GERS: or, A Wayward Exercise in the Capricious
The most important matter regarding GERS is not the output, the ‘figures’, the so-called facts; the most important matter is the methodology used to create them. Without a sound, dependable, stable methodology we have nothing: nothing.
The methodology is provided in ‘Government Expenditure and Revenue Scotland: Detailed Revenue Methodology Paper 2014-15’ (http://www.gov.scot/Resource/0049/00495355.pdf). The Methodology Paper “outlines the various methodologies used to obtain estimates of public sector revenues in Scotland”. The reference to “various methodologies” is a warning that this is going to prove complicated, and so it soon proves.
“In contrast to public sector expenditure, there is no generic best approach to estimating public sector revenue; instead each revenue is estimated using a separate methodology. This paper discusses the methodology used to apportion a share of each revenue stream to Scotland and highlights any significant changes which have been introduced in this edition of GERS. It should be noted that, as the underlying datasets used in GERS have been subject to revisions and updates, estimates may differ from previous editions of GERS even if the methodology remains unchanged” (GERS Methodology Paper p.1).
I can think of few introductions to a methodology less likely to provide confidence in the result. Each sentence in the above preliminary statement identifies both complexity and a shifting base; even the past is subject to “revisions and updates” (numerous?). There is no “generic best approach”. “Even if the methodology remains unchanged” implies the very methodology may change between datasets. It seems that everything is subject to change in this shifty, Heracletean river of subjective financial interpretation. All this qualification from the outset somewhat diminishes the credibility of the methodology that is so apologetically being promoted; and we have not even reached the ‘Methodology Overview’ yet. However the output from this doubtful methodology is claimed by the British Government with extravagant insistence to be meaningful, and to throw light on Scotland’s economy.
The GERS Methodology Overview explains that the majority of public sector receipts raised in Scotland are collected at the UK level by HM Revenue and Customs: “collected at the UK level” we may suspect carries the risk that some fairly rough-and-ready disaggregations are going to be applied to the data; and so it turns out.
“In some cases, revenue figures can be obtained for Scotland directly. Examples include local government revenues and particular elements of public corporation revenues. Such taxes are the exception and separate identification of most other revenues for Scotland is not possible. GERS therefore uses a number of different methodologies to apportion tax revenues to Scotland. In doing so, there are often theoretical and practical challenges in determining an appropriate share to allocate to Scotland. In certain cases, a variety of alternative methodologies could be applied each leading to different estimates.Obtaining an estimate of public sector revenues in Scotland is a two-stage process. In the first stage, the UK outturn figure for each stream of revenue is obtained from ONS Public Sector Finance Statistics. In the second stage, Scotland’s share of the UK figure is estimated according to a specific apportionment methodology. The methodology used differs for each element of revenue. However, in general, the information comes from survey data for the UK” (GERS Methodology Paper p.3).
The public revenues that are directly identifiable for Scotland “are the exception”. Separate identification of many “revenues for Scotland is not possible”. This means that creative, interpretive accounting of one kind or another (with all the hostages to fortune, or worse this may conjure) is going to be required to identify the huge gaps in the information. At this point, wisdom would dictate that ‘the game is up’ and there is a requirement to abandon a bad proposition and begin again with serious intent to identify the real facts: to assemble the underlying, real data by more direct means; not to create the facts by exception, but to find them. Neither of these quite obvious necessities cut any ice in Britain today. This is Scotland, crude shortcuts are not only permissible, but compulsory. What happens is that the existing data is made to fit the purpose, even though clearly, they don’t; no matter what.
Thus figures are being presented for Scotland using this “methodology” (actually methodologies), that do not actually exist. There are also “theoretical and practical challenges” (by implication either insurmountable in nature or, more likely, there is no inclination to overcome them) in determination of the estimates, and different methodologies could lead to different results. Notice also that “the information comes from survey data for the UK”. Survey data for the UK is being used, in order to establish the actual public sector/Government Revenues for Scotland. The methodology is detaching the exercise from reality as a matter of intent. It begins to appear that using the ostensibly authoritative term “methodology” to describe this Westminster inspired and wayward exercise in the capricious, is something of an adventurously exaggerated description of the status of the underlying facts.
