Wednesday, November 28, 2012

Experimental evidence on the effect of taxes

Much of our "fiscal cliff" debate revolves around the incentive effects of raising marginal taxes on high incomes. High tax advocates used to say that taxes won't hurt growth that much, and advocated them for other reasons.  Now they are advocating that even a 91% federal income tax rate, on top of state, sales, etc, as we had in the 1950s, (not counting all the loopholes!) will actually be good for the economy and also raise lots of revenue.

This seems to me like magical thinking, and a great testament to how people can persuade themselves of anything if it suits the partisan passion of the moment.  But wouldn't it be nice if someone would run an experiment for us?

Fortunately, Europe has been running a very useful set of experiments on what happens if you address yawning deficits with high income, wealth and property taxes. Which brings me to a report from the Telegraph
Almost two-thirds of the country’s million-pound earners disappeared from Britain after the introduction of the 50p (percent) top rate of tax, figures have disclosed.
In the 2009-10 tax year, more than 16,000 people declared an annual income of more than £1 million to HM Revenue and Customs.

This number fell to just 6,000 after Gordon Brown introduced the new 50p top rate of income tax shortly before the last general election....

It is believed that rich Britons moved abroad or took steps to avoid paying the new levy by reducing their taxable incomes.

George Osborne, the Chancellor, announced in the Budget earlier this year that the 50p top rate will be reduced to 45p from next April.

Since the announcement, the number of people declaring annual incomes of more than £1 million has risen to 10,000.

However, the number of million-pound earners is still far below the level recorded even at the height of the recession and financial crisis....

Far from raising funds, it actually cost the UK £7 billion in lost tax revenue
That's just one year. Usually, we think that it takes a while for high taxes to have effects. It takes a while for people to move, shelter income, close down businesses, not start businesses, not go to school, etc. Hitting the Laffer limit in one year is pretty impressive.

Update: Thanks to JM Pinder below I went back to the HMRC report which is indeed more detailed. Some highlights:
The 50 per cent additional rate of income tax was introduced on 6 April 2010. It was the first increase in the highest rate of tax in the UK for over 30 years, and was expected to yield around £2.5 billion...

This report provides the first comprehensive ex-post assessment of the additional rate yield using a range of evidence including the 2010-11 Self Assessment returns. The analysis shows that there was a considerable behavioural response to the rate change, including a substantial amount of forestalling: around £16 billion to £18 billion of income is estimated to have been brought forward to 2009-10 to avoid the introduction of the additional rate of tax. ...[This is a suggestion that it's a one time loss. We'll see]
The modelling suggests the underlying behavioural response was greater than estimated previously in Budget 2009 and in March Budget 2010, decreasing the pre-behavioural yield by at least 83 per cent. This result is also consistent with that contained in the Mirrlees review, and suggests the additional rate is a highly distortionary form of taxation.
Don't miss the bigger point here. The US discussion harks back to the great old 1950s, ignoring the much more relevant evidence right before us from Europe: Want to try cutting deficits (very slightly) with high marginal taxes, especially on investment, along with minor "cuts" (declines in growth rates) of spending, but no substantial change in the welfare state? Hey, they just tried it! Their economies sink, and they don't get much revenue.


  1. Apparently, the same thing is happening in France due to a proposed tax of 75 percent on all earnings over 1 million euros, although it hasn't been backed up with the numbers as in the British case (Telegraph article:

  2. Another point (which will be obvious to advocates of Modern Monetary Theory, but less obvious to others) is thus.

    The idea that in order to “pay for” $X of government spending, one needs to collect $X of tax is complete nonsense. The reality is that if government is to raise spending, the inflationary effect of that needs to be countered with an amount of tax which cancels out that inflationary effect. And given that tax increases do not greatly influence the day to day spending of the rich, $X of extra government spending could make it necessary to collect an extra $2X, $3X or $4X from the rich.

    And that of course compounds the Laffer curve effect.

