There’s a plausible case, though not a fully persuasive one, for redistributing income to relieve poverty. Other forms of redistribution should be eliminated.
Miron: Next topic is redistribution. As you know, most societies transfer resources enormously across the different citizens of their populations. They redistribute via welfare, food stamps, housing, healthcare, rent control, progressive taxation, minimum wages, and many, many more policies. As with healthcare, the benefits of redistribution are now taken as given by a very broad fraction of the population. There’s an attitude that all good people believe we should be redistributing and indeed we should always be doing more redistribution if you listen to at least certain members of the chattering classes.
The libertarian response is not so fast. The underlying justifications for doing redistribution are not so compelling in the first place. And even if they’re partially convincing, the impact is potentially quite costly. So we need to think about whether and especially about how much.
I’ll discuss this with the following outline. First think about the different arguments for redistribution. How convincing are those? And then talk about the costs of anti-poverty programs. I will argue the anti-poverty programs are the only kind of redistribution that should survive at all if any, talk about their cost, talk about the design of anti-poverty programs, talk about their generosity. Then think about whether those should occur, if at all, at the federal or the state level, arguing strongly only at the state level at most. And then talk about other policies that governments use that are not aimed mainly at redistributing but have the effect of causing lots of redistribution and how those should really be the main targets for people wanting us to have a more fair or just or better distribution of wealth and income.
What are the basic arguments for trying to redistribute? The first is known as the utilitarian argument. It says that each individual will get some benefit, some satisfaction, some utils that we can measure from having a certain amount of income, and that this relation between how much happiness you have and how much income you have is totally the same for everyone, and that people experience what’s known and diminishing marginal utility. You get an extra thousand dollars; it makes you happier. If you get another thousand dollars after that, it also makes you happier but not as much as the first thousand dollars, and so on and so forth. So the extra happiness Bill Gates would get if he got an extra thousand dollars would be infinitesimal. The extra thousand dollars to someone who is starving or in very poor condition would be quite substantial because of diminishing marginal utility. So under this argument, if we transfer resources from Bill Gates to someone who’s very poor, we will have increased aggregate utility by taking it away from Bill Gates, who got hardly any utility from that last thousand dollars he earned, to someone who has very little income and gets a lot of utility. So the net effect is positive. The poor person had a lot of extra utility. Bill Gates had only a slight reduction in his utility.
That argument can be made totally mathematical and rigorous and logical, but it’s all assumption. There’s no such thing to economists as measurable utility. It’s just an assumption that everyone has utility function, that everyone has the same utility function, even more strongly. So one can use an argument like that to justify almost any transfer across any group in any way by assuming a set of conditions that will lead to that. It’s not a compelling argument. And economists do not accept that argument as a basis for redistribution.
A different argument that has broader support and more ground in economics is known as the public goods or the compassion argument. It says that everyone would like to avoid having really poor people in their society. They don’t want anyone to be starving or in very bad shape. But many individuals will not deliberately engage in private charity to try to help the less fortunate on their own. They will instead try to free-ride on the contributions made by other people. They will wait and see if others give the private charity, hope that the Mellons’ and the Rockefellers’ and the Gates’s and the Warren Buffetts’ will do that, so they don’t have to bother. If that happens, if many people choose to free-ride, then the amount of private charity might be small, smaller than most people would like if they knew everyone was participating. So you might get less provision of charity than is in society’s overall interest. Government could address that by forcing everyone to contribute.
That argument is logical. It doesn’t necessarily imply that we should have huge amounts of redistribution. It says that there’s some benefit from a little bit by helping the very poor people not be very poor. But it in no way, shape, or form says we should redistribute massively across the board, that we should be systematically punishing rich people by trying to have highly progressive taxation that takes a lot of income away from them. It makes a case for a safety net for the very poor and not necessarily even for a generous safety net. So that argument, while logical and perhaps reasonable in moderation, doesn’t justify anything like the degree of redistribution that we currently do in the U.S. and other rich countries. I’ll talk about those amounts in a couple of minutes.
