Since early 2022 there has been a constant stream of media stories and studies claiming that UK exports have been badly hit by Brexit, with more hyperbolic sources claiming the impact has been a ‘disaster’. The volume of these claims, and the lack of critical evaluation most of them have faced, has meant that this supposed disastrous impact has become an accepted ‘fact’ in many circles.
But in reality, these claims are weakly grounded in the available data. They often feature selective use of dates, comparison countries and data series. Sometimes they also feature questionable modifications to the data and in some cases dispense with the data almost entirely and rely on what amounts to guesswork about where UK exports might have gone in the absence of Brexit. Some studies are careless about conflating Brexit impacts with other impacts such as from the pandemic. On top of all this, all such studies suffer from the massive problem that it is very early to be making such strong judgements on this topic. We have only two years’ worth of data, far less than needed for making solid statistical inferences, with the data also subject to a series of measurement changes that further blur the picture.
In this article we show that:
- While underlying UK goods export growth since 2020 looks relatively weak this is more due to a decline in exports to non-EU economies than exports to the EU. Despite some claims to the contrary, there is no evidence the drop in non-EU exports is a Brexit effect
- Looking across a wide range of possible ‘counterfactual’ series that could indicate where UK exports would have gone in the absence of Brexit suggests that UK exports to the EU may only be a few percent below the ‘expected’ level. This is a much better outcome than many critical studies of Brexit claimed would occur prior to the UK leaving the EU customs union and single market
- Looking at past UK export recoveries after global downturns suggests a delayed recovery is unremarkable. It may just be that the UK’s particular export structure and/or corporate strategies produce this kind of pattern with Brexit not an important factor
- If UK exports had indeed been badly hit by Brexit then this should be visible in the performance of UK manufacturing industry. But in fact, UK manufacturing has outperformed that of the other G7 countries since 2020
- UK services exports show no apparent impact from Brexit, in contrast to past claims that they would be hard hit. After a slump during the pandemic, their recovery has been at the top of the range of the recoveries seen among the G7 economies
- The OBR and others claim that Brexit has reduced the UK’s economic openness with a big negative impact on productivity. But the degree of ‘openness’ has at most declined by a very small amount which is not consistent with the claims on productivity. Moreover, the actual data do not show signs of lower productivity in the market sector of the economy – only in the public sector which has nothing to do with Brexit
UK goods exports – the broad trends
A glance at the headline series for UK exports since 2020 might make an observer wonder what all the fuss is about. In Q4 2022, the value of total UK exports of goods was 31% above the average level of 2018 (the best pre-Brexit comparison year due to pandemic effects in 2020 and stockpiling distortions in 2019) while exports in constant prices were 6% above 2018 levels and looked in line with their long-term trend.
However, UK exports in recent quarters have been flattered by two effects – first the rise in global energy prices which boosted the value of UK gas exports and re-exports, and second an unusually high level of precious metals exports. We should not ignore the current price series entirely as there may have been durable positive impacts here which will boost the profitability of UK exporters. But probably the best (and most traditional) measure of underlying export performance is the volume of UK exports excluding oil and erratics. This accounts for both energy price rises and precious metals exports and also removes oil and other erratic items such as precious stones, ships, and aircraft. In Q4 2022, this underlying export series was around 9% below its long-term trend line (Chart 1).
Source: ONS, author’s calculations
On the face of it, this suggests claims of a big impact from Brexit might be correct. But a closer look at the data suggests things are not so clear cut. If we break total exports down into sales to EU and non-EU markets we can see it is sales to non-EU markets that have been the main factor behind the weakness of overall UK exports since the end of 2020, not exports to the EU. UK exports to the EU in Q4 2022 were only slightly below their flat long-term trend while exports to non-EU countries were some 15% below their (strongly upward) long-term trend.
It is far from obvious how the weakness of non-EU exports can be related to Brexit. Kren and Lawless claim it is the result of Brexit ‘spillover effects’ such as higher costs or reduced availability of EU-produced components but produce no real evidence for this.
