The Potential Budgetary Impact
The federal government does not recognize “marriages” of same-sex couples either for receipt of federal benefits or for tax purposes. The 1996 Defense of Marriage Act (Public Law 104-199) provides that the federal government will honor only marriages between one man and one woman. It also stipulates that no state, territory, or possession of the United States or Indian tribe can be required to recognize a same-sex marriage performed in any other jurisdiction.
of Recognizing Same-Sex Marriages
June 21, 2004
The potential effects on the federal budget of recognizing same-sex marriages are numerous. Marriage can affect a person’s eligibility for federal benefits such as Social Security. Married couples may incur higher or lower federal tax liabilities than they would as single individuals. In all, the General Accounting Office has counted 1,138 statutory provisions — ranging from the obvious cases just mentioned to the obscure (landowners’ eligibility to negotiate a surface-mine lease with the Secretary of Labor) — in which marital status is a factor in determining or receiving “benefits, rights, and privileges.”(1) In some cases, recognizing same-sex marriages would increase outlays and revenues; in other cases, it would have the opposite effect. The Congressional Budget Office (CBO) estimates that on net, those impacts would improve the budget’s bottom line to a small extent: by less than $1 billion in each of the next 10 years (CBO’s usual estimating period). That result assumes that same-sex marriages are legalized in all 50 states and recognized by the federal government.
The number of same-sex couples who would marry if they had the opportunity is unknown, but the 2000 census offers some insights. The census does not ask about sexual orientation, but it allows people living with a nonrelative to identify themselves as “partners” instead of “housemates/roommates.” Almost 600,000 households (or 1.2 million people) identified themselves as same-sex partners in 2000, roughly half in male couples and half in female couples. They represented about 0.6 percent of the total adult population and almost 1 percent of people between the ages of 30 and 50.(2) By several common measures of stability — age, home ownership, and length of residence — those 600,000 same-sex couples resemble married couples more than they resemble other cohabiting households, so it seems reasonable to assume that many of them would marry if given the chance.(3) Some would not, of course; but other same-sex couples who did not live together, or who labeled themselves “roommates” rather than “partners” in the census, might choose to marry. The census also contained limited data about the income, earnings, and assets of those 600,000 couples — clues that CBO used to gauge budgetary impacts.
For the purposes of this analysis, CBO assumed that about 0.6 percent of adults would enter into same-sex marriages if they had the opportunity. (That proportion is equivalent to nearly 600,000 couples in 2000, with adjustment for subsequent population growth of about 1 percent a year.) CBO’s estimates reflect significant uncertainty because predicting how many same-sex couples would marry is difficult and because data on their incomes, assets, and participation in federal benefit programs are sparse.
Effects on Revenues
Recognizing same-sex marriages would affect federal revenues through both the individual income tax and the estate tax. Neither effect would be large relative to total federal revenues. Receipts from other taxes — in particular, payroll taxes — would be unlikely to change significantly.
On balance, legalization of same-sex marriages would have only a small impact on federal tax revenues, CBO estimates. Revenues would be slightly higher: by less than $400 million a year from 2005 through 2010 and by $500 million to $700 million annually from 2011 through 2014. Those amounts represent less than 0.1 percent of total federal revenues.
The impact on revenues varies over time in part because, under current law, tax provisions will change in almost every year between now and 2011 and in part because incomes change over time. CBO’s estimates are based on current law and assume that provisions in the Economic Growth and Tax Relief Reconciliation Act of 2001 (EGTRRA) and the Jobs Growth and Tax Relief Reconciliation Act of 2003 (JGTRRA) expire as scheduled rather than be extended.
The estimates are highly uncertain for several reasons. First, data from the 2000 census may not accurately represent the number of same-sex couples, both because of misreporting by respondents and because of misinterpretation of reported relationships by the Census Bureau. Second, how many same-sex partners would marry if allowed is unknown; CBO assumed that age and income would influence their decision, as appears to be the case for heterosexual couples. And third, allowing same-sex marriages could result in behavioral changes that would alter the number of gay and lesbian people in partnered relationships. Despite those uncertainties, however, CBO concluded that any effect of same-sex marriages on federal revenues would be small.
