America has a vast cornucopia of government welfare and assistance programs. And in most cases, they are publicly justified on the ground that the goods and services in question provide “special” benefits to recipients.
Consider food stamps (now SNAP) as an illustration. The primary justification put forward for them is that food is special to recipients. Poor or inadequate nutrition can have many adverse effects, so, proponents assert, it is especially important—that is, more important than for other goods—that poor families have sufficient nutritious food. Consequently, because poor families cannot afford adequate diets, we must subsidize their food.
There are many problems with that rationale for SNAP. Among others, studies find little difference between the nutritional adequacy of the diets of low- and high-income families. Added food spending also often fails to improve nutrition, as less nutritious preprepared food is substituted for healthier home-prepared food. Further, obesity is a more common problem among low-income families today than inadequate food. Other programs have their own problems, as well, as can be seen frequently illustrated in my most recent book, Pathways to Policy Failure.
But here I wish to consider the widespread use of “this good is special” as a central justifying premise. Housing programs, for example, are also important because housing is special. After all, inadequate housing can be physically, mentally, and emotionally damaging. So “obviously” we need to have government aid for it. Similar arguments are applied to health care. They are applied to education. And to heating oil in the winter and electricity for air conditioning in the summer. And transportation. And high-speed internet access and cell phones. Virtually everywhere you turn, Americans are told government assistance is necessary because in every case it provides “special” benefits to recipients.
But the nearly ubiquitous use of that rationale generates a logical contradiction. When it is extended as broadly as it is today, those “special” goods make up the vast majority of low-income family consumption budgets. In that situation, each subsidized good cannot be special relative to the others.
To illustrate, say that (making some heroic assumptions) we decide a dollar spent on food for low-income households is really worth $1.50 to them. That would commonly be taken to imply a dollar spent on their food is more valuable than a dollar spent on other goods and services for them. But when all areas of government assistance are justified as special, that conclusion does not follow. Say that a dollar spent on housing assistance for them is also really worth $1.50. To spend a dollar for food stamps may be worth $1.50 in some sense, but diverting such a dollar from housing assistance will also forego $1.50 of value. Both would be “special,” but neither would be more special than the other.
Applied more generally, if all goods and services provided by government assistance were considered equally “special” for recipients, the argument for subsidizing any particular goods and services disappears. We can get $1.50 in value from spending a dollar in one place, but that dollar would also have generated $1.50 in value if spent elsewhere. We sacrifice as much “special” value as we enable. And if families need subsidies in all these areas, the problem is lack of income, not too little food, transportation, medical care, housing, or anything else. If that is the case, it is an argument for converting those in-kind benefit programs that are truly justified into cash assistance (eliminating the rest).
Converting to cash assistance would lower administrative costs. It would do so dramatically if it was done to replace multiple programs. For a given cost to taxpayers, more would reach the recipients. And the compliance cost they bear would also fall, with fewer administrative hoops to jump through. Both of those, in turn, would raise participation rates among the eligible population.
Given that it is actually highly unlikely the “special” benefits produced by the marginal government dollar of assistance would be equal across multiple programs, even a dollar of assistance in an area with “special” benefits can be worth less than an assistance dollar spent elsewhere. Say that a dollar of food assistance was really worth $1.50 to recipients, but a dollar of housing assistance was really worth $2 to them. Then a dollar of food assistance is actually worth less to them than a dollar of housing assistance. Greater benefits could be created by shifting to less food assistance and more housing assistance, which, in this example, people would choose for themselves if given cash assistance rather than in-kind assistance.
Our in-kind assistance efforts will be inefficient in such cases. Further, government bureaucrats and lawmakers will be determining many of the tradeoffs beneficiaries face, which will, in turn, change many of their choices. That means it would make a great deal more sense to allow recipients to decide for themselves how special each good and service is to them, given their preferences and circumstances, because they not only do they know themselves and their circumstances better than those in government in charge of such programs, they also care more about themselves than those in government. Put another way, to the extent we believe in giving government assistance to those with incomes, we should trust them to love themselves and their families more than the government does, and give them that aid in cash.
That argument is enhanced when, in fact, the effects of many subsidies act like cash subsidies rather than focus on increasing their consumption of the special goods in question. If they act like cash subsidies, but cash programs have lower administrative costs, where is the gain from giving aid in-kind? Examples include SNAP, because food stamp benefits very seldom exceed the amount a family would spend for food if given cash. That means food stamps can simply replace cash that would have been spent on food, freeing up those dollars to spend on whatever the recipient chooses. The same would apply to winter heating oil subsidies.
Other subsidies, such as public housing and public school and universities, can actually lead some recipients to consume less housing or education than they would have if they had been given cash, because the only way to get the subsidy is to accept the quantity and quality of what the government program provides. That may often be less housing or education than recipients would have chosen if they had instead been able to choose the quantity and quality of those goods they desired more instead. And it should be blindingly obvious that consuming less of particular goods than if recipients were given cash aid cannot be squared with a “special” benefits justification for in-kind programs.
Yet even if we moved toward aid in cash, that still leaves out another important “special” consideration. Spending $1 on one of the goods in our example (or on whatever good the recipient would choose to spend a dollar of cash benefits on) may provide $1.50 in benefits, but the costs could still exceed the benefits.
The reason is that, while it is typically ignored in public discussions, raising a dollar of tax revenue to fund assistance costs society more than the revenue raised. Taxation distorts people’s productive incentives, reducing the mutual benefits from voluntary arrangements that would have taken place, but no longer do. Economists call those losses welfare costs or excess burdens.
Tax wedges between what buyers pay and what sellers receive net of taxes eliminate productive trades and the gains they would have created. A 10 percent tax would destroy trades that generated up to $1.10 in value per dollar spent, eliminating the mutual gains that would have been created by those trades. Raising it to 20 percent would further destroy trades generating between $1.10 and $1.20 per dollar spent, and so on. That means that to justify assistance as efficient, not only would there have to be special benefits to recipients, but those would also have to exceed the added welfare costs of raising the funds.
In 2006, Martin Feldstein estimated the excess burden at seventy-six cents per dollar of added tax revenue, when the US government was far smaller than today. Today, with greater tax burdens (especially when we realize that our mountainous deficits are really just deferred taxes) and regulatory burdens (which often act like taxes in raising producers’ costs, on top of what we officially call taxes) causing greater distortions, that estimate would now likely be higher.
Say that Feldstein’s estimate was correct. Even government aid worth 50 percent more than its budgetary cost would not be efficiency enhancing, because the benefits generated by spending a dollar in even a “special” area would have to exceed the sum of that dollar plus the additional welfare costs imposed by raising the added tax revenue. In our illustrative example, assistance generating $1.50 of benefits for $1 of added spending, might seem pretty “special,” but actually costs society more—$1.76—than the benefits. That substantially raises the bar any “special” benefit assertion for giving government aid (taking resources from other citizens) must meet.
Many claims to justify government assistance programs are logically and empirically questionable. But even in cases where there might really be some special benefits, that alone does not justify them, because virtually all assistance programs are claimed as special. Tradeoffs involve one set of “special” benefits for others. That suggests cashing out many in-kind programs.
But that still leaves the question of how much aid to offer. Much of the expressed support is substantially biased upward by overlooking the substantial welfare costs of raising the tax revenue to fund them. If we surveyed people about whether they are getting their money’s worth from domestic aid programs, a large number I know would say “no,” even when only considering budgetary costs. I believe many more would switch to the same answer if they recognized that the real cost was far higher.
