After the failure of almost all economists to predict the banking crisis in 2008 or the weakness of the subsequent recovery, plus their notably poor record on the short-term impact of Brexit, one might have thought that major reforms of the discipline would be in order. Not to mention the scandals of high-profile economists taking money to exonerate damaging economic policies as shown in the film ‘ Inside Job‘. Of course, no such reforms have occurred. The discipline conducts its own hiring and promotions and controls the academic journals which play a large role in career advancement. This makes the discipline immune to radical change however irrelevant its work is to important matters of economic policy. Reform is overdue for economic advice at the Treasury and Bank of England but for whatever reasons politicians seem reluctant to go there, although Chancellors in the current government are preventing Treasury economists from undertaking further reports on Brexit.
BfB contributors have written about the flaws in economists’ exaggerated estimates for both the short-term and long-term impact of Brexit here, here and here. It is the former that are the focus of this article. The Treasury’s embarrassing predictions of a year-long recession and increases in unemployment of between 500,000 and 800,000, when joblessness actually fell by 250,000, are well-known. Some perhaps also remember that the Bank of England and OECD expected negative short-term impacts. Less well remembered is the equally embarrassing letter from ’Economists for Remain’ whose signatories included eight Nobel Laureates and 70 professors of economics. Their predictions were that recession and job losses were ‘significantly more likely’ due to a pro-Brexit vote, together with lower investment and tax rises. This group sloppily put no timescale on their predictions but none of these things have happened yet, indeed employment swelled by a million before Covid. The spectacular success of the post-referendum labour market suggests a resilient economy without much Brexit downside.
After the Treasury’s long-term Brexit forecasts, based on gravity models, were debunked, the Treasury dropped this approach in favour of a black-box ‘general equilibrium’ approach. This produced another set of estimates for the impact of Brexit on the UK economy claiming that in the event of a no deal Brexit GDP will be 8% lower in the long term than it would have been if the UK had remained in the EU. BfB’s Harry Western reacted with incredulity writing that:
The Treasury’s 8% GDP loss for a ‘no deal’ Brexit is … more than double realistic estimates of the long-term GDP loss the UK suffered in [the great depression of] the 1930s. It is very silly indeed. Many will think that it is worse than silly: that it is a deliberate attempt to mislead and alarm.
Economists examine the post-referendum years
You can say this for economists, they never give up trying to demonstrate the downsides of Brexit. With nearly four years of experience between the referendum and the pandemic economists have more recently asserted that the UK economy has indeed grown more slowly than it would have in the absence of the Brexit vote, despite the boom in jobs and rising prices of houses and shares. If these attempts were successful, then it might redeem some of the earlier claims of the economics profession, but we argue here that the attempts are shallow and generally incorrect.
The earliest of these estimates was undertaken by the Centre for European Reform (CER) which estimated that Brexit had already lost the UK economy 2.5% of GDP (i.e. over £50 billion) by mid-2018. An article on the BfB site analysed the methods used by CER to reach this conclusion. The CER estimated that growth in the UK had fallen behind a group of ‘doppelganger’ economies particularly from early 2017, having previously kept pace with these economies prior to the referendum. By the middle of 2018 the cumulative shortfall was calculated as 2.5% (see chart below). The comparator countries were largely the USA and Germany but also included Luxemburg, Iceland and Greece. These countries were selected by a computer programme rather than by human judgement. The latter three countries were of course of little relevance to the UK economy. It is always possible to assemble a collection of countries, suitably weighted, to match UK growth over a defined period, but this need not necessarily have much explanatory or predictive value.