  3. The 50% band was pre-announced, which allowed high income individuals to bring income forward, which would distort these numbers (increasing the number of millionaires prior to the introduction, and decreasing it after).

    The HMRC study on the introduction of the 50% rate does indicate that the Laffer curve was quite a lot flatter than anticipated. The underlying yield was expected to be £2.5bn a year. Ex post, HMRC estimated that the underlying yield (trying to allow for bringing income forward) was probably under £1bn, and may have been negative.

    See executive summary here:

  4. People respond to incentives? Who knew?

  5. This experiment has little relevance for the U.S.m since Britain does not tax its citizens on their worldwide income. It is extremely easy for Brits to declare their income as coming from elsewhere. Say you are a banker. Just have your Swiss or Hong Kong subsidiary pay the bonus instead of UK

    But in the case of U.S., this easy tax avoidance scheme doesn't work. Hence we are back to the interesting fact that U.S. grew pretty gingerly when taxes where a lot higher than today.

    1. Yeah....until the 1970's. Advocates always forget about the '70's. And it did grow ever more when Kennedy cut taxes.

      Nobody ever paid that tax anywat. Would you work an additional hour for only $0.09 of every dollar your toil to earn? If you think people will do that, then we should have a flat tax of 91% and raise lots of revenue. The best way to raise revenue is to tax people who have the least flexibility in arranging their affairs to avoid the tax.

      High income employees were paid in non-taxable perks, effectively allowing them to consume things like cars, apartments and vacations with pre-tax dollars to reduce their taxable money income. People who had the flexibility (which is practically everyone in the target group) either quit working before their income reached the tax threshold or arranged their affairs so that their taxable income was reduced. And let's not forget that practically everything (including interest on car loans and credit cards) was deductible.

      If you're looking to prevent people from creating wealth, a 91% top marginal rate with no loopholes will get you there quickly. if you want to raise tax revenue, forget about it.

    2. Not exactly. Citizenship is not the criterion used by the UK to determine tax status. Most UK citizens who reside in the UK are considered "domiciled" and "ordinarily resident" in the UK and are therefore taxed on their worldwide income, just like US citizens residing in the US are taxed.

      Persons who are domiciled and resident but not ordinarily resident in the UK can be taxed on the remittance basis. UK citizens who don't reside in the UK and are not domiciled there are generally not subject to UK tax non-UK income.

      The UK concepts of domicile and residency are complicated, convoluted and confusing; but, in short, it is easier for wealthy persons to escape UK taxation on worldwide income by limiting their contacts with the UK than it would be for wealthy persons under similar factual circumstances to escape US tax. Fewer persons will want to be domiciled and ordinarily resident in the UK due to the higher effective tax rates there. That's not good for the UK economy, particularly London, which has historically attracted a lot of foreign wealth, talent and investment.

      While it is harder for US citizens and lawful permanet residents to escape the US tax net (short of renouncing citizenship or one's green card), one consequence of higher taxes is that fewer wealthy individuals who are not US citizens (or lawful permanent residents) will want to come to the US to live and invest there. This has significant consequences in a global economy where wealthy persons and their money are highly mobile. Also, persons who cannot legally escape US taxation based on residency or deemed residency due to citizenship can certainly mitigate the effect of higher tax rates through planning or the simple techniques of not working, not investing or indefinitely deferring realization of income.

      The "experiment" is highly relevant to the US (and other countries as well).

    3. Yeah... it also does not apply to US, I guess, because all amounts are in Pounds, and we use Dollars here.

      The point is that people respond to incentives, and that the UK Gov. forgot that, and so the revenues are much lower than projected (and may have been negative). Although the response in the US would be different in some details, I think the end results would not be pretty.

  6. Prof Cochrane - I'd second JM PInder and Anonymous here. A word of warning from a UK resident: don't trust opinion type articles in the Telegraph. They are ideological rather than scientific.

  7. Anon-

    So if it's so easy in the UK to move your income around, why not always do it?

