Jump to ContentJump to Main Navigation
Understanding Poverty$

Abhijit Vinayak Banerjee, Roland Bénabou, and Dilip Mookherjee

Print publication date: 2006

Print ISBN-13: 9780195305197

Published to Oxford Scholarship Online: September 2006

DOI: 10.1093/0195305191.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use. date: 16 July 2020

Globalization and All That

Globalization and All That

(p.85) 6 Globalization and All That
Understanding Poverty

Abhijit Vinayak Banerjee (Contributor Webpage)

Oxford University Press

Abstract and Keywords

This essay begins with a discussion of the Samuelson-Stolper Theorem, which states that if a labor-abundant country is opened to trade with a labor-scarce country, laborers in the first country and the employers in the second will profit at the expense of the workers in the second country and the employers in the first. It then looks at the trade following World War II under the supervision of the newly created guardians of the world trading system: the World Bank, the IMF, and the GATT, as well as trade liberalization in Mexico, Latin America, and Asia. It discusses the issue of reputation: what it means to countries and how it determines what a country ends up exporting to the world.

Keywords:   international trade, high trade theory, Samuelson-Stolper Theorem, inequality, liberalization, reputation

Globalization is one of those strange things that everyone is for or against but no one feels the need to define. It is, of course, all those things that are in the news these days—call centers in Accra, French farmers lying down on the highway, bebop nights in Bangladesh—but what do they all add up to?

It seems natural and tempting to try to define a globalized world as one where there is unrestricted flow of commodities and ideas across national boundaries, and globalization as the process of approaching that ideal. This would mean, however, that Canada would have to move toward U.S. gun laws and Dutch drug laws, which seems unfair. Perhaps we could allow countries to have their own laws about who can consume what (but not where it was made). The bigger problem, of course, is what to do with all the people from India who want to move across the U.S. boundary. Clearly, not even the most enthusiastic globalizer is talking about the unregulated movement of people.

Unrestricted migration is not an option, we are told, because of its potential for enormous displacement and damage to the social fabric. But is there not damage to the social fabric when cotton farmers in India end their lives because, with imports so much cheaper, no one wants their cotton anymore? In the first half of 2004, there were over 300 reported farmer suicides in the state of Andhra Pradesh alone, many of them by drinking the very pesticides that they had bought to protect their cotton crop.1 The sad truth is that we have been taught to privilege the costs of migration by the sheer virulence of the racist reaction that it has sometimes engendered. The victims of freer trade are handicapped by their inability to effectively invoke the irresistible rhetoric of blood and honor.

(p.86) There are, of course, many others who have reason to be grateful for what globalization has done for them. And this includes not only the millionaires in Bangalore and Beijing, but also their chauffeurs and cleaning ladies. The starvation of the cotton farmer ought to be weighed against the potential starvation of his sister, who, after being abandoned by her husband, found a bearable living cleaning dishes in the homes of the software rich.

Indeed, as David Ricardo explained almost exactly two centuries ago, since trade allows the country to specialize in what it does best, total income ought to go up when a country opens itself to trade, implying that the income gains to the winners should be larger than the income losses suffered by the losers. After all, we all understand the advantage of specializing in what we do best. This is the reason why most of us choose to rely on the baker for baking our bread and the potter for producing our pottery.

The past couple of centuries have given us a chance to appreciate the many ways in which this simple and powerful argument is not entirely complete, but it is a rare economist who fails to be compelled by its essential logic. He will concede that there are cases where trade might reduce total income, and accept that there are situations where that might justify protection, but for the most part, when an economist worries about globalization, he is worrying about the losers. He knows that the losers, especially if they happen to be poor to start with, cannot be especially comforted by the fact that there are winners and that these winners are winning big (indeed, it might grate all the more).

