Well fancy that!

Tax Justice Network, looking at the effects of higher income tax rates:

Thompson remarks, based on extensive analysis of the evidence, and other examples of apparently selective use of data, that: “It seems that the prevailing orthodoxy that higher taxes reduce growth is so ingrained that contrary evidence cannot be recognised.”

Prevailing orthodoxy in physics is that gravity exists rather than that the Earth sucks. But for political reasons we’d prefer to insist that the planet is giving our feet a permanent blow job.

So there.

47 thoughts on “Well fancy that!”

  1. Read it. It’s utter rot as one would expect.

    Basically poo-poos one side’s evidence and attaches near certainty to the other. No guesses which side is which.

  2. I bet they are right that the evidence to support the argument that higher top rates of income tax reduce growth is not there.

  3. I thought Murphy had left the TJN due to unforeseen circumstances so theoretically the post should be under ‘Flatulent tosspottery’ perhaps, rather than ‘Ragging on Ritchie’.

    However, the TJN update attached by Arnald is, as Tyler says, so one-sided as to be more like propaganda than a valid contribution – I could no doubt find an IEA, CPS or ASI paper which comes up with similar evidence for the powerful disincentives caused by punitive rates of taxation. I can even guess at the authorship of some of it beyond John Thomson. Fred ‘The Shred’ Goodwin, singled out in Paragraph 31 seems to have been seized upon by the likes of TJN as a pantomime villain – how many Public Sector jobs are there both whose positions and whose job holders are at least as useless as Goodwin? (albeit less well remunerated) – oddly no mention of the need to limit public expenditure in an era of global competition, rapid change and increasing mobility of capital, but then what do you expect from an organization whose very name invites ridicule?

  4. It’s a very limited lit review on the subject. No new work whatsoever, and very selectively choosing who said what to try and make a point that high taxes have no downside – the main argument being that because we can’t calculate exactly what it might be given all the different variables, we shouldn’t bother.

    High tax lefties arguing for higher taxes, saying we can ignore any potential downside. Well I never.

  5. I think there’s a big difference in tone between the obvious Press Release and the report itself. The first blows the trumpets, the second is actually more measured and para 39 even congratulates HMRC on its openness.
    And I think para 40 sounds reasonable as well, like someone wanting an open discussion. So maybe its worth responding with contrary studies and evidence to check that out?
    ” 40. If the policy debates are to be well informed, it’s essential that the caveats, assumptions and uncertainties included in academic papers and technical reports are appreciated more widely. This does not only apply to the direct estimates of tax yields, but also to the belief that higher top tax rates lead to lower growth, a belief which has much less secure foundations than is usually appreciated. Given the huge uncertainties which can be found in the 2012 HMRC estimates, there is a strong case for further analysis.”

  6. From an earlier post, the reason the left hate the Laffer curve is precisely because it does show that tax reduces growth. If you increase the tax rate and the revenue falls, then the economy has shrunk. And the Laffer max is the point of zero growth.

  7. Roue le Jour – it’s more than “If you increase the tax rate and the revenue falls, then the economy has shrunk”. If you increase the tax rate and the increase in revenue is less than linearly proportional, then the economy has shrunk. In most graphs of Laffer the curve is steepest at the origin and starts to fall off immediately, suggesting that tax has an inhibiting effect right from the start.

    Although of course that might be just because Laffer graphs are normally illustrative of a theoretical result, rather than empirical.

  8. Good point, Pellinor. I should be clear about the gist of my argument, which is that tax reduces growth and the point at which growth passes through zero is also the Laffer max. And reason why I say that is because the Laffer max is sometimes seen as a sweet spot, where I would argue it’s a red line you want to keep well clear of.

    To come back to your point, if the increase in revenue is less than proportional to the increase in tax rate the economy has shrunk. In other words, the economy has shrunk before the Laffer max has been reached, invalidating my argument. The answer to this simply timescale. The economy does not respond immediately to tax changes and it is quite possible that a tax rate increase can increase revenue in the short term but that revenue will collapse in the long term, showing that yes, if the revenue is less than proportional you have indeed passed the max and killed the economy.