A more plausible explanation is that UK non-EU exports have been hit by some large sector specific shocks. It is notable in this regard that just a handful of sectors account for the great bulk of declines in underlying exports to non-EU economies from 2018-2022 – cars, mechanical machinery and works of art and jewellery. Of these, car exports have certainly been affected by global disruptions to the computer chip market; the fall in art and jewellery exports cannot, given the nature of the products involved, have much to do with supposed component shortages or cost increases due to Brexit (and notably, art and jewellery sales to the EU rose); and while mechanical machinery exports to non-EU destinations have suffered, exports of machinery to the EU have done well which seems surprising if there is a problem with the availability or cost of EU inputs used in these products.
Counterfactual- what counterfactual?
Some observers have argued that the best way to see if UK exports have been hit by Brexit is to compare the actual data with some kind of counterfactual series that would indicate where UK exports might have gone in the absence of Brexit. But there is little agreement on exactly how to do this. Sampson et al. compare UK exports to the EU with exports to non-EU destinations, reasoning that the latter were unaffected by any new trade barriers with the EU. Using this approach they find only a temporary and small decline in UK exports to the EU related to Brexit. On the other hand, Kren and Lawless compare UK exports to the EU with EU trade with non-EU economies and find a large effect.
Neither of these proposed series have a very strong historic correlation with underlying UK exports to the EU: the simple correlation on levels is around 55% for non-EU exports and just 40% for EU extra-EU imports. One problem with using EU extra-EU imports is that these were strongly boosted in 2020-2021 by imports of consumer electronics from Asia due to changed spending patterns during the pandemic. This is a sector in which the UK has little presence.
In our previous work, we have argued for using EU industrial output (excluding Ireland due its highly distorted data) as a counterfactual on the basis of a better historical fit – with a simple correlation with UK exports to the EU of around 80% (see Chart 2) – and because a large part of UK exports to the EU comprises components and semi-manufactures used in EU industry. Using this approach, we have found a small or even zero impact from Brexit on UK exports to the EU.
Source: ONS, Eurostat, author’s calculations
In truth there is a wide range of possible counterfactual series that could be used for analyses of this type, all of which have their strengths and weaknesses. As a result, it may make sense to look at a big sample of them and try to draw conclusions from that. To this end, we have run a set of regression equations relating UK goods exports to the EU (ex-oil and erratics) to nine different counterfactual series and also to the level of the UK real effective exchange rate for the period 1997-2020 (with a dummy variable in H1 2016 to capture the effects of MTIC fraud in this period). The counterfactual series include the three already mentioned plus Swiss and German total export volumes, French and German export volumes to the EU (from Eurostat), exports by all other G7 economies, and EU industrial production including Ireland.
If we compare the actual path of underlying UK goods exports to the EU since the end of 2020 to the predicted paths from these regressions we find that, during 2022, the different equations predict that UK exports to the EU could have been on average as high as 5% above their ‘predicted’ level or as low as 9% below it. Looking at the average of the predicted paths for UK exports generated by the regressions we can see that since mid-2021, actual UK goods exports to the EU have fluctuated around this average path. For the four quarters of 2022, UK goods exports to the EU averaged just 1.5% below the level predicted by the regressions (Chart 3).
Source: ONS, author’s calculations and regression results
Source: ONS, authors calculation and regression results * Using EU industrial output ex-Ireland and the UK real effective exchange rate (IMF series using relative unit labour costs)
Our preferred counterfactual remains EU industrial output ex-Ireland as the equation using this and the UK real effective exchange rate has the best fit of all of those we used, explaining around 90% of the level of underlying UK goods exports to the EU in the period studied. Comparing the path of actual UK exports to the EU with the predicted path from this equation from 2021 onwards produces some interesting results. There is clear evidence of a large initial disruption to UK exports to the EU, with actual exports in Q1 2021 some 19% below the predicted level. But the effects then fade through the rest of 2021 and 2022 and for the last four quarters there is on average essentially zero deviation (see Chart 4).
