Tax Policy and the Economy: Volume 32

Contents

Copyright

NBER Board of Directors

Relation of the Directors to the Work and Publications of the NBER

Acknowledgments

Robert A. Moffitt

Introduction

Robert A. Moffitt

The Benefits and Costs of Donor-Advised Funds

James Andreoni

Online Postsecondary Education and the Higher Education Tax Benefits: An Analysis with Implications for Tax Administration

Caroline M. Hoxby

Taxing Humans: Pitfalls of the Mechanism Design Approach and Potential Resolutions

Alex Rees-Jones and Dmitry Taubinsky

Implications of Medicaid Financing Reform for State Government Budgets

Jeffrey Clemens and Benedic Ippolito

Means Testing Federal Health Entitlement Benefits

Andrew A. Samwick

Disability, Taxes, Transfers, and the Economic Well-Being of Women

Bruce D. Meyer and Wallace K. C. Mok

Copyright

© by the National Bureau of Economic Research. All rights reserved.

NBER Board of Directors

© by the National Bureau of Economic Research. All rights reserved.

Relation of the Directors to the Work and Publications of the NBER

© by the National Bureau of Economic Research. All rights reserved.

Acknowledgments

Robert A. Moffitt

Johns Hopkins University and NBER

This issue of the NBER’s Tax Policy and the Economy journal series contains revised versions of papers presented at a conference at the National Press Club on September 14, 2017. The papers continue the journal’s tradition of bringing high-quality, policy-relevant research by NBER researchers to an audience of economists in government and in policy positions in Washington and to economists around the country with interests in policy-oriented economic research. The papers in this issue are wide-ranging, from tax issues in online education to the economics of disability transfers to normative issues in optimal taxation when biased responses are present.

The attendees at the conference were also lucky to hear an interesting lunch presentation by Katherine Abraham of the University of Maryland and NBER, who spoke on the recently released Report of the Commission on Evidence-Based Policy-Making, of which she was co-chair.

I would like to thank Rob Shannon of NBER for his usual expertise and organizational acumen in overseeing the logistical details, invitations, and operational aspects of the conference and to Jim Poterba for his continued assistance with the organization of the meeting. I would also like to thank Helena Fitz-Patrick for assistance in many other aspects of the conference, especially the shepherding of the papers toward final publication. I would like to acknowledge the continued financial support of the Lynde and Harry Bradley Foundation. Finally, let me express my thanks to the authors themselves for the hard work they devoted to producing high-quality papers living up to the Tax Policy and Economy standard.

© 2018 by the National Bureau of Economic Research. All rights reserved.

978-0-226-57752-4/2018/2018-0001$10.00

Introduction

Robert A. Moffitt

Johns Hopkins University and NBER

The six papers in this issue of Tax Policy and the Economy are all directly related to important issues concerning the US tax system, its transfer system, and issues related to its design and effects on revenues, expenditures, and economic behavior.

In the first paper, Andreoni provides a study of Donor-Advised Funds (DAFs), which are financial vehicles offered by investment houses, community foundations, and other charitable organizations to provide savings accounts for tax-free charitable giving. A substantial share of charitable giving, more than 10%, flows through these funds every year. Giving deposited in a DAF can be taken immediately as a tax deduction by the giver, but the DAF sponsor can save and invest the donation to dole out to qualified charities later at the discretion of the original giver. Breaking the link between when the donor makes a deductible contribution and when it is disbursed to a recipient provides flexibility in timing and allows smoothing of contributions relative to income. The DAFs also have advantages of convenience and can also serve as commitment devices. They can also be used to reduce exposure to capital gains taxation by strategically contributing appreciated assets to the DAF, thus avoiding both income and capital gains taxes. Because tax savings are higher for those in higher-income tax brackets and who have more noncash assets with significant capital gains to contribute, DAFs are of most advantage to high-income individuals. Using data that allow him to estimate rates of return to funds in DAF savings accounts, as well as data on how long the funds stay in the accounts before being disbursed, Andreoni shows that the relative size of benefits and costs of DAFs to society (rather than to the individual) hinges on how much new charitable donations they generate, as well as on rates of return, discount rates, the percent of assets in capital gains, and other factors. Some simple analyses conducted by the author surrounding the 2013 changes in income and capital gains tax rates suggest that the amount of new donations generated may not be sufficiently large to outweigh the tax cost.

