Monday, November 18, 2013

Money for Nothing: The Negatives of a Negative Income Tax


After a long political hibernation, basic income is back. Switzerland is set to vote on a referendum under which each Swiss citizen would receive approximately $2800 a month from the government as a minimum income. While it's unclear whether this proposal will actually pass (I suspect not), it has provoked a great deal of conversation on the topic. As Annie Lowry puts it, "[g]o to a cocktail party in Berlin, and there is always someone spouting off about the benefits of a basic income." I confess I haven't been to many cocktail parties in Berlin recently, but the idea has been all over the place in the wonkier parts of American media (see here, here, here, and here). 

Compared to the messy, kludgy reality of existing social serviced, the pristine simplicity of the basic income idea has obvious appeal. Unfortunately, much of the discussion of basic income seems to be focused on the wrong things, or is based on misapprehensions. For example, Lowrey says that one of the big advantages of having a basic income is that '[p]overty would disappear."  That may or may not be true, but if it is true, this would have less to do with the actual effects of the basic income program than with how the poverty rate is calculated. As Tim Worstall recently noted, the poverty rate (at least in the U.S.) is calculated based on a person's income pre-taxes and transfers:

[T]he four major poverty reduction programs are Medicaid, SNAP, EITC and Section 8 vouchers. And we include none of them, not one single groat of that money spent, in our current estimates of poverty.
So, while our definition of poverty has not changed (three times a low-cost food budget for a household in the early 1960s upgraded for inflation) what we’re actually measuring is now completely different. The US poverty numbers today do not measure the number of people still in poverty after the aid given: they measure the number of people in poverty before aid is given.

If one were to take account of the effect of current anti-poverty programs, the poverty rate would be nearly zero. Similarly, if we were to replace existing anti-poverty programs with a basic income and didn't include that income in calculating the poverty rate, then the direct effect on poverty of enacting a basic income would be minimal.
  
The real difference between a basic income and the status quo is not how it would effect the poor, but how it would effect the middle class. Medicaid, SNAP, EITC, and Section 8 are all means tested programs, meaning that they aren't available to people who make over a certain income. By contrast, whether billed as a replacement for existing welfare programs or as an add on to it, basic income plans are typically not means tested. If you are a doctor making $200,000, you get the same government stipend as if you are making $10,000 working part time at Wal-Mart.

The fact that the basic income isn't means tested has its advantages. It means, for example, that low income workers don't face the high implicit marginal tax rates that can come from means tested programs. But it also means that any disincentive to work based on the extra income provided by the government will apply across the whole of society.

Which brings us to another issue with the current coverage of the basic income idea. As is typically noted in the recent coverage, back in the 1960s and early 1970s, basic income plans were all the rage. Martin Luther King, Jr. was supportive, as was Milton Friedman. Richard Nixon and George McGovern both put forward basic income proposals. And then, nothing. The idea kind of just faded out of existence. At least, that's the impression one gets from reading some of the recent accounts.

While I'm sure there were many reasons basic income lost its luster, one big factor was the results of a series of experimental implementations of the idea. Between 1968 and 1982, the government sponsored four separate randomized trials, providing $63 million in basic income to more than ten thousand individuals. These studies concluded that a basic income set at the current poverty rate significantly reduced the average amount of time worked by recipients by the equivalent of 2-4 weeks of full time employment, as compared to the existing welfare system. The experiments also seemed to suggest that providing a basic income increased the likelihood of family breakup. While there have been a few smaller studies since then that are more encouraging, it's not surprising that many policymakers reacted to these studies by concluding, in the words of Jim Manzi, of basic income that it "is a fascinating and useful thought experiment, but it's not useful public policy." 
Personally, I remain a fan of the Earned Income Tax Credit, which, unlike basic income proposals, is tied to work (and, unlike such proposals, has been successfully implemented and expanded for decades). But while I admire the simplicity of the basic income idea, I'm afraid I'm with Jim Manzi. 

35 comments:

  1. "basic income increased the likelihood of family breakup"

    Is it positive or negative? If you provide women with certain level of financial independence, then there would be less abuse in our extremely patriarch society. And god forbid families and relationships might be established on completely different grounds.

