My attention was drawn yesterday to a blog by Graeme Roy of the Fraser of Allander Institute, which claims to be 'Scotland's leading independent economic research institute' and is based at the University of Strathclyde. Graeme Roy, who admits he worked on preparing GERS (Government Expenditure and Revenue Scotland, which has been the subject of some discussion here of late) for seven years until 2014, write defending the criticisms I have raised of it, without ever explicitly saying so.
It may be worth starting by recalling what my first (and continuing) criticism of GERS was. I said in a blog that estimated data from the London (which I made clear was the Westminster government) was of little use in deciding the future of Scotland. In that case it's important to note two things Roy says. They are both in this one paragraph:
In short, GERS estimates the contribution of public sector revenue raised in Scotland toward the public sector goods and services provided for the benefit of the people of Scotland. It's important to remember that GERS does this taking Scotland as a mini-UK, and the constraints and protections that the current constitutional arrangements bring.
I could almost finish at this point. First, Scotland is not a mini-UK: it is a separate country within the UK. In that case GERS is prepared on an inappropriate basis. And independence assumes the current constitutional arrangements will change. My claims are, therefore, correct as a matter of fact. But having entirely conceded my argument Roy goes on to defend GERS in ways which do not anyway, in my opinion, make sense, so I will continue.
My main concern is this claim:
It's a National Statistics publication. This means that the statistics — and how they are presented — have been independently judged to be methodologically sound and produced free of political interference.
In what follows I wish to make clear that I respect what Roy has written, and the integrity of his motives in doing so, and his stated beliefs. I am quite sincere when saying that I believe he sees no problems in making a number of assertions with which I will, however take issue. Nor, I stress, is it the case that in saying these things that I am questioning the integrity of those preparing GERS. What I am saying is that they are working sincerely within a system which which, first of all, imposes political interference as a matter of fact and which is not as independent as Roy would like to claim, albeit entirely honestly.
Let me deal with the second issue first. I am afraid that I have little confidence in almost any claims of professional objectivity. I have long challenged the accountancy professions claim to be objective when it comes to standard setting and enforcement. It patently is not. In fact there's strong evidence it does not even understand the law and instead construes it to its own advantage. Economists are no better: I always remember with amusement the claim that an economist once made to me along the lines of "Of course I am objective; I accept all the assumptions of neoclassical economics", to which I fell about laughing, largely because the absurdity of what he'd said was clearly not apparent to him. And when it comes to government statistics the standards are set by one civil service organisation for another civil service organisation and since all such organisations will call upon the same broad pool of talent and operate in the same broad way for the same broad administrative structure, financed in the same broad way it is hardly surprising if there is a convergence of opinion on what is acceptable. The fact that there is a peer review process does not alter this: peer review is almost always designed to reinforce the status quo. I stress I am not saying that the statistics are not prepared in accordance with a standard, but just as winners write history, it's a fact that prevailing power elites write rules to reflect their priorities without always realising that they have done so. As a result saying GERS is acceptable because it meets the standards set by Westminster who quite clearly want Scotland to be treated as if it is part of the rest of the UK is no comfort at all. It just says that's the standard that's been met. It does not say if the standard is appropriate.
And this is the problem Roy faces when claiming there is no political interference in GERS. He seemingly fails to note when doing so that he has made, and noted, two massive political assumptions i.e. that Scotland is just a part of the UK when it would not have its own parliament if it was, and second that Scotland can survive on UK based data and does not need its own economic data despite the fact that it has devolved economic powers. That's not an objective assumption. It's actually an assumption that is, in my opinion, contemptuous of the whole idea that Scotland has the right to exercise discretion by denying it the data it needs to both decide upon appropriate actions and then appraise outcomes. This is a Westminster assumption and is implicit in the data available to prepare GERS. However good the statisticians who prepare GERS are they can't overcome this fact and they should, I suggest, recognise that fact, but Roy does not. In that context the claim Roy then makes that GERS must be right because it looks remarkably like the UK as a whole is, to say the least, mildly absurd. If the data GERS produces for Scotland is an abstraction from that for the UK as a whole the only thing that would be surprising is if it did not look like the UK as a whole. Roy seems to miss this obvious point.
So how might this happen? Roy believes what he has said, I am quite sure. But he has made assumptions that I think are are political and so subjective without realising because, I suggest, they are, to use George Bernard Shaw's definition, the assumptions of a reasonable person. Reasonable people adapt themselves to the ways of the world. Roy is doing that. So too, of course, are those who establish the standards for statistics with which GERS complies. And if you do comply with the ways of the world you rarely realise that is what you are doing precisely because complying seems so normal you do not even realise it is a choice. There is just one problem though: as Shaw also noted, unreasonable people seek to adapt the world to their ways. As a result he suggested all progress is dependent upon the existence of unreasonable people.
