July 29, 2018 by Paul Goldsmith
In today’s febrile political atmosphere, in which information that doesn’t fit a person’s bias is immediately discarded, it is refreshing to have come across a working paper from a group of Remain- voting (yes, Remain-voting, although one of them now supports Brexit) Cambridge University economists entitled ‘How the Economics Profession got it wrong on Brexit’.
The paper (found by clicking here) explains how and why the models used to make predictions during the 2016 referendum campaign, and afterwards, of the impact of Brexit have been so wrong.
It showcases a litany of unlikely and distorting assumptions behind the models, which seem to have been carefully calibrated to get the result that matched the organisation’s agenda. Given most organisations that have made predictions are pro-Remain (The Treasury, the OECD, the Scottish Government), it is unfortunate to say the least that from an economic theory and statistical point of view they worked so hard to ensure they got the results they wanted. It’s unfortunate because we do actually need trustworthy forecasts. Ken Coutts, Jordan Buchanan and Graham Gudgin show us what has been wrong with what has been put out (and how they were reported by the media) and how this can be fixed.
Here is a quick summary of what they found:
1) All the forecasts assume that there is a correlation between the openness to free trade and productivity. They therefore assumed productivity would fall if the UK had some trade barriers with the EU. But the authors pointed out that whilst this correlation has been proven true with developing countries it hasn’t been proven true with developed countries. Several organisations used this link without much questioning.
2) The economic models all assume that if tariffs are put on Britain’s products then the consequent increase in price will result in a more than proportionate decrease in demand for UK products (this is known as ‘price elasticity of demand’). But the authors point out that the goods and services we export to the EU (such as services) tend to have low price elasticity of demand, so demand for UK products will fall nowhere near as much as the forecasters assumed. What’s more, despite prediction a fall in Sterling in their list of ‘immediate impacts’, many forecasters then didn’t make allowance for a long-term lower level of the Sterling exchange rate, which could offset much, or all of any tariffs imposed by the EU under WTO rules.
3) The economic forecasters used the ‘WTO rules’ as their worst case scenario. In this scenario there would be tariffs (taxes on imports) on UK goods and services, but also ‘Non-tariff barriers’. These are rules and regulations such as product and service standards that are currently harmonised in the EU under the Single Market Act. The models assume that these would block UK services in particular from accessing the EU. But, uniquely, under this worst case scenario the UK would be entering WTO rules from a position of being fully compliant with ALL EU rules and regulations, so would still have easier access to the EU single market than any other country in their situation.
Many of the models assume an almost full reversal of all benefits of trade if the UK leaves without a deal, yet the UK starts off from a position of full compliance to be able to trade. This, by the way, is why a ‘Canada-style’ free trade deal is not as much of a disaster as the models are predicting, because they assume that the UK services would not have access to the EU (as many of Canada’s don’t as they don’t meet EU rules and regulations). At the moment of leaving, the UK will have full alignment with the EU’s rules and regulations, and can then choose those they keep, and thus the level of access to the Single Market they would have. This important difference appears to be missing from the doomsayers’ forecasts.
Given the regulatory alignment and no tariffs for goods of a Canada style free-trade deal, it is a bit odd for the Treasury to have predicted a reduction in demand for exports of 33%. The Cambridge economists accepted there would be admin costs of around 2% for border control and 5% for reporting ‘third-country content’ (raw materials that came from outside the UK and EU), but pointing out that even if this was fully passed onto prices and demand for exports was reduced by twice the amount of this price rise the fall in exports would be only 14%. They think this is because the Treasury assumed a ‘knock-on effect’ from loss of trade on productivity, which as point 1) above suggests is unlikely.
4) The forecasts confuse GDP (the value of a country’s output) with standard of living (GDP per head of population). So, they assume GDP will fall, for reasons connected to their inaccurate assumptions above and automatically assume that GDP per head will fall. But all of the models assume that one of the reasons that GDP will fall is that immigration will fall, which by definition mean that the population will fall, which by further definition means it is quite possible for GDP to fall without a fall in GDP per head, and thus living standards. The Cambridge University economists argue that many of the models and papers did mention this, but it got reported differently, possibly because it was briefed to the journalists differently.
5) Many of the models assumed almost immediate economic meltdown after a Leave vote in June 2016, which we now know didn’t happen and hasn’t happened. They predicted a large fall in the pound, which did happen, but also a loss of 500,000 jobs and a serious and prolonged recession and a massive fall in share prices. But the models didn’t include the likelihood that a fall in sterling would increase exports (a cheaper pound led to more demand for UK exports) and a rise in share prices, because the FTSE-100 companies mostly have earnings from abroad, which became worth more in sterling. The models also didn’t take into account that the Bank of England would respond with an immediate cut in interest rates, despite that move helping to stabilise the effect of post-Referendum uncertainty. In fact, both the Treasury and the OECD has used a forecasting model called NiGEM, which holds monetary policies and exchange rates constant. This is unrealistic.
