A father goes grocery shopping for his family and returns with the basics–milk, bread, peanut butter, cereal, applesauce, frozen pizzas. He also comes home with a large steak, which he alone plans to eat, and a bottle of good wine, which his pregnant wife cannot share. Money is a little tight, so he buys fewer vegetables and substitutes Kool-Aid for fresh juice. He uses a credit card, knowing they won’t be able to pay off the full balance next month. No one in the house starves that week, and the father eats and drinks unusually well.

If this happened once, most of us would say the guy was being a little selfish. But if he acted this way year after year, we would be deeply troubled and tell him to get his priorities straight. And yet too many U.S. social programs operate exactly this way: While they serve many people, they often give the most help to those who need it the least. Classic social insurance programs like Social Security and Medicare do, indeed, distribute benefits widely and offer extra help to the poor and the very sick. And means-tested programs like Medicaid and Temporary Assistance for Needy Families (TANF, also known as "welfare") are aimed exclusively at the disadvantaged. Nevertheless, the ability of these programs to fight poverty and inequality is substantially negated by other social programs–mainly tax expenditures like the home mortgage interest deduction and social regulations like the Family and Medical Leave Act (FMLA)–that benefit primarily the middle and upper-middle classes. While these latter policies may have their individual merits, in their current form they often widen the gap between haves and have-nots.

Economists criticize many of these policies for their inefficiency–noting, for example, that the mortgage deduction in the U.S. tax code encourages people to over-invest in large, luxury homes. But an equally powerful objection is rooted in fairness. A number of social policies make a mockery of the goal, enshrined in the Constitution, that government exists to "promote the general welfare." Our long-standing commitment to equal opportunity rings hollow when certain programs help people with good jobs and incomes to get health insurance, housing, parental leave, and retirement pensions, but offer little help to the poor and near-poor. We may disagree over how hard government should try to reduce poverty and inequality. Surely, however, when millions of Americans live in poverty and inequality has reached record levels, we can agree that public policies should not make these problems worse.

Call it phony universalism, Robin Hood in reverse, or socialism for the rich–whatever the name, the U.S. government is effectively targeting tax subsidies and legal protections at the more advantaged members of American society. The level of support is enormous, amounting to hundreds of billions of dollars each year. For every dollar spent on traditional anti-poverty programs, the United States spends almost as much through the tax code helping individuals who are lucky enough to have health and pension benefits at work or rich enough to buy a nice home (these are often the same people). This is how the United States can spend a ton of money on its welfare system and yet make fewer inroads against poverty and inequality than other affluent nations. Imagine a campaign against child obesity that encouraged kids to exercise daily and eat more Cheetos: U.S. social policy is beset by the same kinds of contradictions.

Some policy-makers realize what’s going on. When the Bush Administration proposed new tax incentives for Health Savings Accounts, the Center on Budget and Policy Priorities quickly pointed out that most of these benefits would go to affluent taxpayers. The Democratic authors of the American Dream Initiative, a set of policies designed to expand and strengthen the middle class, were careful last year to propose refundable tax credits for college tuition so that more people with below-average incomes could benefit. But it’s not enough to oppose bad ideas, or layer potentially good new programs on top of dysfunctional old ones. We also need to scrutinize existing programs and figure out how they got started, whom they really help, and what we could do to change them. Otherwise, we may find ourselves repeating these same mistakes as we respond to persistent poverty and growing inequality today. Moreover, if we can find ways to spend less on some of these existing programs, we can free up monies to serve more pressing social needs. The goal should not be to exclude the middle class from these programs but to ensure that more governmental benefits are distributed to those who truly need help.

The Strange Shape of U.S. Social Policy

The programs in question are rarely mentioned in debates over social policy or in studies of the welfare state. The unstated assumption is that U.S. social programs should resemble those in Europe. From this perspective, social programs are supposed to take the form of social insurance and grants. The former is broadly inclusive, and the latter is usually aimed at the poor. Because the United States spends a relatively small share of its gross domestic product on these kinds of social programs, it is considered a laggard or a semi-welfare state by observers on both sides of the Atlantic.

But the big difference between the American welfare state and its European cousins is not so much the level of government involvement as it is the mix of policy tools. After all, social insurance and grants are not the only tools used to make social policy. Governments also employ tax expenditures, social regulations, and loan guarantees, among other mechanisms, and the American welfare state relies on these alternatives more than any other nation. Instead of national health insurance, for instance, we offer tax breaks to those who purchase private health insurance, usually employers. Instead of subsidizing the wages of disabled workers, we require companies to make the workplace accessible to the disabled (via the Americans with Disabilities Act). The historic G. I. Bill extended loan guarantees to help veterans become homeowners; it did not build them homes.

Turning our attention to these stealthy social policies, it becomes clear that the American welfare state has expanded substantially in recent decades. While we haven’t witnessed a "big bang" of innovation comparable to the mid-1930s or mid-1960s, we have seen a steady stream of new social programs since the 1970s. You wouldn’t notice this development, though, if you were looking only for European-style social programs. Notable additions include the far-reaching Employee Retirement Income Security Act (ERISA) of 1974, which established detailed regulations governing the financing, eligibility, and benefits of company pension plans; created the Pension Benefit Guaranty Corporation (PBGC) to guard against the bankruptcy of those plans; and produced a new tax break that gave birth to individual retirement accounts. The Earned Income Tax Credit (EITC), designed to boost the incomes of low-wage workers, became law a year later. Regulations governing employer health benefits passed in 1985 and 1996. The Americans with Disabilities Act (ADA), which President George H. W. Bush has called one of the highlights of his presidency, became law in 1990. The FMLA, compelling many employers to offer parental leave, was one of the first legislative accomplishments of the Clinton Administration, while the Child Tax Credit (CTC) was one of the main domestic initiatives during Clinton’s second term. Tax breaks to offset the costs of higher education emerged in 1997 and 2001. The Medicare prescription-drug benefit, added in 2003, includes tax breaks for employers who offer comparable drug benefits to their retired workers.

Some of these new programs grew quickly. Take the EITC and the CTC, which together function as the equivalent of European-style family allowances. In 1986 and again in 1990 and 1993, officials expanded eligibility and increased benefits for the EITC; as a result, the EITC now costs more than TANF, food stamps, or unemployment benefits, making it the most important means-tested income transfer in the American welfare state. The CTC became just as large as the EITC in much less time. The congressional Joint Committee on Taxation estimates that these two provisions in the tax code cost almost $90 billion combined in 2006.

Other, older tax expenditures have posted significant gains as well. Even adjusted for inflation, the cost of the largest tax expenditures has more than doubled. In most cases, growth has been due less to legislative changes than to demographic and economic forces (e.g., higher health care costs, an aging population). Tax breaks for company health and pension plans have been around for decades. In 1980, these provisions cost $12 billion to $15 billion each. This year, subsidizing corporate health benefits will cost an estimated $100 billion in lost tax revenues ($115 billion if one includes similar tax breaks for individuals and the self-employed). The cost of subsidizing private pensions is greater. And all the tax breaks for homeowners–deductions for mortgage interest, property taxes, and capital gains–now exceed $100 billion, up from $20 billion in 1980. These subsidies dwarf everything spent on rental housing for the poor (all these figures come from the Joint Committee on Taxation; other analysts and organizations, using different assumptions and techniques, put the cost of tax expenditures even higher).