1) At the top you mention "income-tested" and then throughout the body you only speak of "means-tested" so I was shocked that the article just ended! I figured this was a whole long tale about "means-tested" approaches which was then going to be followed by an equally thorough wade through what "income-testing" would look like.
2) Even so... Does this argument (which I very much appreciated reading through and checking the numbers on, so thank you both for being tediously thorough) really argue against "means/income testing" entirely? Or just that it shouldn't be so tight? The concepts of:
1. "rich people should get fewer handouts and benefits from the government" , and
2. "rich people should pay more in taxes"
Don't feeeeeeel like bad concepts? In some of the examples being used here it seems as though our means testing begins taking effect long before we would classify someone as "rich". Must we throw the baby out with the bathwater here? Seems more politically possible to say you're going to raise the accessability of these program to include more people than it would be to say you're doing away with all limits and giving Bill Gates extra money.
3) Focusing so heavily on the marginal tax rate is a very interesting and effective tool but it did sometimes give me a bit of pause when comparing two examples. You both know how progressive tax systems work, but when someone says a millionaire pays more taxes than a fifty-thousand-aire because their "marginal tax rate" is higher I get pulled back to high school civics classes. Sure that next dollar may be taxed differently but the first $50k was taxed the exact same amount for both examples, and every fraction of a dollar you make above that 50k mark is a fraction of a dollar you have that the fifty-thousand-aire doesn't have. You make a strong point about how it's not just the extra taxes that need to be accounted for but also the benefits lost, I just got the creepy crawlies a few times in there.
1. James gave a direct quote from me, in which I used the term "income-tested". He then used "means-tested" for the rest of the article because these programs are often tested on the basis not just of income but also asset levels. (The linked Vox article includes an image of a SNAP application asking about certain types of asset ownership, with someone making a vulgar comment about it.) I was speaking specifically in the context of universal basic income proposals, which is why my quote talks specifically about income testing. However, the argument in many ways applies to a broader concept of personal resource availability, such as assets that could be pawned or sold for cash; see in particular the second paragraph under "How to Fix It": "If the government gives you a benefit because you are poor, it has to take that benefit away when you stop being poor." "Poor" in this sense can mean both low income and low wealth, and eligibility based on wealth (especially if there is a hard cutoff) can have a pernicious effect on people attempting to build up some sort of financial security by having a nest egg of savings to fall back on in hard times or for retirement.
2. The argument is against up-front means-testing altogether, and against caps on payroll taxes such as for Social Security or the Canada Pension Plan. Looking specifically at income, unless you are extremely careful, you will always get the marginal effective tax rate (METR) jumping around when you have income-tested programs, and when you consider wealth you will likely have hard cutoffs where the paradox of the heap (or the sorites paradox) comes into play. (Even if you think that someone with a net worth of $1,000 should get assistance and someone with a net worth of $10,000,000 shouldn't, if you draw the cutoff at, say, $100,000, then you are saying that someone with a net worth of $99,999 is worthy of assistance while someone with a net worth of $100,001 isn't, and that will happen no matter where you draw that line.)
And you're right; the notion that rich people should get fewer benefits doesn't feel like a bad concept. But you can take the math a step further than we did. Consider the final situation, in which both Thomas and Cindy receive the full amount of every benefit offered by Canada and British Columbia for which they are eligible. If you consider taxing all income earned by both of them that is above the Canada Child Benefit's threshold of $36,502 (ideally you would do this benefit by benefit, and also recalculate payroll tax rates to account for removing the caps, which requires other recalculations since CPP and EI premiums are credits against income tax, etc, so this is for simplicity), you will need to levy a tax of about 2.75% on all such income to make up the deficit. Once you run the numbers in this scenario, you find that compared to the means-tested scenario, Cindy is net paying roughly $4,800 more, and Thomas is net receiving roughly that same $4,800. This represents less than 1% of Cindy's pre-tax income, but more than 10% of Thomas's, and after taxes and benefits (which are in many ways the same thing), Cindy's income is still roughly quadruple Thomas's.
In other words, the scheme of universalising benefits while raising general income tax rates above the desired means-testing threshold is still itself a means-test; it is simply done at a different spot in the process.