The CER evidence is based on a trend relative to the ‘dopplegangers’ established over a period of seven years prior to the referendum. Over this period UK growth closely matched the ‘doppleganger’ group and CER expected that the UK would have continued to match the growth of this group for another four years (and presumably for ever) in the absence of Brexit. We regard this as a superficial judgement. The chart below shows UK growth over the longer period since 1990, this time relative to the major G7 economies (excluding the UK). This shows a small improvement in UK GDP relative to these countries between 1990 and 2002. After 2002 the UK grew at a similar rate to these economies but with cyclical ups and downs. The post-2016 downturn looks like a normal cyclical downturn with a return towards the established post-2002 relationship. To arrive at the CER estimate of a 2.5% loss due to Brexit we would have to assume that UK GDP remained at its mid-2016 peak of 4% above the G6 average instead of returning to the post-2002 trend. CER provides no justification for such an assumption, preferring instead to ascribe the downturn entirely to Brexit.
Source of data: OECD
In fact, a number of important factors unrelated to Brexit can account for the downturn in the UK relative to comparator economies. These cause differences in the timing of business cycles between countries. One is that fiscal policy stances were very different between the UK and other major economies from 2016-19. In this period, the UK tightened fiscal policy substantially, which would have weighed on economic growth – but in the euro area the fiscal stance was fairly steady and in the US, President Trump oversaw a large fiscal loosening which boosted US growth.
Also important was a pronounced cycle in consumer credit in the UK. Banks loosened standards for unsecured consumer credit from 2012-2016, leading to a boom in credit growth and consumer spending. But from 2017, banks changed their policy and started tightening unsecured credit availability strongly.. In response, consumer spending growth slowed very sharply, as the chart below shows. So, while it is obvious that GDP growth slowed relative to comparator countries there are several important reasons why this should be so, without any recourse to citing Brexit as the reason.
NIESR and the downturn in investment
A second example of misleading estimates of the impact of Brexit comes from the 2019 General Election Briefing published by the prestigious National Institute for Economic and Social Research (NIESR). The briefing was written by Gary Young formerly of the Bank of England who we regard as having something of an open mind on Brexit but not in this case. He wrote:
“Business investment is estimated to be 15% lower than it would have been in the in the absence of the 2016 Brexit vote (see chart below). This is due to the uncertainty that the decision to leave the EU has created. Uncertainty has led businesses to postpone investment decisions until they know more about the new relationship with the EU.”
Source NIESR (2019)
The NIESR chart above shows a trend established over the short period of four years prior to the mid-2016 referendum. The trend is then projected on for a further three years as what is called a ‘counterfactual’. In practice, business investment did not follow this rising ‘counterfactual’ trend after 2016 but instead flatlined until 2019. NIESR ascribe the difference between the actual path and the counterfactual trend to the Brexit vote. Such uncertainty, largely due to government delays and extensions, was certainly reported in a range of surveys but we do not know the extent to which uncertainty influenced actual investment.
Source of data: ONS National Accounts
If we stand back and examine a longer time trend for business investment a different conclusion is suggested. The chart above shows the trend in real business investment from 1950. The short period covered by NIESR is circled and can be seen to be a recovery from the collapse in business investment during and immediately following the banking crisis of 2008. Once investment recovers to the level of the long-term trend, growth flattens out. It seems reasonable to expect that this would have occurred irrespective of the Brexit vote although there might have been some overshooting and the referendum may have prevented this. Either way the impact of Brexit on business investment appears to have been minor and perhaps temporary.
Several more recent studies have attempted to estimate the impact of Brexit on the UK economy for the period between the referendum in 2016 and actually leaving the EU in 2020. These are now becoming more sophisticated and are reaching top economics journals. Most prominent is a paper by Born et al now published in the Economic Journal. This study estimates that the output loss due to the Brexit referendum result is 2.1% of GDP or a cumulative loss £50 billion, as of March 2019. Using forecasts, the authors predicted a loss of 4% of GDP by the end of 2020.