Income Tax Revenues
Recognizing same-sex marriages for federal tax purposes would require people in those marriages to file income tax returns as couples, either jointly or separately. For almost all married couples, filing jointly rather than separately results in lower tax liability. Depending on the division of income between spouses, marriage can lead to either higher income tax liability (a “marriage penalty”) or lower liability (a “marriage bonus”). The greater the similarity in the two spouses’ earnings, the more likely the couple is to incur a marriage penalty. Conversely, the greater the disparity in earnings, the more likely the couple is to receive a marriage bonus. When one spouse earns all of a couple’s income, the couple always gets a bonus.
Together, EGTRRA and JGTRRA will reduce the number of couples incurring marriage penalties and increase the number receiving bonuses between now and 2010. JGTRRA provided relief from marriage penalties for 2003 and 2004 in the form of a higher standard deduction and broader 15 percent tax bracket for married couples. For 2005 through 2010, that relief is first reduced and then reinstated under the provisions of EGTRRA. Because of those changes and rising real (inflation-adjusted) incomes, marriage penalties would dominate during that period, and same-sex marriages would increase revenues by between $200 million and $400 million each year. After 2010, the expiration of all of EGTRRA’s provisions would raise marriage penalties further, and revenues would be $500 million to $700 million higher each year than they would be if same-sex marriages were not recognized. (Permanently extending the marriage-penalty provisions in EGTRRA would reduce those revenue gains to less than $400 million per year after 2010.)
Estate Tax Revenues
A second effect of same-sex marriages on federal revenues could come through the estate tax, but that effect is almost certain to be small. Little is known about the estate taxes that same-sex couples pay under current law. However, the effect of allowing same-sex marriages can be gauged by assuming that the partners would behave like other couples in terms of leaving inheritances.
The main impact of same-sex marriages on estate taxes would come through the unlimited spousal exemption, which allows a person to leave any amount of assets to his or her spouse without incurring estate tax liability. As a result, wealthy married couples can exempt twice as much wealth from estate taxes as single people can and thus can often pay lower estate taxes than they would if they were unmarried.(4) Furthermore, marriage can defer the payment of estate taxes until the death of the second spouse, thus shifting revenues into later years. Because the estate tax is scheduled to decline steadily through 2010 and then return abruptly to its pre-2001 levels, that shift could increase revenues by moving taxation from a relatively low-tax year (through 2010) into a higher-tax year (after 2010). Extending the estate tax provisions of EGTRRA beyond 2010 would eliminate that possible revenue gain.
Notwithstanding those complexities, under current law, allowing same-sex marriages would have little impact on estate tax liabilities. That conclusion assumes that same-sex married couples would behave similarly to heterosexual married couples in terms of how they bequeathed their estates. If they behaved differently, however, allowing same-sex marriages could have different effects on estate tax revenues. For example, anecdotal evidence suggests that gay decedents currently leave more of their assets to charitable institutions than their heterosexual counterparts do. If allowing same-sex marriages caused that behavior to change — for example, if same-sex couples had more children to whom they left their estates — revenues could rise. Currently, about one in three lesbian couples and one in five gay couples live in a household with their own children.(5) Those proportions might rise if same-sex marriages were legalized.
Effects on Outlays
Marital status has a direct impact on people’s eligibility for some federal payments, such as Social Security benefits, veterans’ benefits, and civil service and military pensions. It can affect other benefits indirectly if a spouse’s income and assets enter into determinations of eligibility. The discussion below focuses on so-called mandatory, or direct, spending — programs like Social Security that make payments to anyone who is qualified and applies — because the budgetary effects on those programs of recognizing same-sex marriages would occur automatically and would not depend on future annual appropriations.
Recognizing same-sex marriages would increase outlays for Social Security and for the Federal Employees Health Benefits (FEHB) program, CBO estimates, but would reduce spending for Supplemental Security Income (SSI), Medicaid, and Medicare. Effects on other programs would be negligible. Altogether, CBO concludes, recognizing same-sex marriages would affect outlays by less than $50 million a year in either direction through 2009 and reduce them by about $100 million to $200 million annually from 2010 through 2014.
With estimated payments of $488 billion in 2004, Social Security is both the largest federal program and the one in which marital history plays the greatest role in determining benefits. Under Social Security rules:
If a spouse or widow(er) has worked long enough (generally 10 years) to earn retired- or disabled-worker benefits on his or her own, Social Security does not pay both benefits. Instead, it pays the larger of the two amounts for which the recipient is eligible. Technically, such people are labeled “dually entitled” and receive their own benefit plus the excess, if any, of their other benefit.