Economists have always had useful things to say about the distributional consequences of international trade. Indeed, some of the most important insights of that most creative era of modern economics, the thirty years starting around 1930, come from attempts to say something about who gains (and who loses) from trade. The theory that builds on these insights, what one might call high trade theory, is as elegant as anything economists have produced. Nevertheless, I will argue, in order to have a useful conversation about globalization today, we need to go beyond this way of looking at the world, not because the theory was wrong to start with, but because the world for which it was conceived is no longer the world in which we live. This is the point of departure of this essay. My hope is that it conveys a sense of why today's world is different from the world of high trade theory, and helps the reader in thinking about who are likely to be the winners and losers in today's world, and how one might try to help the losers.


One of the most insightful results in modern economics is the Samuelson-Stolper Theorem: it tells us that if a labor-abundant country is opened to trade with a labor-scarce country, laborers in the first country and the employers in the second will profit at the expense of the workers in the second country and the employers in the first. The logic is simple: trade gives each (p.87) country a bigger market for the goods it is best suited to produce, goods where it has comparative advantage. We would expect countries to have comparative advantage in goods that make greatest use of its abundant resources. Thus a labor-abundant country should see an expansion of its production of labor-intensive goods, with corresponding increases in the demand for labor and its price. Since nowadays labor-abundant countries also tend to be poor, and laborers are usually poorer than their employers, this implies that freeing up trade helps the poor in the poorer country. Their employers, of course, suffer for much the same reason—higher wages mean higher costs for them—and as a result, inequality in the poorer country ought to fall. And laborers in the rich country are hurt by the fact that their countrymen can go elsewhere for their labor-intensive goods, with the result that wages in the two countries are drawn closer—illustrating the famous link from free trade to what used to be called factor price convergence.2

According to the data we have, these heartening predictions seem to have held up rather well throughout the nineteenth century and until well into the twentieth. In the first half of the nineteenth century, international trade was, as it had always been, largely a matter of exchanging goods that had more value than weight and, as such, a marginal (albeit highly profitable) part of economic life. On the Atlantic rim this was a period of divergence. Wages were growing fastest in the richest countries—England and the United States—and some of the less favored countries, such as Spain and the Netherlands, were in a slump.3 But in each case the driver was at home—U.S. wage growth was propelled by the growing realization of the nation's many natural advantages (the period ends with the California gold rush), and in England, workers were finally getting the benefits of decades of industrialization. Globalization was still a bit player.

Then, suddenly, the floodgates opened. In 1846, England repealed the Corn Laws, which allowed the government to restrict the import of wheat to protect domestic farmers, setting off a spate of unilateral and bilateral lowering of barriers to trade all across Western and Central Europe.4 Transatlantic steam shipping was introduced in 1838, and by 1870 a series of innovations—the screw propeller, the compound engine, steel hulls, among others—had made transporting bulk freight by steamships practical. Railways, introduced in the 1830s, came of age after 1850—mileage all over the world grew by more than five times between 1850 and 1870—completing the integration of the interior with the ports that had begun with the canal revolution of the first half of the century. Also, mechanical refrigeration started being used in the transporting of meat across the Atlantic after 1870.

The combination of these forces led to rapid expansion of trade over the period 1870–1913, with the share of merchandise exports in 1913 across a set of sixteen OECD countries rising to levels that they typically did not exceed until well after 1950, and in some cases (such as the United Kingdom) not until the 1980s.5 It meant that the price of grains and meat in the labor-abundant, land-scarce countries of Western and Central Europe fell to levels (p.88) comparable with those in the labor-scarce, land-abundant countries in the New World—Argentina, Australia, Canada, and the United States. Wheat prices in Liverpool were 57.6% higher than wheat prices in Chicago in 1870, but only 17.8% higher in 1895, and the U.S.-British meat price gap shrank from nearly 100% in 1885 to less than 20% in 1913.6

If nothing else had changed, the fall in the price of food should, by itself, have shrunk the gap in the real (consumption) value of wages between Europe and the New World. And there is indeed some evidence of faster wage growth in Europe over this period. While there was almost no wage growth in Australia between 1870 and 1913, and U.S. wages grew by about 1% per year, wage growth was around 2.5% per year in Scandinavia and around 1.75% per year in Ireland and Italy. But there were also a number of anomalies. Wage growth in the richest European countries—Belgium, France, Germany, the Netherlands and the United Kingdom—was no higher than in the United States; and in the less developed countries in the New World, such as Argentina and Canada, wages actually grew considerably faster.