The method used to construct these estimates is once again to construct a ‘doppleganger’ index based on a group of countries. In this case, the method uses 23 of the 30 OECD economies for which all data is available , with weights chosen by a computer algorithm to maximise the similarity to the growth of UK GDP between Q1 of 1995 and Q2 of 2016. Although 23 countries are included, a few countries dominate the index. In the latest version of the paper four countries account for 94% of the index. The USA dominates with a 50% weight. The other main component countries are New Zealand, Italy and Hungary. New Zealand is a market for only a quarter of 1% of UK exports, Hungary 0.3% and Italy just under 3%. As before, only the USA is likely to have a real influence on the UK economy either through trade or investment. One might have hoped that referees might have questioned the inclusion of some of these, especially Hungary but apparently they did not. Once again, an arbitrary set of countries can be combined to match UK growth during a selected period.
In a previous version of the paper four of them accounted for 84% of the doppleganger index. These are the USA, Canada, Japan and Hungary. The surreal inclusion of Hungary is justified by the authors as reflecting its membership of the EU outside the Eurozone, just like the UK. However, as noted Hungary accounts for a miniscule 0.3% of UK exports and none of its inward investment. It is an export-oriented economy focused on Germany and not the UK. In fact, other than the USA the other 3 main component countries in the index accounted for only 4% of UK exports and little of its inward investment. There is little reason why these should have been used as a benchmark for UK growth.
As noted earlier, the role of the fiscal boom in the USA needs to be taken into account in accounting for the post-2016 downturn in UK relative growth. If we compare the UK to all OECD countries except the USA there is little downturn and what there is can be explained by differences in business cycles between countries caused by tightening credit standards and government fiscal stances. The chart below shows how UK GDP performed relative to all other OECD countries excluding the USA, weighted by the size of their economies. After 2016Q2 the UK reverts from 1% above the OECD countries back to equality with these countries. This is a small movement and would have happened anyway at some point even if one could argue that the timing was brought forward a little by the referendum result.
Source of data: OECD
Regional Impacts of Brexit
Two economists working at Warwick University have used a similar approach to estimate the impact of the Brexit decision on the regional and local economies of the UK. They conclude that the areas which supported ‘Leave’ in the referendum have fared worst in the years since 2016. All regions other than Wales perform negatively after 2016 relative to their comparators, although London recovers to finish no worse by 2019. Easily the worst performing region is Northern Ireland which is judged to have lost 5% of GDP.
The method is similar to the national study of Born et al described above. The authors construct a ‘doppleganger’ index, this time for each region and locality, and use a very wide range of comparator areas including EU regions and US states. There is little detail about how these are weighted but once again the weights are chosen electronically to match the growth trajectory of each UK area. The main problem in this apparently exhaustive study lies in the data used. The main measure is an experimental index of economic performance which is not an official statistic of the Office of National Statistics. The authors describe this as measure of GDP, but it is not since it depends on indices of gross output without information on costs necessary to measure GDP. One financial sector Chief Economist in Northern Ireland regards this data as unreliable and refuses to use it.
If we focus on the worst case, Northern Ireland, it is possible to use a more reliable source of data which is an official statistic of the ONS. This is data for gross value added, a concept close to GDP. Annual data, available up to 2018 in current prices shows Northern Ireland has grown at the same rate as the UK since the start of the data series in 1998 right up to 2018. If we believe that the UK has lost little since the 2016 referendum, the same appears to be true of Northern Ireland.
It is difficult to believe that the exaggeration of the impact of Brexit in all of these studies is accidental. We might reasonably assume that the researchers did not set out with the overt intention of exaggeration. More likely is that, being surrounded by like-minded academics, they ‘knew’ at some level that Brexit must be economically damaging and when they saw a downturn in the relative performance of the UK economy after mid-2016, it must have seemed obvious to them that this was a consequence of the referendum result. After all, surveys showed an increase in ‘uncertainty’ and it was easy to believe that investment would fall as a result. Maybe investment did fall a little, but these economists asked too few questions of their results and did not consider alternative hypotheses, including the role of the Trump ‘fiscal boom’ while austerity tightening occurred in the UK and consumer credit standards also tightened in the UK. This is how bad science is done.