I feel somewhat reluctant to return to the issue of my tax gap estimate, but since it is causing further debate I will.
Let's be clear why I prepared my tax gap estimate, published last week. It was to make clear what are, in my opinion, the serious deficiencies in HMRC's tax gap methodologies. These are numerous. I have explained them before. So let me offer just three here.
One is the failure of all these estimates to be based on top down data, VAT excepted, which is GDP based and in my opinion is pretty reliable as a result. The IMF has suggested that this top down demand is an appropriate request to make as it could give rise to significantly different estimates. I agree.
Another is that I think the bottom up estimates are based on some wholly inappropriate assumptions. For example, on average about 400,000 companies a year do not submit the corporation tax returns requested of them and many hundreds of thousands more are not even asked to submit returns. Sampling on these companies - which I would venture create much of the VAT gap - hardly takes place, the assumption being that they do not trade. I think that quite inappropriate.
Third, the assumption that the shadow economy is made up of marginal, low paid employment is again, I think wrong, and deeply politically biased.
And I would stress, these errors of assumption will not be reflected in the HMRC reliability stats, because they test what is tested and not what is not. So for the record, I also question HMRC's declared error rate.
Might we bear all this in mind? I am doing this to suggest change is necessary. And those who like the status quo - and no one does that more than HM Treasury - will not like that.
Then let me come to the criticisms. They are:
a) I use GDP at face value, subject to my noted discussion on whether it includes the shadow economy or not. The suggestion is that this estimate may contain significant errors.
b) I have used an average of three shadow economy estimates, two published by major international organisations (the EU and IMF) which are probably as good as any estimates are likely to be. The suggestion is, again, that these estimates may be wrong.
c) To estimate the tax lost to the shadow economy I use the average UK tax rate because i) that rate is remarkably consistent across all income levels ii) it is lower than almost all likely marginal tax rates ii) I can see almost no reason why the shadow economy should - when of this aggregate size have a much different rate to the rest of the economy and iv) I am seeking an aggregate rate because as far as I am concerned the tax gap is not a microeconomic measure of the efficiency per se of any one tax but is instead, as modern monetary theory suggests, a measure of the failure of the state to impose its fiscal policy. Others may disagree.
The criticisms arise for a very common reason. Those who do not like the implication of my suggestion, for whatever reason, are using a common technique: they are playing the methodology and not the arguments. The argument made is that statistically I compound the errors in each estimate and therefore any conclusion I come to is much less reliable than anything HMRC offer.
I disagree. First, that is because I suggest the tax bases HMRC do not consider are more significant than any error I might make.
Second, I would argue that unless any GDP based estimate used for any reason is rejected because GDP is unreliable, for reasons I have noted, then to suggest I cannot use the figure at face value is just hyperbole. I am particularly amused that someone who has self-identified as a former Treasury official has made a big play on this GDP issue. It is as if I am not allowed to use GDP data for any purpose but those they have sanctioned when the Treasury not only obsess about GDP but never issue appropriate risk warnings as to its use. Very clearly different rules are being applied to me than are suggested for their own use.
I would actually say the same of using the best available peer reviewed data. Of course there are margins in there. And I sought to reduce those by looking for a variety of sources - plus the HMRC VAT gap data - and they all support each other. I would say that appropriately manages the risk. And that is as much as anyone can do.
Unless we are to say using estimates in economic analysis is not acceptable I think these first two claims of no real significance, and any likely error (and I accept there will be some) are lower than those HMRC create, so comparison with them remains valid.
So what of the use of the average tax rate of 33%, or thereabouts? This seems to be the basis of most criticisms. So let me justify this, from first principles.
First, let me ask why people evade tax by suppressing report of their income? There is only one rational explanation and that is that tax is not paid as a result. So let's not pretend that, as some claim, that many of those in the shadow economy would not be taxable anyway. I suggest that is utterly implausible. In fact, some on very low apparent taxable incomes might have very high marginal effective tax rates if their income was declared because of the interaction with benefits. This argument does then hold no water.
