2.1 The aggregation of individual welfare issue

First, free trade affects relative prices, and these changes in relative prices will in general help some agents within a country, i.e. make them ‘winners’, and hurt others, i.e. make them ‘losers’. For example, an increase in the world demand for beef would raise the relative price of beef in a beef-exporting country such as Argentina. Argentinian beef producers would benefit from this change in relative prices, while Argentinian beef consumers who are not beef producers would lose. Only in special cases, e.g. competitive models with no heterogeneity among agents, will this not be true. This creates a challenge for free-trade advocates: when within a group a change in economic circumstances creates winners and losers, it is problematic to find a non-controversial criterion that allows us to say that the aggregate effect, i.e. the effect on the group as a whole, is good or bad.

Economists have long struggled with the problematic nature of what one can say about aggregate welfare when some people are helped and others are hurt. As documented in Irwin (Reference Irwin1996, p. 179), this problem ‘came to the forefront of academic discussions in the 1930s’.Footnote ⁶

This flurry of intellectual activity led to the development of the Hicks–Kaldor compensation criterion. In essence, this criterion is satisfied for some change in economic circumstances when the gains to the winners are sufficiently large that there exists a compensation scheme from winners to losers that if carried out without use of any resources would make no one worse off and at least some people better off.

The development of this criterion leads us to the second major implication of economic analysis about the effects of trade: In models that assume perfect competition, a change from autarky to free trade satisfies the Hicks–Kaldor compensation criterion. This is sometimes described by saying that free trade represents a potential Pareto improvement.

But satisfaction of the Hicks–Kaldor criterion by and of itself does not make assessment of whether such a change is ‘good for the nation’ any less problematic. For, without actual compensation taking place so as to make no one worse off and, if not everyone, at least someone, better off, economic analysis is still unable to aggregate across winners and losers to calculate a net gain.

Hicks, one of its progenitors, acknowledged this and argued that it must be augmented by an additional argument: if all changes in economic circumstances that satisfy the compensation criterion, either from technical changes or policy choices, are adopted, then, even though a particular individual might be adversely affected by any particular change, on average, over all such changes, that individual would come out ahead.

Of course, the flaws in this criterion have been often pointed out.Footnote ⁷ For example, an elderly worker in the US (or Israeli) textile industry in the late twentieth century would unlikely have lived long enough to be a net beneficiary of trade liberalization in that industry. Samuelson (Reference Samuelson and Rosefield1981) pointed out the implicit assumptions in this use of the criterion as an arbiter of ‘good for the nation’:

As Samuelson's quote makes clear, the use of the compensation criterion as an argument for a policy of free trade should not be thought of as based on a straightforward logical implication of standard economic analysis. Rather, there are numerous empirical leaps of faith (‘quasi-independent’, ‘absence of known bias’, ‘hazy version’ and so on) that have to be made to think of this as a reliable guide to good policies.

His description of the theorem as ‘heuristic’ also implies that he believes this use of the Hicks–Kaldor compensation criterion is not to be viewed as having solved the problematic nature of the problem of what is good for the nation, but rather is to be used as a ‘guide’ to thinking about policy.

Thus, the problematic nature of deciding what constitutes something ‘good for the nation’ when not all members of the nation are helped can be called the aggregation of welfare challenge to free trade advocacy.

2.2 The distribution issue

Of course, there is a non-problematic criterion for evaluating whether a change in economic circumstances benefits a group as a whole: if every member benefits, then surely we can say the group is better off. That is, a change in circumstances that leads to an actual Pareto improvement, i.e. a change that makes no person worse off and makes at least one person better off, is good for the group as a whole. Given the general result that economic analysis does not provide a non-problematic way of aggregating preferences, we ask the question: are there ways in which free trade leads to an actual, rather than potential, Pareto improvement? We consider the two most-discussed possibilities.

2.2.1. The possibility of compensation. One way in which free trade could lead to an actual Pareto improvement would be if the requisite compensation were actually paid from winners to losers and the process of compensation did not use up so many resources as to create a net loss. On the face of it, this seems to be an unlikely actual possibility. First, providing compensation may be politically difficult. Second, actually identifying winners and losers would undoubtedly be difficult, would surely require the use of real resources as a government sets up and runs the necessary bureaucracies, and might induce serious deleterious incentive effects. Nothing in our theory tells us that the gains from trade must outweigh such costs of redistribution.

Samuelson (Reference Samuelson1966) made this point long ago. Within that paper, in Section VII titled ‘A warning about feasibility’, he asked:

He supposed ‘as is the simple truth’, that ideal lump-sum transfers are not available. Then, feasible redistributions give rise to ‘substitution and other effects’ such that autarky could be preferable to free trade. He claimed this shows ‘how difficult must be any rigorous interpretation of ‘potential improvement’’.