To the question of the politics, I would prefer not to address that, as I intended to present this from a purely mathematical perspective. What policies one supports on the basis of it, if any, would depend heavily on one's ideological priors. The Fraser Institute, a conservative Canadian think tank, generally supports relaxing, but not eliminating, means-testing; Matt Bruenig, who writes for, among other outlets, Jacobin, supports full universality and argues that the apparent thriftiness of means-testing is entirely based on accounting tricks (which goes to what I noted above about the means-test occurring at a different spot; he has also written about the income taxation system of the Faroe Islands which is an interesting look into how you can structure such a system to draw no distinction between benefits and taxes, while also having other effects which you may or may not consider salutary); you could also make the argument for full abolition of the social safety net (after all, you can't means-test a benefit that doesn't exist!). The math supports the latter two positions, at least, even though they are very much at odds with each other because they flow from differing ideological positions.
As to giving extremely rich people money via what are supposed to be means-tested benefits, the IRS leak from some years ago showed that in 2011, Jeff Bezos received the child tax credit while paying no income tax, so you could argue that this already happens.
3. The first $50,000 is taxed the same for both people, that is true. (Technically not quite so much any more up here since the basic personal amount at the federal level, and in Nova Scotia provincially, is means-tested, but that's a minor quibble.) My personal suspicion is that the way tax forms present calculating your baseline tax liability obscures the fact that the first parts of your income are being taxed at a lower rate than the later parts, and this leads some people to think that moving up a bracket (say, from 12% federally in the US to 22% federally, or 5.06% provincially in BC to 7.7% provincially) makes their entire income subject to tax at that higher rate, instead of just the part which falls into the higher bracket. But at the same time, having to calculate the tax on the income within each individual bracket would be tedious and create yet more barriers to filing, even if it might make the underlying mathematics more transparent.
But even then this becomes pernicious. In Canada, income taxes are assessed to the individual, but most benefits are based on family income. So in the means-tested scenario, consider the situation where Thomas's wife, whom I'll call Ingrid, goes out and gets a job herself to bring in extra income. This immediately incurs the expense of child care (which Ingrid can deduct from her taxable income, and which would probably come to around $5,000 for their two young children). You'll forgive me for not running through the numbers, but if Ingrid's income (after the child care expense deduction) is around $30,000, then if my memory of the various rates serves, both she and Thomas are facing METRs of somewhere between 50% and 55%, around the same as Cindy's.
But if Cindy's husband Robert decides to get a job, say in order to have a bit of financial independence, then because his family's income is already well past all the points at which the various benefits vanish, if he earns around $30,000, he will only face a METR of a little under 30% (federal and provincial taxes, CPP and EI taxes, and the BC tax reduction). Once he hits $40,000, that drops to about 26% and stays there until $45,654, in 2023 at least, far lower than either Thomas or Ingrid. Meanwhile, a single, childless individual making the same $45,000 Thomas makes faces a METR of around 33%--much lower than the struggling parents of three children, but higher than the husband of the rich woman! So this is another way in which means-testing ends up favouring the wealthy. Not only that, but if you compare Cindy and Robert's situation to Thomas and Ingrid's, it's reasonably clear that the former can much more easily afford child care as a necessity of having two working parents than the latter can.
We focused heavily on the marginal effective tax rate because it's so often overlooked. (Sometimes you'll see it called the effective marginal tax rate, or EMTR.) But it is a sensible way of looking at the tax rate, and as Smith observed over two centuries ago: "The necessaries of life occasion the great expense of the poor. They find it difficult to get food, and the greater part of their little revenue is spent in getting it. The luxuries and vanities of life occasion the principal expense of the rich; and a magnificent house embellishes and sets off to the best advantage all the other luxuries and vanities which they possess. A tax upon house-rents, therefore, would in general fall heaviest upon the rich; and in this sort of inequality there would not, perhaps, be any thing very unreasonable. It is not very unreasonable that the rich should contribute to the public expense, not only in proportion to their revenue, but something more than in that proportion." He is, of course, not discussing general taxation of income, but the principle, I think, applies nonetheless: if it is determined as a policy matter to raise revenue for public expenditure via taxation of income, then it is not unreasonable to do so more from that income which would otherwise be spent on "luxuries and vanities", that is, things which are merely desired, than from income which would otherwise be spent on "the necessaries of life", such as food, which are actual needs. That is one rationale for a graduated income tax system: the more likely it is that a given block of income will be spent on needs rather than on wants, the less heavily it is taxed, if it is taxed at all.
I don't really have anything to add to Daniel's comment (and thank you; he's been waiting for someone to comment so he could type that :) ), so I will just say that I agree with you about this: the position argued in the article is totally counterintuitive and feels *wrong* at a gut level.
But the more I think about it, the more it makes sense to me!
A comment longer than most articles!! I appreciate the time you took to respond.