    ReplyDelete
    Replies
    1. Good point. But you tend to see conservatives favoring the nuclear family as an end in of itself, and not whether or not it actually improves life satisfaction for both parties under all circumstances.

      Delete
  2. "Similarly, if we were to replace existing anti-poverty programs with a basic income and didn't include that income in calculating the poverty rate, then the direct effect on poverty of enacting a basic income would be minimal."

    Entirely true. And to be terribly wonky about it......

    If the basic income were a cash grant then it would be included in the numbers about poverty. So it would pretty much abolish it. However, if it was run through the tax system (a negative income tax) then it would not. For we include n poverty calculations direct cash grants and not anything that operates through the tax system. Odd but true.

    BTW, Charles Murray (yes, that one, but this book is good) went through the argument a decade back in "In Our Hands". Short and worth reading for the insight it gives to the actual numbers, how large a basic income could be without breaking into a sweat etc.

    ReplyDelete
  3. Basic income in Alaska is due to income taks. There is no mentioning of that in your example "If you are a doctor making $200,000, you get the same government stipend as if you are making $10,000 working part time at Wal-Mart." The least we can say is that your remarks are incomplete

    ReplyDelete
  4. I don't understand why a basic income must be non-means-tested. If it is administered as a negative income tax, it is trivially easy to effectively means-test it by bumping up marginal tax rates on low market incomes. This may require regressive marginal tax rates, but that would probably be efficient ( http://www.nber.org/papers/w7512 )

    ReplyDelete
  5. Anonymous9:36 AM

    ". . .a basic income set at the current poverty rate significantly reduced the average amount of time worked by recipients by the equivalent of 2-4 weeks of full time employment, as compared to the existing welfare system."

    Two to four weeks of work, any work, as long as it is work. We are to take, I assume, as a given that work is noble. I do believe that there were many proponents of a guaranteed income that saw it as a way to free people from that very thing, leaving them more free to pursue, at their leisure, the arts, crafts, and relationship with others -- all of which speak more to our innate humanity than does the idea of the noble wage slave.

    ReplyDelete
  6. Negative Income Taxes seriously expose households to economic slumps. Yes, fine for encouraging work, but also assumes work is available to the recipient.

    ReplyDelete
  7. Anonymous10:12 AM

    I'm confused. I would have thought the effects of the randomized trials (increased family breakup, fewer hours worked) would be thought of as successes. If a family breaks up due to reduced financial stress, it wasn't a happy, functional family. Releasing the financially dependent members from the need to remain with the provider is a good thing.

    Further, it is important to remember that the poor have no savings. This means that, when faced with a bad job offer (or finding their working conditions intolerable), they may not be able to negotiate on their own behalf, or to leave: if they can't find a new, better position *fast*, i.e. within days, they are screwed. In a very real sense, at the bottom of the pay scale, all job offers are functionally coercive, even if (as will often be the case), no one involved wants them to be so. About the only vaguely comparable thing for the weller-to-do would be housing: even if you have a reasonable amount of money, it is possible to find oneself only weeks or even days away from being homeless, and that is most certainly stressful. If a study lets the participants give abusive, poorly paying bosses the middle finger, then yay! If the study also doesn't saturate the geographical area, then there are other people who will fill the now empty positions though, so the followup effects (better bosses/better wages) won't appear. So reduced average hours worked for the studies is a good sign.

    ReplyDelete
  8. Josiah, I think "negative income tax" means EITC, not universal basic income.

    ReplyDelete
    Replies
    1. I wanted to point that out too. A Negative income Tax is just a broader form of EITC, and not the same thing as a basic income - I don't see why short-term programs doing NIT experiments would be terribly relevant. Whereas we have the Minincome experiment from Canada in the 1970s, which reduced teenage work but had a number of good social welfare benefits even in the short time that it existed.

      Moreover, part of the argument for a basic income is the "political economy" aspect. A Basic Income might be hard to sell, but once you get it it would be almost indestructible (unlike the EITC and other welfare programs). It would be like getting public schools, firefighter protection, and police protection - people would argue about how much should be spent on it, but nobody except crackpots would want to get rid of it completely.

