The Centre for European Reform has published a series of articles in recent years estimating the impact of Brexit on GDP, trade and investment. These use a so-called doppelgänger methodology which selects a group of countries which had a similar economic performance to the UK prior to Brexit and then compares their post-Brexit performance with that of the UK. The most recent of these ‘What can we know about the cost of Brexit so far?’ was published on 9 June 2022. It estimated the impact of Brexit on the UK economy as a 5.2% reduction in GDP, a 13.7% fall in investment, and 13.6% fall in trade compared to a “modelled ‘doppelgänger’ UK that did not leave the EU”.
It was given major and uncritical coverage in the ITN television news bulletin the following day with its conclusion that Brexit had already had a major negative impact on the UK economy repeated as though it were a fact. The results have been frequently cited by other media outlets and commentators in a similar way, including the Economist magazine.
Analysis by Policy Exchange shows that the CER research has failed in its effort to prove that Brexit has hit the UK’s economic performance and its doppelgänger methodology is fatally flawed.
– The research uses different sets of comparator countries for different variables (and different to its previous report) rather than conducting a comparison with a consistent set of countries.
– It assumes that any divergence between the UK and comparators is due to Brexit, whereas self-evidently it could and probably does come from a wide range of other reasons, notably the cyclical position of different countries at different times.
– Crucially, the massive disruption caused to all economies by the Covid pandemic, and differences in measurement methodology between the UK and many other countries, mean that conclusions based on the pandemic years — nearly half of the CER’s base period — cannot be plausibly substantiated.
A more plausible comparison, between the UK and the other G7 countries, shows no visible impact from Brexit at all. GDP in current prices in the UK fell a little behind the G6 in 2018 but then recovered by 2019. The Covid-induced decline in 2020 was a little deeper than the G6 and, as expected from its earlier ending of lockdowns, the UK’s recovery has been a little faster.
It will take many years to disentangle the effect of Brexit from all the other influences on the UK economy over these years. It can’t be short-circuited by creating an implausible and flawed methodology to draw premature conclusions.
The report’s author, John Springford, described the method as ‘an algorithm [which] compares data on the UK’s economic performance with that of 22 other advanced economies. It selects a subset of those countries and allocates each a weighting, to create a basket of countries that minimises the difference between their data and that of the UK. The algorithm matches the growth rate of real GDP, GFCF [Gross fixed capital formation], services and goods trade, from Q1 2009, as well as the inflation rate, industrial production as a share of GDP, average years of schooling and other variables. We can then compare the performance of this doppelgänger against that of the UK.’
The doppelgänger group for GDP is made up of the United States (31per cent), Germany (15 per cent), New Zealand (14 per cent), Norway (8 per cent) and Australia (5 per cent). The remaining countries make up less than 5 per cent of the doppelgänger each. These countries were selected using quarterly data from the pre-Brexit period 2009Q1 to 2016Q2 with an attempt to match broad economic structures. The UK’s performance is then compared with that of the doppelgänger group from 2016Q3 to 2021Q4. For trade and investment other groups and periods are used.
The result is shown in the CER’s chart 1 (reproduced above) comparing the UK with the CER’s doppelgänger groups. For GDP the UK is 2.9% behind the comparators by the end of 2019 and 5.2% by the end of 2021. The method also generates large negative impacts for Brexit on investment and trade in goods but not trade in services.
This methodology has many problems which mean that its results cannot be taken at face value as representing a reliable counterfactual.
The comparator countries keep changing
The current set of comparator countries is not the same as in previous CER doppelgänger exercises. For instance in its early calculations starting in June 2018 the comparator countries were mainly USA, Germany and Luxemburg but also Iceland and Greece, Similar exercises by academics have put a heavy weight on New Zealand, Japan and Hungary.
The June 2022 CER report also uses different sets of comparators for GDP, investment and trade. Using different comparators for individual variables is rather unsatisfactory and we should note in some cases inappropriate. For instance, matching the UK with Iceland and Greece for trade is hardly comparing like with like.
The report says that, in the past, comparator countries were dropped because some countries were subject to economic shocks unrelated to Brexit. It cites the key example of the USA where major fiscal expansions under Trump and Biden accelerated economic growth (though this very important fact has not stopped it using the USA as a comparator for GDP in the CER report even so). This was a point made in a critique of the earlier CER estimates. Although the growth of the UK economy is correlated with the USA over the long-run, there is no guaranteed correlation over the short-run. Using the USA as a benchmark for post-2016 growth was misleading due to the fiscal expansion generated by the Trump tax-cuts. The IMF fiscal monitor shows that the US fiscal stance loosened by 3% between 2015 and 2019 while in the UK there was a tightening of 1%, a large enough difference to affect economic growth. The point about the USA is acknowledged in the current CER report but it does not appear to be factored into the report.
