Taxing Ontario’s Richest

Ontario’s NDP was out today with a Robin Hood proposal to collect more provincial tax from personal incomes in excess of half a million dollars. The approximately $570 million of additional revenue would increase the Ontario Disability Support Plan, protect childcare spaces and remove provincial HST from home heating.

UBC economist Kevin Milligan has been tweeting his estimate that the NDP’s proposal would collect only $290 million given no behavioural response. He is correct that “there is no public source of data that allows us to know exactly how much income is in an Ontario $500K tax bracket.”

As New Democrats tweeted, Informetrica computed the $570-million estimate using Statistics Canada’s Social Policy Simulation Database and Model (Version 18.1). Whether or not one accepts that calculation, Dr. Milligan’s estimate is clearly too low.

His starting point is the Canada Revenue Agency’s (CRA) income statistics for 2009. While using the most recent available data is reasonable, it’s worth noting that top incomes were temporarily depressed in 2009 by the financial crisis, which cut executive bonuses, capital gains, etc.

Dr. Milligan states, “The CRA statistics do not break out the top 250K by province,” and proceeds to estimate that Ontarians making more than $250,000 had taxable income of $35 billion. In fact, CRA’s website does provide a provincial breakdown, which indicates that this figure was $42 billion. But Dr. Milligan uses the smaller figure as his point of departure to (under)estimate Ontario’s taxable income above $500,000.

He interprets the proposed two percentage points of additional tax as increasing the top combined federal-provincial rate from 46.41% to 48.41%. In fact, the NDP proposal is to raise the provincial rate from 11.16% to 13.16% for income above $500,000.

The existing 56% Ontario surtax would amplify this increase. In other words, the top combined rate would go from 11.16%*1.56+29%=46.41% to 13.16%*1.56+29%=49.53%. So, Dr. Milligan understates both the tax rate and the tax base.

There may be a legitimate debate about whether, and how much, the super-rich might respond to the proposed tax change. But any predicted revenue losses from behavioural adjustments should be subtracted from a reasonable static estimate like $570 million, rather than from a low-ball estimate like $290 million.

UPDATE (April 3): Dr. Milligan now seems to have boosted his estimate.


  • Hi Erin,

    Thanks for pointing out some corrections. The twitter/google spreadsheet exercise was an interactive one. Thanks to many who helped narrow down the difference in the estimates.

    The major difference was understanding that the 2% was to be applied *before* the surtax. It isn’t obvious–this could have been a surtax. It wasn’t stated in the press release. The Globe story said that taxes were going up 2.00%. That’s what I had to work with. I was glad to be informed of the details of the policy by Michael Rosenstock of the ONDP. If there had been even a modest backgrounder with the press release, that mistake could have been avoided.

    I’m encouraged in this post by your openness to considering behavioural responses. I am curious why there seems so much opposition from progressive economists in Canada to this very standard consideration. In the NYTimes today, Paul Krugman made reference to the Diamond-Saez calculations–which use an elasticity of 0.25. Krugman cites Saez to defend against GoP bumf that the elasticity is >1. I’m citing Saez here to say that it is >0.

  • For the record, here’s what appeared in the Globe:

    “Increasing the tax rate for high-income earners by 2 percentage points to 13.16 per cent would raise $570-million a year, according to the New Democrats’ calculations.”

    As I read it, there are 2 things we agree are factually wrong in that sentence, and a third we may or may not agree about.

    I can’t verify what the ONDP said as I can’t find the news release. But since Erin has put it very clearly here, I take it he would regard it as misleading to say that the top tax rate is 11.16%, or that increasing the rate by 2 percentage points is what is being proposed.

    (In other words, can we please get rid of this surtax business and admit we have 5 tax brackets in Ontario, the top one at 17.4%?)

  • I’m not surprised that the 2% increase (that is really a 3.12% increase) has been confusing to people. Erin first clarified the difference since the original CUPE report seems blissfully unaware of the existence of the surtax – it doesn’t mention it at all.

  • By quoting the 13.16% rate, the Globe made it clear that the ONDP was proposing to increase the top rate before the surtax applies. You can’t blame the party or CUPE for addressing the rate structure as printed on the Ontario tax form (page 2). By Dr. Smart’s standard, the Canada Revenue Agency’s list of provincial tax rates is “factually wrong.”

  • Hi Erin,

    I don’t like relying on newspapers and always prefer to go to the party source. Here is what the NDP says; “In order to offset costs those Ontarians making more than half a million dollars would see a modest, two-percentage point income tax increase. ”

    Now, to me, I see a ‘two percentage point increase’. I didn’t know that actually meant 3.16%. If she was talking about the ‘rate’ she should have said the ‘rate’. But, she said income taxes were going up to points. That’s just not true. They are going up 3.16%.

    It could have been a 1.28% increase that, when grossed up by the surtax, led to an actual 2 percentage point increase.

    I understand the needs of political marketing, so I don’t mind that so much. I do think though that we shouldn’t expect other analysts to have to assume that press releases are being slightly fudgey with numbers.