In the second paper, Hoxby uses data from IRS tax returns and other forms combined with data reported by postsecondary educational institutions to analyze the use of tax credits by students enrolled in online postsecondary education. Hoxby finds that the number of students enrolled in online postsecondary institutions is high and growing, and that they account for some of the highest uses of federal tax credits. For-profit educational institutions account for the bulk of enrollments and tax expenditures. Another important finding is that the type of student enrolled in online postsecondary education is quite different than the traditional school-age student studying full time and living residentially at the educational institution, for online students tend to be older, to often be employed, to more often be their own tax filer, and to not reside at the institution. Their enrollment episodes are often short, including cases where tuition is paid but the student does not complete the course or program. Take-up of tax credits varies across the type of student and the type of credit but, on average and over all types of institutions, is fairly high but less than 100%. Hoxby also examines trends in individual labor earnings from before to after taking the credit, finding that the changes vary with type of institution. Based on these findings, Hoxby outlines several practical steps that might improve the administration of the credit, the accuracy of reporting, and take-up.

Rees-Jones and Taubinsky survey some recent literature on psychological biases by taxpayers that lead to incorrect perceptions and understanding of tax incentives. They present an analysis of how such biases affect optimal tax rules, an old and venerable topic in the economics of taxation. The literature they review shows evidence of confusion, adoption of heuristics, and salience, as well as specific forms that result in ironing and spotlighting. They then go on to show that the presence of these biases disrupts the common two-stage procedure of first deriving optimal tax incentives under a direct mechanism, and then reverse engineering the tax system that induces those incentives. Depending on the nature of the biases, their presence can enhance or reduce welfare under optimal policies, and thus provide useful policy levers for the social planner in some circumstances. More generally, they show that optimal tax formulas can have different effects depending on the exact form of their implementation. In one application, they show that the deviations from the classical Ramsey approaches to tax analysis introduced by information asymmetry may be mitigated in the presence of bias. They conclude by suggesting that an approach to optimal taxation using sufficient statistics may be more fruitful to simultaneously solve the problem of optimal tax formulas combined with implementation, optimized over a limited set of tax instruments.

Clemens and Ippolito provide new research on the implications of block grant reforms of the Medicaid program for receipt of federal support for different states. The Medicaid program currently has an open-ended matching structure where a fraction of additional state expenditures on the program are paid for by the federal government, whereas block grants cap the federal subsidy in differing ways. Some block grant programs cap the per-beneficiary amount that the federal government will pay and make that cap the same across all states, but some block grant proposals adjust the federal subsidy by a measure of state need. Block grant proposals also differ in whether, and how, they are adjusted over time and whether they are adjusted during business cycles. The analysis of Clemens and Ippolito shows that there would be extremely large gains and losses by different states in the amount of federal funds received relative to the current Medicaid program, losses in some cases exceeding 10% of states’ own-source revenues in a uniform need-based block grant. They also show that block grant structures not adjusted for business-cycle conditions could significantly increase the exposure of state budgets to stress during recessions. Finally, the authors provide a discussion of how adjustments in block grants affect state incentives to make adjustments on the extensive margin (number of beneficiaries) versus the intensive margin (expenditures per beneficiary).