Of course, in basic models such as the Specific Factors or Heckscher–Ohlin–Samuelson, identification of winners and losers seems more straightforward, e.g. capital and labour in the HOS model. Thus, if these are the relevant models, there is more promise of a relatively costless way to carry out compensation. But current research emphasizes that heterogeneity in more recent models generates winners and losers in more nuanced ways than found through the usual aggregate channels. For example, Artuc et al. (Reference Artuc, Chaudhuri and McLaren2010) found differential effects of trade liberalization depending on the age and human capital of the individuals affected, while Krishna and Senses (Reference Krishna and Senses2009) found import penetration to have important effects on income risk, for which compensation is arguably harder to compute.Footnote ⁹

This presumption that in actual economies it would be difficult to actually transfer the appropriate resources from winners to losers so as to make free trade an actual Pareto improvement provides what I will call the compensation challenge.

2.2.2. The heterogeneity issue and choice of models. Another way in which one could argue that free trade leads to actual Pareto improvements would be to point to specific models for which this would be true. The challenge for economic analysis, though, is that models for which free trade generates an actual Pareto improvement generally have assumptions about heterogeneity that are not likely observed in actual economies. That is, every agent within an economy would have to be identical in every economic aspect – tastes, resources, and productive abilities – which seems on the face of it unrealistic.Footnote ¹⁰

But unrealistic assumptions are never the death knell for a model. What matters are predictions, and the prediction of these models is that every person is made better off by trade, an implication soundly rejected by fact. Thus, observing that in such models free trade generates actual Pareto improvements has nothing directly to tell us about whether free trade in actual economies is a good thing.

Existence of an actual Pareto improvement as opposed to a potential Pareto improvement makes a much stronger case for the benefits of free trade. But clearly, to argue that free trade is better than autarky on the basis of one of these models means that you think the additional assumptions about a lack of heterogeneity among agents are appropriate. Furthermore, one must be able to explain why actual economies that engage in trade liberalization appear to always have some losers as well as winners. I will refer to this as the heterogeneity challenge.

These implications provide a challenge to anyone who would like to judge whether a change in a nation's economic circumstances is good for the nation.

3.1 The starting point: an appeal to authority on the aggregation challenge

I start by seeing how most undergraduate texts and popular writings by economists begin making the case for free trade. As noted, here one would hope for the best analysis the profession has to offer. The starting point is much like that found in Krugman and Obstfeld (Reference Krugman and Obstfeld2009), where early in this best-selling undergraduate textbook they set the tone with the following statement:Footnote ¹¹

That is, they assert that free trade is good for the nation, but note that it may not be good for every member of that nation. By starting the discussion of free trade with such an unqualified assertion, the authors suggest that the concept of ‘gains for a country’ is not problematic, and that there is only the task of proving the assertion in the body of the text.

Are non-textbook writings different? Take, for example, the Op-ed column by Edward Prescott in the Thursday, 15 February 2007 Wall Street Journal, titled ‘Competitive Cooperation’.Footnote ¹²

Prescott starts his column as follows:

I take this statement to mean that he views as a job of the economics profession the teaching to non-economists the ‘fact’ that free trade is good for the society as a whole, even though some people are hurt.

What criterion does Prescott use for deciding when a change in economic circumstance is good for society? He implies that it is something like average welfare when he writes:

Perhaps he really has a more nuanced view, and simply believes that a thorough exposition of such a view is beyond the scope of an Op-Ed column. After all, he later uses fuzzier notions when he writes that openness is ‘the key to bringing developing nations up to the standard of living enjoyed by citizens of wealthier nations’, and that ‘countries that commit to competitive borders will ensure a brighter economic future for their citizens’ (italics mine). But the evidence he brings to bear in support of his view that free trade is good is simply a comparison of per capita GDP growth for areas of the globe that have had different levels of protectionism. I think it is clear that he implies that there is no aggregation of welfare problem for economic analysis.

Or consider what the University of Rochester's Stephen Landsburg wrote in a 16 January 2008 Op-Ed column in the New York Times, titled ‘What to Expect When You're Free Trading’:

Again, I think it is clear that Landsburg is claiming that mere satisfaction of the Hicks–Kaldor compensation criterion is enough to argue that the nation is better off with free trade.

Thus, the standard argument for free trade thus begins with an appeal to authority, e.g. ‘all economists know . . .’, that implies the non-aggregation challenge has been settled by the economics profession. To begin the argument for free trade with such a misleading – or, at best, confusing – claim seems clearly a rhetorical choice that short-circuits critical thinking by the non-economist readers.

To strengthen this criticism, consider how economists in other sub-disciplines grapple with the same issue that arises in trade: what is one to say about society in the face of changes in economic circumstances that help some and hurt others? In contrast to the international trade literature, the work of economists in other sub-disciplines addresses this issue head-on.

For example, Edith Stokey and Richard Zeckhauser (Reference Stokey and Zeckhauser1978) point out that there is an ‘inability to find unassailable criteria for resolving conflicts on social policy’.Footnote ¹³

With this impossibility in mind, they introduce the Hicks–Kaldor compensation criterion, point out the obvious problems with it as a straightforward non-problematic criterion for assessing whether a change in circumstances is ‘good for society’, and finish with a description of Hick's justification based on the grounds that over time things averaged out for a net benefit for any one person.