- The spouse of a retired or disabled worker — assuming that he or she meets
age and other requirements — can receive 50 percent of the worker’s benefit, subject to reductions for early retirement (before age 65 or, eventually, age 67) and, if children are also eligible, subject to a cap on total family benefits. Thus, the basic benefit for a married couple with one earner is 1.5 times that for an unmarried worker with the same work history.
- The widow or widower of an insured worker — again, if he or she meets age and other requirements — can receive 100 percent of the worker’s benefit.
- Divorced spouses can collect either type of benefit described above if they were married to an eligible worker for at least 10 years.(6)
As a general rule, married people fare better under Social Security than single people do, and married couples with one earner fare better than two-earner couples do. One-earner couples get an extra 50 percent of the worker’s check while both spouses are alive and a lifetime benefit if the worker dies first. (In a typical pension plan, by contrast, benefits stop at the worker’s death unless he or she chose a reduced, joint-and-survivor annuity.) Two-earner couples gain less from the spousal benefit because it may exceed the lower earner’s own benefit by little or nothing.(7) But even in two-earner couples, the husband typically earns more and dies first, and his widow gets his higher benefit for life. People who never marry do not gain from those provisions.
Benefits paid to spouses and widow(er)s account for almost one-fifth of Social Security spending. In 2004, $21 billion in benefits will go to 5.5 million spouses and $69 billion to 8 million aged widows and widowers, CBO estimates. Almost half of those recipients are dually entitled.(8)
If permitted to marry, same-sex couples would benefit from those spousal and survivor features. However, their gains would be modest, CBO expects, for two reasons. First, most same-sex couples include two workers, and on average, their earnings are closer to one another’s than is the case for a husband and wife in a two-earner couple. Second, same-sex partners would generally collect survivor benefits for a shorter period. On average, such partners are the same age, and statistically they have the same life expectancy. By contrast, husbands are an average of two to three years older than their wives, earn more, and have a shorter life expectancy. An average married woman can expect to spend six or seven years as a widow.
From analyzing the joint earnings (and Social Security income, if applicable) of the same-sex partnerships in the 2000 census, CBO judges that only 30 percent would receive higher benefits as a retired couple than they would as two single people. And about half of same-sex couples would collect higher benefits after one partner died than they would under current law. Taking into account the age mix and expected mortality of same-sex couples, CBO estimates that additional Social Security benefits would total about $50 million in 2005 and grow to $350 million in 2014 (equivalent to $250 million in today’s dollars, adjusted for intervening wage growth and cost-of-living increases).
That additional cost is small in the near term and grows over time as the couples age. According to the census, the average member of a same-sex couple in 2000 was in his or her early 40s. In only about 10 percent of partnerships were both partners age 62 or older, the earliest age for receiving Social Security retirement benefits. In the next few decades, many more couples will reach age 62, and some members will die, leaving their survivors eligible for widow(er)s’ benefits if their marriages were recognized.
Children — chiefly the minor children of workers who have died — account for about 5 percent of Social Security benefits. Although large numbers of same-sex partners in the 2000 census were raising children, CBO estimates that allowing same-sex marriages would not add significantly to those benefits. Children may qualify for benefits on the earnings record of a biological or adoptive parent; the parent’s marital status does not matter. Even if same-sex marriages led to more adoptions by such couples, the children involved would essentially replace one set of parents (their biological parents) with another (their adoptive parents). The two sets of parents might differ in key respects such as mortality and earnings, but any net effect on Social Security benefits for their children would most likely be small.
Finally, some recipients face marriage penalties in Social Security. Disabled adult children — grown children whose disability (usually mental retardation) occurred before age 22 and who therefore collect on a parent’s record — lose their benefits if they marry. Widows and widowers who remarry before age 60 lose their former eligibility, although they may reclaim it if the remarriage ends in death or divorce. Same-sex marriages would trigger those penalties in a handful of cases, but CBO expects that such effects would be negligible.
Other Federal Programs
Although Social Security is the program that would be most obviously affected by changes in marital status, legalization of same-sex marriage would also change federal spending for various income-support and health programs.