In his paper, Samwick addresses the issue of means-testing Medicare and federal health benefits under recent legislation. He notes that the Medicare Modernization Act of 2003 introduced the means-testing of Medicare and that the Affordable Care Act of 2010 introduced premium subsidies, which are inversely related to income, and in both cases the measure of income is as of the current year. Samwick considers an alternative measure of means, which is a long average of covered earnings, similar in spirit to what is used for old-age retirement benefits. He notes that use of current income provides an incentive to reduce savings and work, an incentive to manipulate the level and composition of income, as well as being a noisy measure of ability to pay. All these effects would be reduced if a measure of average earnings were used instead (although current income could be a better provider of insurance against short-term fluctuations in income). His analysis of data from the Health and Retirement Study shows that current income fluctuates enough so that there is considerable short-term variation in the Medicare premium for those who pay it, much more than if average earnings were used, and that current income is often not highly correlated with lifetime earnings. He finds similar larger variation in current income used to determine ACA subsidies relative to lifetime average earnings. Samwick suggests more research into this issue and for future studies to address possible behavioral responses to the nature of income testing.

Meyer and Mok provide a comprehensive examination of the incidence and effects of disability among US women from 1968 to 2015, studying the impacts of disability on income, consumption, and public transfers, as well as its incidence. They note that most research on this issue has concerned men rather than women and has emphasized the work disincentives of programs. They find that women are actually more likely than men to have experienced a disabling event in the early part of their lifetimes, but the nature of the disability tends to be less severe than that of men. For both men and women they find that a disabling event results in a decline in income and consumption, but the magnitudes are smaller for women and depend on the nature of the disability. However, women who experience disabling events have lower than average levels of income and consumption even prior to disability than men, and hence are drawn from a more disadvantaged population. They also find that reductions in income and payroll taxes, and the receipt of public transfers, after a disabling event has a major effect in cushioning the effect of disability on income. The largest transfers come from the Social Security Disability Insurance and Supplemental Security programs, but workers’ compensation and food stamp receipt play an important role as well.

© 2018 by the National Bureau of Economic Research. All rights reserved.

978-0-226-57752-4/2018/2018-0002$10.00

The Benefits and Costs of Donor-Advised Funds

James Andreoni

University of California, San Diego, and NBER

Executive Summary

Donor-Advised Funds (DAFs) are now a major source of charitable donations in the United States, responsible for 1 in 10 dollars donated to charity in 2015. In 2016, Fidelity Charitable, whose only mission is to provide DAFs, became the largest charity in the United States. Paradoxically, most people have never heard of DAFs or Fidelity Charitable. This leads us to ask, who uses DAFs and why, what is the impact of government tax policy toward DAFs, and could the extra fiscal cost of subsidizing DAFs be balanced out by an extra public gain of new charity resulting from tax policy toward DAFs?

I. Introduction

The largest charity in America in 2016 is one that a vast share of the American population has never heard of. The charity is called Fidelity Charitable Gift Fund, and its mission is to manage Donor-Advised Funds.¹ Remarkably, Fidelity Investments, the parent company to Fidelity Charitable, conducted a survey of its investment clients who could benefit financially from a Donor-Advised Fund and discovered that 64% of those surveyed had no idea about Donor-Advised Funds. Yet, in 2015, 10% of all charitable donations claimed on tax returns were made to Donor-Advised Funds. How can they be so popular and important to charities everywhere, yet so widely unknown to potential donors?

In this article, I will introduce readers to Donor-Advised Funds (or DAFs) and the tax policy toward them. I will explain how the sponsors of DAFs, such as Fidelity Charitable, act as financial intermediaries in the market for charitable giving in order to help donors save more tax dollars as they give money to charity. I will also show how the data can reveal how DAFs can be so dominant in charitable giving and yet so unknown. More importantly, I will argue that DAFs are consequential to all Americans for their impact on government tax revenues and on the number of dollars going to charity.² Gathering these components together brings us to the primary purpose of this article: to evaluate DAFs according to standard concepts of benefit-cost analysis. In particular, the analysis will ask what would need to be true for the policy to create more new charitable giving than it costs the government in forgone tax revenues.

