Heresy on Free Trade

While economists in general are trained to evaluate all arguments sceptically, there is one big exception – Free Trade. Most economists are wedded to the idea of free trade to the point that many will routinely reject the results of mainstream economic analysis in favour of logically incoherent claims about dynamic effects, ‘cold showers’ and so on.

For a variety of reasons, I apostasised from the free-trade religion early on and, for a while, became an outright protectionist in reaction. Now, I don’t have a preconceived position either way, and try to assess the issues on their merits.

One point that comes out of any neoclassical economic analysis is that, at tariff rates below around 10 per cent, the (traditional trade-theoretic) benefits associated with a reduction to zero are trivially small. This is because the welfare loss associated with a tax are proportional to the square of the tax rate, and the square of 0.1 is 0.01 (1 per cent).

Over at Club Troppo, Nick Gruen makes this point, among others, and sets off something of a firestorm. Read, enjoy and comment either there or here as you please.

32 thoughts on “Heresy on Free Trade

  1. John,

    I’d describe myself as having a pretty strong free trade bias. It seems the most sensible starting point – especially since so many of the arguments which complicate the case for free trade and make it harder to prove that is the optimal policy (like the presence of economies of scale for instance) also tend to increase the costs of ill advised restrictions on trade. That’s because with economies of scale there are more gains from specialisation.

    But having started with that fairly strong presumption, you then have to consider the various aspects of the problem on their merits and see where it takes you. Unfortunately however the subject is taken as a litmus test of your sanity as an economist and trying to keep an open mind about the final answer as you work through the issues, let alone coming up with a conclusion that we can improve on free trade is usually taken as a sign that you are mad, bad and/or dangerous to know.

    And even if you agree with the particular case both you and I have suggested for moderate tariffs over zero tariffs on account of a variant of the ‘optimal tariff’ argument, this will generally not vitiate the case for further trade liberalisation in return for better access to other countries’ markets.

  2. The obsesion with reducing tariffs on international trade (trade between people who live in different countries) completely misses the main offence against free trade which is the chunch of tariffs that inhibit trade between domestic households. Domestic taxes are the major destroyer of human welfare, not international trade tariffs.

  3. JQ, that’s pretty much the position that immediately struck me as obvious when I first looked at the issue (I come from a mathematical background, of course). But it leaves out two things. One is that particular tariffs tip things one way or another, making some winners and some losers even though overall wlfare (using that as a technical term) is not materially affected. That means that the welfare matter, since it is not asking tghe who/whom question, is leaving some important questions unanswered – and, furthermore, there is rent seeking involved in prioritising just which material protectionism gets wound back first (it is quite possible for selective reductions in bias to result in more bias overall, and loss of welfare

    The other thing is that we don’t know a priori where the optimum is, if there are imperfections around (and some very subtle ones have been constructed in various times and places). That means that while we should make it a working assumption that the optimum is zero tariffs, it’s not necessarily true – we should check (and bear in mind that what is done with the revenues also counts in all this). Of course, if zero is not near the optimum, the O(2) behaviour isn’t the right approximation for the welfare result.

    Taken together, this is why the slogan “protection all round” is not quite the self evident nonsense it appears to be – things were done with the revenue, and the industrial environment tipped any sub-critical positive network externalities into super-critical behaviour. There are even better ways of doing that, of course, but nevert

    On top of which, rapid winding back – which is what happened – throws out babies with bathwater. Pointing out that Australia would have been better off without protected manufacturing in the first place is not the same thing as justifying chucking away what was in place. It forgets that the costs of the achievements, even if overpriced, then became sunk costs.

    I am always reminded of the Laurel and Hardy film where they try all sorts of ways of shifting a piano up some very difficult stairs. When they finally manage it, someone comes along and points out an easy way round – so Hardy gestures to Laurel and they take the piano all the way back down the stairs again, just so they can bring it up the easy way.

  4. PML,

    In investment terms (and life generally) sunk costs should not determine strategy. Each path forward should be evaluated on its costs and merits given the same common starting point (ie where you are at). Some pathways forward will be cheaper or easier because of past sunk costs but that is another matter, and the key issue is that it represents a cheaper way forward. For instance despite centuries of investment in animal husbandry farmers replaced horses with tractors when it offered a better way forward.

    Abolishing tariffs does not destroy the previous investment in plant and equipment nor does it destroy any skills that exist at the time of abolition. If after the change people find it easier to re-tool and re-skill and do something new then they will, otherwise they will be obliged to be more competitive in what they were previously doing. And retaining tariffs merely distorts the relative incentives in choosing a way forward. Of course this all assumes that there is not significant legislative or institutional barriers to the necessary adjustment processes (which may include wage and salary adjustments).

