In this chapter, we explore the X tax treatment of fringe benefits and work expenses, the implications of the X tax for the financing of Social Security and Medicare, and the X tax treatment of other public and private transfer payments.
We first consider the treatment of nonretirement fringe benefits and then analyze the distinct issues posed by pensions and employer-sponsored retirement accounts.
Health Insurance and Other Nonretirement Benefits. Because the income tax and the X tax are fundamentally similar in their treatment of labor income, the same basic options for the treatment of nonretirement fringe benefits are available under both tax systems. In either case, neutral tax treatment calls for firms to deduct all labor compensation, including fringe benefit costs, as business expenses, and for workers to pay tax on all of their compensation, including fringe benefits.
The current income tax generally disregards this prescription, as health insurance and many other fringe benefits are deducted by employers but are not included in taxable income by workers. This approach is intended to encourage the provision of these benefits and avoid administrative difficulties. Fringe benefits that are exempt from income tax are generally also exempt from Social Security and Medicare payroll tax. The current income tax system also provides an itemized deduction for large out-of-pocket medical expenses and a deduction for health insurance costs of self-employed taxpayers.
The tax treatment of employer-provided health insurance raises a number of complicated issues, including whether such insurance should be encouraged as a good way to pool risks. Further complications arise from the changes slated to take effect under the health care reform law adopted in March 2010. Starting in 2014, refundable tax credits will be provided to workers who do not have access to affordable health insurance at work to help them buy health insurance on newly established exchanges. Penalty taxes will also be imposed on firms with fifty or more employees that do not provide affordable insurance and on certain individuals who do not have health insurance. Starting in 2018, a 40 percent excise tax will be imposed on employer-provided health insurance plans that exceed certain cost thresholds. The health care law also includes many other provisions, including temporary tax credits for some small firms that provide health insurance for their employees.
The appropriate treatment of employer-provided health insurance cannot be discussed in isolation from these provisions, which have important and complicated effects on health insurance obtained both inside and outside the workplace. We need not resolve the intense debate surrounding health care reform. Instead, we simply note that the decision to adopt the X tax leaves open all of the policy options in this area. The health insurance exclusion, the itemized deduction for medical expenses, the self-employed health insurance deduction, the refundable credits, and the penalty and excise taxes can be either maintained or eliminated, as desired, under the X tax, just as they can be under the income tax.
With respect to other fringe benefits, which do not pose some of these complications, we recommend that employees be taxed on the allocated value of their fringe benefits, with exceptions for small fringe benefits that would be administratively difficult to tax. We reiterate, though, that the X tax is also consistent with alternative approaches, such as continued exclusion of these benefits.
Service employees would also be subject to household wage tax on tips that they receive. Compliance would presumably be no better, and no worse, than under the current income tax system.
Work Expenses. Current law allows a tax credit for child care costs, a deduction for moving expenses, and a miscellaneous itemized deduction for large employee business expenses. Education costs, which may warrant tax relief as work-related costs or on other grounds, benefit from an assortment of tax credits and deductions.
These costs can be treated in fundamentally the same manner under the household wage tax as under today's income tax. The same trade-offs and complications arise, such as how to classify mixed business-personal expenses and how to verify employees' expenditures. In general, the appropriate treatment of these costs under the income tax, whatever that may be, is also appropriate under the X tax.
Pensions and Retirement Plans. Pensions raise issues distinct from those raised by employer-provided health insurance and other benefits and receive different tax treatment under current law. Unlike consumption benefits such as health insurance, pensions are a form of saving.
Under current law, firms deduct contributions to pension funds and to employees' tax-preferred savings accounts, and employees are not taxed. The capital income earned in the pension fund or tax-preferred account is exempt from tax. Recipients are taxed on their pension benefits or the amounts they withdraw from their accounts.
The initial deduction by the employer and the lack of employee tax make this treatment superficially similar to the treatment of employer- provided health insurance. Yet, the nature of the tax preference is fundamentally different. Here, the employee actually is taxed on the compensation, which was not true for health insurance. This occurs because the pension benefits or account withdrawals attributable to the employer contributions have the same expected market value as the employers' contributions. Although the tax payment is delayed until the money is withdrawn, tax is ultimately paid on the compensation, albeit at the tax rate applicable to the worker on the withdrawal date rather than the rate applicable when the compensation was earned.