    Again, let me stress that if you want people to analyze your policies accurately, you ought to provide some kind of backgrounder with the details.


  • All that said, I think a strong fairness case can be made for increasing taxes at the top. (As I have said before.)

    I just think the revenue projections (and the marketing of the tax rates!) ought to be honest and realistic if credibility is to be kept.

  • I’m not sure what CUPE report rcp or others are referring to as being “blissfully unaware of the existence of the surtax”; but I suspect it is this one.

    There’s a limit to how much detail you can put in a two pager that deals with six different tax measures and half a dozen other issues. Certainly I’m aware of the existence of the surtaxes and these were factored into the calculations.

    And, as far as press releases are concerned, they also have severe space restraints. Are you holding them to a higher standard than you would a few hundred page government budget?

  • Hi Toby,

    I was asking for a one page backgrounder alongside the press release. This is standard going back through the entire internet era. Policy announcements have press releases and backgrounders.

    Without that, you’re asking the analyst to guess what you mean.

    thanks, Kevin

  • Maybe the correct approach is to employ more precise terminology.

    We know that the thing that Ontario calls a top marginal rate is not what an economist would call a top marginal rate (i.e. how much more tax is paid for an extra dollar of income), due to the existence of the surtax.

    So we could say something like “the NDP proposes that Ontario’s nominal top marginal rate be increased from 11.16% to 13.16%, corresponding to an increase in the real top marginal rate from 17.41% to 20.53%.”

  • Kevin,

    I agree that the NDP should post its media backgrounders along with its press releases. NDP communications staff have a different view.

    If your goal was to better understand the NDP’s proposal, a good approach would have been to contact the NDP directly and/or look at an Ontario tax form.

    If you support higher taxes at the top and want to ensure that such proposals are credible, then trying to publicly discredit the NDP’s numbers seems like a strange approach. Anyway, your corrected estimate validates the NDP-Informetrica revenue figure.

  • Hi Erin,

    My very first tweets did indeed reach out to ONDP people. I @’d them on the tweets directly. I emphasized at the very top of the spreadsheet that my estimates were preliminary and could be wrong. I asked both in the spreadsheet and in the tweets for suggestions and corrections. I was thankful when ONDP’s Michael Rosenstock corrected my assumption about the 2% increase. You are free to think that I should have known that 2% meant 3.12%, but I didn’t.

    Let me reiterate one thing: I can come up with something close to the ONDP estimate only when I assume no behavioural response. What is the empirical basis for that assumption by the ONDP? I was happy to hear lots of talk at the NDP leadership convention about the importance of science in public policy. The best evidence we have on taxable income elasticities bound the elasticity away from zero. Why ignore evidence?

  • Since the updated plausible-elasticity estimate from Kevin Milligan is $254 million, and the NDP estimate was $570 million, it is a bit of a stretch to say that the Milligan estimate validates the NDP estimate. In fact it shows that the NDP estimate is likely overstated by a factor of two.

  • At one point, Kevin had a static estimate of $530 million, which validated the NDP-Informetrica static estimate of $570 million.

    Neither the federal Department of Finance nor the Ontario Ministry of Finance assumes behavioural responses in projecting the fiscal effect of changing tax rates. It is hardly outlandish for the NDP’s budget proposals to follow the same methodology.

    The social “science” of economics is far from definitive on possible behavioural responses to tax changes . It is a legitimate issue to consider, but I am not convinced that the elasticity of taxable income is closer to 0.25 than to zero.

  • Erin & Kevin,

    If you’re going to take into account the taxable elasticity of income, don’t you also need to account for the increased tax revenue from other sources that’s likely to result from the policy (e.g. mom’s who can earn more income because there is child-care available which allows them to enter the work force, increased aggregate demand as a result of upping ODSP payments and taking HST off home heating)? I’m saying that the behavioral responses are likely complex and swing both ways when you look at the proposal as a whole.

  • Darwin O'Connor

    I would have thought that people who only cared about paying the lowest taxes would have already moved to provinces or countries with lower taxes. This increase wouldn’t make much difference.

  • Hi Erin,

    The upper tail is volatile enough that I can say the $570MM isn’t outside the bounds of reason for a 3.12% increase for a *static* estimate.

    You may be right that the Ontario Ministry of Finance doesn’t do dynamic scoring. But:

    a) this is particularly important for top rate taxation where all evidence and economic logic suggests the elasticity is higher

    b) All serious tax analysis in the US and UK (and perhaps other places as well) takes the elasticities seriously. Look at the Mirrlees report. Look at all of Saez’s work. I think we can meet that standard in Canada too.

    You’re welcome to be skeptical about high income folks trying to avoid taxes–it is remarkably generous of you to think so highly of their civic commitment!

    Me, I’ll keep company with the noted knuckle-dragging right-wingers Krugman, Saez, and Diamond who are all happy with a 0.25 elasticity as a reasonable benchmark.

