dani rodrikis the Ford Foundation Professor of International Political Economy at the Harvard Kennedy School. This article is adapted from a study prepared for the IFS Deaton Review.
Illustrations by nate kirsh
Published July 25, 2022
Until quite recently, economists have generally underplayed the role of trade in exacerbating inequality in advanced economies. But in the public imagination, globalization’s adverse effects have loomed, significantly contributing to the backlash against the political mainstream and the rise of far-right populism. The published research on the relationship between international trade and inequality is in fact exceptionally rich, with important theoretical insights as well as extensive empirical findings offering guides for public policy. Here, I summarize a few key takeaways that economic policymakers ignore at our peril.
Redistribution Is the Flip Side of Gains From Trade
The gains from trade derive from the difference in relative prices that prevail in the world economy, on one hand, and in the pretrade domestic economy, on the other. As an economy opens up, domestic relative prices change, producing income redistribution alongside the gains from trade in terms of the expansion of the value of total output. The identities of the winners and losers depend on the nature of social stratification in society (class, occupation, skills, education, gender and the like) and on which side of the change in relative prices each group stands.
None of this constitutes a revelation for anyone who’s been paying attention. Indeed, the Stolper-Samuelson theorem (dating back to 1941), which every econ major learns about, produces one especially stark result in a simplified model of trade. It shows that trade creates absolute losses for a segment of society — and not just varying gains.
The assumptions behind Stolper-Samuelson are extreme: an economy with only two factors of production, only two goods, and perfect inter-industry mobility of the factors. But the underlying logic has powerful implications for more realistic economic environments. If an economy has competitive markets, as long as it does not specialize completely — that is, it continues to produce the goods that are imported — opening to trade must leave at least one factor of production (think of workers with only a high school education or owners of capital in firms that compete with imports) worse off, regardless of the numbers of goods and factors and the degree of domestic factor mobility.
Importantly, this result implies that the benefits to consumers of trade-induced cuts in prices can never fully compensate the losers. There will be at least one group of people (“factor of production,” in economics parlance) whose earnings fall more than the price of the good with the steepest price drop thanks to import competition. So even if, say, less-skilled workers heavily consume goods that can be imported, they are still left worse off when these goods, previously made at home, rely heavily on low-skill labor. Put another way: yes, low-wage workers benefit because they can buy, say, cheap giant-screen TVs at Walmart, but the benefits from lower prices are not as large as their loss in wages.
These theoretical results are important because they run counter to many of the arguments in the public debate — that trade benefits most or all people, that even if trade creates some losers these are merely transitory “adjustment costs,” or that consumer price effects outweigh losses. Essentially, it is inconsistent to argue for freer trade in the name of efficiency gains without accepting that there will be sharp distributional consequences.
Employer- or industry-specific skills allow for a more fine-grained assessment of winners and losers. More importantly, the degree of locational immobility of workers helps determine the distributional effects across regions.
Tracing out these distributional effects requires that we accurately identify the relevant inputs in production. Labor clearly differs from capital, and less-educated workers cannot quickly transform themselves into educated professionals. Early empirical work focused on these broad demarcations — labor versus capital, skilled versus less-skilled workers — but these categories were probably too aggregated to be very revealing.
Employer- or industry-specific skills allow for a more fine-grained assessment of winners and losers. More importantly, the degree of locational immobility of workers helps determine the distributional effects across regions. The research reviewed by David Dorn and Peter Levell (2021) for the IFS Deaton Review identifies significant adverse local labor market consequences of Nafta (in the U.S.) and the Chinese export boom of the 1990s (in both the U.S. and Europe) in regions heavily reliant on jobs that compete with imports. See, in particular, research by David Autor, David Dorn and Gordon Hanson (2013) on the China trade shock and Shushanik Hakobyan and John McLaren (2016) on Nafta. These studies show that regions heavily affected by trade — and workers and industries most directly competing with China and Mexico — suffered significant longterm income losses.
Redistribution Looms Larger in Advanced Stages of Globalization
Another important, but less well-recognized, implication of trade theory is that the gains from removing trade barriers become smaller relative to the redistribution induced as the barriers in question shrink. In other words, as globalization advances, the redistributive impact of trade (good or bad) looms larger relative to the efficiency gains from trade.
This follows straightforwardly from standard economic theory: the efficiency losses linked to a tax on trade, as with all taxes, rise with the square of the tax. Turning that around, reducing a tariff that is half the size creates a gain at the margin that is only a quarter as large. The distributive effects, meanwhile, are roughly proportional to the change in relative prices linked to reducing trade barriers and do not depend on the magnitude of the tariff/tax — or where we are in global economic integration.