The same theory that tells us there should be convergence in wages also tells us that there should be convergence in the price of land, though the gains from trade here go to the landowners in the land-abundant countries. We do find that prices for agricultural land rose by almost 400% in land-abundant Australia between 1870 and 1913, nearly trebled in the United States over the same period, and declined in Britain, France, and Sweden. But in Denmark, Germany, and Spain, which were land-scarce countries, land prices went up.

Of course we are asking for a lot here. Quite so much fealty to a specific theory may be unreasonable to expect, given that there are many other reasons why wages and prices go up—capital accumulation and technological upgrading being the most obvious among them. A more modest goal would be to ask whether the ratio of wages to land prices became more similar in all these countries. If there was something (say political change) that made both wages and land prices go up in one country and down in another, neither wages nor land prices would converge, but their ratio might. And indeed, this is exactly what we find. Every Western and Central European country with the exception of Spain shows sharp growth in the ratio of wages to land prices, and every New World country for which we have data shows a decline.7

For labor-abundant Europe, the period of expanded trade that led up to the First World War was a period of growing equality within. The nervous years between the wars saw the reversal of all that. In these years of growing xenophobia and economic nationalism, tariff barriers were raised and borders were closed to immigrants. And, at least for Western Europe, inequality started to grow once again. Between 1870 and 1913, equality, measured by the ratio of the wages of unskilled workers to the output produced by an average worker, went up in the countries where wages were low and went down where wages were the highest. Between 1921 and 1938, we see exactly (p.89) the opposite pattern—no one really did well, this being the era of the Great Depression, but the poor did the worst in Italy and France and the best in Canada, Australia, and the United States.8

It is not entirely obvious that we can interpret this evidence as support for trade theory. The problem, noted by the O'Rourke and Williamson study that is the source of most of our information about the nineteenth and early twentieth centuries, is that the pro-trade decades were also pro-immigration decades. Enormous numbers of people immigrated from all over Europe to the Americas, creating the melting pot that we know today. The U.S. labor force would have been smaller by about a quarter in 1910, were it not for the flow of immigrants over the previous four decades. This drove wages up in the countries they had left and down in the countries where they went. Indeed, O'Rourke and Williamson do a calculation that shows that immigration alone can explain all the changes in the income distribution in this period and more, though they are at pains to convince us that this cannot be right. Perhaps one day better data will reveal what really happened. In the meanwhile, the fans of trade theory must find the overall picture from this period rather comforting. The problems start later.


After the brutal interlude of World War II, trade resumed, now under the supervision of the newly created guardians of the world trading system: the World Bank, the IMF, and the GATT. Decolonization followed, and suddenly the poorest countries in the world had a chance to make their own trade policies. Since the mid-1950s, trade barriers have been raised and brought down many times in every corner of the world (though perhaps today we live in a world that is more open to trade than at any time since 1930). What can we take away from this experience?

There have been a number of in-depth studies of what happened when specific countries liberalized. Adrian Wood studied the East Asian “tigers” and found that low-skilled workers did relatively well in Korea, Singapore, and Taiwan (but not in Hong Kong) in the decade that followed the lowering of trade barriers. This is as one might have expected, given everything we have been told about how they grew rich by exporting labor-intensive goods.