    By the way, with a 0.25 elasticity you can still get a number for the top optimal tax rate (using the Diamond-Saez argument) of about 71% for Canada.

    There are strong fairness and equity reasons to consider higher rates. But it’s just not a big revenue raiser. You’re not going to get enough revenue there to fund an expansion of social democracy. That’s my point.

  • Hi Darwin,

    I agree. People are not likely to move. They are also unlikely to work any less–CEOs and sports stars respond to positional competition, so tax rates won’t change their effort much at all.

    The argument for behavioural response of the rich is about *accounting* and *financial* responses, not working less or moving away.

    Let me quote Saez again (page 35):

    “As this discussion has shown, there is compelling evidence of substantial responses of upper income taxpayers to changes in tax rates, at least in the short run. However, in all cases, the response is either due to short-term retiming or income shifting. There is no compelling evidence to date of real responses of upper income taxpayers to changes in tax rates.”

    If you raise a rich guy’s taxes by a dollar, the argument is that he’d rather pay his accountant 50 cents to avoid it than face the dollar of taxation.

    Arguing for a zero response is saying that rich people are just going to sit back and happily pay higher taxes instead of following their accountant’s advice.

    I might call that naive.

  • It seems plausible that Ontario’s super-rich are already fully exploiting tax-avoidance opportunities. I doubt that raising the top combined rate from 46.41% to 49.53% would prompt tax evasion from many people who are currently playing by the rules.

    I guess you could argue that a high-income Ontarian with no civic commitment would currently be willing to pay lawyers and accountants up to 46 (after-tax) cents to hide a dollar of income. In Andrea Horwath’s Ontario, someone with a super-high income and no civic commitment might pay up to 49 cents per dollar hidden (although the lawyers and accountants would presumably pay some tax on their additional income.)

    New Democrats and PEF bloggers are not claiming that taxing the super-rich would fund the New Jerusalem. That’s why we need to raise corporate taxes too!


    However, the proposed top tax increase would fund some modest relief for the middle class, income support for the poor, and investment in childcare. That brings me to Rob’s excellent point: if we try to count possible revenue-reducing behavioural responses to tax increases, we should also count the revenue-enhancing effects of spending increases.

    Yesterday’s proposal, like recent NDP platforms, combined a specific tax increase with equivalent spending increases and HST exemptions. If the behavioural response to the former offsets the stimulative effect of the latter, then static costing of both tax changes and new expenditures will show the correct fiscal outcome.

    I have no reason to believe that the revenue gain from increased public and consumer spending would precisely equal any revenue loss from increased income-tax avoidance. But the fact that there would be at least some offsetting effect makes me worry even less about behavioural responses that would probably be rather small anyway.

  • The highest tax bracket right now starts at about $80,000 right? Why does Horwath want to jump over everything in-between $80K and $500K? That seems very ideological, like targeting a small number of individuals. Our tax brackets are not meant to just capture 0.03% or whatever $500K represents. Why not target those above $100K or $150K? Wouldn’t that make a lot more sense and bring in a lot more money? One could introduce two new brackets if one wished. But treating everyone making $80,000 to $499,999 as if none of them could afford to pay any more tax, but those with $500,000 suddenly can, just makes Horwath look unprofessional and superficial in her economics. Which perhaps she is.

  • The last major review of taxation in Canada that had some merit and at least a little bit of lopsided balance was under Paul Martin. There was a lot of research, but mainly with an eye to reducing taxes, from corporate to personal. One thing this thread does bring light to, is we need to put together a comprehensive review of progressive taxation literature as it is undoubtedly going to be needed to substantiate current and future policy openings like this one the ONDP presents.

    The tax modelling software is fairly sophisticated for projecting policy options impacts, but these models do not take into consideration such measures as avoidance and other behaviors . However the tax simulation package used by the ONDP developed and maintained by Statcan, does have a glass box functionality that allows one to attach additional modeling algorithms. Mind you the ad dons must be programmed in C++, so the software is not the easiest to customize. (unless you know C++ which I have been using for years- okay I am a nerdy techs math economist guy). The key is these models can be customized to do the job, but at the end of the day, they are merely models and as sophisticated one get the outputs are merely estimates. Hopefully the expected values have lower error rates, but nothing is written in stone.

  • A good debate about data and measuring income inequality here with James Gailbraith. A good add on to this thread as it gives you some real global perspective of measuring inequality.

  • “That seems very ideological”

    Yes it is. The NDP is dealing with someone who is ideologically opposed to any tax increases. With a barely-a-minority government there are limits to what they can demand.

  • Gentlemen: excellent debate! I learned a lot.

    Dr. Milligan: thank you for pointing me to the Saez articles. I will have a look at them. Am I to understand that there is virtually no Canadian literature that looks at behavioural responses to increasing personal income tax rates for high-income households?

  • Here’s some Canada-specific research on tax elasticity:


    They would probably have benefited from a larger sample size.

Leave a Reply

Your email address will not be published. Required fields are marked *