Divergence can be for many reasons, not only Brexit
The fact that different countries are subject to different shocks at different times, and are at different points in the economic cycle, is a general weakness in the doppelgänger approach. Just because economic growth in two countries has been closely correlated in the past does not mean that they will continue to be correlated in future. This depends on a range of policy changes and other circumstances. The CER interpret any divergence between the UK and the comparators after 2016 or 2020 as being due to Brexit but, as we will show, this need not be the case.
This point is illustrated in the chart above which shows growth in real GDP for the UK relative to the rest of the G7 group of major economies from 1990. Initially the UK performed poorly during a period of sterling overvaluation as the UK shadowed the Deutschmark in an attempt to prepare to join an EU single currency. Following the Black Wednesday depreciation of 1992, the UK grew faster than the rest of the G7 partly due to more competitive currency but also because debt levels from a largely deregulated banking sector once again soared to record levels. After the subsequent banking crisis of 2008, the weakened UK economy grew more slowly than the G6 until 2011 when faster growth resumed until 2016.
Accordingly, the CER’s comparator period (2009-16) largely happens to coincide with a cyclical upturn in the UK economy relative to the G6 over 2011-16. It is implausible, Brexit or no Brexit, that continued relative upswing would have continued after 2016. If it had, and the UK continued its trend outperformance relative to the G6, the UK would have reached an unprecedented level of 6% above the G6 average by the end of 2019. This is very unlikely and indeed what we saw was what was to be expected: a cyclical slowdown between 2016 and 2019 which kept the UK within the broad GDP range established since 2000.
So, the CER’s conclusion that Brexit caused a 2.9% loss of GDP between 2016 and 2020 is based on an assumption that without Brexit the UK would have continued to remain at its 2016 peak relative to the G6, with no cyclical downturn. The CER attribute the downturn that occurred to Brexit; this is simplistic. The UK economy is inherently cyclical for reasons of fiscal and monetary policy, expectations of future profits, and a wide range of other factors. To assume that a cyclical downturn is largely, or even wholly, due to a single factor — Brexit — would require clear evidence to be presented. The CER present none other than the coincidence in timing
The covid pandemic makes comparisons since 2019 essentially meaningless
The covid pandemic massively disrupted all economies from the beginning of 2020, a period which represents nearly half of the CER’s comparator timeline. Countries used different levels and timings of lockdown with varying impacts on their economies and they also used different methods to measure GDP during these unusual circumstances. The measurement of GDP differences before the pandemic is controversial enough, but the CER’s attempt to extend its method into the pandemic and its immediate aftermath is hopelessly flawed. The pandemic massively distorted normal growth trends and performance, very likely to a much greater extent than any changes resulting from leaving the EU single market and customs union (and in any case there is no real way of disentangling the impacts).
Not only are the figures for all the comparator economies massively distorted, there are also specific measurement distortions between the UK and many others. The differences in the impact of Covid on GDP between the UK and other advanced economies have been analysed in detail by the ONS in a February 2021 report. The ONS make two significant points. Firstly, the UK has more of its economic activity in services which require personal contact. These include hotel and restaurant sectors, plus cultural and sports activities. Secondly there is an important difference in the way output is measured in public sectors, including health and education. The ONS has fully applied international standards by using direct output indicators such as the number of operations conducted by the NHS or GP consultations and the number of children in school. Many of these indicators of output which work well during normal periods showed falls in output during the pandemic because normal medical procedures and teaching were disrupted, even though the staff were fully employed on covid-related treatments and on distance learning in schools and colleges. Other countries did not use this methodology and simply used expenditure data deflated for price changes, which showed a much lower loss of output due to covid.
International comparisons thus show a much larger loss of real output in 2020 for the UK, largely though not entirely recovering during 2021. This is principally a measurement effect not an accurate account of output during the pandemic. The CER assume that, once lockdowns ceased and economic activity began to recover, that there would be no further distortion in measurement and hence any further differences will be due to Brexit alone. This can only be regarded as a heroic assumption given what we know about ongoing disruption in all economies and the backlogs in public services almost everywhere.
These factors arguably make the conclusions drawn by the CER, in their updated report covering the pandemic period, essentially meaningless.
A more plausible comparison
It is possible to construct a more plausible version of the CER’s methodology which delivers a quite different result in assessing the impact of Brexit up to and including 2021. We use the ONS recommendation that data on current price GDP be used in place of real GDP since the former largely avoids the differences in international statistical practice in measuring public sector output. We avoid the use of arbitrary comparators including inappropriate countries like Iceland or Greece — instead we continue our comparison of the UK with its G6 major competitors. We also use per capita GDP in place of GDP alone, since this is a more direct measure of living standards, and use purchasing power parity (PPP) to remove the influence of price differences. The use of per capita GDP also ensures that the figures are not distorted purely by population changes, i.e. movements of migrants in and out of the UK and comparators.
The chart below shows that starting in 2015, current price GDP per head in the UK (nominal GDP) fell a little behind the G6 in 2018 but then recovered by 2019. The Covid-induced decline in 2020 was a little greater than the G6 and, as expected from its earlier ending of lockdowns, the UK’s recovery has been a little faster. There is no sign here of any impact of Brexit.