6) The Treasury, in its famed ‘£4300 loss for each household’ report, used a ‘gravity model’ to calculate the amount of additional trade with the EU resulting from the UK’s membership. The gravity model predicts trade flows between two units (in this case the units are the UK and the EU) based upon the economic size and distance and other factors such as a common language between those two units.
Yet the Cambridge economists who produced this report argue that because the UK is the only EU state, other than Malta, which exports more to non-EU countries, UK exports to the EU have often been well below those predicted by the Gravity model methods used. The Treasury model predicted an average impact of EU membership of around 115% extra trade, but the Cambridge economists say it is only around 20-25%. Most of that is encompassed by trade in goods. With services, common language and size of economy are far more important than distance, and some research has thus found that EU membership delivered no significant trade uplift in services.
Intriguingly, the Cambridge economists have become aware of an internal Treasury paper from 2005 with far smaller estimates of the impact of EU membership on intra-EU trade. This paper noted in particular that the impact of EU membership was much smaller for the UK than for other EU members. Somehow, despite the official responsible for the 2016 Treasury report, Sir Dave Ramsden, having been at the Treasury in 2005, the findings of that report were ignored in 2016.
7) The Cambridge economists didn’t just look at forecasts from before the referendum. They also reviewed the Mayor of London’s report – ‘Preparing for Brexit’, published in January 2018. They preferred the economic model, as it used economic relationships estimated using actual historical data rather than the theory and assumptions used by most other Brexit studies. But there were some strange omissions in both the scenarios they analysed and the way the results were reported.
Despite the UK Government indicating they preferred a 2-year transition period followed by a Canada-style trade deal augmented by financial services, the Mayor of London’s report didn’t model this scenario. The media decided to report only the very worst case scenario modelled, with no deal and no transition period, even though that transition is desired by the UK and had appeared to have been accepted by the EU. They then did what is reported in point 4) above, which was to report a fall in GVA (a measure of living standards, without reporting a fall in population that they had in the same table. This meant for instance that in the case of the worst case scenario reported, GVA would fall by 3.0% compared to remaining in the EU, and population would fall by 2.2%, which is a reduction of per head GVA of 0.8%. What got reported in the media was the 3.0% figure.
The falls in per head GVA across the scenarios were well within the margin of error on the model for such a long period (up to 2030). They suggest the loss of output and jobs would be the result of lower migration, with little to no impact on indigenous workers. Did this get reported? Nope. In fact, when the impact on London alone is stripped out, in all cases population falls by more than GVA, meaning a net rise in living standards in London whatever the scenarios. Again, not reported.
8) Finally, the Cambridge economists give particularly short shrift to a report published by the SNP in Scotland. This was a nakedly political report, because for the SNP’s eventual aim of an independent Scotland within the EU to be likely, the UK would have to remain in the European Single a Market and a Customs Union replicating the terms of the current EU Customs Union. Scotland also does four times as much trade with the UK as with the other EU countries, so the prospect of tariffs and regulatory barriers between an independent Scotland in the EU and the UK outside would effectively rule out future independence for Scotland.
So it is no surprise that the SNP’s report used the same model and assumptions as the Treasury has used in 2016. It ‘predicts’ that the UK would lose 50% of its EU export market after Brexit. This despite the average tariff in the worst case scenario being 5% and Sterling falling by around 15%, making exports cheaper and more profitable.
In conclusion, the extent to which economists from so many organisations got their predictions wrong has further damaged confidence in economists’ contribution to public debate.
These Remain-supporting economists (please remember they voted Remain when you talk about this blog), note that the flaws that they have identified in the forecasts on Brexit all pointed in the pessimistic direction on Brexit that most academic and economists and the establishment politicians lean ideologically. The Treasury then refused to discuss their approach until dragged in front of Parliament, which is unacceptable in our open democracy.
Even when the most appropriate assumptions and models were used, which was in the Mayor of London’s report, the most likely scenario wasn’t modelled, and only the unrealistic worst case scenario was reported in the media with the numbers not representing per head living standards.
The biggest problem with this is that the economic impact assessments that have been created, which are vital for the UK’s negotiation position, are being ignored and given no credibility by many politicians.
The economic forecasting profession, and politicians’ and the media’s use of it, needs to change.