This brings me to my second point. This is that for everyone engaged with tax evasion evades tax at their highest marginal tax rate. So, when appraising the likely tax evaded a consideration of marginal tax rates is what is appropriate. Now, I accept there is no one simple answer here, but let's just look at income tax and national insurance to start. No one's marginal tax rate here is going to be much less than 32%, which is basic income tax rate plus basic NIC. I know it is slightly different for the self employed, but not wildly so. I still suggest we can ignore corporation tax, although section 455 has been noted. These evaded sums do not remain in companies. And I will still suggest that marginal relates can frequently be much higher than this. Gabriel Zucman has suggested most evasion is by the top earners in any society. Their marginal rates will exceed 40%. Add in VAT evaded and employers NIC as well, and potential rates go much above that.
When saying so I might add that the claim also being made that tax evasion is mainly by low earners is ridiculous. That is because a) they simply don't earn enough to make up the numbers and b) we know the vast majority of small businesses in the UK are not VAT registered, and therefore have low earnings, and that this is allowed for in VAT gap calculations, but that gap remains about 9% of yield and even after allowing for debt this can only be because more substantial businesses, likely as a result to yield higher earnings, is where the problem arises. Indeed, this has to be the case. Just as most tax has to be paid by high earners, by definition or the sums don't add, it pretty much follows that this must be true of evasion too. Claims otherwise make no sense, and indicate prior research bias if they suggest they are evidence based - as HMRC work in the shadow economy (which as they define it is not the same as the tax evasion base) makes clear.
I admit I have not modelled all these likelihoods, but the fact is that finding arrangements where the marginal tax rates, which are what is overall lost, are less than 32% is fairly hard, and exceptional, whilst the vast majority has to be at significantly higher rates. In other words, I think all rational evidence suggests I have under-estimated this tax rate, which I did, knowingly.
Do I accept the criticisms made then? No I don't. They simply do not stack. Or to put it another way, they look like excuses to me.
Might I suggest that we talk about what we do about this instead, then?
And please note I am not interested in hearing past criticisms again. Please move on, or don't bother commenting. I've dealt with the above issues now, unless new evidence is added.
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:
[…] Sound familiar? A decade on I still seem to be having such arguments. […]
[…] Sound familiar? A decade on I still seem to be having such arguments. […]
Hardly surprising you are standing by your tax gap estimate. Unfortunately that doesn’t make it, or your claims correct.
You claim there are serious deficiencies in HMRC’s methodology, but haven’t actually looked at their methods whatsoever. What exactly is wrong with it? In every place, their work is more thorough and more detailed than yours, with access to better data.
I will address your points in turn.
“One is the failure of all these estimates to be based on top down data”
It is well understood that top down methods are less accurate than bottom up, to the point that they are meaningless.
https://www.irs.gov/pub/irs-soi/11rescontaxgap.pdf
I quote:
Declaration by the world’s statistical institutions (ISWGNA)
“Unofficial estimates [of the shadow economy] are often based on macro economic models.
… The OECD-ILO-IMF-CIS manual on measuring the non-observed economy rejects
such ‘macro-model’ methods because these methods suffer from serious problems that cast
doubt on their utility for any purpose in which accuracy is important.”
and
“Macro model methods have not produced reliable estimates of changes in the shadow economy or of the size of the shadow economy. They are of no value for work on the tax gap. This conclusion on macro model methods is very much in line with a declaration by the world’s statistical institutions.”
“Another is that I think the bottom up estimates are based on some wholly inappropriate assumptions. ”
Bottom up estimates are not assumptions based at all. They sample the actual data, from the real world. The only error comes from the sample taken itself being a proprtion of the whole. HMRC sample large enough sizes to minimise errors where possible. They don’t ignore the companies not submitting returns to them of companies house – they test this data too through random checks and the data from enforcement actions. Not perfect but better than what you have done – which is actually not to test any of the assumptions you have made at all.
“Third, the assumption that the shadow economy is made up of marginal, low paid employment is again, I think wrong, and deeply politically biased.”