I am, I readily confess, unreasonable on this basis. I was told, endlessly, that I could not have and did not need country-by-country reporting. Now it is to be required worldwide.
And I was told automatic information exchange from tax havens would not happen in my lifetime. It is underway.
Time and again HMRC have said I am wrong on the tax gap, but the feeling that it may be them that is wrong is now becoming widespread.
In fact the whole tax justice agenda has been a story of being told that existing data is just fine by a power elite, whether it be politicians, governments, their agencies or professional bodies that claimed we really did not understand just how well the system was working and to leave them in peace. But we did not, and demanded new data anyway, and we now know we were right to do so.
It is my suggestion that the story in Scotland is the same. GERS was created by a Westminster power elite so suit their purpose. It cannot now meet the needs of Scotland, however much it complies with the statistical standards created by that same power elite. I don't apologise for saying so. I reiterate: Scotland needs its own data. I don't think it's unreasonable to say so, but many apparently do. I don't apologise for upsetting them.
Which leads to me to my final question in this blog (which is not aimed at statisticians, but politicians), which is why making this suggestion is so contentious? Could that be because some people do not believe Scotland is a country worthy of its own data? And could it be that they really do not think it should have the information it needs to make informed decisions? That's perfectly possible, but if that's what they do think then I have to ask why do they also think they have the right to suggest they should govern from Holyrood, whether Scotland is independent or not without the data that will increasingly be required to do so?
The question is a serious one. My suggestion is that those so wedded to GERS that they cannot see what their devotion implies are actually not fit to make the necessary judgements that holding office in Scotland would seem to demand. I would presume that a demand for better data for Scotland would have universal appeal amongst anyone who aspired to office in that country. That it does not suggests to me that some are quite determined that in principle Scotland should not have the information its politicians need to govern. At a quite deep level that's worrying because it implies that not only are some opposed to independence but that they have not even embraced the idea of devolved power in which, however, they make the appearance of partaking.
Information is power. Scotland must have the information it needs. Without it, whether within or without the UK it will not have the power to shape its future. And that is no minor issue.
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:
Long ago I recall being involved in moving an armoured division as training for any hostilities that might arise. We had a wealth of data and experienced staff officers for the logistics. It did not stop half of our troops managing to get lost.
🙂
One thing that struck me, apart from the ‘mini UK’, was the use of the decommissioning costs of Dounreay being shared as a plus point for the neutrality of the figures. Clearly though, it also means that Scotland will be allocated a share of the decommissioning costs of the many more nuclear installations in England.
Agreed
And it was also so trifling: sweating the small stuff, I thought
[…] Update, 29 March: Richard Murphy has posted a response. […]
Imagine this scenario:
A company HQ’d in England (for illustrative purposes, let’s call it Westminster) incurs costs (let’s call it HS2) which only benefits that company’s English operation.
The company however cross-charges an element of those costs to a member of its corporate group (let’s call that member company Scotland), and the member company accounts for those costs in its accounts.
The HQ company likewise incurs numerous other costs which it only uses for the English operation (let’s call them Trident, Crossrail, Hinkley etc), and cross-charges all these additional costs to the corporate group member too, Again, although the corporate group entity doesn’t actually use these costs for its own business, it nevertheless accounts for the costs in its accounts.
But, because the HQ has overspent, it needs to borrow a shitload of money, meaning it has a loss in the HQ accounts.
No worries though, it merely allocates some of those losses to the corporate group member, and of course, that corporate group member includes the losses transferred in its accounts.
Oh, and of course, despite the HQ mismanaging the whole corporate group, it makes a management charge to the corporate group member, and yes, the group member includes these in its accounts.
Having run up a series of debts in the prevailing year, the HQ allocates part of the servicing of that debt to the corporate group member.
Now, imagine if you will that, say, Deloitte were asked to sign of the accounts as ‘a fair and accurate statement’ of the group member’s accounts; it would point blank refuse to do so as the costs, losses and interest payable not only relate to the same thing, but are not in any way associated with the running of the corporate member’s business activity.
This, ladies and gentlemen, is GERS.
