Trade and the EU Referendum: Why the Leave campaign has simply got it wrong

The Leave campaign’s arguments about trade are based on out-dated, irrelevant
assumptions and models – following them would be catastrophic for the UK economy, writes Professor Ronald MacDonald of the University of Glasgow.

Trade and the EU Referendum

Trade is at the core of the economic debate on the EU referendum. Numbers relating to the trade from both the Remain and Leave sides have clearly been an important component of the debate, and many have found it difficult to know which set of numbers are the most reliable. On the one hand are the Economists for Brexit who argue that the Single Market is trade diverting and therefore has a negative impact on the UK’s welfare and economic growth. On the other side, Remain have the support of just about every national and international economic institute, and the vast majority of economists, arguing that membership of the EU single market is trade creating and therefore has a positive impact on the UK’s GDP, productivity and GDP growth, and there have been numerous calculations made to demonstrate this.


It is not the purpose of this piece to asses the accuracy of the calculations and numbers that are out there, but rather to use the various trade models in an economists’ toolkit to go behind the numbers to the actual assumptions and models used to generate them. In a recent paper commissioned by the Royal Society of Edinburgh[1], as part of their series on ‘Enlightening the European Debate’, I examined the trade issue from the perspective of trade theory models. Specifically, I tried to give the reader a flavor of what the different trade models predict, so that they could hopefully make an informed decision in the referendum. In this paper I revisit the issue and show that the Leave sides’ argument on trade is simply wrong because they rely on underlying assumptions and models that are not relevant to international trade today. Following their recommendations would, I believe, be catastrophic for the UK economy.

‘Just over 70% of trade today is in intermediate goods, rather than final goods, and understanding this is an important step in understanding why being part of the Single Market is vitally important’

In thinking about the trade issue it is crucially important to recognize a couple of basic points that standard textbook trade models often ignore (and indeed much of the EU debate has ignored). First, just over 70% of trade today is in intermediate goods, rather than final goods, and understanding this is an important step in understanding why being part of the Single Market is vitally important for the UK. Second, trade today is in a whole range of heterogeneous products, many only slightly differentiated from each other. This is also in sharp contrast to standard trade theory where a particular good (a car, say) produced in country A is identical to that product in country B and that trade only takes place in dissimilar goods.


Trade and openness to trade are viewed positively by economists because of the so-called ‘gains from trade’. These are usually expressed in terms of a country that trades having access to a larger bundle of goods – a greater GDP – than in the no trade situation (the static gains from trade). Dynamic gains from trade come from the productivity gains and consequent economic growth that openness to trade can create – the more a country trades the greater these benefits will be.


The crux of the trade question in the context of the referendum is of course whether these gains are better achieved by having free access to the Single Market or by adopting an alternative model (the Norwegian and Swiss relationships with the EU have been touted as alternative trading model for the UK in the event of Brexit). The provision of a single market with a population of approximately 500 million consumers provides the opportunity for any participating country to increase its degree of openness to trade, thereby maximizing the gains to trade. It is worth noting in this context just what the Single Market means.


The Single Market relies on the elimination of tariffs and quotas within the EU, the elimination of non-tariff barriers by the harmonization of regulations, prohibiting artificial state aid and recognizing other countries standards. The other key element is the creation of a customs union and with that the elimination of the border costs of trade (i.e. the elimination of having to clear customs, and the associated bureaucracy, custom duties and custom delays).


Traditional trade theory comes in two packages – Classical trade theory of Adam Smith and David Ricardo and New Classical trade theory of Eli Hecksher, Bertil Ohlin and Paul Samuelson (HOS). Classical trade theory posits that it is productivity differences across countries that drive trade – so if a country is more productive, or relatively more productive (the concept of comparative advantage), in producing a good it should focus resources on that good (or goods) and it will be able to enjoy a larger bundle of goods (or GDP) than in the absence of trade. In the Classical model the value of a good is not determined by the demand for the good but purely on the production inputs.