The veil of ignorance argument – which we came across earlier when we discussed government provision of healthcare – says that not knowing whether you’re going to be rich or poor, you might be willing to pay something in terms of taxes in your future life in order to avoid having very low income. That again is a logical argument. It might be that most people would vote for such a plan if they had the option. But it again implies helping people who are very poor. It doesn’t imply that we just redistribute across the board. Just as with all kinds of insurance, having complete insurance against all possible bad outcomes is extremely costly. What most people have is partial insurance. They pay something to avoid the worst-case scenarios. Likewise, any implication that government should be redistributing based on the veil of ignorance said there’s enough social insurance to make sure no one is really poor, no one is literally starving in the street. But it in no way implies highly progressive taxation or the very generous amounts of redistribution that U.S. and other rich countries engage in.
So plausibly, there’s some argument for some small amount of redistribution – we’ll talk about the details in a second – but nothing like current policy. Now assuming that we were focused only on anti-poverty programs, only on spending that was aimed at helping the very poor, what are the things to be cautious about even in designing those programs? First any type of redistribution is likely to generate disincentives. People know that they’re going to have some income for not working. Some people are going to work less or save less or take less care or work less hard, to marry someone who can help take care of them, or so on and so forth. So there are clear efficiency losses from even limited anti-poverty spending. If it’s truly limited spending, those disincentives might be fairly small, but there’s of course going to be some because you have to raise taxes to pay for anti-poverty spending, there’s going to be some additional distortion because taxes distort economic behavior and reduce economic efficiency. So that should be accounted for in any analysis of government redistribution, even if it’s designed to strictly add people who are very poor.
Government redistribution also has the potential to crowd out private charity. And indeed lots of evidence finds that increases in government support of the poor have tended to crowd out, have reduced private donations and efforts to help the poor – in some cases, close to a one-for-one rate. So that by itself doesn’t automatically tell you that the government redistribution is bad. It just tells you it’s not necessary because it’s mainly just offsetting what would otherwise occur from private charities. So despite the public goods aspect of contributing in charity, lots of people do it quite substantially, and they seem to want to do it less if they know that the government is doing it. So the net amount of charity is not being increased that much.
More speculatively, but I think still very importantly, providing any kind of anti-poverty spending has the potential to create envy from anyone who is not quite poor enough to qualify for that spending. So if you’re making sure people who are very poor are taken care of, people who are sort of poor but not very poor say, “Gee, how come not me too?” And that’s a pressure for those policies to expand. And then the people who are a little bit less poor than that say, “Hey, how come not me too?” And there’s pressure for those policies to expand. And that may lead you to large amounts of redistribution that are way too costly and have big disincentives and lead to lots of losses from taxation, even if a very small amount initially would not have done so. So the risk of expansion is certainly something to think about.
A classic example is when the U.S. adopted Medicare and Medicaid in the 1960s, it did both. The better argument was probably for Medicaid. That was helping people who were explicitly poor. But to get the whole country to go along with it with raising taxes a lot to pay for healthcare for the poor, we had to also introduce Medicare for the elderly, which is something that everybody knew that they would qualify for some day. That bought the political support for Medicaid. So the risk of excess expansion is definitely something to think about. Still one last potential cost even of well-designed limited anti-poverty programs is that they may reduce self-reliance. They may just create an atmosphere that says, “I don’t have to go to whatever lengths necessary to take care of myself. The government is there to do it.” If that applied only to a very, very small number of people who clearly can’t easily take care of themselves, to children, to people who are mentally ill or disabled, of course, compassion would suggest that we might want to do that, setting aside whether it’s government or the private sector that does that. But if it starts to get expansive, then it affects self-reliance broadly. And that’s a potentially bad thing for society to not have people feel that they need to be hard-working, entrepreneurial, and doing their best to take care of themselves, except under the most extreme conditions.
The next point to consider is how we design anti-poverty programs. There are of course many, many components of current anti-poverty spending, from welfare to food stamps to Medicaid to housing to unemployment insurance to Social Security, disability, the Earned Income Tax Credit, Pell Grants, public schools, energy subsidies, and much, much more. So there are tons of different policies. A few of those are much, much bigger than all the rest. But still we do it in this very complicated, very messy way. What are the negatives of this approach? First of all, much of it is being done in kind, i.e. providing free housing projects or subsidize housing projects or free healthcare, not by just giving people cash. And common sense, standard economics, and much evidence suggests that people can make better decisions by getting the cash and being able to choose how to spend it themselves, rather than having government say, “Here’s how you have to spend it.”