Some of the large UK sources of revenue in taxation in GERS (Corporation Tax, VAT and so on) are sourced from HMRC (see for example, GERS Methodology Paper, p.4-5). When we turn to the HMRC ‘A disaggregation of HMRC tax receipts between England, Wales, Scotland & Northern Ireland: Methodology Note’ (October 2015) the better to understand this, we find, by now with a sinking feeling of inevitablility, that the Methodology Overview is carrying us precisely nowhere. Indeed it is preceded by a “Disclaimer”, which I have reproduced in full:
“This publication apportions total UK tax receipts to England, Wales, Scotland and Northern Ireland. It attempts to measure the true economic incidence of taxation, based on the underlying activity, which can often differ from how or where the tax receipts are collected. Actual administrative data is available for Capital Gains Tax, Inheritance Tax, Stamp Duty Land Tax, Child and Working Tax Credits and Child Benefit; for the others, the estimates are arrived at using best available data and statistical techniques, including assumptions and adjustments where necessary. The numbers in this publication do not represent an estimate of the tax revenue that would be raised if each tax was set at the devolved level. All statistical methodologies have an inherent degree of uncertainty and, for this publication, a variety of alternate methodologies could justifiably be applied, each leading to a different estimate” (HMRC Methodology Note, p.2).
The output of this fantastical exercise includes “assumptions and adjustments” (many of which we will never see the detail) and a gratuitous but telling acknowledgement that they “do not represent an estimate of the tax revenue that would be raised if each tax was set at the devolved level”. In other words, what is the point? It seems in practical terms – none. Indeed the last sentence of the above quotation, that introduces a little unexpected, direct candour about what is actually going on in this Kafkaesque exercise (different methodologies would produce different estimates), provides us with a cast-iron guarantee that precisely nothing is being guaranteed by this whole methodological circus.
The HMRC Methodology Overview merely adds layers of potential obfuscation to our perplexity. First we are introduced to ‘complexity’, then we are reminded of the reliance on “survey” data, followed by the sudden introduction of a strange new idea:
“We have also sought, wherever possible, to use HMRC’s administrative data but this is not always available at the required level of geographical disaggregation; in those cases, estimates have been made on the basis of a proxy activity, for which the data is available” (HMRC Methodology Note: Overview, para., 2, p.4).
We now have “proxy” data; presumably a kind of virtual reality. It isn’t real but it looks plausible. No doubt this makes the political masters feel better; a proxy reality is something they understand, live with, manipulate and with the help of a complicit media, rely on day in, day out.
The HMRC Methodology Overview also reveals not only that survey data is used, but in using this source “there are often multiple sources of error, such as sampling, non-response and measurement error, which we assume are consistent across England, Wales, Scotland and Northern Ireland” (HMRC Methodology Note: Overview, para., 4, p.4). All this is the truism of statistics, but it does not add to confidence in the underlying data. In at least one example (LCF Survey) the survey data is not even used “directly”. The problem here is that in disaggregating the data, “the estimates increase in uncertainty”. The Overview seeks to reassure that the alternative used is satisfactory, but has to acknowledge that “there is also a drawback due to imperfect alignment between some estimation dates and survey dates, though the effect of this is typically quite small” (HMRC Methodology Note: Overview, para., 4, p.4-5). All this is no doubt true but it adds nothing to the accuracy of the output, nor does it inform, elucidate or add to our knowledge of the facts in any constructive or substantive way. Each adjustment, each alignment, each proxy, each indirect application, each assumption takes us further and further from reality.
In the GERS methodology much is made of adjusting the data to fit the accrual basis of accounting, because;
“accruals accounting reflects a more accurate picture of when revenue is due and spending occurs than the more volatile alternative of cash, which, for example, records when bills are settled rather than when the expenditure occurs” (GERS Methodology Paper p.3).
It is perhaps unkind to suggest that this reveals a rather quaint and unconsidered formal confidence in the somewhat antiquated metaphysics of financial accountancy theory, but the challenge to comprehension does not end there. The GERs decision on this policy does not appear to sit happily with this observation in the HMRC Methodology Overview:
“Several taxes are normally reported on the basis of the accrual of tax liability. Where possible the estimates here have been adjusted to fit to the receipts profile” (HMRC Methodology Note: Overview, para., 2, p.4).