The magnitude of redistribution is quite dramatic relative to gains from trade that are generated when trade barriers that are already pretty low are reduced further. This is borne out by real-world data.
To see the practical significance of this, consider the following question: how many dollars are shuffled across different income groups per dollar of efficiency gains from trade? The answer to this question is given by what I have called the “political cost-benefit ratio” (PCBR) of trade liberalization. The numerator of PCBR is the sum of the absolute values of gains and losses across identifiable groups, divided by two to ensure there is no double-counting. The denominator is the efficiency gain produced by the trade liberalization.
Based on plausible assumptions, the PCBR rises from around 5 when tariffs are initially at 40 percent to more than 20 when tariffs are below 10 percent. In other words, the magnitude of redistribution is quite dramatic relative to gains from trade that are generated when trade barriers that are already pretty low are reduced further. This is borne out by real-world data. Empirical analyses of Nafta — for example, one by Lorenzo Caliendo and Fernando Parro in 2014 — have concluded that the efficiency gains from trade reaped by the U.S. were tiny compared to distributive effects found in studies such as Hakobyan and McLaren (2016).
These considerations offer an important perspective on the political economy of globalization. Once national markets have become fairly open, attempts to push globalization further will seem to be motivated primarily by the objective of enriching specific interest groups rather than expanding the size of the overall pie — and with good reason! I would hazard the guess that advanced economies had already reached that stage by the late 1990s, if not earlier.
Compensation Is Problematic
Trade induces income redistribution, but it need not exacerbate inequality if the winners happen to be the less fortunate in society. Theory and empirical analyses both suggest, however, that redistribution went in the wrong direction in the advanced economies in recent decades. The losers were poorer workers with less education, along with regions that were already adversely impacted by de-industrialization and the concomitant loss of jobs.
The predictable response from economists and policymakers to such concerns is that trade agreements should include compensation for the losers. In the U.S., compensation is explicitly built into trade policy in the form of Trade Adjustment Assistance. In Europe, compensation is not directly targeted at workers affected by trade, but the regular social insurance and active labor market programs addressing job losses tend to be more generous than in the U.S. In neither case is compensation provided for earnings losses per se, unless workers end up unemployed or fall under income thresholds that trigger public assistance. It is fair to say that compensation is incomplete and imperfect.
There are good reasons why compensation never quite works as promised. First, governments have difficulty in identifying workers whose earnings would have been higher in the absence of trade liberalization. In practice, this problem is “solved” by making public assistance conditional on an observable trade shock — such as job loss due to a specific liberalization of protection. But this misses workers who get to keep their jobs but have to accept lower wages to stay employed. Note, too, that imperfect information rules out lump-sum cash compensation that would allow workers to start afresh. Compensation is conditional on workers following the path set down by the Trade Adjustment Assistance policy, which often creates incentives for displaced workers to delay looking for a new job.
This brings us to the second problem: the resources chewed up in figuring out who gets how much compensation can easily offset a big chunk of the gains from trade. This can make the net gains from trade liberalizationcum- compensation much smaller and can even turn those gains into losses. Research by Pol Antràs, Alonso de Gortari and Oleg Itskhoki in 2017 shows that the uncompensated rise in inequality produced by trade can make a sizable dent in societal gains.
Consider a back-of-the envelope calculation based on my own published analysis. Assume that for every $1 of redistribution, 10 cents of efficiency loss is generated. Assume further that the PCBR (at the margin) of trade liberalization is 10, which is not an extreme number when economies are already pretty open. Then, compensating the losers fully would produce an efficiency loss that exhausts all the gains from trade. Particular groups (export- oriented interests) might still gain — and gain a lot. But the losses incurred by the rest of society might equal (or exceed) that gain.
I have so far considered only the economic arguments for why compensation may be problematic, which are incomplete at best. The assumption is that society values the compensation policies and that political authorities really want to hold the losers harmless. But there are also political obstacles to compensation. If globalization’s beneficiaries are powerful enough to get the trade agreements they want in the first place, they might be also powerful enough to block any redistribution at their own expense. And even if liberalization advocates need a broad coalition that includes advocates of compensation at the outset to get their proposals through, they can often wiggle out of unwanted commitments after the fact.
Suppose signing a trade agreement requires that at least some of the potential losers are on board. In advanced countries, these groups are likely to represent workers in declining industrial regions. To get their agreement, the government will promise compensation.