      More broadly, we have a bunch of policies that theoretically reduce the amount of work people do. Overtime laws, unemployment insurance - people probably would work more without these, but that doesn't mean it's desirable to eliminate them.

      Delete
    2. The original negative income tax proposal by Friedman does not include any work requirements. It is equivalent a basic income together with a flat income tax. The EITC is similar, but is tied to work.

      Delete
    3. Friedman's NIT - if I remember right from Freedom to Choose - was basically a $10,000 tax credit like set-up. If you earned below that, you got the difference between the 10,000 and your income. Earn above that, and you get nothing. It's not the same.

      Delete
    4. Anonymous7:42 PM

      If below, You'd get the difference, but not all of it. Friedman used the example of 50% for simplicity's sake at a point. Also, Friedman was explicit in this example that you'd get 50% of the difference between 0 and the benefit taper off line. Which in his example was around twice the poverty line. (surprise surprise)

      Anyway, sorry if I'm not good at explaining what a taper is, but there you go!

      Now the difference to an always paid out UBI with a flat tax is the non-collection of tax on the first couple thousands that people earn, and the non-payout of any benefit to people who make more than the taper off line. Though they're financially equivalent for state and household finances, unless people unexpectedly lose their labor incomes. Then a UBI is preferable, for the individual.

      And of course there's some policies that give money to people via the tax system based on their incomes, and as such can be called 'negative income taxes', but have little to do with the policy that Friedman was proposing.

      Anyway, I tend to prefer relying on the actual monetary incentive, that is, earning money from the free market, given that this is already how most of the middle class used to function.

      Trying to coerce people to labor as long as they do not do labor just seems like we're preying on the most vulnerable to ensure they work for us and stay down. Of course we're not shy of even 'sorta reasonable' excuses for this situation, given customer spending is indeed quite lacking as of today, hard to pay wages that way.

      Aggregate demand has been stagnant if not declining for decades. I think if we loosely tie aggregage demand to growth rate, we can indeed restore the monetary incentive to a point where we don't need to rely on petty tactics that are so absolutely outside of the free market process. The EITC does in some aspects conflic with the free market process, as well. Work doesn't just double in value, even if you throw an extra state dollar at every person earning a dollar. But at least it does something to get money to people, so I appreciate that aspect, considering the aggregate demand issue. (though a UBI/NIT as suggested by friedman, that grows with GDP, policies like these would make sense in a growth capitalism, to me.)

      Delete
    5. Anonymous7:48 PM

      Oh and keep in mind that while a UBI and a NIT can be financially equivalent for the most part, you're still looking at substantially more 'on paper' state spending and collection of revenue, by doing a UBI with flat tax, as opposed to NIT with taper. Though comes out to be just the same effective marginal tax rates, if using equivalent taper off and tax rates.

      Just confuses people slightly sometimes, not everyone knows what a standard deduction and a personal exemption are (things that a UBI would mostly replace, for one), after all, so might as well give this an extra mention!

      Delete
  9. Money for nothing may be bad, but at least the chicks are free.

    I regret nothing about that Dire Straits reference.

    In all seriousness, what the Swiss are planning on doing reminds me of the Basic Living Stipend used by the People's Republic of Haven in David Weber's excellent Honor Harrington scifi novels. In that, the government created the BLS because they found the disparity between rich and poor disturbing and thought that that would solve the problem.

    It didn't. Instead, it led to mass unemployment and a crappy economy as an increasing number of their populations simply stopped working and lived off their BLS. This led to a rather vicious cycle for the government, as they had to continually conquer other star systems in order to refill their treasury just to keep the government functioning and the stipend payments coming. Unfortunately, once those conquered systems were integrated into Haven's empire, they were added to the BLS system too.

    The BLS created other problems too, since people on the stipend would naturally vote against any politician who tried to curtail the stipend or do anything to get people back to work.

    Now obviously, none of that would happen in real life, especially with Switzerland, but it is an interesting critique of a universal basic income. I don't agree with it, but it's interesting nonetheless.

    ReplyDelete
    Replies
    1. The BLS wasn't really a basic income in the sense we're talking about it. It was just a glorified welfare program - not everyone got it.