  9. @ Arnald
    Thanks – now that I’ve read it my first reaction is that he isdoing exactly what he accuses others of doing – producing a conclusion at direct odds with the evidence, despite cherrypicking the bits of evidence he chooses to look at. For instance he says that statistical evidence fails to support the idea that income was brought forward to avoid the increased rate (I can produce data that some *was*) then argues that HMRC understate the underlying tax gain because of forestalling and that although HMRC assumed lower income elasticity of 0.45 rather than the 0.48 implied by past research they overstated the income effect.
    Arguing against the impact of income tax on FDI he states that tax has a statistically significant negative correlation with FDI – yeah, that’s what we said, not what he wants us to believe. Income tax matters less for FDI than VAT/Sales tax – well, doh! Income tax comes out of profit, VAT out of gross revenue.
    Everyone will go to sleep if I provide a thorough analysis, but basically it’s bullshit.

  10. Howard Reed tries to argue that observed incidence of the Laffer Curve is simply tax avoidance. Tax planning undoubtedly does turn the curve downwards quicker. However, his fundamental point, that there is no economic impact per se, is not correct.

  11. @ Ironman
    I could say: “Well, doh!” again. All this fuss about subsidising childcare is because mothers of young children won’t take a paying job if the net pay is less than the cost of childcare: so the tax rate matters. My standard moan is that thanks to Gordon Brown most unemployed people in my town (or many others – I just know our season ticket rate) cannot afford to take a job in London at a median salary and most have no chance of getting one above median salary levels (OK, one friend went from well over double median salary to unemployed to well over double median salary but he’s exceptional). If Howard Reed tries to argue that there is no effect he is either abysmally stupid or dishonest.

  12. I am very impressed by the Tax Justice Network and their courageous demolition of neoliberal sophistry.

  13. LE

    “I bet they are right that the evidence to support the argument that higher top rates of income tax reduce growth is not there.”

    Do you think a (say) 95% top rate of income tax would have no effect on growth? I suppose if the extra tax was ‘invested’ in schools and hospitals, it might generate some growth for a while as the education and health services employed more people and raised wages, but this would not be sustainable for long.

  14. Adding to Roue Le Jour’s comment, yes indeed the Laffer curve only shows the maximum possible tax extracted. I’ve yet to see a good reason why we should extract as much as possible, when it’s only going to pissed up the wall by the government. The state is a rent-seeking body: it’s up to the voters to rein it in.

  15. So Much for Subtlety

    Bill Sticker – ““Tax Justice” Isn’t that an oxymoron?”

    I don’t know. I think that Tax and Justice are distributed along something like an upside down Laffer curve. If the tax rate is zero, Justice is maximized. But then if the tax rate is 100% it is maximized as well.

    The question is what shape the curve is and where the nadir lies.

  16. Theophrastus

    well I don’t know. I meant that if you try to find rigorous empirical evidence you will find null and conflicting results, because we don’t have enough data.

    in theory the effect is ambiguous. if high-earning individuals reduce their work effort and earn less, perhaps money not spent paying them will be spent paying others to do something more productive. I know nobody here will find that plausible, but at least in theory very high earnings might not be doing much that contributes to economic growth, as in raising aggregate productivity, as opposed to capturing rents or maybe just winning a zero-sum competition.

  17. ““Tax Justice” Isn’t that an oxymoron?”

    This is a social type of justice, ie confiscation and punishment of political enemies.

  18. @ Luis Enrique
    “I meant that if you try to find rigorous empirical evidence you will find null and conflicting results, because we don’t have enough data.”
    That is a blatant non sequitur. The larger the body of data, the easier it is to get conflicting results. Shortage of data leads to non-conflicting results being declared “not statistically significant”.
    As one of those then youngsters in the City who enjoyed schadenfraude at the reports of scores of tax advisers thrown out of work in 1980, I can tell you that we do have enough data. Geoffrey Howe actually collected more higher rate with a 60% top rate; GDP growth accelerated because more effort wasa put into earning money and less into avoiding tax; just as with Woy Jenkins’ experiment with suspending the death penalty the result is clear at any measurable level of statistical significance.
    Thompson tries to say that China and Vietnam with high tax rates have faster growth but that is completely irrelevant because what needs to be measured is whether country A would have higher growth at a lower tax rate not whether country B has higher growth than country A.

  19. john77

    well … you’re wrong. See here under sample size

    if you are trying to identify the effect of one variable amongst many on an outcome like GDP which is noisy and affected by countless other things, you need a large dataset. This is a well known problem with cross-country empirical research: the number of potential explanatory variables is vast and we have too few countries and too few years of data. This is why people do things like Bayesian model averaging and why the entirety of the cross country growth research literature has roughly zero robust results. Do you want me to start citing papers?

    you might attribute increased GDP in the eighties to tax cuts, but a lot else was also going on at the time, and isolating the impact of a tax cut on high earners from everything else going on in the data is nigh on impossible.