2a) I thought the whole purpose of means testing it was so that we didn’t have those cliffs? The 99,999-er vs 100,001-er you mentioned would only live pennies apart in what they received instead of being noticeably (and thus counterencouragingly) different. Having never once thought about this until last week when you both so eloquently tossed it in front of me, I fancy myself plenty qualified to be boneheadedly single-minded here. I’m on team “pro means-testing, but with a high bar and gradual fadeout”. In part because of…
2b) you’re excellent point about Bezos! The “rich” have so many more options at their disposal to use the fact that they have money to make more money (how’s it hanging over there much-lower-than-income-tax capital gains tax?) as well as use the fact that they have money to avoid losing money (welcome to the party financial-advisor-and-tax-accountant!). So to use the blanket (and mathematically sound) statement that we should balance out opening up benefits to all levels of wealth with higher tax rates for those who are freshly gaining benefits to me just sounds like the rich are going to start working harder to better protect their money from higher taxes while being happy to accept the free-to-them benefits. And I’m not trying to paint the rich as mustache-twirlers, but as one ascends to higher levels of wealth it’s natural to start feeling nervous about losing said wealth. Turning to the local tax or financial expert seems like the safe play. The more money one has the “better” (or at least more expensive) expert they can afford to obtain. The better the expert the more ideas they’ll have at their disposal to know how to save their customers a buck or two. Nothing malicious happened to the person who went from working behind the counter when they were 19 to opening their first franchise at 39, but I bet their taxes sure got more complicated to figure out. Why not let you fancy new tax person talk you in to depreciating that asset over 3 years instead of writing off the new truck all in the first year? Doesn’t sound like one of those nasty loopholes to me, but it does sound like something I probably couldn’t do if I was only using the H&RBlock website to file my state and federal taxes for free.
I feel it’s fairer to BOTH tax the rich higher AND remove their benefits. If we try to only tax them higher but then turn around and give them benefits I’m afraid they’ll find a way to minimize the higher taxes. At least with my preferred method when they minimize their taxes they didn’t also get a benefit out of the deal for free.
2a. With that example I was talking about asset testing, not income testing. It's harder to do a phase-out with assets than it is with income. (For instance, if you need regular welfare in my province, there's a hard asset cutoff of $5,000 if you're single, with a few exceptions such as owning one car.)
2b. They are going to work even harder to avoid taxes! That, among many other things, is how someone like Jeff Bezos could have no income tax liability while being eligible for the child tax credit. (Never mind preferential rates on capital gains and dividends; how about not taking any income and instead borrowing at low interest rates against the value of your securities holdings in large, stable companies like Amazon, and rolling over your loans whenever they come due?) I fully agree there. But this goes to a few other things we noted in the article.
In the current Footnote 27 (EDIT: it is now Footnote 28), we note that it's disproportionately extremely poor people who don't file tax returns, and thus miss out on the income-tested benefits that are supposed to go to them. (One rationale for means-testing is that for the same level of taxation and spending, you can increase the benefit to the poor; Bruenig points out how this is an accounting trick, though whether you agree with him may depend on your views on macroeconomics.) But poor people also disproportionately have very simple tax returns, with wages from a job, maybe a little interest on bank deposits, and not much else. One remedy for this would be free clinics for simple tax returns, and arguably the agency best suited to run such clinics is the tax agency itself. So boosting funding to bodies like the Internal Revenue Service, the Canada Revenue Agency, HM Revenue & Customs, etc., could allow them to run such clinics to help poor people file. (Free tax filing software for simple returns is another option; autofilled returns such as are used in countries like Norway would be yet another; and I do recommend reading Bruenig's description of the Faroese system.)
Boosting funding to those agencies would also give them the resources necessary to properly audit tax evasion by high-income and wealthy people who use complex schemes to avoid taxation, instead of auditing low-income people who make innocent mistakes. Of course, this wouldn't deal with the perfectly legal ways high-income people especially can avoid paying taxes; such methods would have to be handled via other legislative changes. And in some cases, as you note, it may even be desirable to leave some of them in place.
To your last point, as I said, the scheme I propose does remove their benefits! My original draft of this piece (which was far dryer; James did a lot of great work to make it readable) used "Then tax it all away again" as its subheader, which we then used as the final footnote. The Faroese system (Bruenig explains this much better than I can) removes the distinction between taxes and benefits by, in effect, paying benefits at the same time that taxes are deducted, through a system whereby employers remit gross wages to the tax agency and that agency then remits payments to individuals that take the wage income, add benefits, and subtract taxes based on their estimated yearly income, corrected throughout the year. (Yes, it's more complicated than that, but the point that structuring such a system in this way makes it clear that benefits are negative taxes.)