Supplemental Security Income. Partners who now collect benefits from SSI — a means-tested program for the elderly and disabled — could lose some or all of their benefits if same-sex marriages were recognized, because their spouse’s income and assets as well as their own would count toward their eligibility. In almost 25,000 (about 4 percent) of the same-sex partnerships reported in the 2000 census, one or (rarely) both partners received SSI benefits. Those participants would be unlikely to marry, but some would. More plausibly, partners who do not now collect SSI benefits would find their future applications rejected because of their spouse’s income. As a result, legalization of same-sex marriages would save the SSI program about $100 million a year by 2014, CBO estimates.
Medicaid. A joint federal/state program, Medicaid provides health coverage to some poor elderly and disabled people, children, and families. The federal share of spending will reach an estimated $174 billion this year and $352 billion in 2014. CBO expects about 58 million enrollees in 2014 — 18 million elderly and disabled people and 40 million other adults and children.
As with SSI, eligibility for Medicaid is generally linked to income and assets, so counting a spouse’s resources could make some individuals ineligible. Participation in SSI generally confers Medicaid eligibility, which means that some people who lost SSI benefits would also lose Medicaid coverage. Other elderly and disabled individuals (including a small number of nursing-home residents) who qualify for Medicaid under current law could also lose eligibility if a couple’s combined incomes and assets were considered. The extent to which people lost coverage would vary among states depending on the degree to which states disregard assets and income. By 2014, about 30,000 fewer elderly and disabled individuals would have Medicaid coverage than under current law, CBO estimates.
Counting a spouse’s income and assets would likewise push some welfare recipients and other poor families above Medicaid’s eligibility limits. Although an increase in family size could boost some families’ chances of qualifying, the prevailing effect of combining incomes would be to reduce Medicaid eligibility. Most of the people losing Medicaid coverage would be children. Because parents face tighter eligibility rules than children do in most states, fewer of them are eligible for the program. In a same-sex couple in which one partner has little or no income, his or her children may qualify for Medicaid under current law. Those children could lose Medicaid coverage if both partners’ incomes were considered in determining eligibility.(9) Furthermore, same-sex marriages might make some children who would otherwise be enrolled in Medicaid eligible for health insurance through an adoptive parent’s or stepparent’s employer. Such children might shift from Medicaid to private coverage. CBO estimates that by 2014, about 100,000 fewer children and their parents would have Medicaid coverage than under current law.
Conversely, Medicaid spending could increase for a small number of nursing-home residents. Under special rules for spouses living in the community — the so-called spousal impoverishment exemption--a noninstitutionalized spouse may shield a home and some other jointly owned assets from Medicaid’s resource limits. Recognizing same-sex marriages would allow couples to protect more assets than they could as individuals (under current law) and thus shrink their expected contribution to the cost of nursing-home care.
In all, CBO expects, federal spending for Medicaid would decline by about $400 million (or about 0.1 percent) in 2014 because of same-sex marriages and by smaller amounts in earlier years. Because states pay about 43 percent of the program’s total costs, they would realize savings of about $300 million in 2014.
Medicare. Savings would also occur in the new Medicare prescription drug benefit’s low-income subsidy program. Under current law, people who meet certain income and asset tests are eligible to receive government subsidies for their cost-sharing payments and premiums for the drug benefit. Some of those people would no longer qualify if the income and assets they shared with a partner were considered for eligibility purposes. The resulting savings for Medicare would amount to less than $50 million a year through 2014, CBO estimates.
Federal Employees Health Benefits Program. By recognizing same-sex marriages, the government would automatically extend health care insurance under the FEHB program to civil servants and civil service retirees who elected to cover a spouse. Under that program, the government pays almost three-quarters of health care premiums, and employees and annuitants pay the rest. The government’s payments for annuitants constitute direct spending (spending that does not require an annual appropriation). CBO estimates that covering the same-sex spouses of retired enrollees in the FEHB program would cost the government less than $50 million a year through 2014. Premiums for current employees, by contrast, come from agencies’ salary and expense budgets, which are funded by appropriations. CBO expects that those additional premiums would cost agencies less than $30 million annually through 2014.(10)
Food Stamps and Other Programs. In the Food Stamp program, the basic unit is the household (people who live together and usually buy and prepare food together), not necessarily the family. Thus, CBO expects that recognizing same-sex marriages between partners who already live together would not affect Food Stamp spending.
In addition, the costs or savings for veterans’ benefits, civil service retirement, and military retirement would be negligible if the federal government recognized same-sex marriages, CBO estimates.
