Why We Know Less Than We Think We Know about the Economic Impact of Brexit
Philip B. Whyman
13 June 2018
Remainers' "Project Fear" is heavily based on doom-laden economic predictions about the impact of Brexit. But there are three key reasons to doubt these predictions: their dubious assumptions, their reliance on outdated "gravity models"; and their exclusion of many potential benefits of Brexit.
One of the main criticisms of Brexit that is regularly repeated in political debate, and reported uncritically in the media, is that there is a consensus amongst economists that Brexit will prove damaging to the UK economy. This was the case during the referendum campaign almost two years ago now, and more recently it has been reinforced by the rather cynical leaking of doom-laden civil servant briefing papers, meaning that the message is in the public domain but none of us can scrutinise how well the underlying research had been designed.
However, this “consensus” proposition is simply inaccurate. Of the 40 or so economic studies reviewed in my recent book, The Economics of Brexit, fully one third suggest that Brexit will lead to either a net gain to the UK economy or that the cost-benefit calculus is finely balanced and dependent upon the form of relationship ultimately agreed between the UK and the EU.
The misleading perception of a consensus stems from the fact that the five most cited studies, from organisations such as the London School of Economics, the Treasury, the Organisation for Economic Cooperation and Development, and the National Institute for Economic and Social Research (NIESR), all adopt a broadly similar methodology, and hence produce broadly comparable results. In turn, these studies have been used as the basis for other reports produced by the Confederation of Business and Industry, the Trades Union Congress and the Office for Budgetary Responsibility, rather than them undertaking their own independent analysis.
These five studies predicted that Brexit is likely to have a negative impact on the UK economy, the magnitude of which would be larger the looser is any future trade relationship with the EU. According to these widely cited studies, participation in the European Economic Area (EEA) would prove to be the least costly to the UK economy, followed by the negotiation of a Free Trade Agreement (FTA), while the most damaging would be the WTO Most Favoured Nation (i.e., “no deal”) option. None of these studies sought to test the impact of a customs union. The estimates of costs varied from 1.3% GDP in the most advantageous case, predicted by the LSE and IMF, to a cost of around 7.5% of GDP as forecast by the Treasury and NIESR under their worst-case scenarios.
Taken at face value, these findings would provide a strong argument for UK policy-makers to push for the closest accommodation with the EU, either by seeking EEA membership or through the negotiation of a customs union. Or, indeed, seeking to overturn the referendum result and remain as a full member of the EU, which is the option all of these five organisations supported during the referendum campaign.
However, three significant problems make their conclusions unreliable.
Garbage in, garbage out
The first arises from the underlying assumptions underpinning their economic models. A degree of simplification is necessary for all studies seeking to predict future behaviour, otherwise the mathematics becomes frighteningly complex. However, assumptions made concerning how the economy is supposed to work, and how people and organisations are likely to react to different stimuli, can influence the predictive power of these same models – thereby introducing bias into these same predictions.
This may sound like a relatively arcane disagreement between different branches or schools of Economics, and in part this is true. However, it matters because, if these assumptions fail to reflect the real world, then these models build inaccuracies into their very DNA.
Given that the five most cited studies all drew upon a rather conventional (new monetarist-New Keynesian) perspective, it is arguable that their results are weakened as a result. For example, they tend to over-emphasise such factors as the importance of migration as a driver of growth, and downplay other factors such as the ability of economic policy intervention to influence expectations and hence the real economy.
Secondly, most of the most cited studies use a similar set of economic models to determine their predictions. “Gravity models” are used to determine the significance of trade flows, with the results then being fed into a macroeconomic model to arrive at the final predictions. Gravity models are based on the idea that one country’s trade with another will be higher with countries that are closer geographically and have larger economies. Partly this is due to transportation costs. However, it could also be because nearby countries are presumed to have closer ties with one another, which could be used to facilitate trade. The size and wealth of individual nations also matters. Luxembourg is closer to the UK than the USA, but we trade more with the latter in part because it is a much larger market. Thus, rather like considerations of gravity upon the movement of planets in the solar system, the economic gravity modelling suggests that size and closeness are significant determinants of trade.
The inference for Brexit is obvious. The EU is closer to the UK and is a more lucrative marketplace than other parts of the world, and therefore Brexit is bound to have a net negative impact.
The problem with this conclusion is that it leaves too much out of the calculation. For example, it ignores the fact that much UK trade depends upon a maritime past. Hence, the UK has a much lower trading density with the EU than is the case for other member states and would be anticipated through gravity explanations alone. Instead, whether through Commonwealth connections, a shared language or long-standing cultural ties, the UK has a greater than expected trading relationship with the USA and other parts of the world. Thus, gravity models are too restrictive – they ignore the significance of other variables more pertinent to the UK than for many other nations.