  20. @ Luis Enrique
    You do not know what you are talking about. An analysis of the frequencies of the different categories of capital murders during and before the “suspension” gives a probnabilitythat capital pounishment is not a deterrent so small that it doesn’t show up in five-figure statistical tables. The increase in revenue from higher rate tax in each year from 1979/80 onwards compared with each year 1974/9 is statistically significant.
    Cross-country empirical research has more relevance to the mud-resistance of boots/running shoes than to the sensitivity of any single country’s economy to tax rises/cuts.

  21. J77

    we’re not talking about the impact of tax rates on tax revenues, we’re talking about the impact on economic growth.

    1000s of cross-country empirical papers have been written on impact of taxes on growth, I am sure your views will be of great interest to the authors.

    this one tends to get cited a lot by people on your side of the debate


    I was speculating that evidence that strong cannot be found in case of high rates of income tax, but I could be wrong.

  22. @ Luis Enrique
    GDP/head grew faster following Howe’s budget despite the really major error in letting sterling float upwards and this was partly *because* – as I said – people spent more time earning money and less on avoiding tax.

  23. You are offended by the idea your assertions don’t qualify as empirical evidence?

    I look at the data, I observe an increase in GDP coincides with a tax cut. That’s certainly supportive of your theory, but because many other things were going on at the same time that could also explain the rise in GDP we don’t really have much evidence based on this one observation. Which is why tax researchers look at data with many tax changes in many countries. Maybe you can include all other plausible explanations in your model, which would help. You haven’t.

  24. John77, to remind you:

    “It seems that the prevailing orthodoxy that higher taxes reduce growth is so ingrained that contrary evidence cannot be recognised.”

    Strange how Worstall can’t be bothered to back up his orthodox-so-right argument, when he clearly doesn’t believe that statement in itself on many other issues.

    Any way Luis Enrique wins this thread. Empirically.

  25. @ Luis Enrique and Arnald
    I don’t have an hour to waste trying locate the raw data from the ONS site so I’ll quote The Guardian which obviously ain’t biased in my favour http://www.theguardian.com/news/datablog/2009/nov/25/gdp-uk-1948-growth-economy
    Real GDP growth Q2 1979 to Q2 1997 61.486% real GDP growth 1973 Q4 to 1979 Q2 12.18% so growth under Labour 1.8% pa, growth post-1979 2.7%.
    “assertion by J77 != rigorous empirical evidence” No it only QUOTES rigorous empiricalevidence
    No, stop being so bloody rude.

  26. somebody who wrote “You do not know what you are talking about” should not whine about rudeness, and really I do not think that either “you are wrong” or ““assertion by J77 != rigorous empirical evidence” is particularly rude.

    and you are completely failing to address the point I made above starting “I look at the data, I observe an increase in GDP coincides with a tax cut ….”

    rigorous empirical evidence != one period of correlation

  27. @ Luis Enrique
    Do you want to be called a liar when when you are merely ignorant? The “Power” of a statistical test is the complement of the probability of a Type II error of wrongly accepting the hypothesis when it is false (see p. 164 of Volume II of “The Advanced Theory of Statistics” by Kendall & Stuart).
    This is *not* the same as the Type I error of wrongly rejecting the hypothesis when it is true. Tests of statistical significance are designed to limit the probability of wrongly rejecting the null hypothesis when it is true: the common standard is to only reject the null hypothesis when it is true 5% of the time.
    When I stated the totally obvious that it is easier to find data that seem to give conflicting results when you increase the number of datasets, you unreasonably stateds that i was wrong and referred me to a wikipedia article that has no relation that point.
    I find being called a liar – either straight out or by implication, more offensive than being called ignorant or stupid. If you don’t, then I suppose I should pity you.
    “and you are completely failing to address the point I made above starting “I look at the data, I observe an increase in GDP coincides with a tax cut ….” That is because I was answering what you had said previous to my post not what you said later. Sadly I am not Doctor Who nor can I read your mind like Gipsy Rose Lee (actually, I don’t want to be Gipsy Rose Lee!!).

