On Scotland, data, data denial and GERS

Posted on

I continued to find myself in the middle of a Twitter and blogging storm yesterday as a result of my comments on the quality of the data in what is called GERS (Government Expenditure and Revenue Scotland) made on this blog over the last couple of days, here and here. Much of the commentary comes from the usual Twitter gutter and makes me think Owen Jones is right to get a life by leaving social media behind. Some has come from, or is about, people who I had not until Tuesday heard of. One is called Wings over Scotland, who appears universally hated by unionists. I have not bothered to find out why. Another is a person called Kevin Hague, who appears to enjoy much the same status with nationalists, who appear all too keen to mention the fact that he makes his living selling dog food, as if that disqualifies him from comment: that I would not agree with. Kevin Hague does however have an unusual style of argument and a surprising following. He has published a blog under the title Richard Murphy, GERS Denier which seems to require somecomment before I then explore bigger issues.

GERS Denier

It’s clear that anyone who questions the quality of data in GERS will be on the wrong side of Mr Hague. It is, apparently, sacrosanct in his opinion. I have already made my opinion clear on that. I am not going to reiterate it. I am, however, apparently quite wrong to question its status according to Mr Hague for three reasons. The first is that it’s right because it says it’s right: he quotes its own stated confidence levels as proof of that. But this is not an argument. No one is right because they are confident that they are.

Then he says I should not question GERS because two academic macroeconomists he has spoken to agree that its right. Professor Ronald McDonald comes to this view because, apparently, GERS is produced in accordance with approved best practice. That provides me with no confidence at all. I questioned the quality of the data, and not what was done with it. Doing good things to bad data in accordance with the best statistical standards in the world still results in bad data. My point has not been addressed by Professor McDonald as a result. I suggested the GIGO (garbage in, garbage out) principle applied here and that issue has not been addressed.

And then Hague quotes a Professor Angus Armstrong who apparently says:

All economic statistics involve sampling and estimates. But when the UK Statistics Authority designate figures as ‘National Statistics’ that’s hugely significant. This is a kite-mark showing they meet international statistical standards. Anybody who says these figures are “easily rigged” or “nonsense data” frankly doesn’t deserve to be taken seriously. The people who work to create these statistics are honest, hard-working and dedicated public servants who aren’t allowed to answer back to defend themselves. Anyone who questions our national statisticians’ honesty and integrity should take a hard look at themselves.

But then note that Armstrong was apparently previously Head of Macroeconomic Analysis at HM Treasury. That hardly suggests impartiality when they were the people who I was criticising for failing to collect the data that Scotland needs and for providing what may be some pretty poor estimates in their place. What the stated opinion actually represents is a very typical view from the Westminster Treasury that suggests that they know what’s right on all occasions and no one has the right to question them. I’d remind anyone that attitudes like that got us to 2008. I’d also suggest that this quote takes us back to Hague’s first argument: the variation is that the data is now right because those preparing it say it is. With respect, I spent many years as an auditor and the whole premise of that profession is that such claims cannot be accepted at face value. I do not then accept Professor Armstrong’s argument: its whole foundation is unjustifiable.

Third, it seems I am wrong because the SNP has used the data. But that was not a point I made, and won’t. If the SNP have got this wrong, then it’s time they put this right in my opinion. In that regard I noticed a Common Weal article by Margaret Cuthbert, a noted commentator on the issue as I now understand it, that said:

It is becoming a trademark of Scottish Government reports that important statements are not backed up by relevant data or by references to where relevant data can be found.

GERS was one of her targets. Another was data on Scottish imports and exports. I’d happily support her argument that much better data is needed. If that implies there has been apparent inaction by successive Scottish governments to achieve this, so be it: I will address why that might be the case below. But what none of this implies is that I am a GERS denier. It does say I’d like GERS to be useful. So let me move on to discuss why data is not available,  data related issues and how progress could be made.

Why data is not available

Let’s be clear, what we measure matters. As importantly, what we do not measure also matters. Both indicate what we think important. Both indicate what we intend to manage. Both suggest how we intend to manage. And such decisions are not free from bias: all decisions do reflect bias because we always come to them with our own prejudices. This point is vital in this debate. As GERS shows, Scotland has almost no real data on its economy. Whether its tax income, savings, investment, imports, exports and much else, what I am saying (and what it seems Margaret Cuthbert is saying) is that there is simply no data available for Scotland. And that means that the decision has been taken by those who might collect this data – which for these purposes would all seem to be London based agencies – not to collect it.

I stress that this point is vital. What those who might have collected this data have said as a result is that Scotland does not matter. And that nor does its economy. They don’t think Scotland is important enough to have reliable data prepared for it. And they very clearly imply as a result that they do not think that the Scottish economy should be managed independently. They are sating that these issues don’t think matter. So they have not allocated resources to them. They have supplied some extrapolations instead. There is no way they know if those extrapolations are appropriate: the data does not exist to test that. The result is that the Scottish economic debate is denied the data it needs because of a decision, or a whole series of decisions, in London, which was my whole initial argument. This is precisely why Angus Armstrong is wrong, for example. London decided Scotland did not need statistics so they did not get them.

