to charities. Not surprisingly, the latter pay little in estate tax. Those who donate 99 percent of their wealth, for instance, bequeathed $5.6 billion to philanthropies and paid only $2 million in estate tax and their heirs received $5 million in inheritances. An interesting question is whether the repeal of $2 million in tax liability would in any way impact the bequests of $5.6 billion.

C. Control for changes in the tax base: the marital deduction

Comparisons of bequest patterns over time may not be very informative. Changes in tax rates are often accompanied by changes in the tax base. An example of this is the introduction of the unlimited marital deduction in 1982 which almost doubled the previous limit on deductible spousal bequests. As a consequence, spousal bequests increased which in turn inflated the future estates of surviving spouses.7

Because tax reforms and law changes commonly entail tax rate as well as base changes, disentangling the effects of rate changes can be difficult. To isolate the effects of tax rate changes on an economic activity, one needs to control for changes in the tax base. The introduction of the unlimited marital deduction is the primary and largest source for the break in the data on reported wealth and bequests. The cleanest comparison of giving is to contrast the bequests of widowed decedents in 1976 to those in 1982.8   The maximum tax rate in effect in 1976 was 77 percent and 65 percent in 1982 (but expected to decline to 50 percent by 1985).

The wealth reported by widowed decedents in the two periods roughly reflects the 1976 tax treatment of spousal bequests, as the full marital deduction took effect for married decedents in 1982. Thus, we observe the pattern of giving to charity in the presence of exogenous variations in tax rates, as well as wealth measures for widowed decedents that are least likely to be contaminated by changes in the marital deduction.

Data on estate tax decedents in 1976 are available only for returns filed in 1977; returns filed in 1976 and after 1977 are unavailable. The sampling rate is 20 percent for returns with gross estate between $200,000 and $500,000; 100 percent for those over $500,000. In contrast, population data for 1982 decedents is available for returns filed in 1982 through 1984 (and beyond) with estates exceeding $1 million sampled with certainty (100 percent). The less wealthy are sampled at an average rate of 30 percent. Although estate tax returns must be filed within 9 months of the date of death, some are filed much later.  Late filers are likely to be different in terms of wealth and sophisticated estate planning.  Thus, to enhance the comparability of the two data sets, I limit the data on 1982 decedents to estate tax returns filed in 1983, discarding those filed in 1982 and after 1983. Also, only estates in both years valued more than $300,000 in 1982 dollars (the SOI sampling threshold for returns filed in 1983) are considered.

To motivate the analysis, I first restrict the sample to widowed and married decedents. Their pattern of charitable bequests over the two periods is summarized in Figure 1; the reported wealth categories correspond to the tax rate schedules in effect over the two periods. Not surprisingly, Figure 1 shows the share of wealth transferred to rise with wealth. But the share of wealth transferred is generally lower for estates in 1982 than their counterparts in 1976 when tax rates were higher. This is particularly true for the wealthiest of estates which may lead us to conclude that lower tax rates depressed giving in 1982, as is commonly reported.

However, and as demonstrated in Figure 2, much of the trend observed in Figure 1 is reversed when married decedents are excluded and the focus is restricted to widowed decedents only. Indeed, in the case of the wealthiest of estates (those in excess of $20 million) the share of wealth transferred almost doubles.  Despite the tax rate reductions, the “generosity” of the very wealthy seems to have increased.

Figure 3 sheds some light on the diverging trends observed above. Married decedents, across all wealth cohorts seem to leave smaller bequests to charity in 1982 compared to the trend observed for 1976.  In contrast, and more telling, Figure 4 exhibits a surge in spousal bequests for all wealth categories, which is likely to have taken place at the expense of charitable bequests. This does not necessarily suggest a reduction in generosity as much as a change in timing, and it highlights the difficulty in identifying the effects of the estate tax on giving.