      Delete
  10. We get many forms of income already from the government. We get free health care through high school. We get free recreation in the form of public parks, playgrounds, and beaches. We get free transportation in the form of free ride times on public transit and free circulator buses. We get free admission to many museums.

    What we don't get is: medical care, housing, food, and clothing.

    What people really need is a good job. When people have good jobs that allow for decent standards of living they don't need basic income. Creating a state of job's demand with high wages should be a policy goal of government. Then the current system of temporary relief is sufficient to handle short term unemployment. Indeed, the current system was designed with the understanding that the normal state is available well paying jobs.

    When is the last time you saw a "Help Wanted" sign in a window.

    For people who are working class or long term unemployed the current situation is a depression. Few jobs and those that are available are not well paying.

    Until the government faces up to the decades of bad policy choices which has led to a structurally dysfunctional economy nothing will change.


    For more thoughts see: http://canonicalthoughts.blogspot.com/

    ReplyDelete
    Replies
    1. Hmm, I meant free education through high school.

      Delete
    2. Public education is a good comparison to a basic income. While funding varies by community (depending on property tax revenues and the generosity of the states), overall everyone theoretically has access despite the fact that they pay wildly different amounts of property tax. And few people aside from crackpots think that public education should be entirely abolished, although you do see people pushing welfare vouchers (wherein people would still get the same amount of money despite paying different amounts of property taxes).

      Honestly, describing a basic income as welfare seems wrong. It's more like a kind of universal security.

      Delete
  11. I'm in accord with Anonymous at 9:36 and 10:12. It seems likely the underlying driver for both less work and "family breakup" (if that actually occurs) is more freedom of choice on the part of recipients. Has this explanation been evaluated?

    From the post it seems that both Josiah Neeley and the previous discussants of the research took as given that both less work and more family breakup are bad. While I can understand how this would play well in sound bite land, it requires justification and deeper analysis in thoughtful social analysis land. In the social democracies of Europe both of these trends seem to be taking place without a consensus among analysts that they make the quality of life worse.

    Finally, Paul Nollen has a good point. Isn't Alaska a big non-randomized experiment in basic income? It seems wrong to ignore it. What can we learn from that experience?

    ReplyDelete
    Replies
    1. There's also Minincome in Canada from the 1970s.

      Delete
    2. Anonymous10:02 PM

      well said, jed

      Delete
  12. Also regarding Jason's allegory from a science fiction series, it helps to remember the cited work is fiction. It tells us what people find fun to read, not how the world does or could work. We could cite the military role of Ents from Lord of the Rings in our discussion about how to deal with Iran but that wouldn't be very helpful either.

    ReplyDelete
  13. Aidan4:35 PM

    I think the point that the poverty rate is calculated based on a person's income pre-taxes and transfers is more relevant when discussing means tested transfers than a guaranteed basic income, and I agree with Anonymous that it's a mistake to definitively conclude that the results of the randomized trials were negative (especially after reading Noah's bullshit jobs article.

    ReplyDelete
  14. I looked at the Jim Manzi article that Josiah cites. Manzi argues (among other things) from "healthy intuitions of natural justice that are essential to maintaining a well-functioning political order". He does not cite any evidence that his intuitions are in fact essential to maintaining a well-functioning political order -- and my intuitions are completely different, would make the political order function much better, and of course are much more healthy. Also I have evidence -- social democracies in northern Europe.

    Manzi also argues that guaranteed annual income would end up like the income tax code -- and then a couple of paragraphs later he starts putting in the sort of social engineering provisions that contribute to the tax code mess. If he was genuinely inquiring into the feasibility of guaranteed basic income, he'd look at how well Social Security works -- since that is a very close parallel. Of course that comparison would show that administrative costs and fraud are very low, and complexity of the rules has not increased notably over many decades.

    ReplyDelete
    Replies
    1. That's pretty typical Manzi - talking about the importance of experimentation while arguing from premises he doesn't really question that condition what follows. As Kevin Drum pointed out on his blog, it conveniently leads him to where his conservative beliefs would point him every time.

      Delete
    2. Anonymous12:21 PM

      jed,

      Social Security has actually become extremely complex. There are more than 2,000 "major rules" that describe the calculation of benefits, with numerous large carve-outs, exemptions and special cases (e.g., veterans, disability components, a special program only for black lung, etc.).