    1. Well, I am sure it involves all sorts of inconvenience, paperwork, legal fees, etc. However, there is no doubt it is quite common. Just go look up how many Brits pay their taxes in Hong Kong or in Switzerland or in Guernsey or Island of Man or Caymans or Singapore. The list goes on.

      There is a reason, after all, why U.S. taxes worldwide income of its citizens.

    2. You're missing something very important: the very high earners that the tax would target can also just quit working as much. At a 91% tax rate, Leisure becomes extremely cheep and very few of these people spend all that they earn, so they can lower their income without lowering their consumption. Operating in the black market also becomes relatively cheap and rent seeking becomes vital. People don't have to leave the country to reduce their taxable income. You're just not going to soak the rich.

    3. yes, in theory. But in practice that did not happen in the U.S. in the 50s and 60s, when we had tax rates between 70-90%

    4. The Federal government has collected a very stable 18 percent of GDP or so in income taxes, despite published tax rates that varied widely, indeed up to 91 percent. The 91 percent came with a huge number of loopholes, so that wealthy taxpayers never paid close to 91% of income in taxes.

    5. Agree. But we only collect 15% of GDP now. So a system with massively higher marginal rates and huge loopholes generated both more tax revenue as a % of GDP than today and better economic growth.

      Look, I agree all things being equal it's better to have lower rates and a more transparent system. But the data seems to indicate that this is by far not the main driver of economic performance

    6. Have you had a chance to look at Thomas L Hungerford's paper? I think it would be another very nice example of how bad things happen when you (over)difference data.

    7. Agree. But we only collect 15% of GDP now.

      Not that I've looked at the numbers myself, but I think if you look you might find that this in large part stems from the 2008 collapse and the loss carryforwards it created.

      Ah, look, I wish they would raise the top marginal tax rate to 91%. I haven't had a vacation in almost 7 years - I am totally undisciplined when it comes to taking time off. The 91% tax rate will affect me and where it will affect me is on a beach in St. Barth's.

      By the time the romance of high tax rates wears off, I'll be rested and ready to work again. What can I tell you? I'm self-interested.

    8. " But the data seems to indicate that this is by far not the main driver of economic performance."

      There seems to be an argument floating around that growth under the 39.6 marginal rate was great, so why not go back to that? Well, the other side of the equation is government spending. As a % of GDP government spending was much smaller in the period of 39.6 marginal rates than it is today, I think around 18% then compared to 25% now. Thus, you can't compare growth rates based only on tax rates and not the level of government spending. That is, I don't see how you can conclude that an increase in marginal rates will have little effect on growth based on historical data, since government spending changed at the same time.

      Moreover, it may be more important what government spending was as a % of GDP before the period of high growth. Like it or not, government spending crowds out private investment. A lower fraction of output going to government spending implies higher investment spending. Generally, this leads to increased economic growth. Are you certain this effect doesn't matter?

      It very well may be that tax elasticities are low enough in the 30-40% marginal tax rate range that increasing this rate will have only a small effect on growth. However, it seems a bit silly to argue that increasing tax rates is going to actually increase growth. Simply pointing to spurious correlations (between rates and growth) is not a good argument.

    9. Right, our spending is increasing in large part due to the aging population. So the question is, how do you pay for this? it's better to pay for this with less deficit, so that investors buy businesses and real estate instead of treasuries. But in order to have less deficit for a given level of spending, we should raise taxes.

    10. No, government spending (it's not "our spending" in any meaningful sense) is increasing because spending other people's money is what governments do.

      An aging population problem can be most easily fixed by allowing a hell of a lot more immigration.

      Why would people start businesses when you keep reducing the reward (via higher taxes) for a given level of risk? Taxes on production are a disincentive to produce and invest. The higher the tax rate the more that's true (see our hosts' post from yesterday). The higher the tax rate, the less incentive. And if there's a disincentive to produce and invest, where will you get the income to tax to pay for these entitlements?

      You won't. No matter how high the tax rate. Let's hope the baby boomers saved for retirement, eh?

