CGE models try to simulate the future, but that future depends on the structure of the model. And that immediately faces the problem that we simply don’t know the “true” model of the economy. PwC’s CGE modelling for the CBI identifies a significant negative impact of Brexit by focusing on the impact of uncertainty and the migratory supply of labour. Academic modelling by professor Patrick Minford, however, produces a significant gain from Brexit because his modelling focuses on – arguably the most important issue – the cold shower boost to economic performance from opening up trade to world prices, outside the protection of the EU common external tariff.
The Open Europe think tank’s CGE modelling produces a net gain from Brexit when allowing for the effects of open trade and politically feasible deregulation. The range between the CBI and Minford models is huge, from a net cost of 5.5 per cent of GDP in the CBI modelling, to a net gain of 3.7 per cent of GDP in Minford’s model.
So what’s the takeaway? In my view it’s three fold. First, moving from the EU customs union to trading at world prices without imposing tariffs on imports is a clear positive. But the scale of this gain is very difficult to be precise about due to modelling and data issues. Second, any gain from world prices is likely to be amplified by moves to shrink the size of the state at the same time. Third, we have to be aware of the threat from intensified uncertainty, even though this is almost impossible to calculate accurately, and is being hyped by Project Fear – to the point of becoming self-fulfilling.
What about backward-looking modelling which seeks to quantify the impact of the EU on trade, and then the impact of trade on productivity and growth? Here again, the problems are immense. Synthetic counterfactual modelling attempts to establish a control group (as in medical research) and then compare its performance with that of the UK after the country’s accession to the then EEC. But no credible control group, however ingenious the methodology, can be devised. So that approach falls at the first fence.
Gravity models also suffer from serious drawbacks. Just as CGE models can’t capture the future, gravity models struggle to capture the past. One example is the extent to which the growth in UK export performance can be attributed to the EU, or simply to economic reforms in the 1980s. These issues are immensely important because the gravity and synthetic approaches produce big numbers for the potential losses from Brexit. Professor Nick Crafts, in his own review of the gravity and synthetic modelling literature, has suggested that UK GDP is 8.6 to 10.6 per cent higher because of EU membership. He compares this with a membership cost (direct fiscal contributions and regulatory costs) of around 1.5 per cent of GDP. If true, that’s a big gap and a big cost from Brexit.
But the numbers could be wrong, very wrong. First, because of the extent to which the estimated benefits of trade are in fact attributable to other factors not captured in gravity models. Second, by examining the gains from EU trade, while ignoring the potential gains from trading at world prices. Finally, because the range of estimates for regulatory costs is actually vast, from zero to 6 per cent of GDP.