You may well be right
Richard Murphy’s argument is sound. The problem is that the technique of statistics is often misused, sometimes in surprising and unexpected places. This is an issue of substance and it is serious. The misuse has been widely discussed in the literature in specific academic disciplines. These include epidemiology and economics. The deficiencies are well-known and understood.
On the specifics of problems in economics with handling ‘national accounting’ statistics effectively, Dr Frits Bos of the CPB Netherlands Bureau for Economic Policy Analysis wrote this in ‘Use, misuse and proper use of national accounts statistics’ (2007: Introduction, p.7):
“Among economic researchers there is a worldwide illiteracy in national accounting. A decade ago, national accounting has been dropped as a separate topic of research on the list of the Journal of Economic Literature. The economic researchers skilled in national accounting have become more and more extinct. Furthermore, the introduction of new guidelines on national accounting has made a substantial part of their knowledge obsolete. However, for the analysis of economic statistics solid knowledge of the merits and limitations of their concepts and measurement is indispensable.
Bos goes on to quote Schumpeter, “History of Economic analysis” (1945, p.14):
“We need statistics not only for explaining things, but also in order to know precisely what there is to explain. … It is impossible to understand statistical figures without understanding how they have been compiled. It is equally impossible to extract information from them or to understand the information that specialists extract for the rest of us without understanding the methods by which this is done- and the epistemological backgrounds of these methods. Thus, an adequate command of modern statistical methods is a necessary (but not a sufficient) condition for preventing the modern economist from producing nonsense”.
The economics profession is unfortunately and just a little too often, given to “producing nonsense”.
Let me be clear, I have no idea what the size of any “deficit” Scotland may have. And that is precisely my point. This is not about proving a political point, one way or the other; it is fundamentally about something I revere – knowledge. Ignorance should not be an option; and cannot be passed off because unsatisfactory data may have been subject to the magical transformatory powers through the application of statistical techniques (with or without the incantation of ‘abracadabra’). The matter of real concern is the failure to establish the existence of adequate observable data, and its accurate measurement. That is the very matter of science and should interest even social scientists.
I agree: the issue is we don’t know
I have not said otherwise
I sometimes think the word statistics is overused. For example, UK VAT is not really a statistic. So there is no such thing as an estimate when considering the UK state that collects VAT. Presumably HMRC must have an actual value for the VAT collected on behalf of the UK state. The same is certainly not true for Scotland, as Prof Murphy has pointed out.
As far as Scotland is concerned, the VAT that would be collected on behalf of an independent Scotland is estimated, and indeed given a 95% confidence interval in GERS.
The estimate is given as about £11bn +/- £250m (95% conf). This is based on consumption survey data and I’ve yet to find a justification for the given confidence interval.
The latest paper on GERS methodology (http://www.gov.scot/Resource/0050/00504656.pdf) describes in broad terms the survey methods used, but does not mention those confidence limits. If there’s another source for the confidence intervals, I’ve yet to find it.
The estimation method may also provide biased data given the uncertainties inherent in the survey methods. The justification for the samples used being representative seems rather weak.
VAT is one of the largest (potential) Scottish revenue numbers in GERS.
Similar doubts exist about how the estimates of Corporation Tax attributed to Scotland in GERS are produced.
Many of these samples aye biased without considerable care being expended: there is serious risk of over sampling some groups is society whikst others are very reticent
I always understood that the use of confidence levels, significance etc required random sampling. Are these samples used in GERS random?
Some will be
Although structuredly random, I hope
Calculating a confidence level depends on knowledge of the characteristics of the underlying population. For some items in GERS, more than one population is sampled to provide the result. This makes the calculation of meaningful results and their confidence intervals more complex, and potentially problematic.
The samples need to be representative of the population (a random sample if you like) – otherwise the results may be biased (in a technical sense).
None of the above means the results are necessarily invalid, but I’ve yet to find an explanation of how the confidence intervals are calculated.
The GERS figures were designed by the Conservative party during the Major government to portray them as more fiscally responsible than the opposition parties. The flaws in the information are most likely to be deliberately designed.
You might say that
I could not possibly comment
Guilty:BBC Reporting Scotland misled viewers about GERS
http://indyref2.scot/guilty-reporting-scotland-misled-viewers-about-gers
Misleading information compounded by further misleading information.
Worth a read.
Interesting
Thanks
Assuming everything you said above is true and to be accepted, why do you think the Scottish Nationalist politicians haven’t come up with, or mooted, another more accurate method of collecting and calculating relevant economic data?
Given that they published a white paper for 2014 based almost wholly on GERS data to convince us to take the leap into independence, doesn’t that strike you as completely irresponsible and contemptuous of the sovereign Scottish people?