But they point out the features of Hicks’ ‘over time’ interpretation that informed Samuelson's description of it as a ‘heuristic theorem’. They then go on to discuss what they see as a consensus view among policy makers of ‘distributional guidelines’ that help them grapple with whether or not a proposed change in economic circumstances should be considered ‘good for the group as a whole’. Among others, these guidelines include:

1. 1. A change should be allowed or implemented when the Hicks–Kaldor criterion is satisfied and the winners and losers are in roughly similar circumstances and the changes in well-being are ‘not of great magnitude’. They illustrate this by a description of a librarian who buys more gardening books than ornithology books because the existing gardening books circulate more than twice as often as the ornithology books.

2. 2. ‘It is not clear’ if a change should be allowed or implemented if such a change benefits some groups only by imposing ‘significant’ costs on others.

They finally point out that, given the problematic nature of resolving conflicts about what constitutes a change in circumstance that is good for the society as a whole, people have focused on whether the process by which such conflicts are resolved is ‘legitimate’.Footnote ¹⁴

Hausman and McPherson (Reference Hausman and McPherson2009) suggest that among most economists interested in public policy the problematic nature of the compensation criterion is well understood, as evidenced by their quote:

Zajak (Reference Zajac1995) discusses the issue in terms of how he thinks legislators and regulators take account of ‘fairness’ issues when contemplating policy. He emphasizes that, along with Hicks–Kaldor compensation, these people take account of such things as the inherent fairness of status quo property rights, societal insurance against large economic loss from exogenous economic changes, and removal of ‘significant’ inefficiencies that benefit ‘special interests’. And somewhat like Stokey and Zeckhauser, he points out that notions of fairness with respect to decision processes are important to people who are affected by and who make economic policy.

Thus, when advocates of free trade choose begin their argument with the dubious claim that they, as authorities, can vouch for the economics profession's success in figuring out what makes an economic policy ‘good for the nation’, they are using rhetorical devices that put them firmly in the ‘wooing women’ camp of argumentation.

3.2 Addressing the distribution issue

Textbooks address the distribution issue by first analysing the Ricardian model, which implies that free trade leads to an actual Pareto improvement. That is, they analyse a model in which there is no distribution issue, and point out that in that model everybody wins from free trade. They then introduce models with heterogeneity, e.g. the Specific Factors model, or the Heckscher–Ohlin–Samuelson model, which requires them to address issues of distribution, which they do with a variety of cognitive devices and off-the-cuff analyses.

3.2.1. The irrelevance of the argument based on Ricardo. After asserting that free trade is good for the nation, most textbooks then develop the two-country two-good Ricardian model and the associated idea of comparative advantage. As usually presented, this model assumes that labour, the only factor of production, is perfectly mobile and that production takes place under constant returns to scale.Footnote ¹⁵

The theorem developed states: if the pattern of trade is determined by comparative advantage, then no one (in either country) can be hurt by trade and some people are made better off.Footnote ¹⁶

The appeal of this model as a basis for promoting free trade is surely based in part on the readily understood individual-model analogies that illustrate the concept of comparative advantage. Almost everyone can understand the logic of why a professional golfer such as Tiger Woods should not weed whack his yard, or why an incredibly good typist who is also a brain surgeon should hire a typist and specialize in surgery. But are these analogies to the point?

Of course not. The analogies help us understand the model, and help us understand the benefits of individual specialization along the lines of comparative advantage. But the model is inappropriate for thinking about actual trade situations in which people are not identical, i.e. situations in which people have differentiated abilities and opportunities for adapting to changing circumstances.

At the end of the day, what is the argument for free trade that is based upon the Ricardian model? It says: we have a model that implies free trade is good for everyone. Of course, anyone who pays attention to current events or is familiar with a little history knows that actual cases of trade liberalization do hurt some people. Hence, if we are looking for a model to address the question of whether, in actual economies, free trade is good for the nation as a whole, this is not a good one.

In an article titled ‘Reconsidering Free Trade’, Hahn (Reference Hahn1998) made similar points. He set out in that paper to ‘reconsider whether on balance economic theory can make a convincing case for free trade’.Footnote ¹⁷

He developed the textbook Ricardian model, and then went on to point out ‘a number of important assumptions and simplifications . . . which need to be firmly kept in mind’. He concluded that only if ‘losers’ from trade can be identified and compensated can there be any propositions such as ‘country A is better off under free trade’.

Two further points. This critique is not a critique of the use of simplified models to help us organize thought and understand the world. The Ricardian model does help us understand some things about the world, e.g. the pattern of trade in a world of perfect competition. But the implications of the model for ‘gains from trade’ should not be used as an argument in favour of a policy of free trade: it is just not the right model for helping us think about this question.