  28. @ Luis Enrique
    What I said in the first place was “GDP growth accelerated because more effort wasa put into earning money and less into avoiding tax;”
    That is a fact that could be *and was observed* by (i) “scores of tax advisers thrown out of work in 1980,” and (ii) the increased tax take by the 60% tax rate.
    [I subsequently pointed out one of Mrs Thatcher’s BIG economic errors (her biggest IMHO) in letting the £ exchange rate drift up like a helium balloon.]
    Your response is to tell me I am wrong using techno-babble like “Bayesian model averaging” and describing the (irrelevant) power of a statistical test as a definition of “sample size” which is at right-angles to the number of samples.
    The effect of taxation on UK GDP growth can either be looked at empirically or considered theoretically as a problem in multi-dimensional vector space. Some people think that they can learn something by introducing a further dimension by considering seriously inadequate data from other countries and assuming that the curved intersection between different datasets is a plane, which is just plain ridiculous – is China’s growth rate despite its nominal tax rate is high or because its effective tax rate is low?
    What is your null hypothesis? The sensible one that people will be more inclined to work if they get paid more (after tax) or the silly one that they will work more if they get paid less (after tax).

  29. @ Luis Enrique
    Does causation work both ways? “Well up to a point, Lord Copper”
    IF a change in taxation is announced in advance then the economy may respond in advance. If the economy is expected to grow, the government may, and probably will, reduce taxes because it can cover its costs at the lower tax rate. BUT UK budgets are set for one year at a time so until Gordon Brown wrecked the UK economy the only anticipation of future economic growth that was allowed to influence the current year’s income tax rates was that in the next twelve months. So it is nonsense to say that the 50% higher economic growth from 1979-97 than under Wilson & Callaghan *caused* Geoffrey Howe’s cut in top rate tax.

  30. nobody implied you were a liar, I think you need to calm down and re-read things without a defensive head on

    also, if you meant estimating the same model separately on many different datasets, then obviously yes you will get conflicting results. I was talking about pooling the data, as in cross-country empirical research. Yes there are problems with that, but if you want to do things like investigate the impact of tax rates on GDP then that’s the way to go.

    you can showcase your knowledge of statistics all you like, but if you cannot grasp the point that one instance of taxes being cut followed by GDP rising does not convincing empirical evidence make, then we are going to get nowhere.

  31. by the way, statistical power, the ability of a test to detect an effect, if the effect actually exists (i.e. to correctly reject the – for sake of argument false – null that taxes have no impact on GDP) is increasing in sample size. I realise this is not the same as wrongly rejecting when true.

    I would not claim to be a statistician, but this seems pretty straightforward to me.

  32. @ Luis Enrique
    Well if you weren’t trying to imply that I was a liar then you need to think a damn sight harder before making offensive posts because that is certainly what anyone would infer.
    “you can showcase your knowledge of statistics all you like, but if you cannot grasp the point that one instance of taxes being cut followed by GDP rising does not convincing empirical evidence make, then we are going to get nowhere.” Firstly, *you* were trying to showcase a (sadly absent) knowledge of statistics in response to my making a point that required no special knowledge of statistics and was phrased in simply language; secondly – WTF? – it *is* empirical evidence, just not enough for a statistician to say it disproves the null hypothesis; thirdly, an Actuary (or, in my case, an ex-Actuary) does not *need* to show-case a knowledge of statistics: it is assumed that we have one. Given the history on this blog of my discussions with GlenDorran, to refer me to wikipedia on statistical theory (even apart from your getting it wrong) was a bit short of some male chauvinist pig telling Paula Radcliffe or Jessica Ennis-Hill that they need training advice from his club coach (and I do need training advice from my club coach, but they don’t).
    You were talking about pooling data with irrelevant data of dubious quality – Mr Thompson claimed that China had a high top rate of tax, but it did not apply to Bo Xilai, did it? The only comparisons that are worth looking at are those that contrast the growth rastes of Country A with a high tax rate and Country A with a lower tax rate. Comparing Senegal with Saudi Arabia, Ireland with India, China with Canada, Gambia wih Germany is stupid because in each case at least set of data isd unreliable and so many of the effects of the other (non-tax) parameters cannot not be estimated withion an ordfer of magnitude. The top cross-country runners wear a range of shoes, depending on the course – I just have one pair with studs which can cope with road/path and aren’t disastrous on mud. You seem to think that inter-nation comparisons are cross-country.
    “if you want to do things like investigate the impact of tax rates on GDP then that’s the way to go” Sez who? There is a case for aggregating the analysis of changes in GDP growth rates following changes in tax rates in each country that has changed tax rates but NOT comparing lemons with pears.