Another set of studies tries to create a counterfactual for UK trade using an algorithm to produce a ‘doppelgänger’ of UK trade performance based on a weighted group of other economies. We have several objections to this approach, which we have outlined in detail previously. To summarise, we are concerned that this methodology uses inappropriate comparator countries (with very different trade patterns), risks picking up a lot of non-Brexit effects such as different fiscal and monetary stances, exchange rate moves, and idiosyncratic shocks (a problem that gets worse the further on beyond the UK’s exit from the EU trade systems we go) and overall may be no more than a statistical artefact. In our view, it makes more sense to compare UK exports with series with which they have been clearly correlated in the past.
Delayed export recoveries may simply be normal for the UK
Several recent commentaries on UK export performance argue that the UK has ‘missed out’ on the upturn in world trade seen since mid-2021 and link this to Brexit. As we have already shown, UK exports to the EU have actually broadly tracked EU industrial output (ex-Ireland) which does not suggest a big Brexit effect. But more broadly if we look back over the last four decades we can see that it is not unusual for UK export recoveries after global downturns to lag those in other G7 economies. In fact, this was the pattern seen in the recoveries of the 1980s, the 1990s and the early 2000s. At the same point during the early 2000s recovery, the gap between other G7 and UK goods exports was similar to the current gap and reached as much as 17% by 2005 (see Chart 5). So the UK ‘missed out’ on this global trade upturn too but presumably that was not to do with Brexit.
Source: ONS, national statistics, author’s calculations * Exports ex-oil and erratics in 2001-2005 and 2020-2022
Global shocks or EU shocks?
One reason that the UK’s trade performance might diverge from other economies is sector-specific global shocks, as already noted above. If we look at goods exports to EU and non-EU economies by sector, we can see strong evidence that such shocks have been affecting UK trade. Across the ten biggest sectors showing declines in exports to the EU from 2018-2022, six of these – organic chemicals, aircraft, iron and steel, cars, crude oil, and oil products – also showed similar large falls in exports to non-EU destinations (Chart 6).
This pattern is more in line with sector-specific shocks affecting exports in all directions than with a story based just on new trade barriers with the EU as there is no obvious reason such barriers would affect exports to non-EU economies. A prime candidate for the cause of some of these shocks is the pandemic, which certainly affected global supplies of components for several sectors (e.g. computer chips as mentioned before) but also crushed global aircraft sales. In addition, the weakness of UK oil exports reflects the trend decline in output in this industry, a factor unrelated to Brexit. The special factors affecting oil and aircraft add to the case for using the UK goods exports series excluding oil and erratics when examining post-Brexit performance.
Source: ONS, author’s calculations
The big exception to the pattern outlined above is the clothing and footwear sector, where exports to the EU have dropped sharply while those to non-EU destinations are up. This is definitely a Brexit effect but is perhaps less dramatic in importance than it first appears. Prior to Brexit, UK ‘exports’ of clothing and footwear to the EU consisted in large part of products largely made in Asia with rather low levels of UK value-added (often little more than packaging). Under the UK-EU trade and cooperation agreement, these goods have insufficient UK content to benefit from zero tariffs and as a result it has become uneconomic to ship them to the EU. But the point here is that the real value of the pre-Brexit exports of these goods to the EU for the UK was much lower than the headline value of the recorded exports showed. In any case, this is a small sector, only around 1% of UK goods exports.
If exports are so bad, why is UK manufacturing outperforming the G7?
If UK export performance since 2021 had really been as terrible as is often suggested, it might have been expected that the performance of manufacturing industry, which is quite heavily export-oriented and potentially most exposed to new trade barriers with the EU, would have been bad too. But the data do not show this. In fact, UK manufacturing industry has outperformed its G7 peers since 2020. Most of the other G7 economies have current levels of manufacturing output below or similar to 2018 levels, while the UK’s is 6% higher (Chart 7). It is very hard to square this with a story of disastrous trade underperformance.
Source: ONS, national statistics, author’s calculations
Services exports – a story of strong recovery
Services exports account for around half of the UK’s exports and our view has long been that these would be minimally affected by Brexit given the nature of the products, weaker ‘gravity’ effects and the limited nature of the EU single market in services. This view was in contrast to that of many other observers, some of whom claimed Brexit has had (or ‘could have’) a large negative effect (see here and here).