Second, many economists claim that the Ricardian model is a necessary antidote to the arguments of Philistines in the world who believe that all trade, whether between two individuals or between two countries, must be a zero-sum game. I would argue that the appropriate use of the Ricardian model is then to counter the zero-sum argument about trade between two individuals. It is simply over-reaching to try and use this model to counter the zero-sum argument about trade between countries, and it destroys the profession's credibility: the student who, having diligently learned the ‘free trade is good’ lesson of the Ricardian model, and who realizes later in life that the individual analogy is a poor one for countries, might be tempted to dismiss all the economics learned at that earlier age.

Finally, one might think that this model is not being used to promote the idea that free trade is good, but is rather being used to develop the idea of comparative advantage. But in textbooks such as Feenstra and Taylor (Reference Feenstra and Taylor2008), we find that the authors first assert that free trade is ‘good for the nation’ in the context of their development of the Ricardian model in Chapter 2. They conclude from their analysis that:

They are not implying that this is a model-specific result that must be interpreted with care, as evidenced by how they segue to a statement about more complicated models:

By simply stating that in these more complicated models in which some people lose there are still ‘overall gains’, they clearly intend for the reader to take away from the Ricardian model that ‘overall gains’ is a relevant lesson about the gains from trade.

My view that the free-trade advocates in the economics profession really want people to take from the Ricardian model an important lesson about the benefits of free trade in actual economies is shared by the noted health economist Uwe Rheinhardt, who in his New York Times blog entry on 4 March 2011, writes the following in a discussion about the argument for free trade:

Of course, as noted, anyone who pays attention to current events – or current research – knows that actual cases of trade liberalization do hurt some people. After developing the Ricardian model, textbooks do acknowledge that, in actual economies, trade creates winners and losers, and thus quickly move on from the Ricardian model to richer models that account for this fact. But by arguing first that the Ricardian model teaches appropriate lessons about ‘gains from trade’, they are implicitly arguing that the problematic nature of the concept of gains from trade in the more realistic models is not something to worry about. Thus, they are encouraging readers not to think critically about the implications of these more realistic models.

3.2.2. Addressing the distribution issue with heterogeneous-agent models. Having acknowledged that more complicated – and relevant – models imply that trade creates winners and losers, textbooks move on to explicate these more complicated models. These might be the specific and mobile factors model, or the Hecksher–Ohlin model, or even the endowment economy model in which residents of a country have different tastes. To come to grips with the distribution issue that arises in these models, they generally make unsubstantiated claims about how one should think about what ‘good for the nation’ means, and introduce cognitive devices with language that misleads readers concerning what lessons should be learned from them.

For example, after introducing models that imply some individuals are likely to be hurt by trade, Krugman and Obstfeld (Reference Krugman and Obstfeld2009, pp. 72–73) then get around to pointing out the problematic nature of the concept of ‘gains from trade’. Their answer to this problem is to assert:

Note the use of ‘better’. They do not elaborate on why this is a better way, or even claim that they will at some point elaborate on why this is better. The reader is left to conclude that the authors – acknowledged experts in economics – have this all figured out.

3.2.2.1 Community indifference curves and national budget constraints At this point, if not earlier, textbooks frequently introduce the concept of community indifference curves, and show how free trade puts a nation on a higher indifference curve. For example, Feenstra and Taylor (Reference Feenstra and Taylor2008) introduce community indifference curves earlier, in their chapter on the Ricardian model, with no discussion of the strong assumptions necessary to make for this concept to be an accurate representation of aggregate preferences. Alternatively, other textbooks at this point argue that there exists a compensation scheme that redistributes goods from winners to losers so that everyone could be (hypothetically) better off under free trade than under autarky.

For example, to prove that it is possible for the winners to compensate the losers, Krugman and Obstfeldt (Reference Krugman and Obstfeld2009) point out that the aggregate budget constraint for the economy as a whole includes consumption pairs that include more of both goods than does the autarkic production/consumption point.Footnote ¹⁹ They conclude:

This of course is not quite right. First, an aggregate budget constraint is not like an individual's budget constraint in the following sense: an individual's budget constraint can legitimately be thought of as describing a menu of possibilities, while an aggregate budget constraint cannot. An aggregate budget constraint reflects individual choices, but to be interpreted as a menu of possibilities requires us to think about how a government or society could move aggregate consumptions along the constraint. As noted earlier, such movements may be politically difficult, surely would require the use of real resources as a government sets up and runs the necessary bureaucracies, and might induce serious deleterious incentive effects. Again, nothing in our theory tells us that the gains from trade must outweigh such costs of redistribution, and the earlier quote from Samuelson (Reference Samuelson1966) remains telling.Footnote ²⁰

In summary, what Krugman and Obstfeld have done is to introduce the idea of the compensation principle as the appropriate criterion – a ‘better way’ – for assessing whether a free trade policy is desirable, but have not made an argument for why this is so. A thoughtful reader is likely to be confused: why argue that the mere existence of hypothetical redistributions of income is a good criterion? For the argument that free trade is a good policy to be persuasive, the development of the compensation principle must be followed by further arguments about why this can inform, but not conclude, thought on this topic.