  8. What happened was complicated structures to avoid taxes and massive tax deductions to avoid high tax rates. And people did just stop working. I recall Ronald Reagan saying that if the next movie would put him over the threshold, he would quit for the year and ride horses at his ranch until the following year. Instead of some reasonable percentage of whatever he would have made acting in that movie, the government got 91% of $0 and there was less economic activity than there otherwise would have been.

    Would you work to earn the next $1.00 if you were only able to consume $0.09? I wouldn't.

  9. John Cochrane,

    I think it is more accurate and less confusing to refer to "wealthy taxpayers" as "high earners". A person earning a high income which is subject to the confiscatory tax rates romanticized by certain former economists may not yet be wealthy in a material sense. In fact, high tax rates would prevent them from accumulating wealth.

  10. "Would you work to earn the next $1.00 if you were only able to consume $0.09? I wouldn't."

    You only would not if you had better alternatives. If the time that you spend making those 9 cents is worth to you more than 9 cents (because you can find a better job, or because you can make more by hiding your taxes, or because you value your non-work time more and can afford not to work) you won't work, but if it isn't - you will.

    1. Sure, ed, but at a 91% tax rate only a miniscule number of alternatives AREN'T better. At the only slightly higher 100% tax rate EVERY alternative is better.

      Since virtually every alternative is better at a tax rate of 91%, how many people do you expect to create taxable income at that rate? If your goal is to raise tax revenue and there is strong disincentive to produce taxable income, would you expect tax revenue to rise or fall by nearly tripling the tax rate?

    2. To nitpick a little - it's not true that at 100% tax every alternative is better - some of us like our jobs, but this isn't important.

      What is important is the assumptions you're making. For instance for sitting on a couch to be an alternative to working, one of the prerequisites is a couch.

      While I agree with the assumption that at infinite tax rate (not 100%, so you'd pay to work) there exist better alternatives, and that at negative infinite tax rate work rules, the assumptions you make about the shape of that curve, in particular that $/time sucks at 91% are unjustified.

    3. Ed, at a 100% tax rate it's no longer a job. It's volunteering.

      I'm one of those human being described in the definition of economics you learn in the first minute of the first day of Econ 1 - one with unlimited wants and needs and limited resources. So, at a 100% tax rate, I'm better off bartering and doing anything else that doesn't produce a taxable income.

      However, if you are willing and eager to work for nothing more than my approving smile, you're the employee of my dreams. Tell me, dear, what kind of work do you do? I can tell we're going to get along very well.

    4. I don't know if you realize that your reply has only touched on the point that was explicitly said to be not important.

    5. Sorry ed, I just responded to the only part of your comment that wasn't a complete, garbled mess.

      Okay, I'll pay you 9% of what you're making now. Come work for me.

  11. Nice try, but the facts are that the wealthy have lots of untapped ability to pay

    US middle class' net worth lowest in decades

    A new report shows that the gap between the rich and the poor in the US is significantly wider than it used to be, with the annual median wage falling to $26,364 in 2010 while the high-earners increased their wealth by 71 percent.
    The New York University report found that in 2010, the median net worth in the US reached its lowest point since 1969 at $57,000. Conducted by Professor Edward Wolff, who has studied Americans’ net worth since 1983, the report provided some surprising revelations about the extent of the income inequality.
    While the middle class lost 18 percent of their net worth as a result of the the housing crisis, the top one percent of the richest Americans increased their earnings by 71 percent, thereby widening the gap between the rich and the poor. Each of the one percent is on average worth 288 times the value of a middle class American, the Economic Policy Institute recently reported.
    The annual median wage decreased in 2010 to $26,364, hitting the lowest level since 1999. A 2011 Gallup poll also found that 20 percent of Americans rated their financial situation as “poor”, which is more than the 16 percent of Americans who viewed their finances this way during the housing crisis.

    What we should be talking about is eliminating payroll taxes of all kinds, go to a VAT/wealth tax

    I would especially like to see real taxes on property & casualty insurance premiums

  12. 2 propositions:

    1. Changing tax rates will not affect tax collections.

    From one my favorite websites:

    How much does the top income tax rate affect how much money the government collects each year?