Why this has not been raised before I do not know, but I can speculate. When Labour was in office in Scotland and London I strongly suspect there was little incentive to change the arrangement. That’s not been true under SNP governments. But I suspect that if they have asked for better data they came straight up against the type of argument already noted: you’ve got data of a high standard and we’re not going to collect anything else for you. And the Scottish government has not got the power to change that at present. I stress, I don’t know this has happened. I hope it has. But the key point is, if better data is not available that is not the SNP’s fault: it is London’s because they refuse Scotland the right to collect it because the agencies that could do so are not devolved. That’s why I think Margaret Cuthbert’s criticism of the SNP may be wrong, but I stress I am speculating and the real issue is not what might have been said: the core issue is that London decided not to collect data so that Scotland could manage its own economy because it did not think that important. But it clearly is.

Data measurement

What data is needed then? I teach my students (yes, I teach this stuff unlike most economists) that decision useful information is:

  • Relevant
  • Reliable
  • Complete (i.e. it covers the whole population at least in principle)
  • Comprehensive (i.e. the subset of data variables being tested is available for all within the population)
  • Comparable (over time, or with other sources)
  • Comprehensible (does it make sense?)

To go back to GERS, the data is not relevant to Scotland because it is an abstract of UK data: it is not Scottish data. It is not reliable because it is not clear that the estimates made meet that criteria (although others disagree, I note). It is not complete because the Scottish population (of taxpayers, company with Scottish profits, sales, etc etc etc) is not defined and not used as a basis for the work. In that case the data is not comprehensive. Nor is it comparable because there is nothing to compare it with, barring itself, on which grounds alone it does work. And as a result it is in my opinion not comprehensible because it is on somewhere that may, quite simply, not be Scotland. For those reasons I fail it.

I stress though that’s not because GERS includes estimates. They say it does. No one denies it. And it is true that all financial measures include estimates. They have to. I suggest it fails because there comes a point where reality fades and estimates take over. That’s my whole criticism of GERS. I reiterate my example hinted at yesterday. Suppose a supermarket presented its accounts and said that the figure for income was based on a random survey of households in which they asked those householders how much they’d spent at the shop in question over a short period. Based on the responses the supermarket had then extrapolated the spend to the whole population and for the year as a whole and said this was its income instead of using the records it had maintained at its tills. What might the reasonable response be? I think we can be sure that the auditors would say this was unacceptable and that the claimed income was not supported by economic fact and that there was a failure of the reported accounts to reflect the reality of the business. There was better data available. And so there is in Scotland, but there’s a refusal to collect it.

Let me offer another example. Suppose that to find out the economic performance of Rumania we took that for the EU as a whole and then apportioned it to that country. Would anyone think that reliable even if some adjusting factors were taken into account when we could instead use the data that Romania itself collects? Of course they would not. But that’s what’s happening in Scotland.

So use estimates by all means. But you can’t use them for almost everything and then claim the result is meaningful. It is not. And I stress, I offer both examples to reiterate my point that suggesting the Scottish national accounts can be prepared on the basis of extrapolation of what is already estimated data for the UK as a whole (which may not always be that reliable in itself) is wholly inappropriate when Scottish data could be measured. Worse, it could be an extremely bad basis for informing such an important decision as whether to become independent or not, whichever way you might be inclined in that debate.

In that case what I suggest is that Scotland start demanding better data that is sourced in Scotland, whether or not there is to be independence. It needs it. So too does Wales. And the other regions of the UK, including Northern Ireland. London may like to think this is one country with one government but that is very obviously untrue. In which case there are a lot of people who should be joining in with this demand.

It is a demand that will meet the contemptuous response of the type Kevin Hague notes has been aimed at me. But I am used to being told such things. Remember, I was told country-by-country reporting was impossible, too costly and undeliverable. But it is happening.

And I was told that we would never get automatic information exchange from tax havens as recently as 2009. But that too is now happening.

Likewise tax havens repeatedly said they were beneficial to the world economy and I did not know what I was talking about. Times have changed.

I’ve fought battles for data because my instincts that we did not know the truth at cost to society at large were right, and change has resulted. The battle for economic data for Scotland, Wales, Northern Ireland and the regions of the UK is based on the same instinct. And remember the absence of this data is not a necessity. It’s a decision made by London that the view of London should prevail and that the only data available should support that view. And that’s no longer acceptable. What is more, it is a massive part of the reason for the alienation of people from government in this country. And their sense of alienation may be justified: the data proves London thinks they do not count. They do. And the data has to change to prove it.