III. Econometric evidence

Because bequests to charitable organizations are deductible in computing the estate tax liability, estate taxation lowers the price of such transfers relative to bequests to children and other heirs. At the very same time, the estate tax lowers after-tax terminal wealth and the potential size of inheritances. Those tendencies raise important policy considerations about how changes in estate tax rates, including the elimination of the estate tax, may affect giving.

All of studies using estate tax data find large tax price elasticities, suggesting that the deductibility is a significant stimulant to giving.9  Many also find large wealth elasticities, suggesting the estate tax, by lowering bequeathable or disposable wealth, has a dampening effect on giving. However, identifying the effects of progressive estate tax rates separately from wealth represents a serious challenge in evaluating the effects of estate taxation especially because only cross sectional data are available, reflecting the uniqueness of death.  Joulfaian (2000b) employs nonlinearities in the tax rate schedule as well as variations in the tax treatment of various assets, among others, to address this concern.  But more recent studies resort to pooling cross sectional or aggregated time series data over a long period where numerous changes in tax regimes have taken place.

Kopczuk and Slemrod (2003), hereafter KS, employ time series analysis of aggregate data on bequests to discern how variations in tax regimes over time influenced the observed trend in giving.  More specifically KS regress annual aggregate charitable bequests reported on estate tax returns on a pseudo measure of the estate tax rate. They conclude that the effect of the estate tax can be larger than what has been reported earlier, an implicit reference to the predicted 12 percent reduction in bequests reported in Joulfaian (2000a), but do not report estimates of this effect.

Bakija, Gale, and Slemrod (2003), hereafter BGS, refine the work of KS and use “pooled” grouped data where estate tax returns of widowed decedents for decedents in 1924-1945, 1969, 1976, 1982, and 1985-1998 are aggregated into five wealth categories: $400,000 to $750,000, $750,000 to $1.25 million, $1.25 to $2 million, $2 to $5 million, and $5 million and over. BGS report that estate tax repeal would increase the price by 77 percent and after-tax wealth by 24 percent reflecting the “progressivity” of the tax, and conclude that “total repeal will cause charitable bequests to decline … unless the wealth elasticity is more than three times as large as the price elasticity.” Like KS, estimates of the potential effects of estate taxation are not reported by BGS.

Using elasticity parameters from BGS and applying them to a sample of estate tax returns filed in 2001, the first two authors, Bakija and Gale, report estate tax repeal would reduce charitable bequests by 37 percent. They also argue that the rising tax rate in the 1980s and 1990s explain the growth in charitable bequests. They are also critical of the work of Joulfaian (2000), in particular for using un-weighted observations in simulating the effect of estate tax repeal.

McClelland (2004) revisits the estate tax effects on bequests by employing samples of estate tax returns filed in 1999 and 2000. Using the coefficients from Joulfaian (2000b) he finds that estate tax repeal reduces bequests by 20 percent and not the 12 percent reported earlier. Similarly, using the coefficients in BGS, and the best measure of the change in wealth from estate tax repeal, he reports a 22 percent reduction and not the 37 percent reported earlier.10  McClelland also arrives at different predictions from equations he estimated for the two years.

Last, Joulfaian (2005) uses comparable data on the estates of widowed decedents in 1976 and 1982 to re-examine the effects of the estate tax on charitable bequests. The findings are almost identical to those reported in Joulfaian (2000b) even after addressing the concerns discussed above. Obviously there are more factors that may explain the variance in the findings than has been discussed thus far.

There are at least three critical elements to estimating the impact of the estate tax in addressing the validity of the findings in the literature. The first, which relates to the structure of the tax, addresses issues related to the progressivity of the tax, the evolution of the tax base over time, and the tax treatment of the beneficiaries. The second concerns model specification and estimation, including the type of data employed. And last, but not least, how one simulates the effects of changes in the estate tax is also an important consideration. Below is an expanded discussion.