Things started going wrong with the Mexican liberalization of 1985–1987. Over this two-year period, Mexico massively reduced both the coverage of its import quota regime and the average duty on imports. Over the rest of the decade, blue-collar workers lost almost 15% of their wages while their white-collar counterparts gained in the same proportion, reversing the trend toward greater equality in the years leading up to the trade liberalization.9 Argentina showed a similar pattern: trade barriers were massively lowered at the beginning of the 1990s, and between 1990 and 1998 the college premium, which is the ratio of wages for those who have a college degree to those of the average unskilled worker, almost doubled for men and almost trebled (p.90) for women, in both cases reversing an equalizing trend that had started in 1980.10 This meant that the lowest wage earners were effectively excluded from the wage growth that took place in this period. In Colombia, trade barriers were massively lowered between 1985 and 1992, leaving the economy more open than it had ever been. Between 1990 and 1998, the college premium went up by over 20%, which is small by Argentine standards, but not negligible. Similar patterns have also been found in Chile in the 1970s, Costa Rica in the late 1980s, and Uruguay in the 1990s.

Lest one think that this is some Latin American perversion, the same correlation between liberalization and increasing inequality showed up in the two Asian giants, China and India. Between 1980 and 1998, while China became increasingly open to the world economy, the income share of the poorest quintile went down from 8% to 7%, while the share of the richest quintile went up from about 30% to about 45%. And in India, all evidence points to an explosion of inequality in the urban areas over the post-liberalization years (1991–2000), after many decades when inequality went neither up nor down.

To check on what we are finding here, we could compare what happened over the last few decades in countries that were open to trade, with what happened in countries that were closed (or we could compare countries that became open with countries that did not). This is really what we were doing with the historical data, and now that we have data from a hundred countries and more, it is actually possible to be more precise and try to eliminate other sources of possible difference between these countries. A number of people have now tried this exercise. Most have found, sometimes to their surprise, that being open to trade either increases inequality or has no effect at all. And the one study that directly focuses on the well-being of the poor (defined as the bottom 40%) finds that their income grows more slowly when the economy is more open to trade, not just in comparison with the rich but in absolute terms.

Correlation, as social scientists never tire of reminding each other, is not causation, and none of this evidence is quite strong enough to entirely transcend that old divide. However, what these case studies do show is that, at the time when countries opened themselves to trade, something happened that was powerful enough to make the countries change course toward greater equality in the nineteenth and early twentieth centuries and toward greater inequality in the last decades of the twentieth. That something may not be the trade liberalization itself, but it is something that operates at a similar time scale; in other words, it cannot be confounded with those other, much more long-term, factors that also affect income distribution, such as the nature of technology and the norms of society.

The studies do offer some clues to what the confounding factors might be (if it is not the effect of trade liberalization) that we have been deliberately ignoring so far. Feenstra and Hanson (1997) argue that trade has very little to do with why inequality went up in Mexico, and they argue that it was (p.91) largely driven by the foreign investment that came in with the liberalization. They suggest the following line of reasoning: when an American firm builds a factory in Mexico to make something it used to make at home, the demand for skilled labor should go up, simply because anything that used to be produced in the United States ought to be much more skill-intensive than most things Mexico used to produce. This is correct, but only if we accept the assumption that once production moves to Mexico, the firm would not switch to a very different (and less skill-intensive) way of doing things. This is perhaps the natural assumption in cases where the overall production process remains centered in the United States and just a few steps get sent to Mexico, but not necessarily when the entire production process is moved to Mexico. When Nike goes to Vietnam, given just how much cheaper unskilled/semi-skilled labor is available in Vietnam compared with the United States, it would seem natural for Nike to try to reorganize production to use more of this kind of labor, thereby benefiting the low-skilled workers relative to everyone else.

On purely theoretical grounds, it seems that the impact of foreign investment on inequality in a relatively poor country like Mexico could have gone either way. The evidence from the Mexican case, at least prima facie, seems to support the Feenstra-Hanson view: it does seem to be the case that skilled workers gained more where there was more foreign investment. Feenstra and Hanson estimate that the impact of foreign investment alone can explain almost half of the rise in the wages of nonproduction workers (a proxy for white-collar workers). The fact remains, however, that half the increase is still unexplained and, in addition, it remains for us to explain why the fall in inequality predicted by trade theory did not happen.