  33. No John. Nobody would infer I thought you were a liar because I claimed you are wrong or because I said your assertion do not equal evidence. Any more than you inferred I am a liar by telling me I don’t know what I am talking about.

    I don’t know anything about his history of debates, I am amused you are so horrified I should dare to disagree with you. Arguing from authority is rarely a good tactic.

    I started by speculating that there is no convincing empirical evidence that cuts to the higher rates of income tax boost growth and have argued that the increase in growth after the Howe tax cuts do not constitute such, because a lot else was going on at the time that could also explain that growth.

    Yes measurement error and pooling data from heterogeneous countries can introduce more problems than they solve. Perhaps you might be better just pooling OECD countries. Because more data does increase prospects of finding compelling evidence. Your denial of this is bizarre. Remeber Tim says we should expect tax cuts to spur growth as a rule, not just in the unique circumstances of eighties Britain. So expecting to find effects in data from many countries is reasonable. If not, advocates from for tax cuts should express more doubt about their effect

  34. @ Luis Enrique
    “Nobody would infer I thought you were a liar because I claimed you are wrong” BUT that is NOT what you said and I did infer that you were suggesting that I was a liar. Cogito ergo sum: I am NOT nobody.
    People who claim that I am wrong may often (not always because there is not always a Euclidean proof) be shown to be in error; very occasionally PaulB finds something where I *am* wrong (my analysis of marathon death rates was 20 years out of date and I have not yet got round to properly answering his criticism). What you said was “assertion by J77 != rigorous empirical evidence” which most people would infer as meaning that I am not to be trusted. Not the same thing. In this particular case I was quoting rigorous empirical evidence – and you have not yet admitted that I was doing so and apologised.
    All this discussion of your impertinence (normally one would say rudeness but someone who doesn’t know the difference between a Type I error and a Type II error pretending to lecture a former Actuary on statistical theory is impertinence) gets us off the subject.
    All the arguments that I have seen on this blog and the study that arnald cites for seeking a wider range of data are actually arguments that only within-country data should be used for comparisons and, at best, these data should be aggregated or averaged. An idiot might look at the Federal income tax rate in the USA and ignore the state income tax rate or look at the nominal income tax rate for a UK entrepreneur and ignore the CGT rate that he anticipates on selling the business.

  35. Again, no John. In context of debate about existence of rigorous empirical evidence, me pointing out your assertions do not constitute it does *not* amount to suggesting you cannot be trusted, this is not about your authority. Although at that point I was unaware I had the privilege of conversing with a titan of the actuarial world. You have *not* presented empirical evidence, you have cited *one instance* of a tax cut being followed by growth. That is data. But to turn it into evidence we’d need to consider possibility that growth was caused by other things that happened around the same time before attributing effect to the tax cut. PThis is the whole point you have such difficulty grasping.

    I’m not sure why you are accusing me of a type 3 error (and if you don’t get that, you haven’t spent enough time around statisticians)

    Out of curiosity. Suppose I were to reveal my authortorial willy is longer than yours on this topic, would you apologise for being impertinent to me? Thought not.

  36. @ Luis Enrique
    Your refusal to admit the truth is getting boring.
    I make no pretence to being an actuarial titan – I am not innthe thousand, BUT we do have moral standards (youngsters may call them “ethical” or “professional” – who cares?). I stated a fact without an internet reference and you implied that it was *not* a fact. When I produced an internet reference to prove that I was right you failed to retract – and you have still not done so.
    I quoted a fact to counter one of your assertions and you complain that it does not deal with a different assertion.
    I could be very rude and call you a liar: I have in fact instanced empirical evidence “you have cited *one instance* of a tax cut being followed by growth.” but I shall assume, once again, that you do not know what you are talking about.

    If you produce evidence that I do not know what I am talking about and you have a unicorn in your front garden and …

    On the other hand, you *have* shown eviudence that you are either abysmally ignorant or deliberately deceitful.

  37. To get back to the subject, DWP cited the number of people in the areas trialling Universal Credit getting jobs was more than 10% higher than those on JSA in comparator areas (those in work after 30 and/or 60 and/or 90 and/or 120 days was 47% vs 41%, those in work after 30 and 60 and 90 and 120 days was 11% vs 8%). There may be some impact from other factors but the reduction in marginal tax rate to 65% is the most obvious factor affecting this increase in employment and, by implication, GDP.

Leave a Reply

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