A. Tax Structure
1. Progressivity

High marginal tax rates may stimulate giving by significantly reducing the tax price of giving. High average tax rates, however, may discourage giving by reducing disposable wealth. The disparity between the marginal and average (or infra-marginal) tax rates is an important determinant of how estate tax repeal may affect bequests. If the marginal tax rate is higher than the average rate, than estate tax repeal would proportionately reduce the tax price by more than it increases wealth, thereby ensuring that bequests will decline, other things equal.

BGS rely on the progressivity of the estate tax and the disparity in the effect of repeal on changes in the tax price and after-tax wealth (77 vs. 24%). This point is further amplified at great length in Bakija and Gale (2003) who report a large disparity in average and marginal tax rates, and make the case for a large reduction in bequests in case of repeal. Using data from estate tax returns filed in 2001 they report an average tax rate of about 11 percent and a marginal tax rate of 45 percent. This disparity suggests that estate tax repeal would reduce the price disproportionately by more than it increases after-tax wealth, in absolute terms, and it leads to the conclusion that estate tax repeal would significantly reduce charitable bequests.

However, the disparity in the measured average and marginal tax rate is only in part explained by the progressive tax rate schedule, and not at all for the wealthiest estates. For instance, the marginal tax rate in Bakija and Gale is computed before applying the credit for state death taxes while the average tax rate is net of the credit. Roughly speaking, Bakija and Gale compute the marginal rate to reflect the combined state and federal taxes while the average tax rate reflects the net federal tax only. At a maximum estate tax rate of 55 percent, for instance, the net federal tax rate is only 39 percent as it is reduced by the credit for state death taxes.11  For a taxable estate valued at $100 million, estate tax repeal raises the tax price from 0.45 to one. It should also increase after-tax wealth from about $45 to $100 million, but would incorrectly increase it from $61 to $100 million per Bakija and Gale as well as McClelland; wealth increases by 122 percent instead of 64 percent in case of repeal.

The difference is also in part explained by the spousal and charitable bequests as they reduce the size of bequests to heirs. And finally, it is explained by the rate structure including the size of exempted estate. The top panel of Table 4 replicates the work in Bakija and Gale and expands on it in an attempt to isolate the effect of tax progression on the average tax rate. Column (6) is a rough approximation of the figures reported in Bakija and Gale. The average tax rate here is slightly over 11 percent for filers in 2001 and less than 16 percent for estates over $20 million.12  If we exclude spousal bequests from terminal wealth, the average tax rate rises to 25.6 percent for the top estates as in column (7). If we further exclude charitable bequests, the average tax rate rises to 40 percent for the rich, and to 19 percent overall. If we add the tax credit back to the estate tax liability, the wealthiest estates would face an average tax rate of 54 percent on bequests to their heirs and 24 percent for all estate tax filers as shown in column (9). In other words, for every dollar in saved bequests, the very wealthy will pay $0.54 in taxes on average, leaving a net of tax inheritance of $0.46. A similar pattern is observed for widowed decedents, as shown in the bottom panel of Table 4.

In short, the progressivity story in Bakija and Gale is at best incomplete. BGS do not face all the shortcomings of Bakija and Gale as they only consider widowed decedents and account for state taxes. But nevertheless, they also overstate the case of progressivity by overlooking the impact of charitable bequests and the endogeneity of tax rates. Many leave the bulk of their estates to charity thereby lowering their average tax rates. This is evident from column (8) in Table 4 for all estates, as well as for those of widowed decedents.

2. Changes in the tax base

The size of terminal wealth is a critical determinant of bequests (indeed a binding constraint) as well as the ensuing tax liability. As such, properly measuring wealth, the tax base, is important in modeling donor behavior. Pooled data over time may exploit changes in statutory tax rates and aid in the identification process. But often tax reforms bring about changes in the tax base and not just in the tax rate schedules. And as such we need to be concerned with the definition of wealth and its evolution. For example, the unlimited marital deduction was introduced in 1982, almost doubling the allowed deduction at the same time as lower tax rates were being phased-in. The expansion in spousal bequests led to an increase in the size of estates of widowed decedents as well as the postponement of bequests.