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/654905/P11533_HE_Final_report_MASTER_draft_v6_0_PC_FINAL.PDF
I quote:
“Income varied within this group, with 44% of Hidden Economy participants earning less than £250 in the past 12 months.”
“At least 53% of those in the Hidden Economy had an overall income above the tax threshold, which means at least some of their Hidden Economy income is assumed to be taxable”
This research from real data (not assumptions) shows that the largest part of the shadow economy is small, and at low marginal rates of taxation.
“And I would stress, these errors of assumption will not be reflected in the HMRC reliability stats, because they test what is tested and not what is not”
HMRC explore and research what might be missing from their data. You do no such thing.
Regarding the criticisms of your work:
“a) I use GDP at face value, subject to my noted discussion on whether it includes the shadow economy or not. The suggestion is that this estimate may contain significant errors.”
Nobody is criticising your use of the GDP statistic. People are saying that GDP has an error margin.
“b) I have used an average of three shadow economy estimates, two published by major international organisations (the EU and IMF) which are probably as good as any estimates are likely to be. The suggestion is, again, that these estimates may be wrong.”
Yes, they all are. All of them have an error margin of 15%, because they all use the same type of model to estimate them. See above where the IMF criticise their own model.
“c) To estimate the tax lost to the shadow economy I use the average UK tax rate because i) that rate is remarkably consistent across all income levels”
See above, but tax rates are not consistent at all income levels, particularly so at low marginal rates.
“ii) it is lower than almost all likely marginal tax rates ”
You have just assumed this, for which you have no data, no evidence.
“ii) I can see almost no reason why the shadow economy should – when of this aggregate size have a much different rate to the rest of the economy”
Just because you don’t see it doesn’t mean it isn’t true. The research has been done, as linked to above. The shadow economy is at lower marginal rates.
“Those who do not like the implication of my suggestion, for whatever reason, are using a common technique: they are playing the methodology and not the arguments”
Nobody is claiming there is no tax gap. People are claiming that your methods grossly overestimate it, because your methods are simplistic and flawed. They also object to your claims that your work is more accurate than HMRC’s.
“The argument made is that statistically I compound the errors in each estimate and therefore any conclusion I come to is much less reliable than anything HMRC offer”
Yes, by definition. HMRC have at worst a 10% uncertainty.
You have a 5% from GDP, 15% from the size of the shadow economy and a large but undisclosed error from the size of your average take rate. You calculate this average tax rate using GDP, so the error cannot be smaller than 5%. This means, just be definition, your error rate is around 30% as a MINIMUM.
Unless you simply don’t understand statistics, and they work differently for you than anyone else, this means your data is MORE uncertain and therefore LESS reliable than HMRCs by default.
“I disagree. First, that is because I suggest the tax bases HMRC do not consider are more significant than any error I might make”
Presented without evidence. HMRC take care to estimate the whole tax base – investigating what is missing. You do no investigation into anything at all.
“Second, I would argue that unless any GDP based estimate used for any reason is rejected because GDP is unreliable”
You are trying to create a straw man and deflect attention from the real problems in your work. GDP is the most accurate part of the data you use. Nobody is saying you shouldn’t use it, people are saying that you need to account for the errors and uncertainty, which you do not do at all.
“I would actually say the same of using the best available peer reviewed data. Of course there are margins in there. And I sought to reduce those by looking for a variety of sources”
I think this just underlines your lack of understanding of statistics. Looking at a few different sources does not reduce your error, if all of those sources have the same level of uncertainty. Especially as in your estimate you have just averaged the data, or for some just picked the one you liked the most.
“Unless we are to say using estimates in economic analysis is not acceptable I think these first two claims of no real significance, and any likely error (and I accept there will be some) are lower than those HMRC create, so comparison with them remains valid.”
Nobody is saying using estimates is unacceptable, but you have to account for the errors estimate create. By multiplying three estimates together you compound that error.
Unless your error is less than 10%, which we know it cannot be, your data cannot be more reliable than HMRCs.
“So let’s not pretend that, as some claim, that many of those in the shadow economy would not be taxable anyway”
That is what observed data suggests. You are just making a claim you have no evidence for.