Clearly, Krugman and Obstfeld understand that there is a problem with a hypothetical criterion, because they go on to immediately say that while everyone could gain, not everyone actually does. And they point out that, given this reality, economists still ‘do not generally stress the income distribution effects of trade’ (p. 74). By feeling the need to give additional reasons beyond the compensation criterion, they imply that the compensation principle is not reason by itself to argue that free trade is good. And their discussion of why economists do not generally address the distribution challenge is, I claim, likely to confuse readers even more.

3.2.2.2 The arguments for why economists stress potential gains rather than possible losses As representative of other (and most) international economists, Krugman and Obstfeld give three reasons why economists, while aware of the losses imposed on some members of the nation by a change from protectionism to free trade, generally support free trade. The first two reasons are inconsistent: they cannot both be good reasons. And the third reason only has force if in fact the problematic nature of ‘gains from trade’ has been settled. I think these reasons are not persuasive justifications for the support of free trade by economists.

Point one starts by noting that every change in a nation's economy creates winners and losers. Krugman and Obstfeld then note that if every change were only allowed after examination for distributional effects, ‘economic progress could easily end up snarled in red tape’. By equating these changes with ‘economic progress’, I assume that they are referring to changes that are potential Pareto improvements, i.e. changes that satisfy the compensation principle. Of course, as argued above, this concept of ‘economic progress’ is not persuasive without a further development of why existence of a potential Pareto improvement is a good criterion for assessing whether a change in economic circumstance is a good thing. But note that this point does seem to be arguing that actual redistribution may more than offset the improvements that might accrue from the economic change (‘snarled in red tape’).

Point two argues that it is better to allow trade and compensate those hurt. They say this would be true of other forms of economic change as well. This is subject to the critique developed above about why the compensation criterion needs to be thought of as a hypothetical redistribution: we have no theoretical assumption that the costs of redistribution don't outweigh the benefits of the change in economic circumstances. And furthermore, this point stands in direct contradiction to point one!

In addition, the logic of this point is that economists should always advocate a joint policy of free trade and compensation schemes that cushion losses. But in reality, such choices of joint policies may not be available because of political costs. In practice, adoption of freer trade policies frequently seems to entail simultaneous adoption of other inefficient policies. For example, legislative support of CAFTA by representatives of districts where sugar beets were an important crop was ‘bought’ by subsidies to production of sugar beets into ethanol.Footnote ²¹

Thus, the logic of these two points seems to be: on the one hand, we should ignore the problem of losers because of the costs of compensation; on the other hand, the problem of losers is so important that we should only advocate free trade as part of joint policy which insures potential losers. These can't both be correct. Again, it appears to me that non-economists can only make sense of this dichotomy if they have seen the ‘fleshed out’ arguments about why the compensation principle should inform one's thinking about policy.

Finally, the third point says that losers from free trade are better organized politically than those who stand to gain. The economist's role is to provide a counterweight to this political bias by ‘pointing to the overall gains’. This has force only if the case for ‘overall gains’ has been made, which, I argue, has not yet been done in the traditional arguments as presented in popular writings by economists and in textbooks.

What could account for the juxtaposition of these partially-contradictory points? It seems to me that it is simply an uncritical view of the rightness of a policy of free trade, a view that no longer invites critical thinking from the promulgators of this view.

3.3 Other rhetorical choices

3.3.1. Cherry-picking implications: the ‘lower price’ fallacy. Consider the following quote from Krugman and Obstfeld (Reference Krugman and Obstfeld2006):

Or consider the discussion between ‘Dave’ (Ricardo) and ‘Ed’ in Russell Roberts’ fable of a return by Ricardo to the USA in 1959, The Choice: A Fable of Free Trade and Protectionism (Reference Roberts2001). Ed owns a television-producing firm, and is considering supporting a presidential nominee (in 1960) who favours protectionism. A key part of the story is a discussion by Ricardo (illustrated by a trip to the future in the year 2005) of how resources that were used to produce televisions have been released to produce pharmaceuticals, which are exported. At one point, Ed and Ricardo engage in a conversation about the effects of a tariff on televisions (Chapter 7, ‘Do Tariffs Protect American Jobs?’). Dave convinces Ed that a $25 tariff will raise the price of a television by $25. Dave points out that this makes consumers of televisions worse off, and argues that the efficient policy would not do this, but would allow TV imports that free up resources to work in a pharmaceutical plant.