As we will see, a foundational reason for a giver to use a DAF is to save additional taxes on a household’s current giving—no increase in giving is required to claim the extra tax savings of DAFs. If they are used this way, they may create no benefits for society, but add significantly to the tax costs to the US Treasury. On the other hand, if DAF donors are motivated to dedicate all of their additional tax savings to their charitable giving, then DAFs would break even as a policy. If donors give beyond this, or create other socially valuable returns, DAFs will be a net benefit as a tax policy.

Although they have become popular only recently, DAFs have been with us since the 1930s, shortly after the introduction of the charitable deduction.³ Had they been introduced as new legislation this year, the Congressional Budget Office would be required to score the legislation to estimate whether the social value of the proposal outweighs the shared costs, and would make conjectures about the incidence or distributional aspects of DAFs. This is difficult, however, as the legal rules surrounding DAFs protect the individual DAF accounts from public scrutiny. But, as I hope to convince you, we can learn quite a bit about the flows of benefits and costs of DAFs to make meaningful comparisons of the two.

As we continue, it is important to keep in mind that the objective of this exercise is to look at things from the point of view of a disinterested taxpayer. That is, we should not concern ourselves with how this institution of Donor-Advised Funds affects our own giving, tax bill, social esteem, fundraising goals, prestige, or self-image. Our job is to learn whether our country as a whole has made a good bargain when extending extra tax preferences to those who give through Donor-Advised Funds, or whether DAFs reduce the efficiency of the current system of subsidies to giving.

The next section will review charitable tax policy, including DAFs. Section III will discuss how DAFs save tax payments and loosen constraints set by other tax policies toward giving. Sections IV and V will discuss the concepts of benefit-cost analysis and derive the parameters for our analysis. The benefit-cost calculation will be presented in section VI and discussed in section VII. Section VIII is a conclusion.

II. What Is a Donor-Advised Fund?

Before talking about DAFs, it will help to first discuss standard tax policy toward charitable giving. We can then contrast that with how DAFs expand the possibilities for giving and tax savings.

A. Tax Policy Toward Giving without DAFs

In US tax law, a qualified charity must gain an IRS tax classification as a 501(c)(3) organization. Individual tax filers who itemize deductions can deduct their donations to 501(c)(3) organizations from their taxable incomes. If one is facing a marginal tax rate on income of 35%, then a $1,000 donation will reduce a donor’s tax bill by $350, resulting in a net cost of $650 for each $1,000 given. Most states with income taxes also allow a deduction, further lowering the price of giving.

In addition to cash, one can also give appreciated assets, such as equities, artworks, and real estate. Imagine giving something easily valued, such as stock in a publicly traded company. If the asset were to be liquidated before giving, the owner would pay capital gains tax of as much as 23.8% on long-term gains (assets held for a year or more). Thus, stocks worth $1,000 that had been purchased for $400 would first generate $143 in capital gains tax (that is, 23.8% of the gain of $600), leaving the donor with $857. Giving this net amount to charity then earns a tax deduction, which reduces income taxes by $300 (that is, $857 times the marginal tax rate of 0.35). In sum, the $1,000 asset yields $857 for charity and a net tax savings of $157. However, if one gives the asset directly, then the charity gets the full $1,000, the $143 tax on capital gains is forgiven, and the full $1,000 face value of the asset can be deducted from income. Given this way, the $1,000 asset yields $1,000 for the charity, earns a tax deduction on the full $1,000 (now worth $350), and a swing in the donor’s bank account of $193 (= $350 − $157).

Clearly, giving the asset directly is more advantageous for tax purposes. But the difference between giving most assets and giving cash is only a technical one. The irony in giving assets is that most charities follow a policy of liquidating any noncash gifts, like equities, as soon as possible upon receipt.⁴ So any difference between the donor or the recipient liquidating the asset is of little practical consequence, yet the consequences are very real for the donors’ tax payments and, potentially, for the charities’ receipts.