Instead, the tax preference arises because the employee is allowed to save tax-free, as pensions and tax-preferred accounts effectively receive the same front-loaded treatment that the PET provides to all savings. Under a PET, the contributions and capital income would be excluded from the tax base because they are saved rather than consumed; tax would be imposed only when withdrawals are made from the saving vehicle. In this area, therefore, the current system follows the same consumption tax approach as the PET.
As we discussed in chapter 2, however, the PET's consumption tax approach follows different timing rules than the approach followed by the household wage tax in an X tax system. The household wage tax adheres to the back-loaded Roth approach, which features no deduction for saving and no tax on withdrawals from saving. Under normal X tax rules, therefore, the employee would be taxed on the employer contributions when they are made, with no subsequent taxes on withdrawals or pension benefits. As explained in chapter 2, the two approaches are equivalent in present discounted value when the tax rate remains constant, but the front-loaded approach is more generous if the worker's tax rate is lower when he or she receives the money. The "Zero Revenue from Taxation of Risky Returns" box in chapter 3 (pages 51–52) establishes that the presence of risky returns does not change this analysis.
To minimize disruptive changes from current law, promote simplicity, and preserve a role for employer-sponsored retirement plans, we propose that pension benefits and withdrawals from employer-sponsored retirement accounts continue to be taxed when workers receive the money, as recommended by Hall and Rabushka (1995, 55–59). The employer should deduct contributions into pension funds and accounts when they are made. To avoid the need to distinguish them from employer contributions, employee contributions into employer-sponsored plans should also be deductible.
Under the household wage tax in an X tax system, most households will be in lower brackets in their retirement years than during their working years, making this front-loaded treatment more generous than the back-loaded treatment given to other tax savings. This generosity will give workers some incentive to save through employer-provided savings vehicles and help to maintain these institutions, thereby promoting an environment conducive to saving.
Under the X tax, Roth IRAs and Roth 401(k) accounts are unnecessary because all saving enjoys Roth treatment at the household level. We propose that holders of existing Roth accounts be allowed to withdraw their balances tax-free, as they can under the current system, and that penalties for early withdrawals be abolished. We also propose that conventional front-loaded IRAs be abolished; we will discuss the treatment of existing accounts in chapter 8. We also propose to eliminate the savers credit provided by Internal Revenue Code section 25B; the credit has been ineffective under the current tax system and will be unnecessary under the X tax.
We now turn to public retirement programs.
The federal government imposes a 12.4 percent payroll and self- employment tax, limited to the first $110,100 of wages and self-employment earnings in 2012, which is earmarked to finance Social Security old age, survivor, and disability benefits. The ceiling is adjusted each year, based on a two-year lagged measure of wage growth in the national economy. The federal government also imposes a separate payroll and self-employment tax earmarked to finance Medicare Part A, which has no ceiling on taxable earnings. As of 2013, this tax will be imposed at a 2.9 percent rate, with a 0.9 percent surtax on wages and self-employment earnings in excess of $200,000 ($250,000 for married couples). The highest earners will therefore face a zero marginal tax rate under the Social Security tax and a 3.8 percent marginal tax rate under the Medicare tax.6
Although half of the payroll tax is legally imposed on employers and half on employees, the economic effects are the same as those of a pure employee tax. The Social Security tax does not apply to the dwindling group of federal government employees hired before January 1984 or to employees of some state and local governments (those that opted out of Social Security when states and localities were still allowed to do so). Even these groups are subject to Medicare tax, however, except for the dwindling group of state and local government employees hired before 1986.
We recommend that the Social Security and Medicare payroll taxes be maintained when the X tax is adopted. Because a payroll tax is a wage tax, it, like the household wage tax in an X tax system, imposes no saving disincentive; the tax does not apply to interest, dividends, capital gains, or any other income from saving. So, the economic gains from replacing the payroll taxes with the X tax would be slight.
In contrast, the costs of replacing the Social Security payroll tax with the X tax would be substantial, for two reasons. First, such a replacement would eliminate the use of this tax as an earmarked financing source for Social Security, an arrangement that helps restrain the program's spending. Second, it would disrupt the current linkage between the benefits each worker receives from Social Security and the taxes he or she has paid into the program.