There are many, many more changes that would need to be made than just "make every benefit universal and raise rates above the thresholds". That is certainly the case, but was beyond the scope of the article. But as long as you have a system where benefits are clawed back on the basis of income, and perhaps even more egregiously where certain taxes aren't assessed above a given income level, then at some point you will face this issue of a fluctuating METR unless you are extremely careful. (The federal basic personal amount up here gets around it by linking the means-testing interval to the start and end of a tax bracket, so it's effectively an additional tax on income in that bracket. But it's decidedly the exception.) Not everyone will have the same policy response to this issue, and it's far from unreasonable to say that any realistic implementation of a policy rooted in the math we presented would be worse than what we've currently got because of considerations outside the scope of the calculations.
EDIT: I will freely admit to not having read it in full (it's extremely long) but there is the 1962 Report of the Royal Commission on Taxation, headed by Kenneth Carter, which among other things noted the following:
"A dollar gained through the sale of a share, bond or piece of real property bestows exactly the same economic power as a dollar gained through employment or operating a business. The equity principles we hold dictate that both should be taxed in exactly the same way. To tax the gain on the disposal of property more lightly than other kinds of gains or not at all would be grossly unfair."
I'm going to add to this to say that in some cases, means-testing can even fail the wealthy. I am stealing from Bruenig here; he explains part of this better than I can. (Link to his article in Jacobin: https://jacobin.com/2021/12/universal-benefits-means-testing-build-back-better-childcare Keep in mind that this is someone on the left criticising the centre-left's policy proposals, but the math at least holds regardless of ideology.)
Consider a program for subsidised child-care, such as was, by my understanding, at least proposed for the Build Back Better plan in the US, or that exists in my province as the Affordable Child Care Benefit.
If you income-test eligibility for such a program, then you fail the poor, who disproportionately do not file tax returns, and you fail middle-income families, who see their METR increase, because their subsidy decreases as their income increases. (This gets even worse; I'll get back to that.) But you also fail the wealthy.
If you have a family making, say, $200,000/yr after taxes and benefits, above the final phaseout for the subsidy, and child care or preschool costs $25,000/yr, then this family faces income volatility by having children, because each young child reduces their effective income by that $25,000 in child care expenses.
If, on the other hand, you universalise the subsidy and impose a tax on income above the threshold, and apply it to everyone, not just families with children, then the amount of such a tax on that $200,000 is much lower than $25,000 (never mind $50,000 or more if you have more young children!), and it is levied consistently every year. (Bruenig gives actual figures.) This creates income stability for high-income families, regardless of how many children they have. (And as I noted, even people who have chosen to remain childless still have a strong interest in ensuring that children are healthy and well-educated, because who else is going to fund the retirement benefits?)
Getting back to the parenthetical point above, means-testing can get even worse for that middle-income family. In my province, the maximum child care subsidy is $1,250 (for very young children attending daycares with the highest certification), which is reduced for income above $45,000. Between that and $60,000, the amount of the reduction in the subsidy for a $100 increase in income is $2.27. (In fact, if you put your children in a worse daycare, your subsidy is lower but so is the reduction in your subsidy, so you can lower your METR by putting your kids in a worse daycare, which certainly seems like a perverse incentive.) These are per-child subsidies, not per-family benefits (like the Canada Child Benefit), so if you happen to have enough young children you can end up facing a METR above 100%. (That would occur with 12 young children at the above rate. This is extremely unlikely, but there are genetic conditions that can cause hyperovulation, such as occurred with Mariam Nabatanzi.) It should be noted that my province does correct for this somewhat by reducing the income considered for the benefit by $2,000 per child (except for the first child in a single-parent family, I believe), but it still can occur, simply at a higher income range. Without running the numbers, however, it's possible that because of this exemption the cascade of income-testing ranges and capped taxes does not quite overlap sufficiently to push the METR above 100% except at extremely high numbers of children (probably requiring well more than 12), but the fact nonetheless remains that it is theoretically possible to have a METR above 100% as a result of programs like this, and there are known medical conditions which can result in someone having a very large number of children.
As a caveat to the preceding paragraph, it's possible I've completely misread the regulations, but tinkering around with the benefit estimator the provincial government website provides makes me think I have not.
And whatever you might think about the merits of means-testing as opposed to universal benefits, on whatever grounds, I hope at least that you can agree that a means-tested program which has the potential to push a family's METR above 100%, in this case simply for having more children, something that government policy should at a bare minimum not disincentivise and really should incentivise (lest you face a population implosion, such as China is facing due to, among other things, the aftermath of the one-child policy), is one that should be strongly reconsidered.