A less interesting theory of why things did not work out the way they were supposed to argues that in Argentina, Colombia, and Mexico, it was the industries with a high concentration of low-skilled workers that had the highest tariffs to begin with and where the tariff cuts and the resulting fall in prices were the largest. But tariffs matter only if you are in an industry that is a net importer, and indeed most of these studies do find that prices fell more in industries where the tariff cut was deeper—confirming that the barriers to trade in these industries were actually blocking imports. It follows that these Latin American countries must have been importing all these labor-intensive goods, despite the fact that nearly 90% of Latin America's external trade is with rich, labor-scarce countries.

The puzzle deepens when we push this argument a step further. When these studies directly look at changes in employment across sectors, the striking thing is the lack of correlation between changes in tariffs (or the resulting change in imports) and changes in employment. After all, our faith in the therapeutic qualities of trade comes from the belief that trade helps the country specialize in what it is good at, and we might have expected that removing the barriers to trade would set off the process of reallocation of labor in that salubrious direction.

(p.92) One might wonder, especially if one comes from a certain political persuasion, whether the absence of labor reallocation is a result of draconian labor regulations that make it impossible to fire anyone. As it turns out, manufacturing employment fell dramatically in Argentina, considerably in Colombia, and more modestly but still significantly in Mexico. It also does not seem to be true that people in what would have been shrinking sectors protected their jobs by taking huge pay cuts (say out of a love of their jobs). Wages did fall, but the correlation between the fall in wages and the changes in protection is never very large. The removal of protection hurt all workers and not just the workers in the protected industries.

It is not that trade theory cannot find a way around these new puzzles. One can always invoke special explanations, something that just happened to coincide with the shift in trade policy. But there is a general sense that things are not fitting together, that one has to do too much work these days to defend the view, dear to trade economists, that trade always helps the poor in poor countries. It looks like we might be missing something important.


The key missing ingredient, I believe, is reputation. Reputation is an omnibus word that economists use to describe a range of things, from the expectation of quality associated with a brand name, to the trust that two people who have been doing business for a long time have in each other.

Brand names are clearly hugely important in today's trade. This makes perfect sense, given that the average buyer in the north (who has all the money), is rich enough today to afford to indulge his sense of quality. Clothes must not shrink or bleed; machines must not fall apart in a week; and software must be ready when promised. Consumers would rather pay something extra for a brand name because they feel assured of the quality.

Trust, less obviously, is also vital. Retailers are willing to pay more to trusted suppliers because they do not want to offend their customers by selling duds. And manufacturers want suppliers who are known to be reliable, because what they make and sell is only as good as its weakest link.

The obverse of quality being important is that the price is not quite as important as it used to be. In 1997, when the Indian software industry was just beginning to become a household name, I happened to be in the offices of one of the leading exporters of customized software. We were talking about the industry, and the CEO said, almost casually, “We just raised the price per line of code by 50%, and our clients did not seem to mind. I think we will raise it by another 50% in a few months.” I happen to know that he was not bragging. Indian software in those days (we are now in the time scale of the software world) was so cheap that it did not make sense for the average overseas buyer to give up someone they knew and trusted for someone new, even at half the price. After all, the new supplier could be incompetent or worse, and chances were that the buyer would find out only after (p.93) the software was ready, many months hence. For the typical buyer of customized software, a Fortune 500 company, the cost of the lost time was probably worth as much as it was paying for the software, or more.

The logic here applies as much to T‐shirts as it does to software. The famous New York department store, Macy's, orders its T‐shirts from its overseas suppliers in the fall of each year, but the T‐shirts go on sale the next spring. If, when the T‐shirts arrive in early spring, Macy's discovers that they are not what it had asked for, it would be too late to do much about it for that year. The loss of face vis‐à‐vis its customers could be enormous. It has to be worth paying extra to suppliers that it can trust.

