How a neutral economic impact assessment of Brexit should be done

The Treasury and the OECD have produced assessments of the economic impact of Brexit. A number of think-tanks and private sector bodies have done the same. Virtually all of these studies have one key feature in common: they assume that, post-Brexit, the UK’s trade deals with non-EU countries are no better if we leave the EU than if we remain.

Many studies, in fact (e.g. that of the OECD) assume the UK’s trade deals with non-EU countries would be worse if we leave than if we remain. There are also all kinds of things to be said about the assumptions in those studies regarding the nature of deals with the EU (e.g. they typically assume that the trade deal in respect of services would decline materially, whereas I believe the most significant impacts would be on agriculture and certain key manufacturing sectors such as motor vehicles).

Be that as it may, for now I want to highlight the assumption about non-EU trade, because it points at something important about these studies. Very often, when one performs an “economic impact assessment” the point is not to provide a balanced account of the possible impacts of whatever one is studying. Instead, the goal is to quantify the impacts one believes most likely to occur. It’s an unpacking of a line, not a neutral analysis.

That’s fine, in its way. But suppose one did want to do a neutral analysis of the impact of Brexit. How should that work? I think it should be like this.

What you should do is to consider the claims each side (Leave and Remain) makes about how things will work under its favoured option (what Leave says will happen if we Leave, what Remain says will happen if we Remain) and the claims each side makes about how things will work under the other option (what Leave says will happen if we Remain, what Remain says will happen if we Leave). Then produce quantitative results for each of these scenarios. Then produce the authors’ judgement about how likely each of these scenarios is to come to pass.

This last point is key. I’m not saying a neutral judgement should not come to a conclusion. That’s not my complaint. There’s every chance that, if, say, the OECD had produced the kind of methodology I’ve described above, it would still have come to the conclusion that Brexit would lead to lower GDP in the short-run and by 2030.

To see what I’m saying would be different, let’s set things out in a bit more detail. Some of those advocating Leave say that if we leave the EU we will be able to enter into new trade deals with countries the EU would not have trade deal with and perhaps also deep preferential trade agreements that would be incompatible with EU membership. (As an illustrative example of the latter, suppose the UK set up a customs union with Canada and Australia post-Brexit. One cannot be both in a customs union with the EU and also a customs union with Canada and Australia unless those countries are also in a customs union with the EU (which they aren’t and won’t be). A customs union with Canada and Australia would therefore be a new deep trading relationship that would be at least technically feasible post-Brexit but could not be done if we Remain in the EU.)

The way a neutral Brexit impact assessment should proceed, in my view, would be to include a scenario in which the UK post-Brexit had new trade deals with illustrative countries the EU does not have trade deals with (e.g. Japan) and new preferential trade deals with illustrative new blocs such as Canada and Australia. Those deals need not be defined such that they would lead to the UK making a net GDP gain from trade deals alone, relative to whatever Leave-defined scenario for Remain is assumed, provided that there were other compensating advantages (e.g. perhaps under the Leave-defined scenario for Remain, it is assumed that regulation moves by 2030 in a direction that is damaging to the UK economy relative to being outside the EU). But ultimately, once everything was combined, the authors of a neutral study should do their utmost to construct the most plausible scenarios they can in which the Leave-described Leave scenario is better in economic terms than the Leave-defined Remain scenario or at least roughly equal to it (if one favours the Leave school of thought – represented by Roger Bootle, Gerard Lyons and myself – whereby the economic case for Brexit is that there is relatively little gained or lost, in GDP terms, by leaving).

Having devised quantifcations of the scenarios believed by the Leave and Remain sides, the authors of a neutral study would then be entitled to ascribe probabilities to them. They might, for example, say their judgement is that it is very unlikely that the UK will be able to set up a customs union with Canada and Australia, or that it is very unlikely that the Eurozone will grow faster if the UK is outside the EU, allowing the Eurozone to take control of EU institutions and use them to solve the Eurozone’s governance problems, or whatever other economic gains the Leave side claims of Brexit. I repeat: I’m not saying a neutral impact assessment should not come to a conclusion. The authors could say they believe the most likely combination of scenarios would be a, b and c and therefore they expect the net impact would be in the range X to Y.

But the key difference between the kind of approach I have described above and what has been done in other studies up to now is that there has been no attempt by allegedly neutral authors to place themselves in the mindset of those favouring the Leave camp, to see what gains they believe their might be. Instead, these models simply assume away all the sources of gain Leave campaigners claim – there are no gains from any new trade deals, no gains from the Eurozone growing faster post-Brexit, no gains from avoiding new damaging regulation the EU might impose in the future, no gains from resolving the UK’s internal constitutional problems (e.g. with Scotland).

To be truly neutral, an impact assessment needs to include some attempt to quantify the gains, as those on each side of the argument see them, before saying that those gains are unlikely to arise. Then we could have a better and more technically-informed debate.