General Accounting Office, Defense of Marriage Act: Update to Prior Report, GAO-04-353R (January 23, 2004).
By comparison, the 2000 census counted 54 million married-couple households and 4.9 million households with unmarried, opposite-sex partners. See Bureau of the Census, Married-Couple and Unmarried-Partner Households: 2000, Census 2000 Special Reports, No. CENSR-5 (February 2003).
Before the 2000 census, researchers James Alm, M.V. Lee Badgett, and Leslie A. Whittington used a variety of sources (the National Health and Social Life Survey, the General Social Survey, and data from the National Opinion Research Center) to reach a similar estimate: that about 550,000 same-sex couples might marry. See Alm, Badgett, and Whittington, “Wedding Bell Blues: The Income Tax Consequences of Legalizing Same-Sex Marriage,” National Tax Journal, vol. 53, no. 2 (June 2000), pp. 201-214. That study estimated that male couples would make up almost two-thirds of those marrying. By contrast, about two-thirds of the same-sex couples who wed in San Francisco; Portland, Oregon; and Massachusetts in early 2004 were female. See Evelyn Nieves, “The Women’s Marriage March: Majority of Same-Sex Couples Who Took Vows Are Female,” Washington Post, May 25, 2004, p. A3.
That effective doubling of the exemption occurs as follows: when the first spouse dies, he or she can pass on to heirs other than the surviving spouse an amount equal to the single exemption without owing estate tax. The balance of the estate goes to the surviving spouse, also without tax because of the unlimited spousal exemption. When the second spouse dies, an additional amount equal to the single exemption goes to heirs, again without tax. The couple thus has an effective exemption equal to twice the single exemption. If each spouse has assets of his or her own exceeding the exemption, marriage will have no effect on estate tax liability, other than on the timing of tax receipts. But if the assets of one spouse exceed the exemption and those of the other spouse do not, marriage will result in a higher combined exemption.
Bureau of the Census, Married-Couple and Unmarried-Partner Households: 2000, Table 4, p. 9.
Relatively few people collect on an ex-spouse’s record: about 375,000 spouses and 625,000 widow(er)s did so in December 2002 (including those who were eligible for smaller benefits in their own right) out of a total of 46 million Social Security recipients. The 10-year requirement clearly limits that number. More than half of marriages that end in divorce do so before the couple’s eighth anniversary, according to Bureau of the Census, Number, Timing, and Duration of Marriages and Divorces: 1996, Current Population Reports, P70-80 (February 2002), Table 6, p. 12.
As a rule of thumb, the lower earner — usually the wife — will not receive a spousal benefit if she earned at least one-third as much as her husband over their lifetimes, because her own benefit will be higher. That outcome stems from the weighted formula used to calculate benefits. Social Security bases benefits on a worker’s highest 35 years of earnings and aims to “replace” more earnings for a lower-paid worker than for a higher-paid one. Thus, a worker who earned an average of $4,500 a month (in today’s dollars) might get a benefit of about $1,655 a month (before any reductions for early retirement), and a worker who earned only one-third as much ($1,500 a month, on average) would qualify for a basic monthly benefit of $835 — more than half of the higher earner’s amount.
The most readily available figures from the Social Security Administration show far fewer spouse and widow(er) recipients — about 2.8 million and 4.6 million, respectively. That is because dually entitled people are already included among retired workers and disabled workers, so listing them again would constitute double-counting. Adding an estimated 2.7 million dually entitled spouses and 3.6 million dually entitled widow(er)s yields the totals cited above. For more information, see Social Security Administration, Annual Statistical Supplement (various years), Table 5G2.
Even if a child is related by blood or adoption to only one of the spouses — for example, if the child was born during a previous marriage — most states consider a stepparent’s income and resources when determining the child’s eligibility for welfare and Medicaid. Some stepparents, though, could newly gain Medicaid coverage, depending on their states’ rules.
In 2003, CBO analyzed the Domestic Partnership Benefits and Obligations Act of 2003, a bill that would expand certain fringe benefits — notably health insurance — to “domestic partners” of federal civilian employees. CBO estimated that 83 percent of the potential beneficiaries would be people in opposite-sex rather than same-sex partnerships. At the sponsor’s request, CBO confined its analysis to current federal employees, not retirees. See Congressional Budget Office, Cost Estimate for H.R. 2426, the Domestic Partnership Benefits and Obligations Act of 2003 (August 4, 2003).