Several consequences follow from recognizing the centrality of reputation. First, reputation is a bit like a fixed cost—the advantage it gives you is independent of how much you sell, though if you expand beyond the point where you are able to maintain quality, your reputation will eventually collapse. Over some range, however, it clearly pays for firms that have a reputation to keep expanding. It is no accident that Gucci now markets everything from perfumes to polo shirts, and Amazon.com is no longer just the bookseller to the world.

Second, when firms that have a brand name shift production to a Third World country, they are often reluctant to change the process of production, even though the local cost structure might favor reorganization. This is because they are worried about whether they will be able to exercise the same degree of control over quality when they change their processes, not necessarily because the new processes are inherently harder to control but because they are unfamiliar. Given that so much is riding on predictably delivering quality, experimenting with their processes is not something they do lightly.

This reluctance to change processes means that when a First World firm relocates to a Third World country, it demands the same kind of labor that it was using at home. This may be why foreign investment is associated with a rise in the skill premium.

Third, the urge to capitalize on their brand name is an important part of what drives multinationals to set up shop in LDCs that are willing to have them. And since services are where brand names count the most (would you trust your money to a bank that you have never heard of; or get advice from a consultant on his first day on the job?), a lot of multinationals are headed for the service sector. Since service industries tend to employ large numbers of white-collar workers, this is one more reason why the skill premium goes up when the multinationals come in.

Fourth, we would expect a great deal of inertia in the choice of from whom to buy. Combined with the fact that those who have a reputation will tend to want to grow, this implies that it could take a long time before cost advantages show up in the pattern of trade. This may explain why the Argentines and the Colombians import labor-intensive goods from the labor-scarce north. It also explains why so many domestic firms shrink or shut down when their countries are opened to trade—they cannot compete with (p.94) the reputation of the best-established multinational suppliers. The firms that survive and expand, apart from the few that already have an established international reputation, will be the ones that are in sectors where reputation does not matter so much—manufacturers aimed at the lower end of the local market or exports to other Third World countries, for example. This may be one way to interpret the fact that in all the Latin American examples, the informal sector expanded following trade liberalization.

Finally, and perhaps most important, all this suggests that it may be some time before those who lose their livelihoods with the liberalization find comparable jobs in the industries that start to grow after the liberalization. In the long run, we would expect the country to broadly specialize in industries where its factor endowments give it a natural advantage, but success in building a reputation will clearly be crucial in determining exactly what it ends up exporting to the world. This process is likely to be slow because building a reputation takes time: Esther Duflo and I found that in the Indian software industry it took about five years to be really trusted by foreign buyers, and in software everything happens quickly.11 It is likely to be doubly slow because capital markets in developing countries work very poorly, and even visibly successful firms have trouble getting enough capital to expand as fast as they want. In another joint project with Esther Duflo, we found that even firms in India, where the return on the extra dollar is 80 cents or more, cannot get enough capital.

This problem is made worse now by the fact that China and India are in the process of expanding their exports to the level that their enormous reserves of skilled manpower would justify. They have cheap wages, a large and wealthy émigré population that provides them with the necessary contacts, and a growing reputation. Most other countries would have to build their own reputations in competition with these two established players, just when they are gathering steam. It cannot be easy.


If I am right in looking at the world in this way, where does it leave us? Relative to the benign vision of globalization offered by the Samuelson-Stolper Theorem, at least for the poor in poor countries, all this is obviously bad news, though recognizing the problem does help us control the damage. It tells us, for one thing, that countries need to think hard about reputation.

There are a number of steps that a country can take to make reputation acquisition less of an issue. More effective court systems make it easier to sue for damages if a supplier misbehaves, which encourages buyers to consider sellers who are yet to make a name for themselves. Developing trade relations with other developing countries has a similar effect, because poorer buyers may be more willing to trade off quality for low prices. It is perhaps no accident that Guido Porto found that Mercosur, which integrated the (p.95) economies of Argentina, Brazil, Paraguay, and Uruguay into a single (almost) free trade zone, benefited the poor in Argentina.