    I think that it was a good move to reduce tariffs and open up the economy, and it possibly had some healthy political implications as well. However since then we have had huge hikes in the tariffs on inter-household trade and these should be of far more concern for “free traders” than any of the remaining tariffs on Australias international trade. In so far as they both raise revenue and both destroy welfare, I kind of prefer tariffs on imports over tariffs on inter-household domestic trade.


  5. P.S. Given that one of the motivations for Federation was to elliminate domestic tariffs it is ironic that the Federal Government is now the largest source of domestic tariffs.

  6. On optimal tariffs, mainstream economic analysis does say that welfare may be maximised by a tariff that raises a country’s terms of trade. But it also says that this only works if other countries don’t do likewise. Why would we think that we could get away with levying an optimal tariff on another country’s exports and they wouldn’t notice, and wouldn’t retaliate if they did notice?

    On the the estimated welfare costs of protection, these were small even when tariffs were large, and now they are trivial.

    But is that because the costs really are small or because Harberger triangles which are the conventional measure don’t capture all the costs? Not all appeals to dynamic effects and so on are incoherent expressions of free market ideology and half-learnt economics (though a lot are). Paul Romer, who is an extremely accomplished economist, and will probaby win the Nobel Prize, once wrote a paper arguing that the welfare cost of proetction is an order of magnitude bigger than conventionally measured – a rectangle rather than a triangle – because of dynamic effects. (Sorry, don’t have a reference.)

    It’s true that all sorts of charlatans and ideologues waive their hands and proclaim dynamic effects but it doesn’t follow that there can never be dynamic costs to protection.

  7. “Abolishing tariffs does not destroy the previous investment in plant and equipment nor does it destroy any skills that exist at the time of abolition.”

    I have to say that I think this view is incorrect. The plant and equipment may physically be there, but the businesses that own them will not. If the plant is cannot be used profitably its value is destroyed. Without active use, skills atrophy surprisingly quickly. The economies of scale required to keep and an indurstry going can disappear pretty quickly as well.

  8. I think your post starts off on teh wrong foot:” While economists in general are trained to evaluate all arguments sceptically…”

    Oh really?

  9. Swio,

    Lets say you invest a million bucks in a plant to earn a return of 20 percent then a change in the terms of trade lowers that to 2 percent. That may lead to regret but it does not mean that you discard the capital. You merely downgrade its NPV and live with the capital loss.


  10. There may be some non-zero level of tariffs which makes sense. Once the protection option is on the table for politicians to use, what chance is there that the tariffs they impose will be sensible? Looking around the world I’d say it’s low.

    Isn’t it a bit like giving special counter-terrorist/drug dealer/etc. powers to the police — OK in theory, but very likely to be misused.

  11. The argument that Nicholas has propounded and which John seems to have supported is just wrong.

    I think they have in mind a simple two good model where the Lerner symmetry condition implies that it doesn’t matter whether you levy an export tax on goods in not perfectly elastic supply or levy tariffs on imports. In this simple two good model that it doesn’t matter which traded good that you tax to exploit the optimal tariff argument. Levying a tariff on imports in a two good world will generally drag resources away from the export sector, reducing output there and potentially advantaging the economy from an optimal tariff viewpoint.

    But in a model that even broadly captures the features of the Australian economy this argument breaks down completely. For example if you have market power only in wool exports and you put an import tariff on car imports, then in general, that tariff will draw resources away from all sectors, not just the wool sector. Lerner symmetry breaks down entirely.

    And given the vastly different production technologies in play here a tariff on car imports may not draw any resources from the wool sector – instead it will draw resources away from industriies with no export market power at all. In this case the economy will be worse off with any non-zero tariff on car imports.

    In general you can’t draw any conclusion without paying attention to the nature of complementarities and substitutabilities in the use of inputs in the economy as a whole. To suggest there is some general argument for low tariffs is an erroneous conclusion.

    And to present the rejection of this fallacious argument as JQ does as a kneejerk rejection of any argument against free trade is unfair. This is a very specialised argument that will almost never be correct. The difficulty with these fallacious arguments is that they create the impression – on the basis of an inapplicable 2*2 trade model – that there is a general argument for low levels of tariff protection in any economy with some market power in exporting. Since this would apply to almost any economy it would suggest a general case against free trade.

    There is no such general case and , in particular given the lack of input substitutability between the car industry and those industries where Australia does have market power there is no specific argument for low levels of import tariff protection in this particular setting.

  12. Harry, even on the informal logic of your argument the position is ambiguous and depends on what resources are drawn from where and where they go. To the extent that the economy adjusts to reduced tariffs through a lower exchange rate, then the result will be an expansion of export sectors. Many export sectors can expand without depressing their export prices, but some can’t and their expansion depresses their prices.