The individual G6 countries are shown in the second chart below. The UK ends up ahead of Canada and Japan but a little behind the USA, France, Germany and Italy.
For completeness the UK to G6 data is shown at (the mismeasured) constant prices as in the CER report in the third chart below.
Source of data: OECD
Source of data: OECD
Source of data: OECD
The trade impacts
The CER undertook a similar exercise for trade performance, but with a different set of comparator countries. Norway and New Zealand are dropped and Italy, Iceland and Greece are added. The main conclusion is that trade in goods (exports plus imports) is reduced by 13.6% and that this is largely due to Brexit. A parallel exercise for trade in services uses yet another set of comparators, this time dropping the USA and adding Spain and Luxembourg. The conclusion in this case is that the UK outperforms the comparator countries but only because of the inclusion of, and high weight given to, Australia.
The trend for trade in goods is more complex than is described in the CER report and its conclusions are simplistic. Firstly, the CER report shows trade in goods broadly following the trend for the comparators until the beginning of 2021 when the TCA trade agreement came into force, and then lagging behind during 2021. What the report does not say is that the fall in trade has until the most recent quarter been largely on imports despite the lack of either tariffs or full customs checks on UK imports from the EU. As for exports, it is exports to non-EU destinations which have fallen most, with only a slight fall in exports to the EU, as the chart below shows.
The data in the chart below is from the ONS and is the volume of trade excluding oil and erratics with an ONS adjustment for changes in recording methods. Exports in 2022Q1 are 8.4% below the average of the three years before the pandemic. But within this figure exports to the EU are down by only 1.2% while exports to non-EU destinations are down by 14.2%. It is hard to see how this pattern of exports can be interpreted to conclude that Brexit is a prime cause. The fall in exports to non-EU countries suggests that the influence of Covid is still very much with us, and in this context, exports to the EU appear to be holding up well. Moreover, because the CER’s analysis for imports runs only up to 2021Q4, it misses the recovery in 2022Q1 which brings total imports back up to pre-Covid and pre-Brexit trends. In this last quarter the faster recovery is in imports from non-EU countries.
Source of data: ONS
A similar method is also applied to investment. The countries that make up this doppelgänger group are different again: the US (48 per cent), New Zealand (15 per cent), Iceland and Denmark (9 per cent each) and Japan (8 per cent) although the report says that it makes little difference if the USA is dropped. The data is described as gross fixed capital formation. In the national accounts GFCF includes the investment of companies (including in housing and other property), but also households (in housing) and government. There are a range of influences on household and government investment, so an attempt to assess the impact of Brexit would best focus on company investment, but it is unclear whether this study does this.
The CER analysis shows UK investment falling below that of the comparators in the 2017-19 period with little further deterioration during and after the pandemic (see CER chart 1 above). The slowdown in company investment after 2016 is sometimes cited as reflecting companies’ expectations of slower growth in a post-Brexit world. There may be something in this (though it could at least as plausibly be a reaction to the general chaos and misgovernance of the Theresa May period), but here too cyclical factors are probably more important: a longer time-series of data on company investment (see chart below) shows that by 2017 investment had recovered its 1.3% per annum long-term trend following the large fall during the banking crisis of 2008. Since investment overshot the trendline in 2016 it was due for a slowdown and eventual decline. Only in exceptional circumstances would it have continued to grow at an above-trend rate. There is little reason to ascribe the observed slowdown to Brexit or any other exceptional circumstance. Since the pandemic led to a sudden drop in investment in 2020 without much recovery in 2021, it is now very difficult to separate any impact of Brexit from that of Covid.
Source of data: ONS
The CER’s method of constructing a counterfactual or doppelgänger simply cannot be used to draw conclusions which can be justified economically. It will always be possible to construct a weighted series of GDP or other data for western nations which closely mimics the UK’s past performance over a limited period, but this in no way means that the matched performance will last forever or even for a limited period, because all economies are subject to a range of influences which affect them in different ways.
A more straightforward comparison with the major economies of the G7 reveals that there is no obvious Brexit-related lag in the UK’s economic performance measured by per capita GDP. In short, the UK economy has grown at much the same rate as the G7 average since the Brexit referendum in 2016.
The CER’s work appears to follow an unfortunate trend for analysts to use inapplicable techniques or inappropriate data points to reach the inevitable conclusion that Brexit has had a negative impact on the UK economy. The CER report follows a series of commentaries arguing that Brexit has reduced trade in goods. It is reasonable to expect that the primary impact of Brexit will show up in goods trade rather than services since the main change has been to impose customs barriers on UK:EU trade, but the fact that UK exports to the EU have held up well suggests that any wider impact on GDP is in fact likely to be small. More time and patience are needed to make an accurate assessment of the economic impact of Brexit and a doppelgänger approach cannot be used to short-circuit this process.
Dr. Graham Gudgin is Chief Economic Adviser to Policy Exchange. He is currently Honorary Research Associate at the Centre for Business Research (CBR) in the Judge Business School at the University of Cambridge.