It is far from clear that trade is the most important factor behind the distributional woes of advanced societies in recent decades. Among other potential villains: rapidly rising returns to specialized skills, regional decline and falling income shares of labor.
In the U.S. context, this takes the form of enhanced Trade Adjustment Assistance. Once the agreement is signed, however — and as long as the trade agreement cannot be easily abandoned — there will be little incentive to ensure the compensation is actually delivered by Congress. And, indeed, there is evidence that the Trade Adjustment Assistance program has generally been underfunded and its effectiveness quite limited.
Fairness Versus Inequality: Trade Is Different
But why should governments try to undo the redistributive effects of trade and globalization in the first place? Market-based economies undergo continuous changes, many of which have implications for relative prices and for income distribution. Changes in consumer demand, new technologies and a variety of other idiosyncratic shocks buffet economies without necessarily giving rise to concern about inequality or calls for compensation.
Moreover, it is far from clear that trade is the most important factor behind the distributional woes of advanced societies in recent decades. Among other potential villains: rapidly rising returns to specialized skills, regional decline and falling income shares of labor. There is broad consensus within the economics profession that technology and perhaps broad institutional changes (such as the decline in unionization) have played a larger role than trade. Yet somehow the adverse effects of globalization have become politically salient in a way that many of the other drivers have not. Indeed, a large body of research shows that globalization shocks have played a causal and significant role in the rise of right-wing populism.
The outsized effects of trade shocks in shaping public attitudes are demonstrated in an experiment that Rafael Di Tella and I ran in the U.S. in 2020. We presented online survey subjects with a “newspaper article” on the impending closure of a garment plant. Our test subjects were divided randomly into groups, with each group presented with a different reason for the plant closure. The scenarios covered a negative product demand shock, the introduction of labor-saving technology, management mistakes and different types of international outsourcing (i.e., trade). The respondents were then asked about their preferred response: do nothing, provide cash to the displaced workers or impose trade protection.
The main takeaway was that people distinguish among labor market shocks according to what propels them. While non-trade disruptions such as technology and unanticipated declines in product demand did generate sentiment in favor of intervention, trade shocks elicited a much more drastic protectionist response — doubling or tripling the share of respondents who sought trade restrictions.
Economists have traditionally resisted entertaining such fairness concerns in discussions of trade policy. If labor standards are weak or nonexistent in low-income countries, why should that not count as just another source of comparative advantage?
Moreover, our subjects were especially sensitive to trade with developing nations. Simply changing the name of the country to which production is outsourced from France to Cambodia increased calls for import protection by more than half.
The Economic Nobelist Angus Deaton, among others, has argued that public reactions to economic trends are shaped less by inequality per se than by perceptions of unfairness. As Deaton puts it:
Inequality is not the same thing as unfairness … it is the latter that has incited so much political turmoil in the rich world today. Some of the processes that generate inequality are widely seen as fair. But others are deeply and obviously unfair, and have become a legitimate source of anger and disaffection.
Foreign trade is particularly prone to charges of unfairness because it entails transactions among entities that operate under different rules. Consider the difference between a market exchange that is domestic and one that crosses national borders. In the former, all firms in a given industry are subject to identical rules, and the expectation is that the government does not favor one over the other. In other words, there is a “level” playing field.
In the latter, circumstances facing firms may vary a lot. A firm in country X might be subsidized (explicitly or implicitly) by its government, or may face much weaker environmental and labor standards than prevail in country Y. And even if regulations on the books are similar, the firm in country X may find it easier to evade them.
From a formal economic standpoint, the resulting variation in comparative costs across countries is no different from those that arise from differences in relative factor endowments (which country has cheap oil) or productivity (which country’s factories are more automated), and hence may even be the source of additional gains from trade. But the opportunities to trade that arise from such unevenness in rules may appear unfair.
Economists have traditionally resisted entertaining such fairness concerns in discussions of trade policy. If labor standards are weak or nonexistent in low-income countries, why should that not count as just another source of comparative advantage? Besides, would the workers displaced from “sweatshops” if trade of this kind were banned not be even worse off in the absence of the trade opportunities? Does it not make economic sense to accept pollution-intensive activities in jurisdictions where the political pressure for cleaner air is lower and hence environmental regulations are weaker?
But let’s look at these concerns from the standpoint of the affected groups, particularly labor, in the importing country. After long struggles, workers in most advanced countries have achieved a significant expansion of social rights, including freedom of collective bargaining and prohibition of child labor. A key feature is that they make it illegal (and illegitimate) for firms to compete on the basis of cost advantages derived from violating these standards. A firm cannot outcompete another firm by employing workers who are willing to exempt themselves from national labor regulations.