“I suggest that is utterly implausible.”
You suggesting does not make you right.
“This is that for everyone engaged with tax evasion evades tax at their highest marginal tax rate.”
Again an assumption, but marginal is important here.
“No one’s marginal tax rate here is going to be much less than 32%, which is basic income tax rate plus basic NIC”
Which you don’t start paying until you hit the tax band thresholds, so the rate won’t be as high.
“Gabriel Zucman has suggested most evasion is by the top earners in any society”
http://gabriel-zucman.eu/files/AJZ2019.pdf
His data shows that only the top 0.01% of earners might be evading 25% of taxes owed – already lower than your 33% average tax rate. For the rest of the 99% the rate is lower than 5%.
This blows a huge hole in your own argument. Even with this evidence are you unable to admit your 33% rate is going to be way too high?
“Their marginal rates will exceed 40%. Add in VAT evaded and employers NIC as well, and potential rates go much above that.”
According to Zucman, who you refer to, this isn’t true.
“When saying so I might add that the claim also being made that tax evasion is mainly by low earners is ridiculous. That is because a) they simply don’t earn enough to make up the numbers”
Again, you are just making a claim without evidence but the second part of your statement should have given you pause for thought. If the bulk of evasion is from the low income deciles, then maybe it is your estimate (“the numbers”) that is wrong, not the data.
“and that this is allowed for in VAT gap calculations, but that gap remains about 9% of yield”
Top down VAT gaps are known to be inaccurate and are known to produce a MAXIMUM theoretical VAT gap. The actual VAT gap will be lower.
“Claims otherwise make no sense, and indicate prior research bias if they suggest they are evidence based – as HMRC work in the shadow economy (which as they define it is not the same as the tax evasion base) makes clear”
The whole point of acquiring real world data is to remove bias. You get a picture of what is really happening from the real world. HMRC’s shadow economy work does not claim that the shadow economy and tax evasion bases are the same – quite the opposite. You ARE claiming that the shadow economy can be used as a proxy for tax evasion – it is the entire basis for your estimate.
“I admit I have not modelled all these likelihoods”
Have you modelled ANY of these likelihoods?
“but the fact is that finding arrangements where the marginal tax rates, which are what is overall lost, are less than 32% is fairly hard, and exceptional, whilst the vast majority has to be at significantly higher rates.”
Despite not having modelled it, you then go on to make a claim that average tax rates will be 32% or higher. It is very easy to see this is false, from the data provided, as well as arguments people have already made. In fact, your own example from the last post showed that the average tax rate of evasion was much lower than this, even when you used a much higher rate of VAT evasion. I actually doubt you are able to provide a single worked example where the average tax rate of evasion is as high as the 33% you claim.
Regardless, the data we have available shows that the 33% rate you use is dramatically high.
“Do I accept the criticisms made then? No I don’t. They simply do not stack. Or to put it another way, they look like excuses to me.”
Hardly a surprise you don’t accept criticism. The criticisms made are valid. You haven’t looked at any data, haven’t done any research, have ignored the statistical errors in the few pieces of data you have used and your methods are suspect – both the top down approach and your arguments with regard to average tax rates.
I think the only person making excuses here is you, in a King Canute style effort to turn back the tide.
Jono
I said I would not accept repetition
And that I am not really interested in ad hominem attacks, which you appear to make a speciality
If you add new evidence here I can’t see
What I can see is this:
Bottom up estimates are not assumptions based at all. They sample the actual data, from the real world.
Oh, dear, was my reaction to that, closely followed by curiosity as to what you really knew
What we do know is you do not understand double entry, you have a blind faith in HMRC and you are not good at actually dealing with evidence.
Next time you will be deleted for time wasting – as this comment was since it added nothing new
Richard
Jono’s input is still looking at the wrong thing – your estimate – instead of digging deeper into HMRC’s methods and estimates in detail given the effects austerity has had on the service. We also know that we had a party in power who are hostile to income tax.
All you have done is ask questions in my view. And then certain people have asked questions about your questions and HMRC just get let off lightly in the process.