Regards
I was not making party political points
I do indeed hope they will replace GERS with real tax data in particular
And indications as to what is imposed and what is chosen spending
Your question is best directed towards and answered by the SNP. I do not know why, but I will hypothesise there may be two plausible reasons:
1) Before the oil price collapse the optics of GERS presented a publically very presentable picture of revenue/expenditure from a Scottish government perspective, whether or not the data collection methodology was sound. It was much easier and uncontroversial to use GERS than change or even challenge the underpinning principles, over which the SNP had little power.
2) The SNP rose to power at Holyrood by presenting themselves to the electorate as responsible and competent managers of the existing system and structures in the devolution settlement: SNP credibility was built on a reputation for relative managerial competence (compared with the other political parties/talent on offer in Scotland). This included using GERS, and only making small adjustments to the GERS methodology, over which in any case they had limited independent influence in many areas. Having used the system for years without even or adequately challenging the fundamentals, the SNP do not now find themselves in the best position to critique the substance of GERS.
QED
Various attempts have been made to improve the estimates used in GERS. The actual data for most of the line items is not available as Prof Murphy has pointed out, so you can hardly blame the Scottish Government. For example, the real data on corporation tax on commercial activity in Scotland in not collected anywhere.
From April 2017, the collection of income tax will become the responsibility of the Scottish Government, so that will provide real data for use in GERS.
As I mentioned in a comment above, VAT is a large source of revenue. In April 2019, the Scottish Government will be assigned one half of the estimated VAT, compensated by a reduction in the so-called block grant. This clearly does not improve our knowledge of VAT data.
The Scottish government should demand real vat data
Prof Murphy may not be impressed with how the VAT assignment will work –
“The methodology for apportioning VAT revenues to Scotland will be used to estimate receipts in the year immediately prior to assignment. This will be reconciled to outturn using the same methodology once the relevant data is available.”
Those magic words – apportioning and estimate, with a built-in year’s delay to add further uncertainty.
The UK has argued they could not give Scotland powers over VAT because the EU does not allow sub-national variation. This would not have prevented the collection of real data on VAT and other parameters relevant to Scotland. As Prof Murphy has previously suggested, real data is essential if Scotland is to make rational decisions, but the UK has no incentive to provide such data.
I’m not sure about the QED at the end of the contribution above from John S Warren, but I agree with some of what he says. However, does he really imagine the UK Government would put in place the systems required to provide the real data on Scottish revenues and balance of trade?
Thanks George
Interesting that Gordon Brown has suggested VAT powers for Scotland
As it is these terms are about as poor as the VAT sharing agreement with the Isle of Man and I have history on that: it was rewritten because I exposed how dire it was. But it’s now just bad, not dire
Interesting stuff sure enough. Anyway, two questions for you if I may be so bold.
1) Since your original post a few weeks back there seems to be a body of opinion from Nationalists that you are saying that the GDP deficit has been overstated in GERS, but I cannot find anywhere where you actually say this. Could it be that if that you are correct in your opinions re the veracity of the data used in compiling GERS and that our figures used are actually way-off estimates, could it be that the GDP deficit is actually worse than what is stated in GERS, not better?
2) I believe that you have been invited to have a debate with Kevin Hague on GERS on Radio Scotland, and I am to understand that Mr Hague accepted. Is it your intention to accept this invitation?
Re a) I will be posting sometime on this issue but so far I have not said that
Re b) I do not have an invitation – no one has contacted me to suggest dates or times or even to actually ask. I would accept if it was possible, of course, although it’s always a little odd debating with someone who has blocked you on twitter
Richard’s point about Scotland having control of the data is absolutely key here for a safe independence if is should come about.
If I were a Scottish politician serious about knowing my counttry’s net worth I would have set up my own capacity to measure the Scottish economy long ago and not relied on data from over the Border. Surely this has already occurred to someone?
Anyone decrying this genuine effort to get to the truth can only habe an agenda of their own to hide. Scotland does not need gung-ho nationalism as this moment in time.
Almost all the real data is only available to the UK Government, and is not split in a way that is useful to Scotland (as far as we know!). The clearest example of this is VAT, which would be one of the largest revenue sources for an independent Scotland.
Scotland cannot set up its own capacity to measure the economy since it doesn’t collect any of the big taxes. Although Scotland appears to have a positive trade balance, it is again impossible to be sure since the proper data is not collected – not even by the UK Government.
Precisely