At this point a fair minded citizen attempting to understand the meaning of this exercise may well decide the effort is wasted, but this is not the end of the problems. Annex B (HMRC UK Tax Recepts Methodology) reveals under the un-nerving heading “Data Quality” that recondite adjustments are made to reconcile approaches of HMRC with the NAO (who audit, it seems on a cash basis):
“At the end of each year the statistics are reviewed against the audited annual accounts and adjustments are made to bring the statistics in line with the cash based account” (HMRC Methodology Note: Annex B, para., 140, p.61).
Not content with confusing the reader (or themsleves?) about cash or accrual accounting, HMRC note that “data is not available for each year” (para., 4, p.5) and then provides a whole Annexe A to explain why they are using the wrong year to cover the gap. We are then informed that the sum of the four nations in the UK also aggregates to a figure that turns out to be less than the UK total revenues because of “non-identification, incomplete data or non-UK activity”, which however is nowhere explained or reconciled. It is just a fact. What we are supposed to make of this is not explained.
When we turn to the detail of HMRC treatment of major sources of tax revenue we enter a world that is completely impenetrable. Corporation Tax may be disaggregated on the basis of geographical location and employment numbers, or the location of registered office. Of course such wildly different bases could produce radically different results, and in many cases registered office may have been established by the business concern for long-lost historical reasons or reasons completely unconnected to the underlying economics of the enterprise. What this does offer is the opportunity to move the results in large aggregations in any direction required, simply by choosing betweeen one of two diametrically different alternatives.
We are not told how this basis of disaggregation is selected for major companies (BP, RBS, Diageo just for example). Diageo is one of the largest whisky distillers, producing very large profits, foreign exchange and balance of payments benefits to the UK Treasury; much of it from whisky. Its Registered Office is in London. The point here is not how Diageo has in fact been treated by GERS but the fact that we have no way of knowing the basis of the selection process case-by-case from the information supplied. If we add all the possible adjustments together we could envisage producing results that could mean just about anything at all.
I began by writing that the methodology was more important than the figures, and without a sound methodology we have precisely nothing. We have nothing. We cannot expect that GERS or HMRC would claim their efforts were in vain, but the weasel-worded explanation offered for this slapdash method of disaggregation tells us something different. Their efforts are in vain.
The real criticism of this failure is not to be found in the pointless politcal debate GERS will spawn among politicians. The real criticism is that Scotland has been allowed to progress throughout the 20th century and into the 21st century without even the adequate collection of economic data to allow for the prudent, wise and progressive government of the country, or an understanding of the problems created by this gross want of knowledge. This is an appalling indictment of the Government of Scotland within the Union. The GERS and HMRC disaggregation methodology would not pass muster in the ‘hard sciences’, but while this may set too demanding a standard, the situation is much worse. The disaggregation method would not pass muster in a business, big or small.
Notice that I do not claim that it matters whether the knowledge provided by the disaggregation method reveals positive or negative information about the prospects for the Scottish economy; such a conclusion misses the point. We can do nothing about remedies until we understand the facts, and we do not possess “the facts”. For generations in Scotland we have not even bothered to gather the facts; and Westminster has been casually, wantonly indifferent to the need. We are therefore in no position to understand the economy of our own country. This is a fundamental failure of administration and government.
We are now attempting to recover from the intellectual and political failure that has extended to Scotland’s great and real cost for over a century. This is a fundamental failure of basic good governance of Scotland that rests, ultimately with Westminster (and the ill-informed, ill-equipped Scots representatives there over many, many decades). This is the matter that should be exercising the minds of politicians, and not whatever the current methodologically unsatisfactory GERS figures purport to prove. They prove nothing, save the demonstrable want of care for Scotland that the inadequacy of the “methodology” amply endorses. The disaggregation of data for the ‘four nations’ that everyone claims to be concerned about should be established at source; by which I mean at the point the data is initially captured. Scotland deserves nothing less: the GERS methodology standard, and the current UK collection of the data is not good enough for Scotland. GERS therefore invites us to debate the wrong issues, for the wrong reasons.