    It seems to me to follow that the optimal tariff is therefore positive. (Always allowing that the optimal trade policy response is a targeted export tax). The remaining issue relating to this argument then becomes whether this postive number is very small (in which case it’s best to forget about it) or just moderate to small – say 15, 10, 5%. Of course that’s not clear from informal reasoning.

    But isn’t that why we have CGE models? To try to add up all the columns and give us the best answer to this question we can in our ignorance get? Well the Monash and the Murphy Models say that the positive tariff is not very small – it’s moderate.

  13. Nicholas, It is ambiguous and that’s the difficulty. It is not an argument for retaining tariffs in the auto industry. I can’t imagine any resource transfers that would justify your claims since cars and wool/resources have completely different technologies and resource dependencies. What is most likely is that a tariff on cars will drag resources away from industries without any monopoly power and will thereby reduce welfare.

    Thus without detailed knowledge of the economy you get back to the standard idea that even low levels of protection are a bad idea. They won’t cost much but won’t generate advantage.

  14. That’s what the model is there for – to try to track those relationships. Why aren’t the CGE models the Commission used the best way to determine that?

  15. But the modelling you refer to does not support your case. It rejects it. GDP rises when tariffs are cut contrary to your claims. You were presenting a general argument for setting low tariffs on imported cars on the basis of monopoly power in exporting wool and minerals. My claim is that this argument is logically unsound and empirically not supported by the PC modelling. Moreover I understand entirely why the PC modelling does not support your claims – the argument you develop requires that resources will flow from wool and minerals into the car industry. But their are only the most indirect input substitutions here. What is more likely is that the resources will come from industries without monopoly power in exporting and such effects will be immiserising.

    And your argument really causes problems when JQ tries to turn it around as a general argument favouring low levels of protection on the grounds that economists have a blindspot with respect to their belief in the case for free trade. We don’t – we just won’t be misled by fallacious arguments based on resource transfers that won’t occur.

  16. Tom Davies hit the nail on the head – granting special privileges is just creating a lobby for more privileges (just like granting privileges to cops and spooks), and it’s also an incentive for others to lobby for privileges for themselves. It’s like putting blood in the water – a sniff that the government might hand out some rents brings the sharks from all over. It’s a lot easier to stitch up a deal with an Industry Minister than actually compete in a market.

    I lean to Harry’s view that we just don’t have that much influence on our ToT. But I’m not inclined to chase it up and test that view seriously because I think the net welfare effects of a 10% tariff in isolation aren’t really the main point at issue.

  17. JQ, harry clarke, Nicholas Gruen, PML,

    I hesitate to make comments on this topic because I believe Nicholas Gruen makes a valid point when he writes: “Unfortunately however the subject is taken as a litmus test of your sanity as an economist and trying to keep an open mind about the final answer as you work through the issues, let alone coming up with a conclusion that we can improve on free trade is usually taken as a sign that you are mad, bad and/or dangerous to know.”

    I assume CGE stands for ‘computational general equilibrium’. I would agree that nothing less than a G.E. approach will do.

    Potentially, there are many such models. The computational aspect is important for specific policy decisions but not for the analysis of general arguments.

    My questions and comments pertain to the underlying theoretical models of the CGE models.

    How does the model(s) you have in mind treat multinational firms? How does the model treat negative externalities (air, water, acoustic pollution) associated with the transportation of internationally traded commodities?

    The gains from ‘international trade’, however measured, are overestimated in the absence of full-costing of negative externalities (pollution).

    In my theoretical general equilibrium model of ‘competitive private owership economy with partially segmented markets and multinational producers’, the solution is difficult to analyse regarding the distribution of gains from trade because they depend on the relative values of technological parameters (‘technological development’), shareownership distribution and a few other parameters. A general conclusion is that even though ‘everybody’ in this model behaves ‘nicely’ (in terms of neo-classical assumptions) and all commodities are marketable, there is ‘unintended exploitation’ (shareholder unanimity breaks down; some sub-sets of consumers may lose while others may gain; these sub-sets may correspond to ‘countries’ under specified scenarios; permanent uneven development).

    Even though all commodities are ‘marketable’, there is an externality problem, which might be similar to the one PML referred to.

    I presented this model once to a wider audience (ANZIBA) in 1998, a time when globalisation was promoted. It wasn’t a ‘nice’ experience – I wasn’t called mad or bad, but the disappointment on many faces ..

  18. Just a quick question. If having some tariffs is, or may be, a good idea, why should the optimal area covered by those tariffs be related to the nation state? Is there some other area to which a tariff should be imposed? Would having a tariff at the (Australian) state level be a good idea? At the local government level? Or should it be imposed at a regional level, like over the whole of South East Asia?