Source: ONS, national statistics, author’s calculations
The data we have contradict these negative views. The UK’s service exports suffered, along with those of other major services exporters, during the pandemic but have since recovered strongly. Indeed, the UK’s service exports recovery is at the top of the range of recoveries seen among the G7 economies and looks much better than that in the US (see Chart 8). Notably, financial services exports have not seen the collapse that was widely predicted before Brexit. They grew in value terms by 8% in 2022 and were 22% higher than in 2019.
‘Openness’ and productivity – a story that doesn’t stack up
Much attention has focused on claims by the OBR that Brexit will cause a 4% decline in long-term UK productivity due to reduced trade intensity. We have repeatedly argued that this claim is not supported by the evidence: the performance of trade since Brexit is not in line with the OBR’s claims and crucially the evidence for a strong link between trade intensity and productivity for the UK simply does not exist.
Recent trade data make the OBR’s argument even harder to stand up. ‘Openness’, measured as the share of exports and imports in GDP is up since 2016 and 2019 when measured in current prices and is unchanged when measured in constant prices. Even if we exclude precious metals from goods exports (for which there is a good case), ‘openness’ measured in constant prices is higher than in 2016 and is less than 1 percentage point below the level of 2019 (see Chart 9). There is no way that such a tiny fluctuation, which may in any case be temporary, is consistent with a large 4% fall in long-run UK productivity.
Source: ONS, author’s calculations
Moreover, the actual UK productivity data does not support the OBR’s claims. Any negative impact so far should be seen in productivity in the market sector of the economy, especially tradable goods sectors. Yet the numbers do not show this – productivity per hour in this sector is not down either in absolute terms relative to pre-Brexit levels or relative to trend (Chart 10). Productivity declines are only visible in the public sector, which has nothing to do with Brexit (but an oversized public sector may be a serious drag on UK economic performance, see here).
Source: ONS, author’s calculations
The analysis above suggests that a careful look at the available evidence does not support the view that UK export performance since the start of 2021 has been a ‘disaster’. The underlying level of exports to the EU may only be around 2% lower than it would have been in the absence of Brexit. This much lower than some recent claims of exports being 15% or so down due to Brexit, and contrasts even more starkly with Treasury claims before the 2016 referendum that UK exports to the EU might be 40% lower due to Brexit. Our findings are similar to the estimated effect on UK exports due to Brexit of Kee and Nicita (which we have long flagged as being a realistic estimate) and also close to estimates in Hearne. Like the latter study, we find evidence of a sharp initial drop in UK exports to the EU which has waned since. Nor is the oft-quoted claim that Brexit has caused (or will cause) a 4% slump in productivity consistent with the available evidence.
Attempts to stand up claims about disastrous Brexit impacts rely increasingly on selective use of data and questionable methodologies. Recently, a further questionable element has been introduced in the form of dubious upward revisions of historic trade data in an attempt to make post-Brexit data look worse by comparison. This one-sided approach excludes adjustments that could easily be made in the other direction, relating for example to the ‘Rotterdam effect’ which historically inflated UK exports to the EU perhaps 8%, VAT-related fraud which in the mid-2000s was inflating UK exports to the EU by as much as 10% at one stage and which was probably never fully documented or eliminated, and the existence of a chunk of UK exports to the EU which were effectively re-exports with low UK value added content (e.g. clothing and footwear as noted earlier, but also some foodstuffs – for more see here). Given these distortions, and the several measurement changes made to UK trade data since Brexit, making selective adjustments to back data looks arbitrary.
Indeed, a major objection to all the recent attempts to show a big negative effect on UK trade from Brexit is the paucity of data. Extraordinary claims such as those we are seeing require extraordinary evidence and a couple of years of noisy data subject to significant measurement and interpretation issues simply does not give you such extraordinary evidence. There is something of an air of desperation in the way that the available data is being tortured and over-interpreted at every turn to try to show the disaster scenario that so many analysts confidently predicted prior to Brexit but which has not materialised.