« Federal Budget 2016: Quick Comment | Main | Economics can't be encapsulated into knowledge pills. And that matters for research funding. »


Feed You can follow this conversation by subscribing to the comment feed for this post.

The only one I know that has worked through the details is Ed Dolan. In general, low administrative costs and highly targeted programs mean they would cost more, though universality means fewer falling through the cracks.


I'm not sure about your calculations, but here is one way to evaluate the impact of a minimum income policy (negative income tax scheme):


It is a theoretical analysis that explicitly takes into account the incentive effects.

If you look beyond the accounting, there are economic effects, for example on precautionary savings and labor supply that should not be neglected. In fact, they are huge: here is my paper with Fabre and Pallage on this.


Efficiency is a big issue here. Which is more efficient: an electricity rebate for low income families or basic income? Politicans confuse the allocative and distributional roles of prices but they never confuse what buys votes.

Hi Nick,

Thanks for offering this mind-teaser. Here's how to reconcile this with the way I've been thinking about it:

From your 'current' system to your 'basic income' system, you are making two changes:
a) increasing transfers from 2.5B to 10B, for an increase of 7.5B
b) getting rid of personal exemption which raises taxes by 7.5B.

So, I'd cost your plan by adding up the amount of new taxes from the elimination of the basic exemption (7.5B) plus the amount saved by getting rid of SA (2.5B).

So, I don't see a flaw in the way we've been talking about costing basic income on twitter.

I quite like the way you are thinking about it--especially integrating it with the tax system as one fluid system. That's the right way to do it.

To note, the advocates of these plans typically spend their time arguing about whether the transfer ought to be 15K or 20K, not how to cost it or how to integrate it with taxes. They typically conceive of this as outside the existing tax system as a replacement for Social Assistance.

Another note--I've often heard people claim that the administrative costs for social assistance represent some kind of big savings that could fund a basic income. Ontario spends $285M administrating all of its social assistance programs (including drug benefits; disabled benefits). That's $20 per Ontarian per year. That's about $14,980 short of funding a basic income of $15K per year!

Thanks Christian for posting your paper. The findings about the advantages of being able to target better to those in need with UI than with UBI make sense to me.

To note, the UBI in your parameterization is 2K per year, which is quite a different thing than is being spoken about in the Finnish pilot program and other media reports talking about 15K/year transfers....

But isn't there a problem, in that the $20,000 exemption only provides cash by reducing income tax payable. For the working poor the Basic Income would be better and this improvement has positive effects for the macro economy i e better incentives. You ignore the positive incentives of a Basic Income plan for low income workers (or is this unimportant?).

Lord: I forgot about the "falling through the cracks" argument. Seems true to me, though I don't know how big it is.

David and Christian: I have read the abstracts to your papers. My hunch is that you are doing something similar (which is why I am replying to both of you at once). My hunch is that the intuition for the results in your papers has something to do with Jensen's inequality, and the non-linearity of the tax system. Because a linear tax-transfer system (where the marginal net tax rate is the same for everyone at all time periods) should neither encourage nor discourage saving and investment and risk-taking. Am I right?

Avon: yep. Though you can get an example where it would maybe be efficient to provide free stuff, like glasses for example, if the demand for glasses is perfectly inelastic, and only people with bad eyes buy glasses. It's like a lump sum transfer from those with good eyes to those with bad eyes, which has no disincentive effects, and is better than giving free glasses only to poor people, because that raises the marginal net tax rate for people with bad eyes, and has disincentive effects as well as being unfair. (That was the argument behind my and Frances' universality paper (pdf).)

Kevin: Thanks.

A lot of what I read in support of UBI does indeed sound like "Let's give everyone a free pony!", and you and Stephen are right to slam pieces like that. (I first thought "they can't be that daft", then I read more, and think "yes they can".) But I wondered if you weren't going too far the other way, with your costing exercises. And Twitter's 140 just wasn't enough to explain why I felt very uneasy, and I wanted to get my own head clearer on the issue too, so I wrote this post.

When I think about switching to UBI I "see" a graph with disposable income on the Y axis and market income on the X axis, and I see a total mess representing the current system, and I see UBI as using OLS to fit a line through the existing plot. And so it's self-financing by construction assuming no behavioural changes. Like in my old post.

But I think you are right that another way to think about the same thing is that UBI lets us eliminate the personal exemption (and maybe eliminate the lower banks so everyone pays the top tax rate). Because the argument for zero or low tax rates on the first $X of income was always a bit re-distributionist "Poor people shouldn't have to pay tax, or should only pay low tax". And if UBI handles the poverty problem, those aren't needed any more. So another way to get a benchmark UBI is we eliminate welfare etc., and we make everyone pay the existing top marginal tax rate on all income, and we see how big a UBI that would finance.

