After shock

How economic models can help predict the winners (and losers) of post-Brexit trade.

Steering a path

Brexit is probably the greatest political challenge and shock that the UK has experienced in several generations, and it may well turn out to be one of the greatest economic shocks too. It has already fractured the UK politically and continues to do so. It has and will impact on firms, workers and regions – in the UK, the EU and beyond. In steering a path through this, the UK will need to develop its own independent trade policy and negotiate trading terms with new and old partners.

In this context, and “…with a gamut of options for new trade policy and practice, economic models can be extremely useful to help inform public policy,” but, “you need to use them appropriately,” cautions Professor Michael Gasiorek. Before and since the Brexit referendum there have been numerous criticisms made of economic models, of the views of ‘experts’ and the supposed inaccuracy of their forecasts. However, these critiques are often based on misunderstandings of what a model can do.

“Models are not designed to provide accurate predictions or forecasts of future reality,” explains Michael. “Economic models involve simplifications. Each model will have its own objectives, and degree of simplification – in the same way as do different maps. Each model will shed light on particular characteristics and mechanisms and by design leave others out. This is intentional. Of course, in setting some things aside, the model cannot fully capture all the underlying economic mechanisms and therefore can never provide a completely accurate prediction of the future. It is not designed to do so.”

For example, in November 2018 the UK Government published a set of results on the impact of different Brexit options. To do so it used a fairly standard, and to use the jargon, “Computable General Equilibrium” model. This type of model takes into account all the linkages between both goods markets and factor markets. Hence, if the price of plastics goes up, that will increase the costs of plastics for all those industries that use plastics as an intermediate. In order to take into account that linkage, you need information on how much plastic an industry – say, the car industry – is using, so you can model all these linkages. It also takes into account factor markets – namely, labour markets. For example, negative effects on the car industry may lead to skilled workers being laid off, which lowers the cost of wages in the car industry and possibly in other industries. A CGE model tries to take these linkages into account.

“To do that sort of modelling, you need to know about all those input-output linkages – and that sort of data can be difficult to obtain. You need to know how much each industry uses from every other industry. And… because you are looking at trade with other countries you also need this information for every other country you are interested in. There is some information on this – but not at a great level of detail. The most commonly used CGE models not much more than 50 sectors. The Brexit impact results the Government produced ended up with nine sectors, and that covers the entire economy!”, continues Michael.

“These models are really useful, but have their limitations – just like any map,” he adds. “We start from the other end and ask ‘what happens if you just take each industry individually, and you don’t worry about all those linkages?’. That ‘industry’ could be very broad e.g. the textile industry or very narrow as, for example, jeans. And because you are not trying to capture the linkages between industries, you can model that industry on its own and ask ‘what happens to trade, prices and production if tariffs change?’ ‘What happens if we leave the Single Market and Non-Tariff Barriers to trade change?’”

“This type of model (jargon again) is called a Partial Equilibrium (PE) model. On the face of it, it is not as good as a CGE because, surely, it is ‘more realistic’ to take the linkages into account – so why use such a model? There are two very good reasons. One, because you don’t need all those input/output linkages, you can work at a much more disaggregated level. In our research looking into the possible effects of Brexit on the UK manufacturing sector, we had 122 sectors. That meant we could do a much more detailed analysis than anybody had done previously with these CGE models. The second reason is that, because the model is less demanding of the data it needs, we can work with much more up-to-date data and hence in that way, the model is more relevant.”

So, the advantage with a PE model is that you can do much more detail; the disadvantage is that you can’t take into account those linkages.

“Through our university spin-out company, InterAnalysis, and working closely with the UK Trade Policy Observatory, we won a contract from the Department for International Trade (DIT) to provide them with such a model, which they could use to evaluate all the free trade agreements they may well be negotiating in the near future. The model could help them, say, when they’re talking about negotiating with the US, Canada, Australia, to be able to understand ‘what happens if we change tariffs, or Non-Tariff Barriers?’ The model enables them to do this at a much more detailed level than a CGE model. Other UK ministries such as the Department for Environment, Food and Rural Affairs, the Department for Business, Energy and Industrial Strategy, and the Department for Exiting the European Union are also using the model,” says Michael.

“Since the initial contract, and over the past year or so, we have continued to develop the model. In particular, we have figured out a way to incorporate some of the intermediate input cost linkages described above. So, if the price of plastic goes up the model can assess the impact this may have on the car industry – even though we are working at a more detailed level. In this way, it combines aspects of the CGE modelling with a PE model,” he reveals.

Let’s go back to how this sort of modelling is useful. A PE model is better at capturing short-run, direct impacts of changes in tariffs and Non-Tariff Barriers. Changes in the costs of labour or firms adjusting the mix and sourcing of the intermediates they use, tends to take place over a longer time period. The PE modelling basically asks ‘what happens if prices – or relative prices – change because of changes in tariffs or trade costs e.g. if the price of cars goes up by more than the price of something else’ or ‘the price of meat goes up because of changes in tariffs – how might production and trade adjust to that?’.

Hence, through the UK Trade Policy Observatory Michael and other Fellows of the Observatory have been using this innovative model to understand the impact of ‘No Deal’ on tariffs, prices, the cost of living and jobs. For example, their work showed that it is probably the higher skill and higher R&D-intensive manufacturing industries that may be most vulnerable to a hard Brexit. The researchers used the model to consider the impact on prices which, when combined with information on household consumption, could be used to think about how different income groups might be affected. They found that the average annual household spend could rise by £260, and that poorer households would be most affected by a ‘no-deal’ scenario in which tariffs and prices rose.

The team also used the model to assess the possible consequences on jobs for different parliamentary constituencies. The results suggest that failing to secure a close trading relationship with the EU could give the UK economy a shock equivalent to losing a total of about 750,000 jobs (only about half the value implied by the Government’s own estimates of 28 November 2018). Moreover, while those job losses will tend to be concentrated in cities and large towns, the people whose jobs they are, tend to live over much larger surrounding areas.

“In each of the above examples we are considering the consequences on industries, households and regions given the possible changes in tariffs and Non-Tariff Barriers. Remember, however, that as I said in the beginning, models will shed light on particular characteristics and mechanisms and by design leave others out. Our model doesn’t take into account, for example, changes in investment (most similar models don’t either). Neither does it take into account anything else that might shock the economy, or exchange rate changes, or changes in productivity. Nor do we include government policy responses to the shock,” states Michael.

“Yet, our model does provide a first pass at who might be the ‘Winners and Losers’ of the Brexit-induced changes to trade, or to put it slightly differently, who might be vulnerable from such a shock. Understanding that vulnerability, providing those sorts of insights, is important for evidence-based policymaking. As the UK tries to navigate its way through the politics of Brexit, providing clear, consistent economic analyses should help inform our choices and the decisions that are made,” concludes Michael.

A good model can provide such insights, it can be used to inform the debate and decision-making, and yet, it does not provide the last word. Models cannot predict the future. But they can provide useful inputs to guide policymaking.