A public scoring system for buyers and sellers, where buyers report their experiences with specific suppliers and vice versa (for example, on CNET.com or Dealtime.com), may be one way to speed the process of building up a reputation, simply by making it easier to identify malefactors. We do, however, have to careful about the possibility of abuse here. Sellers may try to blackmail buyers by threatening to give them a harsh report.

Creating an environment that is friendly toward multinational investors also helps, though most countries seem to trust domestic capitalists more than their foreign counterparts. My sense is that this is largely unwarranted (though perhaps politically necessary). The prima facie evidence seems to be that within any sector multinationals pay higher wages to people with the same observable qualities, compared with domestic firms. Indeed, the simple fact of being in the public eye probably makes them behave better.

Multinationals are useful because their reputations can intermediate between domestic sellers and foreign buyers, and also because they are better at negotiating with regulators in the north, who are often all too willing to protect domestic producers from cheaper imports, in the name of protecting the health of domestic consumers. Much of what China now exports is produced by nameless firms in China, but marketed under some global brand name. In the typical case, the Chinese firm sells to some other established firm, often based in Hong Kong, and it is this firm that sells to the eventual marketer. This double intermediation is useful because the firm in Hong Kong has a reputation in the world of other firms, not among the eventual buyers, while the firm that does the final sale does not need to learn all the nuances of Chinese supply chains. Indian customized software firms have a similar system. The firm will often have a U.S. front office that the buyers know and deal with, which in turn will pass on the actual work to Indian partners, whom the buyers know about only vaguely.

Finally, improving capital markets is yet another way to make building a reputation easier. Firms that have been doing well need to go to scale quickly, and inefficient capital markets slow their growth.

All this, of course, does not answer the really basic question: Is trade worth it? The conventional answer, at least among economists, is that trade is worth it as long as it makes the pie bigger. If the income gains to those who gain from trade are larger than the losses to those who lose, we should be in favor of trade, because we could always redistribute the gains to make everyone better off. On the other hand, if I am right in my diagnosis of what drives trade today, the losers will include large numbers of poor people in poor countries, and I very much doubt that they will all be compensated for their losses. The problem is that targeted redistribution is never easy. Identifying the losers is hard, taxing the winners is harder, and making sure that the money does not somehow get lost along the way is sometimes well nigh (p.96) impossible. The best way to make sure that no one loses from liberalization is almost surely not to do it.

On the other hand, I know of no example of a labor-abundant developing country that has had sustained growth without being outwardly oriented in some broad sense of the term. The problem is that growing countries need technology and capital from abroad, and to pay for these imports and to take full advantage of what they are good at doing, they need markets abroad.

It is true that none of this directly requires being totally open to imports: China, after all, is only starting to get there. But it does require a credible commitment to remain open for the foreseeable future, to convince foreign investors that the country will earn enough foreign currency to repay what it is borrowing now. And it requires being sufficiently open to imports from the north to keep the WTO happy, if for no other reason than keeping the markets in the north open to its exports. In the end, the choice is between cutting oneself more or less completely off and giving up on fast growth, and accepting a very substantial degree of openness.

Having thought about it, I think I come out on the side of openness. We live in a world where people have come to see the good life as something to which they can legitimately aspire. There is no way to put that genie back in the bottle (even if we could, I do not know that I would want to). The explosion of ethno-religious conflict around the world since the 1980s, I believe, is a product of an era in which hopes were created and then denied. To go back and convince people that growth is not worth it, is to invite a bloodbath.