Classic example would be housing. If you give somebody subsidized housing by saying, “You have to live in this project,” then they’re stuck living in that neighborhood, which may be far from where they work. They’re stuck living in that neighborhood with some people who might be relatively engaged in a criminal activity. They may be stuck living in a place which is very unsafe, so on and so forth. If they get something which is not quite cash but is in that direction, a housing voucher, they can then choose to live in any neighborhood without voucher, have lots more flexibility in choice. And there’s evidence that in fact people make good use of that flexibility in choice to get better outcomes than if they just got the subsidy in kind. So a lot of the current policies are in kind. That’s a problem.
The total amount that we’re redistributing is not at all obvious under the current system. There are so many policies that are being provided partially in cash, partially in kind, partially in vouchers. Adding that all up and figuring out exactly how much we’re redistributing is a mess. So people don’t think about it. And that at some level makes it easy for us to spend too much because there’s not one big figure that could be put in a headline in newspaper. But that’s problematic for anyone who believes in democracy and letting people decide how much we should be spending and on what.
Still further, the current design is a problem because there are all sorts of huge interactions between these different policies, in particular something that economists call high marginal tax rates. In many cases, if you earn a little bit more, that reduces your eligibility or even eliminates your eligibility for some of these anti-poverty programs in such a way that you’re worse off by having worked harder and earned more income. If that’s the case, many people are quite rationally going to say, “Well, forget it. I’m not going to go out and do more babysitting, work at McDonald’s, get a low-wage boring, uninteresting job because it’s going to mean I lose my Medicaid benefits or I lose my housing voucher or so on and so forth. That’s an incredibly bad feature of the current system. Whatever you think about whether we should be redistributing and whatever you think about the amount, it should be designed in a way that doesn’t create those adverse incentives.
There is an alternative approach that is not perfect but much, much better, known as the negative income tax that would simply give people some lump sum amount, just to illustrate, $5000 a year, and say, “Here it is. Spend it however you want. Earn as much as you want. If you earn income, it’s of course subject to standard income taxation but at a low rate. So you will absolutely be better off if you go out and generate more income.”
Okay, so the costs are large. The design is currently horrific. Another thing to think about is just how generous current welfare spending is in the United States. If we took all the programs that I’ve mentioned, from Medicaid and Social Security and Medicare and food stamps and welfare etc., if you take all of that expenditure and thought about it just as a big cash amount and divided it up and gave it in equal proportions to the poorest 20 percent of the population, that would provide an income of about $60,000 per household in the lowest 20 percent of the income distribution. That’s a lot of guaranteed income. Now that’s not exactly the way it works. I’m looking at it hypothetically. But it illustrates what you could do and what on average we are in fact spending on all this redistribution. The fact that it’s so substantial means there is significant potential for disincentives. Really, really, really stingy welfare programs don’t have much disincentive, but generous ones certainly can have large disincentives. It means there are substantial losses from all the taxation we use. And it means there’s substantial potential for envy and reduced work ethic and so on because of the large amount of redistribution in which we’re currently engaging.
Next thing to discuss, taking all of that as given for the moment, is if we’re going to engage in anti-poverty spending, should it be federal or state governments that do this? And I’m going to push the view that it should be entirely left to states, it should be zero at the federal level, because any program like this has the potential to expand. There are always incentives for any government bureaucracy, for any entity of any kind to want to survive. And one way you survive is by getting bigger. So if we do this at the federal level, we have one system of redistribution, and it always is going to be pressured to make it more generous and to spend more. Say that we leave it to the states. The standard concern is if we left it to the states to have welfare programs or not and how generous each state decided to make its program, then states would engage in a race to the bottom. That’s the standard fear. And there would be almost no redistribution, even at the state level. I think that’s probably wrong. And in addition, I think it’s good that states would worry about whether they’re being too generous relative to their neighboring states because if they worry about that, that puts some counter pressure on the tendency for redistribution to become truly excessive. In addition, states are clearly altruistic in many ways. Many states undertake policies that are way more generous already than what they’re mandated to do by the federal government – states and cities that have much higher minimum wages, states that have more generous unemployment insurance, states that have more generous Medicaid programs, and so on and so forth, states that have much more aggressive environmental policies. So the claim that all states would simply say, “Oh, we don’t have to do this anymore; we’re not going to have any redistribution,” is completely consistent with what we currently observe. It’s inconsistent with the fact that states were developing their own state-level Social Security and welfare programs long before the federal government started doing it in the 1930s. So we should rationally expect, based on all that evidence, that states would probably engage in this anti-poverty spending to some degree. But we would get a better balance than we get leaving it all to the federal level.