1) At the top you mention "income-tested" and then throughout the body you only speak of "means-tested" so I was shocked that the article just ended! I figured this was a whole long tale about "means-tested" approaches which was then going to be followed by an equally thorough wade through what "income-testing" would look like.
2) Even so... Does this argument (which I very much appreciated reading through and checking the numbers on, so thank you both for being tediously thorough) really argue against "means/income testing" entirely? Or just that it shouldn't be so tight? The concepts of:
1. "rich people should get fewer handouts and benefits from the government" , and
2. "rich people should pay more in taxes"
Don't feeeeeeel like bad concepts? In some of the examples being used here it seems as though our means testing begins taking effect long before we would classify someone as "rich". Must we throw the baby out with the bathwater here? Seems more politically possible to say you're going to raise the accessability of these program to include more people than it would be to say you're doing away with all limits and giving Bill Gates extra money.
3) Focusing so heavily on the marginal tax rate is a very interesting and effective tool but it did sometimes give me a bit of pause when comparing two examples. You both know how progressive tax systems work, but when someone says a millionaire pays more taxes than a fifty-thousand-aire because their "marginal tax rate" is higher I get pulled back to high school civics classes. Sure that next dollar may be taxed differently but the first $50k was taxed the exact same amount for both examples, and every fraction of a dollar you make above that 50k mark is a fraction of a dollar you have that the fifty-thousand-aire doesn't have. You make a strong point about how it's not just the extra taxes that need to be accounted for but also the benefits lost, I just got the creepy crawlies a few times in there.
1. James gave a direct quote from me, in which I used the term "income-tested". He then used "means-tested" for the rest of the article because these programs are often tested on the basis not just of income but also asset levels. (The linked Vox article includes an image of a SNAP application asking about certain types of asset ownership, with someone making a vulgar comment about it.) I was speaking specifically in the context of universal basic income proposals, which is why my quote talks specifically about income testing. However, the argument in many ways applies to a broader concept of personal resource availability, such as assets that could be pawned or sold for cash; see in particular the second paragraph under "How to Fix It": "If the government gives you a benefit because you are poor, it has to take that benefit away when you stop being poor." "Poor" in this sense can mean both low income and low wealth, and eligibility based on wealth (especially if there is a hard cutoff) can have a pernicious effect on people attempting to build up some sort of financial security by having a nest egg of savings to fall back on in hard times or for retirement.
2. The argument is against up-front means-testing altogether, and against caps on payroll taxes such as for Social Security or the Canada Pension Plan. Looking specifically at income, unless you are extremely careful, you will always get the marginal effective tax rate (METR) jumping around when you have income-tested programs, and when you consider wealth you will likely have hard cutoffs where the paradox of the heap (or the sorites paradox) comes into play. (Even if you think that someone with a net worth of $1,000 should get assistance and someone with a net worth of $10,000,000 shouldn't, if you draw the cutoff at, say, $100,000, then you are saying that someone with a net worth of $99,999 is worthy of assistance while someone with a net worth of $100,001 isn't, and that will happen no matter where you draw that line.)
And you're right; the notion that rich people should get fewer benefits doesn't feel like a bad concept. But you can take the math a step further than we did. Consider the final situation, in which both Thomas and Cindy receive the full amount of every benefit offered by Canada and British Columbia for which they are eligible. If you consider taxing all income earned by both of them that is above the Canada Child Benefit's threshold of $36,502 (ideally you would do this benefit by benefit, and also recalculate payroll tax rates to account for removing the caps, which requires other recalculations since CPP and EI premiums are credits against income tax, etc, so this is for simplicity), you will need to levy a tax of about 2.75% on all such income to make up the deficit. Once you run the numbers in this scenario, you find that compared to the means-tested scenario, Cindy is net paying roughly $4,800 more, and Thomas is net receiving roughly that same $4,800. This represents less than 1% of Cindy's pre-tax income, but more than 10% of Thomas's, and after taxes and benefits (which are in many ways the same thing), Cindy's income is still roughly quadruple Thomas's.
In other words, the scheme of universalising benefits while raising general income tax rates above the desired means-testing threshold is still itself a means-test; it is simply done at a different spot in the process.