Given that, our job as economists should be to think of ways to make the whole process as painless for the many losers as it can be made to be. There seems to be broad agreement these days that openness should go with a strong commitment to an acceptable guaranteed minimum living for all. While I am certainly in favor, two remarks are in order. First, the popular view seems to be that the guarantee should primarily take the form of a right to education and health. This is in many ways very appealing, but my sense is that we know very little about how to deliver these services effectively to the poor in poor countries. An income support program, targeted toward the elderly and women with young children, combined with a negative income tax program targeted toward those who are working for low wages (modeled on the earned income tax credit in the United States), has the obvious advantage that we know it helps the poor. Second, many of the biggest losers will be the owners of very small firms and farms and workers in the organized sector, rather than the very poor. The minimum guarantees that are being discussed are clearly not going to compensate them for what they will have lost. While in some ways this is fair—after all, they do have more to start with—we should recognize that at the very least, they think that they have a right to be compensated. There is very little work that I know of that thinks hard about how to target the nondestitute.

Finally, there is the issue of how to share the benefits of trade fairly across (p.97) countries. The economics of how the gains from trade get distributed across the trading partners is always subtle, and I have no way to say who gains more. The distribution of costs, on the other hand, seems clear enough. Every rich country has a well-established system of welfare that will automatically compensate its losers. Every poor country will have to set up new mechanisms to do so, and pay for them now. In addition, migration today is explicitly a system by which rich countries raid the poor countries for talent. With every engineer or doctor who settles in the north, a poor country loses a potential entrepreneur or a public servant, and it is hard for me to believe that the money they send back is compensation enough. Globalization, by the simple fact of increasing contact, can only make it easier for the north to pick off the people it wants. It is true that globalization has encouraged some of these people to take a second look at the opportunities in their own countries, but on balance I suspect it is still a cost (China and India may be exceptions here, though I would not be sanguine).

I therefore see no reason why the poor countries should not demand an explicit price from the rich for agreeing to remain open. There are many ways the price could be paid, but let me end by mentioning one that appeals very much to the economist in me. Given that rich countries are aging fast and need labor, why not a system of quotas for immigration to rich countries that are open to everyone (i.e., not just the doctors and engineers) in poor countries that agree to remain open.12 The quota would be allocated by a lottery, and would allow one person between the ages of twenty and thirty-five to have a work permit in a rich country for a fixed period, say five years. His/her travel expenses and some resettlement costs would be paid for, and precautions would be taken to make sure that in most cases, he/she did go back, so that the home country would benefit from his/her savings and the skills he or she had acquired.

This will not be easy: It will require us to confront the forces of racism and xenophobia that will surely try to make the most of new opportunities. But in the end, if we cannot ask the rich democracies to do what is both just and economically rational, what is left for us to hope for?


(2.)  Because labor, land, and capital used to be called factors of production, and it is their prices that converge.

(3.)  Kevin O'Rourke and Jeffrey Williamson, Globalization and History (Cambridge, Mass.: MIT Press, 2000).

(4.)  Ibid.

(5.)  Based on ibid., Table 3.1.

(6.)  Ibid.

(7.)  Ibid., Table 2.2.

(8.)  Ibid.

(9.)  Robert Feenstra and Gordon Hanson, “Foreign Direct Investment and Relative (p.98) Wages: Evidence from Mexico's Maquiladoras,” Journal of International Economics 42 (1997): 371–394.

(10.)  Sebastian Galliani and Pablo Sanguinetti, “The Impact of Trade Liberalization on Wage Inequality: Evidence from Argentina,” mimeo, Universidad di Tella, Buenos Aires, 2002.

(11.)  Abhijit Banerjee and Esther Duflo, “Reputation Effects and the Limits of Contracting,” Quarterly Journal of Economics 115 (3) (2000): 989–1017.

(12.)  I am certainly not the first person to think along these lines. See, for example, Jagdish Bhagwati, “Borders Beyond Control,” Foreign Affairs 82 (1), (2003): 98–104; and Dani Rodrik, “How to Make the Trade Regime Work for Development,” mimeo, Harvard University, 2004.