An interesting question for libertarians is say we agree that any government redistribution was going to be limited strictly to anti-poverty programs and was going to be done via simple, well-designed negative income tax. It was only going to be done at the state or even the local level, nothing at the federal level. And it was going to be small. So imagine you’re a voter as a libertarian in your state and the question is what to do about a proposal like that for your states. Would libertarians support that level of government redistribution? I think across the range of libertarians, you get some division of opinion. You might get some people who would vote in favor of that type of very well-designed, very limited, a very small amount of state-level redistribution. And I don’t think too many libertarians would say that’s totally horrific, that that’s the worst thing governments had ever done. My own hunch is that even under the best conditions, government redistribution, even if it’s small, even if it’s state-level and so on, is still not a great idea, mainly because we can’t keep it small. So once you have it, it will get bigger. Once it’s at the state level, there will somehow be pressure to make it federal, which is exactly what we’ve observed over the last hundred or so years in the United States. Nevertheless, libertarians, I think, could reasonably agree over whether the optimal amount of government redistribution is literally zero, or maybe it’s something that’s slightly bigger than that.
Last item to discuss are policies that are not explicitly designed to redistribute but do so in ways which make no sense – minimum-wage laws, rent controls, union protections, immigration restrictions, bans on peak load pricing, many, many more policies. That you could hardly justify under any circumstances making the economy more efficient. They’re clearly about redistribution, whatever people say about them. So minimum-wage laws are not meant to make the labor market more efficient. They’re meant to raise the wages of some people who work in labor markets. Union protections are not meant to make the labor market more efficient. They’re meant to raise the wages, to transfer income in particular ways. Bans on peak load pricing are meant to avoid some people having to pay high prices by using electricity in certain times of the day that is inefficient in the times of the day that have high demand when electricity prices should be higher.
The key thing is that all these programs and many, many more generate winners and losers from a distributional perspective. Minimum-wage law helps the people who keep their jobs at the higher wage but hurts the people who lose their jobs entirely. Same thing with unions, same thing with immigration quotas and so on and so forth. So the libertarian perspective says very clearly if our goal is to redistribute – and of course we’ve discussed that we don’t really endorse that goal very much, if at all – then just redistribute, then just say we are having a policy that takes money from people in the middle and up and gives it to the people who are very poor. Don’t interfere in individual markets in a way that sort of tries to redistribute but does so in a kludgy, inefficient way. That just makes a mess, makes for complication, instead of first letting the amounts that you’re redistributing be very clear, because if you say you have a negative income tax of $5000 per person, you know exactly what you’re doing. As opposed to a minimum-wage law, we have no idea exactly how many people you’re causing to be unemployed or not. And it means that you’re not distorting all these individual markets with all of these policies that are mainly trying to help some poor people or some middle-income people not have to pay the full price but are simultaneously generating all sorts of deficiency losses, distortions in those markets rather than simply trying to redistribute. So if we are going to redistribute, call it what it is. And then we can have arguments about how much and exactly how.
To summarize this discussion, the arguments for general redistribution are not compelling, but they’re not completely illogical. There’s a plausible case for perhaps some degree of anti-poverty spending, just on ex-ante grounds. But current anti-poverty programs are extremely generous, poorly designed. And any anti-poverty program is going to have some distortionary effects, some costs, and potentially large cost if small programs don’t stay small. There is a compelling argument for zero redistribution at the federal level, even if one thinks perhaps there should be some at lower levels of government. If it’s all at the state level, it might not be a horrible policy. But it’s not obviously desirable policy either. I think that most libertarians would come down for a little, if any, even at the state level. But some might advocate for at least a modest amount, as long as it’s done under exactly the right conditions. Thank you very much.