Two seldom-discussed constraints on giving are often quite important for DAFs. First, gifts of noncash assets that do not have any easily identified fair market value, such as real estate, works of art, or shares in closely held corporations, are required to have professionally conducted appraisals if their values are of any significance.⁵ Second, there are limits on the fraction of income that can be claimed as a charitable deduction each year. Donors who are giving cash can deduct up to 50% of adjusted gross income ([AGI], which can be thought of as income net of standard adjustments, such as subtracting IRA contributions and adding in unemployment benefits and IRA distributions). If appreciated property is given, the limit is 30% of AGI.⁶ Deductions that exceed these caps can, however, be carried forward up to five years.

Finally, it should be noted that only those who itemize deductions on their tax returns can claim a charitable deduction. This means those with lower incomes who live in states with small or nonexistent state income taxes or pay no home mortgage interest cannot benefit from even the charitable deduction, to say nothing of the DAFs.

B. Tax Policy Toward Donor-Advised Funds

Imagine a donor wishing to give $100 to a small local charity, say, a food bank. Coincidentally, the donor owns shares with substantial capital gains selling for $100 per share. Ideally the donor would like to give one share of stock to the food bank in order to get the maximum tax savings. Unfortunately, the cost of transferring and liquidating the single share of stock would be so high that the food bank would likely refuse the gift of the noncash asset. Wouldn’t it be convenient, therefore, if the donor could give the shares easily to another charity who can accept them, and for a small fee, send the food bank a check for $100? This is what Donor-Advised Funds do.

DAFs are brokerage-like accounts that are sponsored by qualified 501(c)(3) charities. These charities accept the donors’ funds into the sponsored account. The sponsor legally owns the money donated, but acts only as an intermediary by a allowing the DAF donor two important advising rights. The donor can advise the sponsor on how to invest the donation and when funds in the account should be liquidated and sent to another 501(c)(3) charity, or used for other charitable purposes. While the sponsor, who is the legal owner of the fund, can stipulate its own spending and investment limits, such constraints would be voluntary—none are required by the law. Giving money to charity through a DAF is very much like giving money directly from a brokerage account, with three important differences. First is timing. The tax consequences of a charitable deduction are absorbed when money goes into the DAF rather than when it is granted out of the DAF to a traditional charity. Once in the DAF, however, there are no tax consequences of trading or reinvesting funds, and all gains and losses accrue to the eventual charitable recipients.

Second is convenience. It is easy to avoid capital gains taxation by donating securities, artwork, or real estate to a DAF before liquidating them, whereas this could be impossible for some smaller donations without a DAF.

Third is commitment. Once in the DAF, any funds account can only be withdrawn in the form of grants to charitable organizations.⁷

How do DAFs differ from trusts or private foundations? First, the typical foundation is much larger. In 2015, private foundations averaged about $9.5 million in assets, while individual DAFs averaged $292,000. However, DAFs can be opened and ready to operate in a matter of a few hours, and at low cost. Private foundations, by contrast, can take months or years to establish, involving great expense. As a result, in 2015 there existed around 82,000 private foundations, but nearly 270,000 DAFs. Importantly, foundations do much more than grant money to other organizations, and often pursue agendas of their own, employing staff and affording allowances to trustees. DAFs primarily make grants to existing charities, although they can pay management fees to the sponsors of the DAFs.

A contentious difference between DAFs and foundations is the 5% payout rule. The rule states that private foundations must distribute at least 5% of their assets annually as either grants or as eligible administrative expenses. When adopted, the point of this rule was to guarantee taxpayers a fair rate of return on money subsidized when donated to the foundations, while at the same time not bleeding so much money from foundations that it would force them to eventually disappear. In debating the law, the Treasury argued that 5% was justifiable because carefully invested endowments would on average yield 6.75%.⁸ In principle, since DAFs allow the advising rights to be given away or bequeathed upon death, DAFs can legally live forever. Just how long money lingers in a DAF will be an important feature to be explored here.

C. How Important Are DAFs and DAF Tax Policy?

Do DAFs involve enough money for policymakers to really worry about? Perhaps surprisingly, the answer is a resounding yes. Figure 1 illustrates recent trends in DAFs. From