I posted this as long ago as 2010. I am reposting it because in essence nothing has changed in the UK on this issue since then. But yesterday representatives of the Institute for Fiscal Studies were all over Twitter saying that VAT is a progressive tax, and that is not true. It is regressive. I countered their argument in 2010 and it profoundly annoys me that they persist in perpetuating the same untruth now.
The UK government has proposed increasing the standard rate of Value Added Tax (VAT) from 17.5% to 20% from 4 January 2011. They are not alone in proposing increases in VAT or equivalent taxes to address deficits in government budgets. The States of Jersey currently has a proposal to do much the same thing — increasing their rate of Goods and Services Tax (which is a VAT in all but name) from 3% to 5%. These rises will be contagious.
In this case though there is a curious link between the two proposals. A paper issued by the House of Commons library[i] on this issue and commentary in Jersey on the same issue[ii] both rely on work by the Institute for Fiscal Studies to support their claim that any increase in VAT is only mildly regressive at most, or might actually be progressive — as the IFS have claimed[iii].
A new Tax Briefing from Tax Research UK examines that Institute for Fiscal Studies claim and finds it is a statement of political dogma, but not of fact.
As the Tax Research briefing argues, a regressive tax is almost universally agreed to be one where the proportion of an individual's income expended on that tax falls as they progress up the income scale[iv]. VAT is a regressive tax. This is shown, quite dramatically, in the graph below which is based on UK official data[v] :
By chance the VAT and total direct tax burdens on the bottom 20% of households ranked by their income is the same. Direct taxes then rise steadily as a proportion of income as incomes rise and both VAT and all indirect taxes combined do the exact opposite, falling as a proportion of income as income rises. So marked is the trend that the overall progressive effect of income tax is not enough to counter the fact that the poorest households suffer such a high rate of overall indirect tax that they end up with the highest average tax rates in the economy as a whole.
The message from this data is unambiguous: the poorest 20% of households in the UK have both the highest overall tax burden of any quintile and the highest VAT burden. That VAT burden at 12.1% of their income is more than double that paid by the top quintile, where the VAT burden is 5.9% of income. VAT is, therefore, regressive.
The IFS dispute this. They produce the following data in evidence:
They say of this:
It shows that the percentage of net income paid as VAT varies relatively little across most of the income distribution, with the biggest exception being that the bottom decile group does pay a higher fraction of its net income on VAT than do other income groups.
And they then use this claim to justify the fact that in their opinion VAT paid is not regressive with regard to income.
The slight problem for them is that this overlooks the very obvious fact that it is. Replotting their data and excluding the bottom decile as they would like the following graph can be drawn:
The linear regression shows a clear downward trend that makes very clear VAT is regressive.
Surprisingly the IFS ignore this obvious fact and go on to claim:
However, looking at a snapshot of the patterns of spending, VAT paid and income in the population at any given moment is misleading, because incomes are volatile and spending can be smoothed through borrowing and saving. Consider a student or a retiree: their current income is likely to be quite low but their lifetime earnings could be relatively high. The student may borrow to fund spending, whilst the retiree may be running down savings. Similarly, many people in the lowest income decile will be temporarily not in paid work and able to maintain relatively high spending in the short period they are out of the labour market. Because their spending is higher than their current income, these people will be paying a high fraction of their current income in VAT. Similarly, those with high current incomes tend to have high saving, and so appear to escape the tax, but they will face it when they come to spend the accumulated savings. Because of this ‘consumption smoothing', expenditure is probably a better measure of living standards (and households' perceptions of the level of spending they can sustain).
And they then claim that comparing VAT with spending shows that VAT is progressive:
However, this requires that a number of further conditions hold. First, the poor must have savings, and as I show, they don't. Second, they must have access to borrowing, and as I show, they don't (except for doorstep lenders). Third, the consumption patterns of the rich must be the same as the poor, and they're not. In fact, the consumption patterns of the rich (for school frees, private health, leisure travel, second homes and financial services products) are all VAT free, unlike the consumption patterns of the poorest. In addition, the IFS has to abuse all known notions of measure for progressivity to reach this conclusion.