Figures 3 and 4, combined, make the case that potential findings from longitudinal data on the effects of estate taxation can be biased if spousal bequests and their consequences for the evolution of wealth are not properly controlled for. BGS exclude married decedents. But because increases in spousal bequests increase the wealth of the surviving spouse, and potentially the concomitant tax price, they may also influence giving in the future. Consequently, errors in measuring the budget constraint and the tax price faced by widowed decedents don’t go away.13

B. Model Estimation
1. Aggregation bias

A casual look at Table 3 should show the potential bias of using observations units that lump together estates with different levels of bequests. Given the diverse pattern of giving, aggregation bias is a potentially serious problem in studies that rely on aggregate data.

2. Frequency of tax changes

Large donors are likely to be very wealthy and may also face high tax rates by virtue of the progressive tax rate schedule. Hence it may be difficult to disentangle the effects of wealth on giving separately from those of high tax rates. Resorting to pooled cross sectional or time series aggregate data is one way to address this identification problem because they exploit variations in statutory tax rates, changes that are independent of wealth variations. But this may also introduce a number of other biases, or at the very least exacerbate them.  As Clotfelter (1985, pp. 240) points out, the price term is likely to be measured with error during periods of frequent changes in tax rates because it is not clear whether reported charitable bequests are influenced by current or past tax rates. There is also the question whether planned bequests reflect future taxes, as estate planning, by its very nature, is forward looking.

3. Model specification

Estimates can be sensitive to the functional form of estimated equations. They may also depend on the definition and measurement of variables. For instance, should after-tax wealth be defined after setting charitable bequests to zero, as is the common practice, of assuming full deductibility as in BGS? McClelland provides a textbook discussion of the appropriate definition.

4. Data coverage

Estimates can also be sensitive to the exclusion of certain groups. Joulfaian (1991), for instance, limits the analyses to observations when less than 100 percent of wealth is bequeathed to charity thereby biasing the results in favor of large estate tax effects. Other bias may be introduced by examining periods where the sample design is not representative of the filing universe (1985) or provide a snap shot of decedents in partial years (e.g. returns filed in 1969 and 1977).

C. Simulations and Predictions
1. Weights

Because estate tax data typically are available in the form of stratified random samples where the wealthy are overrepresented, simulating the effects of law changes without using appropriate weights may bias estimates. To their credit, Bakija and Gale highlight the importance of weights. Yet they provide no discussion on the sample designs and how weights may have evolved over time, nor do they provide any evidence on the bias created by not using weights. It is left up to McClelland to provide a measure of the extent of the bias created.  Using the coefficients in Joulfaian (2000b) he simulates the effects of repeal using weighted and un-weighted data for returns filed in 1999 and 2000. He concludes that bequests will decline by 11 and 16 percent in the unweighted 1999 and 2000 data, respectively, and that they’ll decline by 19 and 20 when weighted; these compare to 12 percent in Joulfaian (2000b) and 37 percent in Bakija and Gale.

2. Data

Accurately simulating the effect of tax law changes requires data that allows us to distinguish between donors and non-donors and their respective wealth and tax liabilities. Using aggregate data to simulate the effects of repeal or lowering tax rates is obviously not feasible, or, at the least, cannot be done with any reasonable accuracy.

IV. Concluding Comment

Simulating the effects of estate tax repeal on charitable bequests is a difficult task as our data and models are at best appropriate to evaluate the effects of small changes in tax rates, and not changes as draconian as doing away with the tax altogether. We have little direct evidence on how the wealthy have behaved prior to the enactment of the estate tax. One is tempted to rely on the treatment of spousal bequests and the comparison of charitable bequests pre and post the introduction of the unlimited marital deduction, which effectively repealed the tax for effected estates. But to the extent that spousal transfers are viewed as deferred bequests to children and other heirs, the estate tax liability is simply deferred as well. An alternative line of inquiry and research would examine joint bequests of husbands and wives over time. That exercise would show that the effective tax rate and the generosity of donors are likely to be much greater than published data usually show.14  However, a conceptual difficulty is that the surviving spouse's estate planning may not harmonize with the wishes of her predeceased spouse, which makes it difficult to treat both spouse as one household unit separated by time.