To the question of the politics, I would prefer not to address that, as I intended to present this from a purely mathematical perspective. What policies one supports on the basis of it, if any, would depend heavily on one's ideological priors. The Fraser Institute, a conservative Canadian think tank, generally supports relaxing, but not eliminating, means-testing; Matt Bruenig, who writes for, among other outlets, Jacobin, supports full universality and argues that the apparent thriftiness of means-testing is entirely based on accounting tricks (which goes to what I noted above about the means-test occurring at a different spot; he has also written about the income taxation system of the Faroe Islands which is an interesting look into how you can structure such a system to draw no distinction between benefits and taxes, while also having other effects which you may or may not consider salutary); you could also make the argument for full abolition of the social safety net (after all, you can't means-test a benefit that doesn't exist!). The math supports the latter two positions, at least, even though they are very much at odds with each other because they flow from differing ideological positions.
As to giving extremely rich people money via what are supposed to be means-tested benefits, the IRS leak from some years ago showed that in 2011, Jeff Bezos received the child tax credit while paying no income tax, so you could argue that this already happens.
3. The first $50,000 is taxed the same for both people, that is true. (Technically not quite so much any more up here since the basic personal amount at the federal level, and in Nova Scotia provincially, is means-tested, but that's a minor quibble.) My personal suspicion is that the way tax forms present calculating your baseline tax liability obscures the fact that the first parts of your income are being taxed at a lower rate than the later parts, and this leads some people to think that moving up a bracket (say, from 12% federally in the US to 22% federally, or 5.06% provincially in BC to 7.7% provincially) makes their entire income subject to tax at that higher rate, instead of just the part which falls into the higher bracket. But at the same time, having to calculate the tax on the income within each individual bracket would be tedious and create yet more barriers to filing, even if it might make the underlying mathematics more transparent.
But even then this becomes pernicious. In Canada, income taxes are assessed to the individual, but most benefits are based on family income. So in the means-tested scenario, consider the situation where Thomas's wife, whom I'll call Ingrid, goes out and gets a job herself to bring in extra income. This immediately incurs the expense of child care (which Ingrid can deduct from her taxable income, and which would probably come to around $5,000 for their two young children). You'll forgive me for not running through the numbers, but if Ingrid's income (after the child care expense deduction) is around $30,000, then if my memory of the various rates serves, both she and Thomas are facing METRs of somewhere between 50% and 55%, around the same as Cindy's.
But if Cindy's husband Robert decides to get a job, say in order to have a bit of financial independence, then because his family's income is already well past all the points at which the various benefits vanish, if he earns around $30,000, he will only face a METR of a little under 30% (federal and provincial taxes, CPP and EI taxes, and the BC tax reduction). Once he hits $40,000, that drops to about 26% and stays there until $45,654, in 2023 at least, far lower than either Thomas or Ingrid. Meanwhile, a single, childless individual making the same $45,000 Thomas makes faces a METR of around 33%--much lower than the struggling parents of three children, but higher than the husband of the rich woman! So this is another way in which means-testing ends up favouring the wealthy. Not only that, but if you compare Cindy and Robert's situation to Thomas and Ingrid's, it's reasonably clear that the former can much more easily afford child care as a necessity of having two working parents than the latter can.
We focused heavily on the marginal effective tax rate because it's so often overlooked. (Sometimes you'll see it called the effective marginal tax rate, or EMTR.) But it is a sensible way of looking at the tax rate, and as Smith observed over two centuries ago: "The necessaries of life occasion the great expense of the poor. They find it difficult to get food, and the greater part of their little revenue is spent in getting it. The luxuries and vanities of life occasion the principal expense of the rich; and a magnificent house embellishes and sets off to the best advantage all the other luxuries and vanities which they possess. A tax upon house-rents, therefore, would in general fall heaviest upon the rich; and in this sort of inequality there would not, perhaps, be any thing very unreasonable. It is not very unreasonable that the rich should contribute to the public expense, not only in proportion to their revenue, but something more than in that proportion." He is, of course, not discussing general taxation of income, but the principle, I think, applies nonetheless: if it is determined as a policy matter to raise revenue for public expenditure via taxation of income, then it is not unreasonable to do so more from that income which would otherwise be spent on "luxuries and vanities", that is, things which are merely desired, than from income which would otherwise be spent on "the necessaries of life", such as food, which are actual needs. That is one rationale for a graduated income tax system: the more likely it is that a given block of income will be spent on needs rather than on wants, the less heavily it is taxed, if it is taxed at all.
I don't really have anything to add to Daniel's comment (and thank you; he's been waiting for someone to comment so he could type that :) ), so I will just say that I agree with you about this: the position argued in the article is totally counterintuitive and feels *wrong* at a gut level.
But the more I think about it, the more it makes sense to me!
A comment longer than most articles!! I appreciate the time you took to respond.