The result is that far from the IFS claim being justified, it is very obviously wrong, and very poor quality research. As a matter of fact VAT is regressive.
The IFS claim is, however, consistent with persistent IFS recommendations that VAT be increased (to replace corporation tax, for example, and on food and children's clothing to pay for “desirable tax reductions”) all of which, together with their recommendations that Inheritance Tax be abolished and tax on interest income be abolished suggest a systematic bias towards making recommendations that favour redistribution of taxes from those who work for a living or who are the poorest in our country towards those with wealth and who enjoy income from capital.
None of which makes it easy to see how the IFS can sustain the claim that [vi] it:
maintain a rigorous, scientific approach to research, while offering scope for timely, independent, well-informed contributions to public debate.
The full paper is available here.
_______________________
[i] http://www.parliament.uk/briefingpapers/commons/lib/research/briefings/snbt-05620.pdf
[ii]http://www.gov.je/SiteCollectionDocuments/Tax%20and%20your%20money/ID%20FSR%20GREEN%20PAPER%2020100621%20MM.pdf
[iii] http://www.ifs.org.uk/budgets/gb2009/09chap10.pdf
[iv] It is, for example, defined as such in the Oxford Dictionary of Economics.
[v] [ii] http://www.statistics.gov.uk/downloads/theme_social/Taxes-Benefits-2007-2008/Taxes_benefits_0708.pd
Thanks for reading this post.
You can share this post on social media of your choice by clicking these icons:
You can subscribe to this blog's daily email here.
And if you would like to support this blog you can, here:
IFS =
Initiating
Fiscal
Stupidity
As one who spends much of my day looking at graphs, the first IFS figure (labelled as fig 10.1) still presents a clear negative correlation even if the lowest decile point is ignored.
Just to satisfy my own curiosity, I looked at the relative difference on the percentage of VAT paid between the 2nd and 3rd deciles and the 10th. The relative differences between the percentages paid come out at ~25% and 28% respectively wrt the 10th decile.
For a more smoothed comparison I took the average of 2,3,4 and 8,9,10 and calculated the relative difference. It does reduce slightly, but there is still ~16% relative difference between them. It seems that the description of “varies little” is misleading at best.
As an aside, I’m trying to work out how people who earn more can spend a greater proportion of their income as VAT (as VAT is a flat rate of 20% on many purchases). For instance between the 2nd to the 3rd decile or the 6th to the 7th decile. Are those increases caused by different spending behaviours? Could that reasonably be linked to more disposable income?
This is differences in spending patterns as more spend goes on VAT standard rate items
Johan G “Are those increases caused by different spending behaviours? Could that reasonably be linked to more disposable income?”
I was wondering about that (apart from the obvious exempt items like food which would impact on the distribution), maybe a big ticket item like a (brand) new car is more likely to be bought by people in a specific income group whilst others would opt for a second hand one (with no vat). Also some people still get a company car, how does that affect the data (as that isn’t an individuals expenditure as such).
BTW Richard, this is fantastic reseach. I don’t know what IFS are thinking about here but what they have produced isn’t statistical analysis.
Thanks
And yet, somehow the IFS is regarded by many as a reliable source of information and opinion.
Go figure.
Well, even I ‘enjoyed’ reading this piece, and I’m only used to various education related data analyses.
Seems to me the IFS aren’t doing a very rigorous analysis…
By design or by fault? I wouldn’t know. But if I’m right, they are the reference, so someone needs to point out their misleading/incomplete results and demand a better job be done.
Is the political agenda allowing for that to happen?
If not, is the press allowed to publicise the information and the misleading/incomplete nature of the IFS study?
It may be done somewhere already?
I have long suggested the IFS has an agenda, which is, of course, the mainstream economic one
This just illustrates it