Regulatory differences between countries need not always be problematic. They can be based on differences in circumstances or preferences, and need not reflect clear-cut violations of social rights.
But international trade renders what is illegal (and illegitimate) in a national setting to be legal (and, in the eyes of many economists and technocrats, fully legitimate) elsewhere. A firm may not be able to import 12-yearolds and put them to work at home, but can achieve the same outcome by employing those children abroad either directly or through local subcontractors. An economist looks at this and sees gains from trade. A social reformer, by contrast, sees undercutting of domestic labor standards. Domestic workers are effectively being told that if they want to compete with imports, they need to sacrifice hard-earned labor rights.
In some cases, international trade laws recognize the need to pay at least lip service to considerations of fairness. That is why export subsidies and “dumping” (selling below cost) by exporters are generally punishable by remedies such as tariffs in the importing country even though the purely economic case for doing so is weak.
But what about child labor, exploitative work practices or blatant repression of collective bargaining rights? In all these cases, many people see trade without regulation as unfair. Yet current rules generally do not allow countries to restrict imports on the basis of such considerations. Indeed, restricting imports because they were made under conditions that break the laws of the importing country would violate World Trade Organization rules and could legally be met with retaliation.
Regulatory differences between countries need not always be problematic. They can be based on differences in circumstances or preferences, and need not reflect clear-cut violations of social rights. For example, an exporting country may have a comparatively low minimum wage, reflecting depressed levels of labor productivity. Clearly, this should not raise concerns about unfair trade — though in practice it often does.
In other cases, countries may choose weaker social and labor protections because of perceived trade-offs with other social objectives (e.g., higher levels of employment). This will put competitive pressure on producers in these countries even if there are no rights violations. Indeed, this is one of the considerations that weighed heavily with the European Union when negotiating the terms of Brexit with the UK.
Fairness considerations in trade do not call for uniformity in labor or social rules — regulatory diversity reflecting cultural diversity is a value in itself. But in general, the more complete and deeper the integration, the greater will be the need for harmonizing regulations. Within the EU, divergence in labor rules between some countries at the periphery (e.g., Poland) and the more advanced nations (e.g., France) has often created tensions. In the Brexit agreement, the EU received assurances from the UK that its industries would not be undercut by weaker labor and environmental rules in the UK.
Economic integration comes with a tradeoff between the gains from trade, on the one hand, and the gains from regulatory diversity, on the other. Economists should acknowledge that trade can, indeed, raise questions without obvious answers where values diverge.
The Benefits of Deep Integration Are Ambiguous
Economists typically think of international trade policy in terms of tariffs and quotas. But as the analysis above suggests, trade policy has become less about such textbook frictions and more about behind-the-border barriers that raise the costs of penetrating markets. The idea is (or at least was) that as traditional barriers came down, further gains from trade could be reaped by removing the costs created by policies or regulations that were traditionally considered domestic policies.
Agriculture, services, subsidies, health and sanitary rules and intellectual property regulations were some of the new areas that were incorporated into the WTO in 1994. Subsequent trade agreements negotiated bilaterally or regionally went even further, entering additional areas such as banking, finance and labor regulations.
The trouble is that domestic policies in these domains often served important distributive roles or were the outcome of historical social bargains. When they became part of trade negotiations, the result was the perception (often accurate) that trade agreements were being hijacked by specific interests seeking to overturn long-standing social contracts. Trade agreements became more divisive and contentious.
But this is not merely a question of perception: the political backlash against deep integration does have reasonable economic underpinnings. International agreements that constrain domestic regulatory autonomy produce aggregate economic benefits that are far more ambiguous than is the case for lowering traditional border barriers. They may well reduce “trade costs” and boost increases in the volume of trade and cross-border investment. But their efficiency and social welfare impacts are fundamentally uncertain.
Consider regulatory standards designed to promote consumer safety, the environment or health. The harmonization of such standards lies at the center of today’s trade agreements. The justification is that reducing differences reduces the costs associated with doing business across borders. Tough regulatory standards that impede market access by foreign producers are sometimes labeled “non-tariff barriers.” And there is little question that governments sometimes do deploy regulations to favor domestic producers. But, as I discussed earlier, these differences often reflect varying consumer preferences or divergent regulatory styles.
European bans on genetically modified organisms and hormone-fed beef, for example, are rooted not in protectionist motives — the same bans apply to domestic producers as well — but in pressures from consumer groups at home. The U.S. government considers these to be protectionist barriers, and WTO dispute-settlement panels have often agreed.