Richard: You can't have UBI the same as existing welfare and at the same time reduce marginal net tax rates for everyone. But what you can do is have UBI the same as existing welfare and *smooth out* the marginal net tax rate schedule. And that should have beneficial incentive effects, but it's a second-order effect from eliminating spikes and flat spots in the tax/transfer system. In other words, the incentive gains from cutting a 100% marginal tax rate in some spots will exceed the incentive losses from raising a 0% marginal tax rate in other spots.

And the argument *against* UBI is that some people (e.g. the severely disabled) might have a lower labour supply elasticity than other people, so it makes sense to pay them bigger welfare and face a higher marginal net tax rate, because the disincentive effects will be lower. Maybe not the best example.


In the model I study, the tax rate on labor income is linear. There is no financial market. People accumulate human capital a la Ben Porath. And, people differ with respect to their pure rate of time preference (so they have different marginal propensities to consume.) Not sure if this last feature is what's responsible for the results.


Kevin: in my paper, a UBI of more than $2K is really bad. But then, the paper only seeks to replace EI. I suspect that if you add other social programs, results carry through with larger numbers, assuming the use of the various programs is correlated.

Thanks Christian. That's very helpful. When discussing these schemes I tend to over-rely on the "it would cost 25% of GDP" argument because it's easy to make. But that makes it sound too much like "gee this would be great if only we could afford it". In fact, because of the mechanisms you quantify, I'm skeptical of the whole concept of replacing all existing social insurance with such a transfer. It's not just about cost..

Hi Nick, I may be going too far the other way in leaving the impression that *all* UBI-tyoe schemes are bad. The mechanisms of transfer and phase out are very useful to smooth out sharp MTRs as you point out. The Simpson and Stevens scheme to make the basic exemption refundable is a sound idea, for example, and is effectively a mini basic income.

But I dont think my costing thumbnails I've been tweeting are wrong. If a scheme pays for itself by increasing taxes (like yours does with removal of basic exemption) I would include that in the costing. But when this is discussed, every scheme I see talks only about SA (and possibly other transfers). When pressed, many will offer up "admin savings" which are in fact trivial--in the case of Ontario $20 per capita per annum.

Exemption isn't redistributionist but a recognition of costs of income generation, but a refundable ubi could cover this.

Hi Lord,

I agree--if more UBI proposals focused on folding in the basic exemption (as Nick's proposal does) then they would be much more credible proposals.

Christian Zimmerman, it's no big surprise that UBI didn't work very well in your model as a replacement for unemp. insurance alone. Generally, UBI is viewed as a redistribution framework (much like a low-wage subsidy, e.g. EITC), with unemployment insurance seen as a sort of desirable side effect. In this framework, the transfers on employed workers are not "wasted", they're almost the whole point! The flip side is that the level of UBI that's appropriate for _this_ may be low enough that you do need some actual unemp. insurance on top - but the incentive problems associated with insurance would be reduced accordingly.

As an aside: AIUI the reason your model does not reflect such a rationale is because the agents are not heterogenous in productivity - the wage of employed agents is simply normalized to 1. This is a nice simplification if all you care about is employment shocks, but the drawbacks are also obvious.

> is that UBI lets us eliminate the personal exemption (and maybe eliminate the lower banks so everyone pays the top tax rate).

Nick, AIUI in non-linear taxation models, the optimal solution involves roughly U-shaped marginal tax rates, so not quite linear. It may seem surprising that low incomes should in fact face _higher_ marginal rates, but my intuition for this is that this ensures that the breakeven income post-taxes and transfers does not 'drift' too high, so redistribution is in fact targeted to the lowest incomes. (You also get mildly progressive rates on higher incomes, but the rationale for that is fairly standard of course.)

Kevin: that makes sense. It seems like the "UBI" I've always had at the back of my mind (my benchmark UBI) is very different from the UBI that some of its supporters seem to be talking about and that you were responding to, which is why it seemed to me you were missing the point. We are now on the same page. I see the benefits of UBI as: marginal tax rate smoothing; making sure no-one falls through the cracks; and on the admin cost side I think we need to remember the admin costs of people applying as well as the government (though only the latter should be included when we are checking for budget-neutrality, of course). I'm really useless at filling in forms and stuff to apply for things, and I'm supposed to be one of the intelligent educated ones. If the people who really need the cash could easily do the paperwork, they probably would be able to get a job and wouldn't need the cash. The whole family vs individual tax-splitting question is mostly orthogonal to the UBI question, but if a flat tax were seen as part of the UBI a lot of those questions go away. At most we would face the question of whether a couple should get twice the UBI of two singles, for marriage-neutrality.

anon: I can't get the intuition for U-shaped marginal tax rates. I thought there was a standard argument for *falling* marginal tax rates at the very top end of the income distribution, and I can get the intuition for that (though I am very dubious about the generality of that result). My *guess* is that it might have something to do with the assumed shape of the distribution of abilities.