This line of argument that says ‘consumers’ benefit from free trade because of lower prices is ubiquitous: news stories frequently make this point, as did a Wall Street Journal article that pointed out the benefit to consumers of lower prices for baby clothes that would arise if restrictions on Chinese imports were eliminated.Footnote ²²

The question not asked, in Krugman and Obstfeld, in the imaginary dialogue between Dave and Ed, and in newspaper articles, is: would other prices be higher without protectionism? The answer is most likely yes. Consider the question not asked in the dialogue about the higher price of TVs under protectionism: what is the price of pharmaceuticals under TV protectionism, relative to free trade? The answer, given the assumptions Russell makes about increasing marginal costs and upward-sloping supply functions, is that they are lower.

This of course is a fairly general proposition. Tariffs, by changing relative prices, move an economy along the PPF, leading to a lower quantity supplied of some other goods. This implies that the relative price of these other goods has gone down. Put another way, the Lerner symmetry theorem alerts us that exports pay for imports. Under most conditions, this will mean that fewer imports will lead to fewer exports, and a lower price for these exports, i.e. leads to a movement towards lower quantity along the export-good supply curve.

Economists are fond of invoking the ‘exports pay for imports’ argument to point out that protectionism might save import-competing-sector jobs but also logically entails loss of export-sector jobs. But the ‘exports pay for imports’ argument also implies something about prices as well. The observation that the prices of televisions are higher with protection does not necessarily imply consumers are worse off. Surely there are some consumers who don't care about televisions but do buy medicines. Likewise for the case of clothing: consumers of clothing pay a higher price for clothing under protectionism, but a lower price for some exported good, e.g. food.

Again, Samuelson (Reference Samuelson1939) made this point long ago. In this paper, he first showed gains from trade for an economy with identical consumers. But he went on to discuss ‘more realistic cases where individuals are not all alike’ (p. 203). About this more realistic case, he said:

What is especially disconcerting is that our textbooks do point out that changes in relative prices in general hurt some consumers. For example, textbooks generally point out that in both the Specific Factors and HOS models some identifiable groups, e.g. workers, or capitalists, lose from relative-price changes: their real incomes fall, whether measured in units of the good whose relative price has risen or in units of the good whose relative price has fallen. But this general result seems forgotten when a partial-equilibrium (apparent) result that supports free trade is available.

3.3.2. The use of analogies and metaphors to persuade. A favoured analogy used to make the argument for free trade comes from Adam Smith:

As with analysing the usefulness of any analogy as an aid to thought, we must ask in what ways families and countries are similar and in what ways different. The key insight of this analogy concerns specialization and the division of labour: a family that was forced to be self-sufficient would undoubtedly be worse off than that same family in a world in which members could specialize. The Ricardian model gives us a formalization of when the same is true of a country.

But families and countries are different in ways that makes false the glib assertion that ‘what is prudent . . . of every family . . . can scarce be folly . . . in . . . a great kingdom’. Families are composed of relatively small numbers of intimately connected individuals. A move from autarky to trade might create some ‘losers’, but these could be easily identified and compensated. This is just not the case for a nation with many millions of individuals.Footnote ²⁵

Another often-invoked metaphor is that found in Bastiat's Petition of the Candle Makers. In this satire, candle makers were said to petition the legislature for relief from ‘unfair competition’ from the sun. The position of the candle makers is undoubtedly supposed to be analogous to the position of advocates of tariffs and quotas in that their narrow self-interest is opposed to the interest of the general public.

But notice the subtle framing of this satire: the status quo here is free light from the sun for everyone. In the satire, candlemakers are petitioning to take away from non-candlemakers something important that they have historically been given for free. The candlemakers are attempting to improve their position vis-à-vis the status quo. They epitomize a greedy special interest that is attempting to impose significant losses on many people.

But in most cases in which import-competing industries ask for protection, the status quo is a level of employment and profits – and prices – that are now being threatened. The petitions in these cases are trying to protect the status quo benefits of a small concentrated group, at slight potential individual cost to many people. This is certainly different from the Bastiat scenario.

There is much about which this metaphor can teach, but it is not that ‘free trade is good’. It fails to grapple with the fundamental question that must be addressed about the beneficence of free trade, namely, what can one say when some members of society are helped and others hurt?

3.3.3. What about all those empirical studies?. Most empirical studies quantifying gains from trade rely on some version of showing the consumption-possibilities frontier moves out, or implicitly assume there exists some representative citizen, i.e. Samuelson's aforementioned ‘less realistic’ case of identical individuals (Samuelson Reference Samuelson1939). Is this sound? Consider the analogous idea of saying the USA is better off in 2007 than four years earlier because GDP has grown. If, as some assert, all of this growth has been accounted for by growth of income for people in the upper parts of the income distribution, partly at the expense of those in the lower parts, is GDP the right measure of ‘good for the nation’? Reasonable people surely can disagree on this question.

Of course, a claim has been made that there is much empirical evidence about the gains from trade. Irwin (Reference Irwin2002) wrote:

Some of the empirical evidence that Irwin (Reference Irwin2002) reviews is subject to the above critique. Some, though, is evidence about the ‘refined and updated’ insights. For example, he points out that trade might be expected to lead to an expanded range of consumer and intermediate varieties available to the domestic economy. He also points out that trade contributes to productivity growth because it serves as a conduit for the transfer of foreign technologies and because it forces domestic industries to become more efficient, and he provides references to studies that support this view.