2a) I thought the whole purpose of means testing it was so that we didn’t have those cliffs? The 99,999-er vs 100,001-er you mentioned would only live pennies apart in what they received instead of being noticeably (and thus counterencouragingly) different. Having never once thought about this until last week when you both so eloquently tossed it in front of me, I fancy myself plenty qualified to be boneheadedly single-minded here. I’m on team “pro means-testing, but with a high bar and gradual fadeout”. In part because of…
2b) you’re excellent point about Bezos! The “rich” have so many more options at their disposal to use the fact that they have money to make more money (how’s it hanging over there much-lower-than-income-tax capital gains tax?) as well as use the fact that they have money to avoid losing money (welcome to the party financial-advisor-and-tax-accountant!). So to use the blanket (and mathematically sound) statement that we should balance out opening up benefits to all levels of wealth with higher tax rates for those who are freshly gaining benefits to me just sounds like the rich are going to start working harder to better protect their money from higher taxes while being happy to accept the free-to-them benefits. And I’m not trying to paint the rich as mustache-twirlers, but as one ascends to higher levels of wealth it’s natural to start feeling nervous about losing said wealth. Turning to the local tax or financial expert seems like the safe play. The more money one has the “better” (or at least more expensive) expert they can afford to obtain. The better the expert the more ideas they’ll have at their disposal to know how to save their customers a buck or two. Nothing malicious happened to the person who went from working behind the counter when they were 19 to opening their first franchise at 39, but I bet their taxes sure got more complicated to figure out. Why not let you fancy new tax person talk you in to depreciating that asset over 3 years instead of writing off the new truck all in the first year? Doesn’t sound like one of those nasty loopholes to me, but it does sound like something I probably couldn’t do if I was only using the H&RBlock website to file my state and federal taxes for free.
I feel it’s fairer to BOTH tax the rich higher AND remove their benefits. If we try to only tax them higher but then turn around and give them benefits I’m afraid they’ll find a way to minimize the higher taxes. At least with my preferred method when they minimize their taxes they didn’t also get a benefit out of the deal for free.
2a. With that example I was talking about asset testing, not income testing. It's harder to do a phase-out with assets than it is with income. (For instance, if you need regular welfare in my province, there's a hard asset cutoff of $5,000 if you're single, with a few exceptions such as owning one car.)
2b. They are going to work even harder to avoid taxes! That, among many other things, is how someone like Jeff Bezos could have no income tax liability while being eligible for the child tax credit. (Never mind preferential rates on capital gains and dividends; how about not taking any income and instead borrowing at low interest rates against the value of your securities holdings in large, stable companies like Amazon, and rolling over your loans whenever they come due?) I fully agree there. But this goes to a few other things we noted in the article.
In the current Footnote 27 (EDIT: it is now Footnote 28), we note that it's disproportionately extremely poor people who don't file tax returns, and thus miss out on the income-tested benefits that are supposed to go to them. (One rationale for means-testing is that for the same level of taxation and spending, you can increase the benefit to the poor; Bruenig points out how this is an accounting trick, though whether you agree with him may depend on your views on macroeconomics.) But poor people also disproportionately have very simple tax returns, with wages from a job, maybe a little interest on bank deposits, and not much else. One remedy for this would be free clinics for simple tax returns, and arguably the agency best suited to run such clinics is the tax agency itself. So boosting funding to bodies like the Internal Revenue Service, the Canada Revenue Agency, HM Revenue & Customs, etc., could allow them to run such clinics to help poor people file. (Free tax filing software for simple returns is another option; autofilled returns such as are used in countries like Norway would be yet another; and I do recommend reading Bruenig's description of the Faroese system.)
Boosting funding to those agencies would also give them the resources necessary to properly audit tax evasion by high-income and wealthy people who use complex schemes to avoid taxation, instead of auditing low-income people who make innocent mistakes. Of course, this wouldn't deal with the perfectly legal ways high-income people especially can avoid paying taxes; such methods would have to be handled via other legislative changes. And in some cases, as you note, it may even be desirable to leave some of them in place.
To your last point, as I said, the scheme I propose does remove their benefits! My original draft of this piece (which was far dryer; James did a lot of great work to make it readable) used "Then tax it all away again" as its subheader, which we then used as the final footnote. The Faroese system (Bruenig explains this much better than I can) removes the distinction between taxes and benefits by, in effect, paying benefits at the same time that taxes are deducted, through a system whereby employers remit gross wages to the tax agency and that agency then remits payments to individuals that take the wage income, add benefits, and subtract taxes based on their estimated yearly income, corrected throughout the year. (Yes, it's more complicated than that, but the point that structuring such a system in this way makes it clear that benefits are negative taxes.)