OK, maybe it's because you want to have low marginal tax rates around the mode of the distribution, to reduce the disincentive effects on more people?

> I thought there was a standard argument for *falling* marginal tax rates at the very top end of the income distribution

I think the only "standard" argument is the argument from contract design that if you could figure out what the very top incomes are, the _marginal_ rate on these should in fact be zero. The best intuition I can give is that the optimal setting is one where the top-earning agent pays all of her taxes in the 'lower brackets' (which have no incentive effect), and any marginal tax higher than zero would just incent her to work less. But AIUI this has very limited relevance at best, given what we know about the actual distribution of income-earning potential among top earners.

> I see the benefits of UBI as: marginal tax rate smoothing

As an aside, note that marginal rate smoothing becomes extremely important if you do assume that lower incomes should face higher marginal rates. (Because the distortion effect rises with the square of the tax rate - and with a patchwork of programs it's even very easy for clawback rates to add up to over 100%.) So the argument for high clawback rates is not an anti-UBI argument - far from it!

Nick: "I can't get the intuition for U-shaped marginal tax rates."

Imagine that there's a distribution of pre-tax income that's at least partially exogenous, and it's roughly normal. The tax/benefit system has to give support to people with no other means of supporting themselves, and collect revenue too, so it's got to have some kind of structure of marginal tax rates. If you put high marginal tax rates places where there aren't many people, and low marginal tax rates places where there are lots of people, that reduces the marginal tax rate that the typical person faces. So a roughly normal distribution of income would give you a roughly U-shaped marginal tax rates. (Of course it's much more complicated than that, but that gives you something to go on). IMO, whenever we start seeing pressure to reduce MTRs at the bottom end of the income distribution, it's because incomes/wages are shifting in such a way that there are a lot more people potentially affected by those MTRs.

E.g. in the 1960s it was pretty easy for an unskilled worker to get a job that would lift him/her out of poverty. So it didn't matter so much if the marginal tax rate on welfare was 100%, most people would be better off with a job than without, and most jobs were full year/full time. Another way of thinking about this is that the decision was being made on the extensive margin, not the intensive one. The introduction of Earned Income TAx Credits, welfare reform, etc was in many ways a product of Springsteen economics: those steady jobs are going boys, and they ain't coming back. When more jobs are part time/part year/low wage, the intensive margin (and MTRs at the bottom) become more important.

(Not that you'll see this for a few days anyways - have a great weekend).

> those steady jobs are going boys, and they ain't coming back.

I agree that this is an important issue, and this goes back to the distinction between unemp. insurance and UBI in reply to Zimmerman. When unskilled jobs are broadly available and relatively high-paying, it's really hard to make a case for either UBI or wage subsidies. Thus, all you really need is unemp. insurance, plus maybe a few patchwork programs for the odd folks who, for whatever reason, cannot fill these good jobs.

Things change quite drastically when unskilled jobs become low-paying and unsteady, _and_ it's difficult to retrain for a better job. Then the case for something like UBI gets quite strong. Some countries (UK, US to some extent, maybe Germany?) try to manage simply by having no UBI persay, but aggressively slashing taxes on low incomes. But there's no real guarantee that this will suffice to guarantee a minimally decent quality of life - so plenty of people in the UK are in fact still living on 'benefits' of some sort, with all that entails.

I know you abstract from behavioral consequences (which makes this more a flat tax post in my view). Still I think their role in even this UBI version shows how fragile this idea really is.

The public budget spirals out of balance if only one guy reduces his work income after UBI. Not because the transfer is different, but because the associated hassle is. Then you have to raise taxes or reduce the UBI, and off it goes.

I stole this point from my colleague Florian Habermacher:

Thanks Frances. Got it. Am sipping double double in Tims in London Ont. Lovely drive down, especially seeing sun reflecting off the ice on all the trees.

Joannes: I will take a look when I'm on a better connection. But I would have thought the same problem applies to any tax/transfer system?

The comments to this entry are closed.

Search this site

  • Google

Blog powered by Typepad