On the first point, Social Security is legally prohibited from paying benefits greater than the levels that can be financed from current and past earmarked taxes, as tracked by a trust-fund accounting mechanism. On the whole, this earmarking arrangement appears to promote fiscal responsibility and spending restraint. Legislation to restrain Social Security benefit growth was enacted in 1983 when earmarked taxes proved insufficient to support scheduled benefits, and the current discussion of Social Security reform is driven by the trustees' projections that earmarked taxes will again fall short around 2036. To maintain fiscal responsibility, therefore, Social Security should continue to be financed by an earmarked tax, rather than being given a blank check on general revenues.
On the second point, each worker's Social Security retirement or disability benefits are based on his or her Average Indexed Monthly Earnings (AIME), which is a measure of the worker's lifetime wages and self- employment earnings. AIME includes only earnings subject to Social Security tax and therefore excludes earnings that exceeded the taxable maximum or were earned in employment not covered by Social Security. A worker with higher AIME receives higher monthly benefits, although the increase is less than proportional; if Smith earns double the wages and pays double the taxes of Jones throughout their lives, Smith receives a higher monthly benefit than Jones, but less than double Jones's benefit. This benefit formula reflects a delicate compromise between competing objectives. Paying higher monthly benefits to those who have paid higher taxes (and awarding no benefits based on untaxed earnings) gives the program a contributory feature by linking benefits received to taxes paid. At the same time, paying benefits that rise less than proportionately with taxable earnings gives workers with lower lifetime earnings a higher rate of return on their tax payments, making the system redistributive. This delicate balance, and the associated tax-benefit linkage, would be upset if Social Security was financed by general revenue from the X tax rather than by the current payroll tax.
In view of these two points, we recommend that the Social Security payroll tax be retained when the X tax is adopted. The second point is largely inapplicable to the Medicare payroll tax, because there is almost no link between Medicare Part A taxes and benefits at the individual level.7 The first point applies, however, because Medicare Part A spending is also limited to the amount that can be financed by its current and past earmarked tax revenue, as tracked by a trust fund mechanism. Accordingly, we recommend that the Medicare payroll tax also be retained. As we will discuss in chapter 5, we propose folding the self-employment tax into the payroll tax.
Another small payroll tax, legally imposed only on employers, is earmarked to finance the unemployment insurance (UI) program. The program's spending is limited to the revenue raised by this earmarked tax, so we recommend that this tax also be retained, for reasons similar to those discussed above.
As stated in chapter 1, we propose to replace the Unearned Income Medicare Contribution tax, which is slated to take effect in 2013, with the X tax. Despite the name of this tax, all of its revenue is paid into the general treasury, and none of it is earmarked for Medicare. Accordingly, the repeal of this tax has no implications for Medicare financing.
We now turn to broader issues that affect transfer payments.
The first issue is whether transfer payments should be taxed.
Current Tax Treatment. Most public transfer payments are excluded from taxable income. Long-standing Internal Revenue Service (IRS) rulings stipulate that government social welfare benefits are excluded from taxable income, unless they are specifically included by a provision in the tax code. Specific code provisions require the inclusion of UI benefits and a portion of Social Security benefits, but most other public transfer payments are excluded. Exclusion is the rule for all means-tested transfer payments and also for many non-means-tested payments, such as veterans' benefits.
The current income tax system generally disregards private transfer payments, providing no deduction to the payer and imposing no tax on the recipient. This treatment applies to gifts and bequests to individuals and organizations (other than gifts to charities, for which the payer receives a deduction) and to child support payments. The major exception is that alimony payments are taxable to the recipient and deductible to the payer. Due to the difference in tax treatment, litigation sometimes ensues over whether payments between ex-spouses should be classified as child support or as alimony.
X Tax Treatment. Of the potential tax treatments of transfer payments, exclusion is simplest and most compatible with the philosophy of the X tax. Recall from chapter 2 that the X tax is generally imposed on real production activity; unlike the PET, it disregards financial transactions and eschews tracking cash flows to their ultimate recipients. This philosophy suggests that transfer payments should be similarly disregarded.
Accordingly, those public and private transfer payments that are currently excluded should continue to be excluded under the X tax. Exclusion should also be extended to alimony, UI benefits, and Social Security.