Leamer (Reference Leamer2007) also addresses the contribution of trade to what he calls ‘the mobility of ideas’.Footnote ²⁶ He points out that ‘stowaway ideas’ travel with goods and services and lead to technological progress. He claims that after the Second World War, one-third of the globe formed a trading network and the other two-thirds looked inward. He argues that the technological progress created in the trading group left the other two-thirds so far behind that they eventually abandoned their inward-looking strategies.

Finally, in his Op-Ed column, Prescott (Reference Prescott2007) argued that research provides evidence that competitive barriers are the reason that one part of Leamer's inward-looking group, Latin America, had growth over the period 1950–2001 that lagged far behind that of Europe, the USA and Asia.

I would not claim that increased variety and ‘stowaway ideas’ are not part of the gains that countries reap from participation in the global economy. I would point out, though, that there have been thoughtful criticisms of the evidence about these and related gains, and that the evidence is more qualified than is sometimes claimed.

First of all, consider the wide criticism of the inward-looking import-substitution industrialization (ISI) strategies of Latin-American post-Second World War economies, in which tradition are Leamer's (Reference Leamer2007) and Prescott's (Reference Prescott2007) points. As noted by Diaz-Alejandro (Reference Diaz-Alejandro and Kenen1975), growth rates for many Latin American economies that followed ISI strategies were, at least until 1973, quite good. Furthermore, he argued that there was persuasive evidence that many LDC import-substituting industries were not as ‘inefficient, uncompetitive, and economically stagnant’ as the many critics of ISI portrayed them to be. Cardosa and Helwege (Reference Cardosa and Helwege1992, p. 11) assess the evidence as follows:

Indeed, if we look at the 1950–1980 sub-sample of the data used as evidence by Prescott (Reference Prescott2007), the period for which ISI policies were most prevalent, we find average annual per capita GDP growth rates for Latin America to be 2.7% over this period, better than that for the USA and roughly on par with the outward-looking part of the globe referred to by Leamer (Reference Leamer2007).Footnote ²⁷

Diaz-Alejandro (Reference Diaz-Alejandro and Kenen1975) also suggested that the evidence we had at the time was in an important sense too fragile for the making of confident policy prescriptions. As he put it:

A related point about the evidence of the trade-growth nexus has been made more recently by Rodriguez and Rodrik (Reference Rodriguez, Rodrik, Bernanke and Rogoff2001). They argue that econometric problems make the conclusions that trade begets growth also unreliable. And the World Bank's The East Asian Miracle: Economic Growth and Public Policy (1993) offers some support for the idea that ISI strategies were important precursors of the export-led growth policies eventually followed by Japan and the East Asian ‘tigers’.

The point is not that there is no evidence that good things happen to many people as a result of trade. Rather, the question is why evidence that is less supportive of the idea that there are gains, or that these gains are due to a combination of policies and not just trade liberalization, gets short shrift? Diaz-Alejandro (Reference Diaz-Alejandro and Kenen1975) believed that part of the problem came from the disproportionate influence of Anglo-Saxon and Scandinavian economists, who were overly influenced by their own countries’ positive experiences with trade liberalization.

Perhaps another reason is that, even though they are scientists, economists suffer from cognitive dissonance. For example, Magee (Reference Magee and Kenen1975) noted that empirical researchers of the 1950s and 1960s who estimated trade elasticities, ‘free traders’ that they were, had what he suggested may be ‘a subconscious desire for empirical justification of these normative judgments’. That is, they easily saw reasons to dismiss estimates that did not correspond to their subconscious desire to find high trade elasticities.

4.1 Be clear about what economic analysis can say about the matter

First, it would seem necessary that economists stipulate at the outset that advocacy of any policy, be it free trade or protectionism, involves, at some level, value judgements. This point of view is not new, but has not gained many adherents over the years. For example, Robbins (Reference Robbins1938, p. 637) wrote:

This was reiterated by I.M.D. Little (Reference Little1960, p. 257):

The necessity of such an up-front statement about the inescapable use of value judgements in discussions of free trade was also addressed by Little (Reference Little1960, p. 256):

Put more bluntly, an argument that defends or advocates a policy of free trade because it ‘increases the size of the pie’ cannot by and of itself be a good argument in favour of such a policy. But this does not mean that the ‘increased size of the pie’ result cannot be part of an extended argument that points out what the implications of policies that increase pie size can mean for societies over time.

Let me be clear about what I think is not a clear statement of what economic analysis cannot say about ‘gains from trade’. In a textbook, we find early on, immediately after a brief description of net national gains as measured by changes in surplus, the following statement:

The book goes on to say that ‘economists have tended to resolve the matter by imposing the value judgement that we shall call the one-dollar, one-vote metric . . .’