      See the compilation by the governing laws for the SSA (which themselves require 2 volumes to compile) followed by an entire volume of regulations promulgated by the SSA under these laws, followed by a third section with hundreds of pages of further special rulings: http://www.ssa.gov/regulations/

      Here is an article in a professional journal for financial planners describing how the complexities of understanding SS eligibility can befuddle advisors: http://www.fa-mag.com/news/social-security-complexities-can-befuddle-advisors--expert-says-14309.html

      I've lived, worked and paid taxes in more than one country in Northern Europe. Not one of these societies systematically violates my intuitions of natural justice (though obviously I am not therefore endorsing every policy ever executed in Sweden as my ideal).

      Best regards,
      Jim Manzi

      Delete
    3. Anonymous12:31 PM

      Brett,

      Some obvious counter-examples form my book on the subject include:

      1. Private voucher schools have proven no material gain in test scores in randomized trials beyond what has been achieved by public-sector charter schools.

      2. Family caps in welfare programs have demonstrated no material improvement in any material outcomes in repeated RCTs.

      3. Abstinence education has been repeatedly shown in RCTs to have no material effects on teen pregnancy.

      4. More aggressive police preventive patrols have been shown to have no significant effect in reducing crimes through RCT.

      I could go on at great length.

      A key poitn I tried to make in the book was that very few programs (whether "liberal" or "conservative") actually demonstrate gains in targeted outcomes when subjected to rigorous, independent testing.

      Best regards,
      Jim Manzi

      Delete
  15. "But it also means that any disincentive to work based on the extra income provided by the government will apply across the whole of society."

    It need not mean any such thing; whether or not that is true is entirely determined by the extra taxes needed to pay for a non-means-tested benefit.

    Given all that we do to means-test taxes, there is rarely any economic reason to means-test anything else. Means-testing is a way of reducing the apparent size of government without actually saving anyone money. The real question we should be asking when it comes to means testing is administrative complexity: is it more of a pain to determine who qualifies for a benefit, or is it more of a pain to administer a benefit to everyone when it could be given to just a few. So for example, if food stamps were provided to everyone, and I payed more taxes by an amount roughly equal to my food stamps, my budget and incentives would not be affected, but I'd have the hassle of having two payment methods at the grocery store, which I'd prefer not to have to deal with; better to pay less taxes and give food stamps only to those who need them as far as I'm concerned. With a universal basic income, I don't have any strong preference as to whether it is means tested. Obamacare would probably be simpler and could work without the mandate if subsidies were not means tested, but the CBO would say it was more costly if the government taxed more and gave everyone subsidies instead of mandating that people pay insurance companies, but the real cost is the same.

    ReplyDelete
  16. "These studies concluded that a basic income set at the current poverty rate significantly reduced the average amount of time worked by recipients by the equivalent of 2-4 weeks of full time employment, as compared to the existing welfare system. The experiments also seemed to suggest that providing a basic income increased the likelihood of family breakup. "

    I'm not so sure that these things are negatives. I see them as positives.

    ReplyDelete
  17. (Because you are giving people choices they didn't have before and they are taking them. This is welfare enhancing.)

    ReplyDelete
    Replies
    1. Perhaps this is a case of confusion about the aims of a policy - or of the economy in general. We do tend to live in world where measurability distorts completely our view of the world.

      "We can only manage what we can measure" - so we manage only what we can measure. Sowhere the logic got twisted here.

      Delete
  18. Anonymous12:41 PM

    OK, it's picking a nit. But your prose would be easier on my poor overworked nerves if you would learn the difference between "effect" and "affect."

    The location(as) of the problem are left as an exercise.

    ReplyDelete
  19. keith9:40 PM

    "These studies concluded that a basic income set at the current poverty rate significantly reduced the average amount of time worked by recipients by the equivalent of 2-4 weeks of full time employment" so? That is not very much. Economic growth tends to reduce working time in developed societies as leisure is consumed more. No one argues for less economic growth for that reason. It would be interesting to know if wages were forced up by a Basic income if the working time effects would diminish. Small pilot studies cannot answer that question.

    ReplyDelete