The exclusion for alimony received should be accompanied by non-deductibility for alimony paid, which would avert litigation by bringing its treatment into line with that of child support. This policy also frees family courts from the need to consider taxes in setting payment amounts. Current- law treatment (deduction by payer and inclusion by recipient) should continue to apply to payments under existing alimony decrees, however, until and unless a court modifies the alimony amount for any reason.
As discussed in the "Taxation of Gambling" box (pages 66–67), gambling winnings and losses can also be viewed as private transfer payments. In keeping with the above principles, we propose that the X tax ignore those amounts, imposing no tax on winnings and allowing no deduction for losses.
Charitable Contributions. As mentioned above, current law treats charitable contributions to 501(c)(3) organizations more favorably than other private transfer payments, allowing the donor to claim an itemized deduction while imposing no tax on the recipient. The corporate income tax also allows a deduction for charitable contributions. Tax relief for charitable giving has wide support, although controversy sometimes occurs over the appropriate magnitude of such relief. The Congressional Budget Office (2011a) discusses recent proposals to limit the deduction. For the most part, the decision of whether to provide tax relief for charitable contributions and the generosity of any such relief can be resolved separately from the decision of whether to adopt the X tax.
The X tax design may have some impact, though, on the manner in which such relief is provided. A deduction for contributions under the household wage tax would not only be of no value to households too poor to pay tax (who also receive no benefit from the current itemized deduction), but would also be of no value to households with only investment income. Screening out the latter group could significantly diminish the effectiveness of an incentive for charitable giving. It may therefore be best to provide a refundable credit, offering a uniform subsidy to giving by all households. Firms' contributions could be given similar treatment under the business cash flow tax. Batchelder, Goldberg, and Orszag (2006) and Jenn (2008) note that a uniform refundable credit is the best way to subsidize a socially beneficial activity, if the social benefits do not depend on which household performs the activity. Seidman (2011) further argues that tax credits are more transparent and flexibile than deductions.8
Social Security. As stated above, we recommend exemption of Social Security benefits, in line with our treatment of other transfer payments. But the tax treatment of these benefits raises distinctive issues that merit separate discussion.
Social Security benefits were originally excluded from taxable income, but a 1983 law required the inclusion of up to 50 percent of benefits, and a 1993 law increased the maximum includable fraction to 85 percent. The includable fraction rises with the recipient's income, with benefits remaining fully excludable for recipients with incomes below $25,000 ($32,000 for couples). Because earning additional income can cause additional benefits to become taxable, the provision increases effective marginal tax rates on some recipients' labor and capital incomes.
The income tax revenue due to the 1983 law is earmarked to the Social Security trust fund, and the incremental revenue from the 1993 law is earmarked to the Medicare Part A trust fund. In 2010, $38 billion of income tax was paid on $702 billion of benefits, with $24 billion going to Social Security and $14 billion to Medicare. Because the income thresholds are not indexed for inflation, a larger portion of benefits is slated to become taxable in upcoming decades.
Excluding Social Security benefits from the X tax base promotes simplicity and reduces effective marginal tax rates. Moreover, the current income-based inclusion regime cannot be easily administered under the X tax in any event, because household wage tax returns do not provide information on any income other than wages.
Repealing the taxation of benefits will reduce the resources available to the Social Security and Medicare Part A trust funds. If this result is considered undesirable, it can be offset through general revenue transfers.9
Design and Administration of Means-Tested Transfers. Another issue concerns the administration of means-tested spending programs and tax credits in a world without the income tax. Steuerle (2005a, 239), who offers important insights on this matter, complains with some justice that consumption tax advocates "never deal" with this issue. In the discussion below, we seek to correct this oversight of our fellow consumption tax advocates.
Means tests are employed by many programs, including Medicaid, Temporary Assistance to Needy Families (TANF), food stamps, Supplemental Security Income (SSI) for the elderly and disabled poor, subsidized Medicare Parts B and D premiums, the earned income tax credit, the refundable tax credits for private health insurance slated to take effect in 2014, and housing assistance. These means tests generally rely on a mixture of labor income, capital income, and assets to determine eligibility and benefit amounts. The income information is sometimes obtained, or at least verified, from income tax returns. As Steuerle (2005a) notes, replacement of the income tax therefore poses complications for these means tests. Yin (1995) discusses this issue in the context of a PET.