In the Neo-Classical trade model of HOS demand or preferences feature but since preferences are assumed to be the same across countries they are essentially irrelevant in determining trade patterns, which are driven by relative factor endowments – capital and labour. The prediction here is that trading patterns are determined by relative factor endowments – a capital rich country will export capital-intensive goods and a labour intensive country will export labour. The HOS model in its textbook form is presented as a perfectly competitive model in which firms exhibit constant returns to scale and the goods produced in one country have a homogenous counterpart in the foreign country. It is noteworthy that there is very little empirical support for the HOD model.


It is also noteworthy the eight economists who form ‘Economists for Brexit’ base their predictions that the Single Market is trade diverting rely on these traditional models and particularly the Neo Classical trade model of HOS. But how much credence can their view of trade diversion be given once it is realised that these models are unable to explain the vast majority of today’s trade? For example, these models do not address the intra industry trade referred to above and the fact that the firms we observe trading to today operate in oligopolistic or monopolistically competitive markets, rather than in the perfectly competitive markets, of the HOS models. Also the goods entering international trade today are often very similar to each other rather than the dissimilar goods predicted in the traditional models.


In order to understand modern trading patterns we have to turn to the New Trade Theory (NTT) of Paul Krugman and others. In NTT, geographical location and the extent of the market are crucially important factors in satisfying consumers’ preferences for a whole range of products, many of which are close substitutes for each other. Specifically, in NTT businesses tend to congregate geographically or agglomerate, a concept initially noted by Adam Smith and discussed extensively by Krugman. That is, the opening of similar companies in a particular location attracts workers with skills in that business, which, in turn, draws in more businesses seeking experienced employees and the network effects (i.e. externalities from other firms) available. The agglomeration, in turn, is decided by the size of the market.


The firms operating in the agglomeration are monopolistically competitive and are able to reap increasing returns to scale in contrast to the assumed production pattern of constant returns to scale in the traditional trade models (i.e. no dynamic productivity gains). A key feature of NTT is its focus on intra industry trade in goods, services and capital. The agglomerations or clusters of activity predicted in NTT help to explain this phenomenon in that they crucially rely on cross border supply chains for their existence. For example, in the UK context pharmaceuticals, the aerospace and automotive industries are all crucially dependent on the supply of intermediate goods from other cluster centres within the EU.


Perhaps the most obvious, and one of the most important, clusters in the UK is the London based financial sector which crucially depends on the existence of the Single Market, and the complex interaction with other sectors, plus having access to the free movement of labour that the Single Market provides. In that regard it is often overlooked in the current debate that approximately 10% of those employed in the London-based financial sector come from elsewhere in the EU. The South East of England has its own clusters in, for example, motor-sport and BMW mini plant cluster and those in nanotechnology and cryogenics. In the North West of England there are important clusters in aerospace and bio-manufacturing and many others could be identified in other parts of England. These clusters are clearly dependent on their interaction with other clusters in Europe and having an unencumbered pitch on which to conduct their business.


In a Scottish context the existence of the above noted successful clusters elsewhere in the UK is crucially important for its trade and economic well being since the vast majority of Scotland’s trade is with the rest of the UK and, of course, Scotland also benefits from direct access to trade in the Single market too. In that regard Scotland has its won important clusters in, for example, financial services, pharmaceuticals, the games industry, life sciences and food and drink.

‘The key question is whether the clusters of industry in the UK today would be sustainable if barriers were placed upon them’


In the context of the referendum debate the key question therefore is whether the cluster and agglomerations we observe in the UK today would be sustainable if barriers were placed on both final and intra industry trade and on the free movement of people? To answer that question we turn to some of the empirical evidence that seeks to measure the role of frictions in trading relations.


The empirical workhorse in International trade is the so-called gravity model which captures the effect of geography on trade and other frictions introduced by having barriers to trade. This model derives its name from Newtonian Physics, where the attraction force between two objects is proportional to their mass and inversely proportional to the distance between them. Similarly, trade between two countries can be modelled as an increasing function of the size of the two economies, and a decreasing function of their cost of transport measured by their distance. So other things equal, the further apart two trading nations or blocs are the greater the transportation costs and the less they are expected to trade. Trade is also seen to be directly proportional to the mass of the two countries.