This is a step in the right direction, but needs to expand upon why economists have ‘tended to resolve the matter . . .’ Without this expansion, readers are left to their own devices in figuring out why economists have tended in this direction. Surely an expansion along the lines found in the above Public Policy texts would help readers understand this ‘tendency’.

4.2 Make a distinction between policies and one-off choices

Over the years, I have asked many economists the question: ‘Why do you think economists favour free trade?’ One response that seems to resonate with many of the economists with which I have discussed this topic goes as follows: Imagine two distinct economies, one of which embraces free trade along with other changes in economic circumstances that satisfy the Hicks–Kaldor compensation criterion, and the other which doesn't. These economies persist through time. If you did not know when you were to be born, into which of these economies would you like to be born?

From such a Rawlsian ‘veil of ignorance’ perspective or Harsanyi-esque ‘initial position’, the power of a policy that follows the Hicks–Kaldor criterion when thought about through time in contrast to a policy that exalts the status quo through time is most clear. In contrast, when thought about in a static context, the Hicks–Kaldor criterion has little force.Footnote ²⁹

A policy is a rule that tells one what is to be done in certain situations that occur in different times and places. It appears that many people can understand that a policy may help one over the long haul, even though in any instant one might dislike the application of the policy. I favour enforcement of speed limits as a policy, even though there are times at which I wish they would be suspended.

Of course, rules are meant to be broken. Under extraordinary circumstances, an argument can sometimes be made that an otherwise good policy should be suspended. For example, one might argue that if offshoring is going to be as disruptive to as many US residents as is sometimes claimed, then a policy of free trade should be suspended. Of course, one could argue the other side: suspending the rule destroys its credibility, and leads to opportunistic behaviour.

For another example, one might argue that trade liberalization should be a policy of ‘shock therapy’, while others might argue it should be a policy of piecemeal reform. Shock therapy might have the advantage of not allowing vested interests to coordinate and act to thwart reforms, while piecemeal reform might reduce adjustment costs.Footnote ³⁰

The point here is that the subject is open to discussion: one can disagree on whether the premise of vast dislocations is correct, or one can argue about the incentive effects of breaking the rule, but there is no a priori ‘correct’ answer.

4.3 Bring fairness issues into the debate

A key principle of fairness about which most people agree is that equals should be treated equal. Oftentimes, policies adopted towards economic interactions between home and foreign countries end up treating essentially similar citizens of the same country differently.

For example, why should workers in import-competing industries that are hurt by import penetration be given special consideration in comparison to workers who lose jobs because of changes in demand, or changes in technology? Or, why should domestic firms that allege harm from foreign firm predatory pricing, i.e. foreign dumping, be treated differently from domestic firms that allege harm from predatory pricing by domestic firms?

These arguments are straightforward and resonate with most people's sense of how a society should treat its members. They are made by economists in their defence of free trade, but usually as part of a bundle of arguments about why economists don't worry about income distribution.Footnote ³¹

There is no reason to entangle these fairness claims with more problematic issues.

4.4 Focus on the dangers of protectionism

The development of organizations such as the WTO that are designed to promote free trade was in large part a response to the terrible tragedy of the Second World War, which in turn was thought to have been brought on in a significant way by the tariff wars and competitive devaluations of the interwar years. Indeed it is commonly believed that the formation of the European precursors to the European Union were in fact started more as a way to insure against another European war than as a way to reap the traditional economic advantages of free trade.

While it is beyond an economist's warrant as an economist to be the arbiter of what criterion is appropriate for deciding what is ‘good for the nation as a whole’, it seems that war is such a terrible thing that there is general agreement that policies that reduce the probability of war are indeed good for the nation.

Research by political scientists reinforces and expands these ideas of the founders of the post-war free-trade-promotion organizations. As argued in Russett and Oneal (Reference Russett and Oneal2001), among the good things that accrue to nations that have trading relations with their neighbours is the possibility of a lower probability of war. They persuasively argue that evidence shows that trade is one of the three legs that support the ‘stool ‘of Kantian peace.

4.5 Do what economists do best: emphasize trade-offs

When asked by non-economists, ‘What should I say about the benefits of free trade to a worker who has lost his job to import competition?’, I sometimes respond by telling them we as a nation should also care about the export worker, or the import worker, or the worker in a downstream industry, who would lose her or his job if imports were curtailed. Or perhaps I mention that we should also care about the yet-to-be-born individual who benefits from the bigger pie. What economists are good at and trained for is to see the trade-offs involved in different policy choices. These trade-offs are not obvious to non-economists. In fact, this area is where the textbooks critiqued here and books such as Irwin's Free Trade Under Fire or Robert's The Choice shine.

President Truman allegedly got so tired of hearing economists tell him ‘on the one hand . . .’ that he wished for a one-armed economist. But frequently the best advice we can give is a menu of effects that flow from different choices. Trying to come up with a valid measure of the net effects is above our expertise.