We recommend changing the means tests for Medicare Part B and Part D premium subsidies to base them on lifetime labor earnings, as measured by AIME. We recommend that the other programs' means tests remain largely unchanged, while recognizing that the loss of capital income information from income tax returns may create some administrative difficulties.
One might expect us to recommend means tests based on recipients' current consumption. Because a means test is effectively a form of taxation, our argument in chapter 1 for why consumption taxes are more neutral than income taxes may suggest that consumption-based means tests are more neutral than income-based or asset-based means tests. Indeed, Hubbard, Skinner, and Zeldes (1995) emphasize the savings disincentives created by asset-based means tests. Moreover, consumption may best measure the recipient's standard of living and need for assistance. But, as Steuerle (2005a) notes, consumption-based means tests are not administratively practical, because household consumption cannot be accurately measured.
Interestingly, consumption-based means tests would not actually be neutral with respect to the saving decision. As we explained in chapter 1, the neutrality of the consumption tax relies on the tax rate remaining constant over time. Means-tested transfer programs apply steep marginal tax rates to households in years in which they receive benefits and zero marginal rates in other years. A consumption-based means test would therefore discourage saving during good times in preparation for possible bad times; the means test would offer no reward for reducing consumption during good times, when the individual is not on the means-tested program, but would penalize the resulting additional consumption during bad times, when the individual is on the program. As Steuerle (2005a) observes, households could receive transfer payments simply by reducing their consumption. As discussed below, however, such savings disincentives are unavoidable in this context. We do not reject consumption-based means tests because of such disincentives; we reject them because they are administratively impractical.
Basing means tests on lifetime labor earnings is another option. This method faces two potential problems, though, and we recommend its use only in contexts in which both problems can be avoided. First, a satisfactory measure of lifetime earnings cannot be obtained for young and middle-aged workers, because their expected future earnings cannot be observed. A means test based on lifetime labor earnings is therefore appropriate only for programs for the elderly or the disabled, who have little or no future labor earnings. Second, we consider it morally and socially unacceptable to deny minimum safety-net benefits to a person with high lifetime labor income who, for whatever reason, failed to save. Of course, our unwillingness to condemn such a person to total destitution precludes the complete removal of saving disincentives from transfer programs. A means test that reduces transfer payments based on lifetime labor earnings is therefore appropriate only if a back-up mechanism guarantees minimum benefits to those individuals with high lifetime earnings who failed to save.
Under these principles, means tests based on lifetime labor earnings can be used in a few contexts. As discussed above, the current Social Security benefit formula provides higher rates of return to workers with lower lifetime earnings, as measured by AIME. SSI, which has a means test based on current income, provides the necessary safety net for high lifetime earners who failed to save.
Also, under current law, higher Medicare Part B and Part D premiums are charged to recipients with high adjusted gross incomes. We recommend linking the premiums to AIME rather than adjusted gross income. Steuerle (1997) proposed this approach to avoid saving disincentives, even with the income tax system in place. Biggs (2011) also notes the advantages of using lifetime earnings rather than the retiree's current income in means tests. With the X tax replacing the income tax, the proposal is even more useful because adjusted gross income, except for wages, is no longer available from household tax returns. Although high lifetime earners who failed to save will pay higher Medicare Part B premiums under this proposal, they will escape destitution by receiving benefits from Medicaid and SSI, which will continue to have means tests based on current income.10
Under the X tax, means tests for other programs should remain largely unchanged, continuing to rely on income and assets. Steuerle (2005a) comments that consumption tax proponents seem to implicitly and indirectly propose the continuation of income-based means tests in transfer programs; we now explicitly advance that proposal, subject to the exceptions noted above. The lack of capital income information on household tax returns may make it somewhat more difficult for administrators to check recipients' reported capital income and assets, possibly requiring additional enforcement resources. Although these means tests create saving disincentives, that outcome is unavoidable, given our commitment to protect from destitution those who fail to save. The disincentive should be minor for most households, who have a low probability of receiving means-tested transfers.
