Dr Huw Edwards on the ‘Economists for Free Trade’ report
Economists for Free Trade: Extraordinary Claims Demand Extraordinary Evidence, but they have provided none.
In discussing the ‘Economists for Free Trade’ (EFT) ‘report’, I am unfortunately unable to comment upon the numerical and modelling work in any depth, since no technical documents have, to date, been released. (The absence of serious data, model details or literature references is one reason why we need inverted commas around ‘report’). This does rather hamstring discussion, and, in my view, raises serious ethical concerns. Why should people take numbers seriously when no supporting evidence has been produced?
We do, however, have some idea as to Professor Minford’s thinking on trade. And, as Andrew Walker of the BBC states, there is some academic background to what he says.
The key paragraph from Professor Minford is the following:
What many people do not realise is that the biggest gains from free trade come from a country eliminating its own trade barriers against imports from the rest of this world. Indeed, most people think the opposite: that the big gains come from other countries lowering their trade barriers against our exports. But this is quite wrong for a country like the UK, which though the fifth biggest in the world, is still rather small relative to the world economy – about 3% of it. Why is this?
Let’s summarise this by saying that there are two ‘Minford Propositions’ here (I guess I am flattering the EFT report, as they are quite well-known). 1. Tariffs hurt the country raising them, 2. Tariffs do not hurt other countries.
There is indeed a trade model which supports this view: it is called the Heckscher-Ohlin-Samuelson model, which is well-known to most students of international trade, and was seen as the standard workhorse model of international trade up to the late 1960s.
In this model, markets are perfectly competitive, technology is at the same level globally, goods are homogeneous and factors can move freely within (but not between) countries. There are also zero transport costs, and we have to assume that a country is too ‘small’ to affect World traded prices. We also have to assume that all countries produce all goods, and that the number of factors of production is identical to the number of traded goods.
This model produces a number of important theorems (such as Stolper-Samuelson), and is analytically well-developed. It is true that, if all the above assumptions are met, a country will face fixed prices at its borders for imports and exports.
In this very restrictive model, we do indeed get the result that, if Britain were to impose a tariff – any tariff – upon its imports, it would simply drive up prices at home, and the losses to consumers would exceed the gains to the taxman and producers.
It is also true that, in this model, if the EU put a tariff on British exports, we could sell elsewhere at the same price, and so would not lose.
Every trade modeller wants to have a model that can deliver the above propositions if the right set of restrictive assumptions is made. But equally, the HOS model is far too restrictive to be used for analysis of trade protection by itself.
The reason is that the HOS model is inconsistent with the stylized facts of international trade. In particular, the HOS model says that, if country A exports a good to country B (because A produces it more cheaply), then B cannot export the same good to A. Yet this is exactly what we see all the time: Germany exports cars to France, but France also exports cars to Germany. In fact, ‘intraindustry trade’ dominates much of the trade that actually occurs between advanced countries in particular – and yet the model EFT are assuming would rule out such trade!
Hence, trade theory has moved on. New models have been developed, taking account of differences in the products different countries trade. This may reflect firm-level differences in products (the Krugman-Melitz models), national differences (Armington), very specific factor endowments or technological differences (Eaton-Kortum) or even just perfect competition but with far more products than factors (Deardorff).
While all these models may differ in subtle ways, they are in fact remarkably similar in many key respects: in fact, they are all observationally equivalent, and, when applied to trade data between countries, all are consistent with a ‘structural gravity’ formulation, which can be estimated (see Head and Mayer, 2014). Moreover, the welfare implications – whether a country gains or loses from a particular tariff, and how much – do not depend on the assumed underlying structure (although there may be distributional differences). (Arkolakis and Costinot, 2014).
This is important. Even though the EFT economists claim to dismiss the gravity model, their own model is, in fact, a special case of the gravity model (with a near-infinite trade elasticity). Given tariff data, we can, in fact, test whether this strong assumption holds up, and the overwhelming conclusions of the literature (Head and Mayer’s meta-analysis, 2014) are that it does not. In fact, the majority of studies estimate a long-term trade substitution elasticity of about 5 – which suggests that trade is ‘moderately flexible’ in response to tariff or exchange rate changes, but nowhere near flexible enough to produce the kind of results that EFT claim.
In this regard, I am afraid that Andrew Walker of the BBC was far too generous to the EFT ‘report’: there is indeed ‘a theory’ which supports Patrick Minford’s conclusions, but ‘economic theory’ (meaning the set of theories which are consistent with real data) rejects it outright.
The point is this. Simply running simulations on a HOS model, which imposes infinite price elasticity of trade, is nonsensical. We know this is inconsistent with the data. We need to estimate – or at least simulate – a more general model, and then see whether there is any model which is consistent with the ‘Minford Propositions’. Now, I have to say that I am not aware of any model which is data-consistent and consistent with the results EFT are claiming. If they think they have one, they should publish it – preferably after exhaustive peer review.
In the absence of their producing evidence of this sort, I am afraid that, even though EFT have released very little of their technical material, we really don’t need it. Patrick Minford’s discussion of the theory is enough for us to draw a very firm conclusion: their model is simply not credible, and any numbers or policy advice based upon it are junk.
So the best advice on what the press and policymakers should do about the EFT ‘report’? Unless the EFT economists can come up with some very surprising new evidence (and I will not hold my breath), the only thing worth doing with their ‘report’ is to ignore it.