According to the latest estimates from the Centre for European Reform (CER), the UK economy was 2.5% smaller in the fourth quarter of 2018 than it would have been had Britain voted to remain in the EU referendum. The Bank of England has recently published similar work by the independent MPC member Gertjan Vlieghe, which suggests that the level of GDP is around 2% (or £40bn) lower than it would otherwise have been.
How useful are these figures? In my view, not very. This is partly because there are many good reasons why the UK would probably have under-performed its peers over the last couple of years, regardless of the outcome of the referendum. In particular, the US economy has benefited from the substantial fiscal boost under President Trump. UK consumer spending was also already looking over-stretched.
At least as importantly, these numbers tell us very little about the longer-term impact of Brexit once the UK has actually left the EU. Indeed, even the initial hit to UK GDP may be at least partially reversed if, or when, uncertainty lifts. But let’s start by explaining what the CER has done.
The CER report uses a similar methodology to that presented in an academic paper published in November 2017 (and also borrowed since then by the investment bank UBS). In short, these studies use a computer algorithm to select a weighted combination of countries whose growth best matched that of the UK economy before the 2016 referendum.
The actual performance of the UK economy since the referendum is then compared to this control group, or synthetic ‘doppelganger’, and the difference taken as a proxy for the impact of the vote for Brexit. In other words, the performance of the doppelganger is assumed to the ‘counter-factual’ (or what would otherwise have happened) if the UK had voted to remain. The chart below shows the results from the latest CER paper.
Gertjan Vlieghe’s work is very similar – the chart below is taken from his recent speech. The widely reported 2% figure is the central estimate based on a control group of the other G7 countries, albeit within a wide range.
This approach is fine as far as it can go. The author of the CER report, John Springford, has explained his models clearly and published tests of their robustness. I also agree with the conclusion of this study (and many others) that the UK economy has grown more slowly, thus far, than it would otherwise have done. This is mainly due to the additional inflation prompted by sterling’s fall, which has undermined real incomes, and the heightened uncertainty, which has clearly held back business investment.
However, this should not be a surprise. People were warned ahead of the 2016 referendum that the initial economic impact of a vote to leave would probably be negative. The Remain campaign obviously made a lot of this. But it was also acknowledged by many Brexit-supporting economists, such as my former colleagues Roger Bootle and Gerard Lyons. Despite this, the UK still voted to leave.
I’d also argue that the latest CER estimate of 2.5% for the hit to GDP is too high. There are two fundamental problems with the ‘doppelganger’ method, especially over relatively short periods.
First, the results obviously depend on the choice of countries in the control group (and the weights on them). Indeed, there have already been some large changes in the specification of the CER model. These changes might make for a better fit, but they also illustrate the difficulty here. Initially, the CER model gave a suspiciously large weight of 23% to Hungary, which was widely questioned at the time. But the latest iteration still gives a relatively high weight to two tiny countries, namely Luxembourg (11%) and Iceland (10%). Interestingly, when the June 2018 version of the CER model was restricted to 22 advanced economies that were more obviously peers of the UK, the estimated hit then shrunk to 1.6%.
While this is probably getting closer to the mark, I still think it is too large. This is because of the second problem: the doppelganger approach assumes that all the under-performance of the UK economy is due to Brexit, rather than other unrelated factors affecting the UK and/or the control group. In fact, they may be some very good reasons why the UK would have slipped down the growth league table anyway, regardless of the outcome of the 2016 referendum.
In particular, the US economy has been benefiting from a large fiscal stimulus under President Trump (see the chart below, which also comes from the Vlieghe speech). This is crucial, because the US has a 50% weight in the latest iteration of the CER model, and accounts for more than half the GDP of the G7 control group excluding the UK. The euro-zone was probably overdue a period of catch up too, after several years of under-performance in the wake of the debt crisis there in 2011-12.
On the other side, there were also some good reasons to expect the UK to struggle to match its strong performance from 2013 to 2016. In particular, there had been an unsustainable consumer boom. It is worth remembering that the volume of retail sales rose by 4.7% in 2016, a growth rate that was never likely to be maintained.
Similar points also apply, by the way, to earlier testimony from Bank of England Governor Mark Carney, in May 2018, when he also suggested that the level of GDP was ‘up to 2%’ lower than otherwise as a result of the Brexit vote.
This estimate simply compared the actual path of UK GDP with that implied by the Bank’s own pre-referendum forecasts (with some further adjustments to take account of stronger-than-expected growth in the rest of the world and the additional monetary stimulus at home). In effect, this approach assumes that the Bank’s 2016 forecasts were sufficiently good to use as the starting point for counter-factual analysis more than two years later, which is obviously questionable.
For what it’s worth, my own estimate is that the fall in the pound and the additional uncertainty following the Brexit vote means that the level of GDP is now around 1.0-1.5% lower than it would otherwise been. The continued strength of the UK labour market (illustrated below) is one good reason to doubt that the impact has been much greater than this.
Of course, a GDP loss of 1.0-1.5% is still painful, and implies a sizeable hit to the public finances too. But there are three further points to bear in mind. First, this is still far smaller than the damage that many had predicted. In particular, HMT’s now infamous pre-referendum analysis had suggested that GDP would be between 3.6% and 6.0% lower than otherwise by now. Few people voting in 2016 could have been unaware of this.
Second, I also think that if the UK government had handled the negotiations better, the hit could have been less than 1%. For many Brexiteers, the counter factual should have been a smooth exit on 29th March transitioning quickly to a Canada ++ trade deal, rather than the political chaos we are seeing now.
Third, and at least as importantly, whether the initial hit has been 1%, 2%, or something a little bigger, it is wrong to assume that the impact on GDP can only get worse.
Here, Brexit pessimists typically say things like ‘the economy is already x% smaller and we haven’t even left yet‘, implying that the bulk of the costs are still to come. But by the same logic, the fact that we haven’t left yet surely also means it is too early to see the potential benefits of Brexit, notably any gains from independent trade and regulatory policies, and from reduced payments to the EU budget.
What’s more, even the initial hit could be partially reversed. For a start, as long as the ‘Project Fear’ scenarios are avoided, investment is likely to rebound as uncertainty lifts. Much of the ‘lost investment’ to date will be projects that have been postponed, rather than cancelled completely, and which could soon be revived.
More speculatively, sterling may recover some ground too. But even if it doesn’t, the vote to leave the EU may simply have brought forward a correction in the value of the pound that would have happened anyway, given the large imbalances in the economy (including a huge current account deficit). In these circumstances, many would regard a more competitive currency as a ‘good thing’. (For a longer discussion of the pros and cons of devaluing the pound see this debate between John Mills and myself, hosted by the IEA.)
Of course, the initial impact of the fall in the pound has been negative, due to the rise in the sterling cost of imports. But over time this could still be offset by the benefits as the volumes of both imports and exports adjust (the classic J-curve effect). It may just be that this process is another victim of Brexit uncertainty, which has deterred exporters from investing to take full advantage of a more competitive currency. This boost could therefore still be coming too.
The CER will continue to update its estimates regularly and it will be interesting to see whether these show the ‘cost’ of Brexit rising, or falling. With the euro-zone now stagnating, and the Trump boost likely to fade, it would not be a surprise to see the gap narrow in the coming quarters. But whatever the correct numbers are now, they can’t tell us much about where the UK economy will eventually end up.