    * * *

    The chart below plots the ratio of total government Revenue Per Household (RPH) to the Median Household Income (MHI) for the U.S. for each year from 1967 through 2010. The chart also plots the maximum income tax rate that the topmost income earners in the United States have had to pay for each year from 1967 through 2010.

    [chart omitted]

    * * *

    What we do see indicates that the maximum tax rate has little to no bearing on how much money the federal government collects per household in any given year. Since 1967, the government's RPH to MHI ratio has risen steadily on average, indicating that the U.S. government is collecting more and more money per household over time, with the changing level of the topmost income tax rate having little to no effect on the rate of that change.

    With that being the case, there is no legitimate reason to set higher income tax rates today, as they are now demonstrated to have little to no effect on how much money the government collects in any given year.

    Check it out at the link.

    The correlation between MHI and RPH is very tight. R^2 = .996584

    2. No matter what Congress does with the tax code, go over the fiscal cliff or do what Prez wants, there will be a recession. At least three factors determine that:

    a. The implementation of Obamacare will cause employers to reduce hiring, lay employees off, make full time employees part timers. It will also cause insurance premiums to skyrocket and millions of people to drop existing coverage.

    b. The ongoing cluster**** in Europe will get worse not better.

    c. China will continue to struggle to maintain any semblance of growth, and fail.

    My new motto:

    Lasciate ogne speranza, voi ch'intrate.

    1. These graphs are great! Everyone go click. It looks like blogger won't let you put graphs in a comment.

      I'll match your motto with the first line, strangely appropriate for our times and country as well

      Nel mezzo del cammin di nostra vita mi ritrovai per una selva oscura ché la diritta via era smarrita..

    2. Ignore GDP and the Fiscal Cliff, U.S. Is Already in Recession: ECRI’s Achuthan

    3. Very interesting post and comment. I share much of what is said here on the negative effects of taxation but I think the second graph on "The correlation between MHI and RPH" is misleading. There is obviously a unit root problem here and a high R2 would be expected even if no relation were to exist between these variables.

  13. The thing is that, on the left, the effects of marginal tax rates on growth, respect for rule and a lot of other things are routinely denied as matter of pure irrational and illogical faith even by very knowledgeable people no matter the evidence or the common sense on the issue, much like communism is still perceived, especially by those leftish who didn't know it : very good in theory, but very badly applied!

  14. I meant "very good in theory but very bad when applied".

  15. Professor Cochrane,

    I enjoyed this post, but believe you're forgetting the key evidence reached by Slemrod, Goolsbee, and others that the short run elasticity of taxable income might be high, but after about the first year it's a lot lower. One would expect there to be a lot of income shifting in the year of a tax change, but it doesn't stay that way permanently. Would be interested to hear your thoughts on this.

  16. It is not an "experiment" but rather a "case study" so I think the post is somewhat misleading. The point may as well be valid, but the evidence is not able to fully support it.

  17. Your figures are in error. The taxes for both Clinton and Bush were calculated using the maximum rate for that selected income. For instance the Clinton 1999 tax rate on 30K was 28%, which is what they used to get the 8400 figure. However taxes are not calculated that way. The first 25K of income would have been 2013 tax brackets at the lower 15% bracket first, thus yielding a much lower figure than what you show.I am not arguing that Bush doesn't have lower taxes. He certainly does. Of course he obtained his lower tax brackets by using deficit spending and increasing the national debt. Add back in the interest payments we'll be making and I bet Bush actually cost taxpayers far more than Clinton ever did.


Comments are welcome. Keep it short, polite, and on topic.

Thanks to a few abusers I am now moderating comments. I welcome thoughtful disagreement. I will block comments with insulting or abusive language. I'm also blocking totally inane comments. Try to make some sense. I am much more likely to allow critical comments if you have the honesty and courage to use your real name.