  19. Andrew Reynolds, are you talking about ‘tariffs’ imposed by multinational firms in the form of ‘transfer pricing’? If so then you’ll get many ‘optimal geographical areas’, potentially one for each multinational corporation. Each one of these MNC-optimal areas would be technology determined and changing over time.

  20. Ernestine,
    While not specified in the post, I thought this thread was on government imposed tariffs. If you wish to go off on a tangent, up to you, but I would prefer to keep on topic – if only to avoid having you later accuse me of being off topic.

  21. JQ,

    Please advise if this thread is limited to ‘government imposed tariffs’ at the exclusion of ‘multinational corporations imposed tariffs’ in the form of transfer pricing?

  22. Andrew raises a good point. How it is answered may depend on whether you have an agenda of nationalism or one of human welfare.

    A follow on question might be whether such tariffs improve the welfare of Australians at the expence of the welfare of foreign people and whether this is automatically a good thing.

    We do have tariffs on trade between Australian localities. Divide the county into households and you will note that tariffs on multilateral trade between households is very high. Which is why many services in my home (gardening, renovating, accounting, food preparation, cleaning, investment advice, painting, plumbing and other household repairs) are produced domestically rather than being imported from more skilled members of other households.


  23. As a sometimes semi-Marxist I sympathise with the dynamic efficiency argument. Another point that puzzles me is how often the argument is made that by increasing prices to consumers tariffs are regressive. This argument seems a bit demagogic. Would this actually be the case in a general equilibrium analysis after all the effects had worked through the system? In an overall analysis tariff reductions are linked to deunionisation (see AWIRS) with consequent regressive impacts on income distribution.

  24. Geoff R,
    Do you have any support for the contention that deunionisation reduces real wages across an economy? I remmeber seeing some analysis some years ago (I will have to dig it out) that indicates that a high unionisation rate correlates with low overall wages growth.

  25. Andrew,

    Which causes which. Does low unionisation cause high wages growth or is it the other way around.


  26. Harry,

    This is what I claim.

    1) The PC modelling – its use of the normal specifications of the Monash model – came up with the result that consumption would fall if car tariffs were reduced and quantitatively illustrated my qualitative argument.
    2) GDP would rise, but the PC took consumption to be the correct measure of welfare – see p. 302 of its report. (Are you able to provide us with any reputable source for the argument that GDP should be prefered to consumption as a measure of welfare?)
    3) If GDP is the correct measurement of welfare – you win. The PC modellig doesn’t show what I claim it does – that cutting tariffs reduces welfare.
    4) If consumption is the correct measure of welfare then on your argument above, the PC modelling illustrates and quantifies my case. (I did not say and never said that the PC supported my argument. I said it’s modelling did.)
    5) The PC then added a ‘cold shower’ effect in which cost pressures elicited productivity improvements in the car industry. I’m unaware of the PC having done this in any inquiry before this one. In this new modelling the right trade policy would be an import subsidy and one might reasonably argue that taxes should be raised on business to make that shower that bit colder – improving productivity further. If this further modelling means it is no longer legitimate to refer to the rest of the PC modelling as PC modelling – then I guess you win again.

  27. Terje’s hypothesised company that sees its margins fall from 20 per cent to 2 per cent when tariffs fall has little basis in reality.

    Firstly, very few manufacturing companies earn gross margins anything like 20 per cent; more normal companies with a net profit of 5 – 10 per cent of revenue would be plunged into loss by an 18 per cent fall in prices; worse, increased imports would cut their market share leaving them with fixed costs larger than their earned margin.

    The latter effect might be ameliorated if the tariff reduction was mutual and lost domestic sales could be made up by increased exports; but unilateral tariff reductions do destroy jobs and businesses.

  28. Sorry, read “net” for “gross” in second line of second para.

    Manufacturing gross margins are typically 30 – 50 per cent with admin, marketing, prodict development, finance, depreciation and tax chewing up most of this and leaving a 5 – 10 per cent net margin.

  29. John Legge,

    My statement was more a point of logic than an observation of reality. When it comes to reality I think that your point is fair enough. Perhaps I could argue that destroying the business is not the same as destroying the physical capital, however that would be ignoring the significant social capital tied up in the structural aspects of any business. And the latter is probably more significant than the former anyway.

    My broader point of logic was that rationally it is not sunk costs that should determine the way forward but rather a comparison of the future costs and the future rewards that would flow from alternate given decisions (allowing for risk also). Of course due to sunk costs the status quo is usually a lot cheaper to implement than change, so the benefits of change (or else the future cost of the status quo) needs to be quite high in order to justify such a choice.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s