In testing the gravity model researchers also include other terms which may create impediments, or frictions, to trade such as whether the two countries share a common language and crucially in the current context whether they share a common border, which introduces the border costs mentioned above.


There are a plethora of papers using sophisticated statistical and econometric methods that support the basic premise of the gravity model – distance does matter in international trade, the mass of the countries or country blocs you are trading with also matters as do the additional frictions that add to the costs of trade. One of the pioneering tests of the gravity model is by John McCallum who studied bilateral trade flows between Canada and the USA and found that inter province trade is 20 times larger than trade between the two countries. He attributed this ‘border effect’ to the totality of barriers between the countries, including tariffs and non- tariff barriers, and the ‘pure’ border effect (that is the customs costs) mentioned above.


A number of authors have suggested that the large border effect that McCallum finds is a result of statistical deficiencies in his modelling. But it turns out that even when these have been addressed the border, or home bias, effect is still very large indeed (6 to 12 times). The existence of this border effect has been replicated for other country pairings and it would be expected that the existence of the Single market in the EU would have considerable trade enhancing effects along with the associated gains from trade.


A number of studies have applied the gravity model to calculate the effects of EU membership on intra EU trade and they find large and statistically significant effects; that is, the Single Market is trade creating rather than trade diverting. Furthermore, these studies find that EEA and EFTA membership on their own (i.e. separate from the Single Market) are not significant determinants of trade creation. This would mean that the alternative models in the event of a Brexit such as Norway and Switzerland, would not provide the gains that the single market does to the UK.


Modern trade theory has been pushed further in the ‘New’ New Trade (NNTT) Theory of Marc Melitz and others. In contrast to the traditional trade models, this approach emphasises firm level differences in the same industry of the same country and in understanding the challenges and the opportunities countries face in the age of globalisation. The key point in this model is that as trade barriers are lowered competition is stimulated and low productivity firms that had been protected with barriers, within the very same industry go out of business and are replaced by increased volume production of high-productivity firms. In NNTT the resulting intra-industry reallocations of market shares and productive resources are much more pronounced than inter-industry reallocations driven by comparative advantage. As a result of these reallocations the average productivity of the country as a whole increases, the gains from trade in the New NTT model. So NNTT clearly supports the existence of a single market in Europe – the productivity gains to a country will be maximised in such an arrangement – but the kind of dramatic productivity gains of this class of model are missed in the standard productivity model of Ricardo in which country differences in productivity dominate.


We have argued that the traditional trade models used by economists on the Leave side are not well suited to an understanding trading patterns today. This is perhaps why there is in fact very little empirical evidence for the traditional models of Ricardo and HOS. However, it is interesting to note that nearly all of this work has been undertaken for country pairings outwith the Single Market. A recent study of bilateral trade flows between 21 EU-member states during the single market period 1994-2004 (see my recent RSE paper for further details) finds that productivity differences have good explanatory power for EU trade (statistically significant and positive in 67% of country pairs). It would seem therefore that the EU’s single market has been a powerful means of unlocking the Ricardian prediction, but not for countries outside it and this may well be due to the core argument of NNTT.

‘The trade argument proposed by Leave is based on very shaky foundations indeed’

There have been many numbers and calculations regarding the economic benefits of either remaining in or leaving the EU. In terms of the essential issue of trade, we have argued in this short paper that those who argue for Brexit on the basis of trade are basing their arguments and numbers on models and assumptions which do not reflect the reality of modern day trade in terms of the kinds of goods traded, the nature of trade today and the importance of geography and industrial structure. Once these issue are recognized it is very clear that the trade argument proposed by Leave is based on very shaky foundations indeed and although an analysis of various trade models cannot confirm the exact magnitude of the trade numbers from the Remain side, it is unquestionable that these numbers are on the correct side of the debate. From my own reading of the situation, especially given the considerable complexity of modern trade as given here, such numbers may if anything understate the huge mistake that the UK would make in leaving the EU and being denied access to the Single Market, especially if a Brexit vote led to a sudden stop in the financing of the UK’s external deficit.


Ronald MacDonald
Professor of Macroeconomics and International Finance
University of Glasgow

[1] The paper is available at: A full list of references to the literature mentioned here can be found in the paper.