Means Tests within the Tax System. As we noted in chapter 3, the X tax can provide either refundable or nonrefundable credits to low-income households, much as the income tax system does today. But the lack of capital income information on household returns will pose administrative challenges for these credits, just as it does for the transfer programs discussed above. Under the current tax system, for example, low-wage workers lose part or all of the earned income tax credit to which they would otherwise be entitled if they have too much capital income, as shown on the tax return itself. If something like the earned income tax credit is maintained under the X tax system, as we recommend, recipients will be required to certify that they do not have large amounts of capital income as a condition of eligibility. We recognize that the IRS will have to rely on outside information to check the accuracy of the certifications.
In addition to low-income credits, the current tax system includes dozens of income phase-out provisions that reduce or eliminate the benefit of credits or deductions as taxpayers' incomes rise, as detailed by Brill and Viard (2008). Most of these provisions are unwarranted within the context of the current tax system, as progressivity is better promoted through the rate schedule rather than by phasing out specific tax breaks. Such phase-outs will be even harder to administer under the X tax, as household tax returns do not provide a complete picture of income. Accordingly, income phase-outs, other than those related to low-income credits, should be removed from the tax system when the X tax is adopted.
The X tax offers essentially the same options as the income tax with respect to the treatment of nonretirement fringe benefits and work expenses. Employer-sponsored retirement plans should be taxed in line with current law, which already adheres to consumption tax principles, although with timing rules that differ from the basic design of the household wage tax in the X tax system. The Social Security and Medicare payroll taxes should be maintained. Public and private transfer payments should be excluded from the household X tax. Most means tests should remain largely unchanged, but those for Medicare premium subsidies should be based on lifetime labor earnings rather than current income.
BOX
TAXATION OF GAMBLING
Although it may not be immediately obvious, the taxation of gambling raises similar issues as the taxation of financial transactions and transfer payments. Gamblers as a group enter into wagers with gaming firms, in which they expect, or should expect, to lose money. Their expected losses, which presumably equal their actual losses in the aggregate, are payments for the services that they receive from the gaming firms. To tax these services, the X tax requires gaming firms to pay business cash-flow tax on their net winnings, which are the gamblers' aggregate net losses.
But some gamblers lose more than the expected amounts, and others lose less than expected or even win. In effect, all gamblers pay the firms for services, while unlucky gamblers make transfer payments to lucky gamblers. The question is how to treat those implicit transfers; the appropriate answer is to ignore them.
For example, suppose that "the house" has a 10 percent edge on each bet. A large number of gamblers each places $100 bets, and each receives $10 of services from the gaming firm in exchange for his or her expected losses. But in the actual event, half of the gamblers lose $25 apiece, and the other half win $5 apiece. Conceptually, each gambler pays $10 to the firm for services and each losing gambler makes a $15 implicit transfer payment to a winning gambler. Taxing the gaming firm on its winnings yields the correct aggregate tax base, $10 per gambler, and accurately measures the services provided to the gamblers.
It is possible, though, to take the further step of taxing each winner on $15 and allowing each losing gambler a $15 deduction. Note that these amounts are equal to the amount won or loss plus the $10 expected loss; this expected-loss adjustment is necessary to prevent gamblers from deducting the cost of the services that they received from the gaming firm. As with financial transactions and explicit transfer payments, this further step does not change the aggregate tax base but merely allows each individual's final consumption to be accurately measured. As with financial transactions and explicit transfers, the X tax should omit this step because this tax system generally eschews attempts to measure each individual's final consumption. Accordingly, we propose that there be no tax on gambling winnings and no deduction for gambling losses.
Although it may seem desirable for the tax system to spread risk by taxing the winners and providing relief to the losers, it is unclear why the tax system should spread risks that gamblers voluntarily undertook or why gamblers would not undo such risk spreading by simply placing larger wagers. In any case, attempting such a task is a fool's errand. The tax system cannot accurately track each individual's winnings and losses, let alone apply the appropriate expected-loss adjustments. The current income tax adopts a complex and compliance-proof approach to gambling; in principle, taxpayers pay tax on each winning wager and deduct losses on losing wagers, with losses claimed only as itemized deductions and with each year's loss deductions limited to that year's winnings. The expected-loss adjustment needed to measure the implicit transfers correctly is not even attempted. In practice, of course, gamblers pay tax only on the large winnings that gaming firms are required to report to the IRS and deduct as many losses, up to the amount of reported winnings, as they can document or fabricate. Pollack (1997) and Raby and Raby (2001) discuss the problems posed by the current income tax treatment of gambling. The X tax offers a better approach.