In addition, gravity equations depend upon historical data and so they are, naturally enough, backward looking. This makes them imperfect predictors for future trade trends. While the EU has been a historically wealthy area in international terms, this advantage is slowly eroding as other parts of the global economy, like China and India, have far higher growth rates. Likewise, technological advances have significantly reduced the cost of transporting goods over long distances. Hence, in the future, “gravity” may influence trade patterns less than in the past. This is particularly true for services, one of the UK’s areas of specialism, because the cost for transporting people and ideas is far less significant than for goods.
As a result, gravity models are likely to exaggerate the importance of trade with the EU while underestimating other areas of potential trade expansion in high growth regions.
Missing variable bias
The third problem with these studies concerns not what they contain but what they omit. In Economics terminology, the models suffer from “missing variable bias”, which significantly skewed their findings.
If an economist started with a blank sheet of paper and sought to list all of the variables that might be reasonably expected to be impacted by Brexit, this should include factors likely to be either negatively or positively impacted by Brexit. Factors indicating a net economic cost might be expected to include trade with the EU, inward foreign direct investment (FDI) from the EU, the quantity of net migration, and the level of uncertainty arising from Brexit. In contrast, likely benefits arising from Brexit might include the greater potential for increasing trade with the rest of the world, inward FDI from outside Europe, outward FDI, quality or productivity effects from a better-managed migration regime, the cost of regulation, exchange rate changes leading to greater competitiveness, and government policy actions.
Remarkably, the five dominant studies focused almost exclusively upon the variables from the first list and excluded most of those from the second. As a result, their predictions over-estimated likely costs and under-estimated likely benefits.
A further problem with these forecasts is that they extrapolate current trends into the future. They make their estimates of what is likely to happen on the basis that nothing else changes other than Britain’s relationship with the EU. Given that Brexit is all about change, this does not seem like a safe assumption. Many other variables are likely to change in the wake of Brexit, not least the economic policy of post-Brexit UK governments.
The consequence of these weaknesses is that we unfortunately know less than we think we do about what is likely to happen in the future.
This means that our political leaders are setting policy with an inadequate understanding of the likely consequences of their actions. It is difficult enough to enter negotiations with a strong hand of cards, but it is doubly difficult if you are, in effect, unsure how strong your position may actually be in practice. Similarly, policy-makers like the Bank of England and the Chancellor of the Exchequer are currently setting macroeconomic policy on the basis of analysis which is flawed in certain key respects. And leaders of the opposition are setting economic and Brexit policy on the basis of a set of forecasts which may be more inaccurate than normal.
The media do not help in this regard, since most commentary uncritically accepts the predictions made by these models as fact, rather than estimates based on tendentious assumptions about both the past and the future. A more cautious and circumspect treatment from key opinion formers would be helpful. It might, for example, help to prevent companies from making the wrong decisions over the timing of investments, or policy makers erring over the design of economic strategy. It would be all too easy for a self-fulfilling prophesy to develop whereby negative predictions, whether justified by objective facts of not, could result in economic under-performance.
In this sense, recognising that we know less than we think we know about the likely economic impact of Brexit would be a good starting point.
Beyond that, well there is an obvious need for new economic studies to be commissioned, by a range of research teams adopting a range of different approaches and methodologies, in order to generate better and more varied results estimating the likely impact of Brexit. Without this information base upon which to draw, it is likely that those in whom we place our trust to manage the economy and smooth the Brexit transition will miss a few tricks and make more mistakes than might otherwise have been the case.
We know less than we could about the likely economic impact of Brexit, and surely that is not good enough. We need to have a better evidence base than this if we are to take advantage of those opportunities that Brexit will deliver, and strengthen our economy against the equally inevitable challenges that we will face.
 Philip B. Whyman and Alina I. Petrescu, The Economics of Brexit: A cost-benefit analysis of the UK’s economics relationship with the EU (Basingstoke: Palgrave, 2017), pp. 35-40.
About the author
Philip B. Whyman is Professor of Economics and Director of the Lancashire Institute for Economic and Business Research at the University of Central Lancashire. He has been researching the economics of Britain’s relationship with the EU for over twenty years and has written or edited 14 books on the topic. The most recent are The Economics of Brexit: A cost-benefit analysis of the UK’s economics relationship with the EU (Palgrave, 2017), co-authored with Alina I. Petrescu, and Rethinking Economic and Monetary Union in Europe: A post-Keynesian alternative (Routledge, 2018).
This work represents the views of the authors only. It is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.