There are many, many more changes that would need to be made than just "make every benefit universal and raise rates above the thresholds". That is certainly the case, but was beyond the scope of the article. But as long as you have a system where benefits are clawed back on the basis of income, and perhaps even more egregiously where certain taxes aren't assessed above a given income level, then at some point you will face this issue of a fluctuating METR unless you are extremely careful. (The federal basic personal amount up here gets around it by linking the means-testing interval to the start and end of a tax bracket, so it's effectively an additional tax on income in that bracket. But it's decidedly the exception.) Not everyone will have the same policy response to this issue, and it's far from unreasonable to say that any realistic implementation of a policy rooted in the math we presented would be worse than what we've currently got because of considerations outside the scope of the calculations.
EDIT: I will freely admit to not having read it in full (it's extremely long) but there is the 1962 Report of the Royal Commission on Taxation, headed by Kenneth Carter, which among other things noted the following:
"A dollar gained through the sale of a share, bond or piece of real property bestows exactly the same economic power as a dollar gained through employment or operating a business. The equity principles we hold dictate that both should be taxed in exactly the same way. To tax the gain on the disposal of property more lightly than other kinds of gains or not at all would be grossly unfair."
I'm going to add to this to say that in some cases, means-testing can even fail the wealthy. I am stealing from Bruenig here; he explains part of this better than I can. (Link to his article in Jacobin: https://jacobin.com/2021/12/universal-benefits-means-testing-build-back-better-childcare Keep in mind that this is someone on the left criticising the centre-left's policy proposals, but the math at least holds regardless of ideology.)
Consider a program for subsidised child-care, such as was, by my understanding, at least proposed for the Build Back Better plan in the US, or that exists in my province as the Affordable Child Care Benefit.
If you income-test eligibility for such a program, then you fail the poor, who disproportionately do not file tax returns, and you fail middle-income families, who see their METR increase, because their subsidy decreases as their income increases. (This gets even worse; I'll get back to that.) But you also fail the wealthy.
If you have a family making, say, $200,000/yr after taxes and benefits, above the final phaseout for the subsidy, and child care or preschool costs $25,000/yr, then this family faces income volatility by having children, because each young child reduces their effective income by that $25,000 in child care expenses.
If, on the other hand, you universalise the subsidy and impose a tax on income above the threshold, and apply it to everyone, not just families with children, then the amount of such a tax on that $200,000 is much lower than $25,000 (never mind $50,000 or more if you have more young children!), and it is levied consistently every year. (Bruenig gives actual figures.) This creates income stability for high-income families, regardless of how many children they have. (And as I noted, even people who have chosen to remain childless still have a strong interest in ensuring that children are healthy and well-educated, because who else is going to fund the retirement benefits?)
Getting back to the parenthetical point above, means-testing can get even worse for that middle-income family. In my province, the maximum child care subsidy is $1,250 (for very young children attending daycares with the highest certification), which is reduced for income above $45,000. Between that and $60,000, the amount of the reduction in the subsidy for a $100 increase in income is $2.27. (In fact, if you put your children in a worse daycare, your subsidy is lower but so is the reduction in your subsidy, so you can lower your METR by putting your kids in a worse daycare, which certainly seems like a perverse incentive.) These are per-child subsidies, not per-family benefits (like the Canada Child Benefit), so if you happen to have enough young children you can end up facing a METR above 100%. (That would occur with 12 young children at the above rate. This is extremely unlikely, but there are genetic conditions that can cause hyperovulation, such as occurred with Mariam Nabatanzi.) It should be noted that my province does correct for this somewhat by reducing the income considered for the benefit by $2,000 per child (except for the first child in a single-parent family, I believe), but it still can occur, simply at a higher income range. Without running the numbers, however, it's possible that because of this exemption the cascade of income-testing ranges and capped taxes does not quite overlap sufficiently to push the METR above 100% except at extremely high numbers of children (probably requiring well more than 12), but the fact nonetheless remains that it is theoretically possible to have a METR above 100% as a result of programs like this, and there are known medical conditions which can result in someone having a very large number of children.
As a caveat to the preceding paragraph, it's possible I've completely misread the regulations, but tinkering around with the benefit estimator the provincial government website provides makes me think I have not.
And whatever you might think about the merits of means-testing as opposed to universal benefits, on whatever grounds, I hope at least that you can agree that a means-tested program which has the potential to push a family's METR above 100%, in this case simply for having more children, something that government policy should at a bare minimum not disincentivise and really should incentivise (lest you face a population implosion, such as China is facing due to, among other things, the aftermath of the one-child policy), is one that should be strongly reconsidered.