The study of the effects of estate taxation is also hampered by a number of data limitations. Donors, the wealthiest in particular, do not share common attributes that may influence the pattern of charitable bequests which vary over time for reasons we are unable to account and control for. There is little that can be introduced to help explain the pattern of giving observed in Table 2.  We have no way of explaining why donors gave 17.3 percent of their wealth in 1995 and 20.5 percent in 2001. Similarly, we are unable to explain why some leave their entire estate to charity and others leave next to nothing. Overall estimating the effects of estate taxation is a very humbling experience and one that requires constant review of our models and methods.


1. Bakija, Jon and William Gale. “Effects of Estate Tax Reform on Charitable Giving,” Urban–Brookings Tax Policy Center No. 6, July 2003 and Tax Notes, July 23, 2003.

2. Bakija, Jon, William Gale, and Joel Slemrod. “New Evidence on the Effects of Taxes on Charitable Bequests,” paper presented at the January 2005 AEA meetings.

3. Bakija, Jon, William Gale, and Joel Slemrod. “Charitable Bequests and Taxes on Inheritances and Estates: Aggregate Evidence from across States and Time.” American Economic Review, May, 2003.

4. Barthold, Thomas and Robert Plotnick. “Estate Taxation and Other Determinants of Charitable Bequests,” National Tax Journal 37, June 1984, 225-37.

5. Brunetti, Michael J. “The Estate Tax and Charitable Bequests: Elasticity Estimates Using Probate Records,” National Tax Journal 58, 2005, 165-88.

6. Eller, Martha Britton. Federal Taxation of wealth Transfers, 1992-1995, Statistics of
Income Bulletin, Winter 1996-97, Internal Revenue Service, Washington, DC, pp. 8-63.

7. Joulfaian, David. What if Estate Taxpayers Were Viewed as Households Rather Than as Individuals? April, 2007, available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=980294

8. Joulfaian, David. "Estate Taxes and Charitable Bequests: Evidence from Two Tax Regimes," U.S. Department of the Treasury, OTA Paper 92, March 2005.

9. Joulfaian, David. “Taxing Wealth Transfers and Its Behavioral Consequences,” National Tax Journal Vol. 53:4 Part 1, December 2000a, pp. 933-958.

10. Joulfaian, David. "Estate Taxes and Charitable Bequests by the Wealthy," National Tax Journal 53:3, Part 2, September 2000b, pp. 743-763.

11. Joulfaian, David. "A Quarter Century of Estate Tax Reforms," National Tax Journal 53:3, Part 1, September 2000c, pp. 343-360.

12. Joulfaian, David. "Charitable Bequests and Estate Taxes," National Tax Journal 44:2, June 1991, pp. 169-80.

13. Kopczuk, Wojciech, and Joel Slemrod. 2003. “Tax Impacts on Wealth Accumulation and Transfers of the Rich.” In Death and Dollars: The Role of Gifts and Bequests in America, edited by Alicia H. Munnell and Annika Sundén (213–57). Washington, D.C.: Brookings Institution Press.

14. McClelland, Robert. Charitable Bequests and the Repeal of the Estate Tax, CBO Technical Paper 2004-08, July 2004.

  Tables and Figures









Add comment

Login or register to post comments


Group details



This group offers an RSS feed.
7520 Rates: November 2.4% October 2.2% September 2.4%

Already a member?

Learn, Share, Gain Insight, Connect, Advance

Join Today For Free!

Join the PGDC community and…

  • Learn through thousands of pages of content, newsletters and forums
  • Share by commenting on and rating content, answering questions in the forums, and writing
  • Gain insight into other disciplines in the field
  • Connect – Interact – Grow
  • Opt-in to Include your profile in our searchable national directory. By default, your identity is protected

…Market yourself to a growing industry