But unlike the case of tariffs and quotas, there is no natural benchmark that allows us to judge whether a regulatory standard is “protectionist.” Different national assessments of risk and varying conceptions of how business should relate to its stakeholders — employees, suppliers, consumers, local communities — will produce different standards, none obviously superior to others.
Continued membership in the EU implied that the relevant trade-offs would be made favoring the single market rather than national difference. This was distributional conflict revolving less around material interests and more around values and broader social/political preferences.
In the language of economics, regulatory standards are public goods over which nations (and groups within nations) can have different tastes. Nations need to trade off the benefits of expanding market integration against the costs of harmonization. The resulting decisions are inherently political and distributional. And they remain contested as preferences and political coalitions shift.
The European acquis communautaire (the body of pan-EU laws) represents the apex of regulatory harmonization. The European single market is the intentional result of pursuing not just free trade but deep integration. This, in turn, has required an extensive and detailed body of laws and regulations — going so far as prescribing, for example, the size of cages for egg-producing hens — that apply, for the most part, to all member-states.
These trade-offs featured heavily in the UK debate on Brexit. One (perhaps charitable) way to understand the pro-Brexit case is that it was a demand for such decisions to remain in the hands of domestic politicians and policymakers rather than with European technocrats. Continued membership in the EU implied that the relevant trade-offs would be made in Brussels and would likely favor the single market rather than national difference.
This was perhaps a different kind of distributional conflict, revolving less around material interests and more around values and broader social/political preferences. For those with significant commercial, economic or professional stakes in accessing the European market, it was natural that material interests would predominate. For others, for whom the economic prospects were less bright, political and regulatory autonomy could rise to the surface.
“Dynamic” Gains From Trade Are Uncertain
The standard gains from trade are static effects that can be measured by the more efficient allocation of domestic resources created by cross-border exchange. It is also possible to envisage dynamic growth effects or productivity benefits that go beyond standard allocative efficiency gains. In particular, freer trade could produce an increase in the underlying rate of productivity growth of the economy instead of a one-time increase in consumption possibilities.
The advocates of trade agreements often rely on such growth or productivity effects to claim large economic gains. Many of the distributional issues I have discussed would not loom as large in the presence of a sustained increase in economic growth — a continuously rising tide is much more likely to eventually lift all (or most) boats.
In many middle-income and advanced economies, import competition has accelerated the process of de-industrialization. The key question is what happens to the labor that has to be re-allocated to other sectors as manufacturing shrinks.
Welfare-significant growth effects are most likely when trade enhances productivity within sectors, either within firms or by reallocation among firms, and when there are positive externalities associated with innovation. One common mechanism is increased trade facilitating technology transfer from frontier firms in other countries. Another is import competition forcing less efficient firms to exit while others are forced to rationalize their operations. Such effects are extensively documented, and, in general, trade is associated with increased productivity growth within sectors most exposed to the global economy, such as manufacturing. Remember how much better American cars became after the Big Three faced competition from Toyota and Honda?
What has been less well recognized is that trade-induced productivity growth within manufacturing does not necessarily translate to what really matters for aggregate gains, which is economy-wide productivity growth. In many middle-income and advanced economies, import competition has accelerated the process of de-industrialization. The key question is what happens to the labor that has to be re-allocated to other sectors as manufacturing shrinks.
When labor moves to lower-productivity services where technological externalities are less significant, or employment levels remain depressed in adversely affected regions, the economy-wide effects are considerably less salutary. Local economic decline and deindustrialization have been linked not only to poorer productivity but to a variety of negative externalities — social ills ranging from family breakdown to rising rates of addiction and suicide.
What about low-income exporting countries? It is undeniable that growth in many of these countries — and China in particular — has benefited from the openness of markets in Europe and the U.S. So even if trade may have aggravated inequality in advanced economies, it likely reduced global inequality — thanks in large part to China’s economic performance.
However, two points are worth making here, lest one draws too tight a link between post-1990 globalization and global economic convergence. First, successful industrializers relied on a wide range of policies that violated deep integration rules. China promoted industrialization not only by shielding its state enterprises from import competition for a long time, but also through subsidies, forced technology transfer, domestic-content requirements, currency manipulation and lax patent and copyright practices. Second, aside from China, the most prominent examples of export-oriented industrialization (Japan, South Korea, Taiwan) took place before the 1990s, when trade restrictions in the advanced economies were generally higher and trade liberalization was confined to border barriers.
Globalization and Capital Mobility Aggravate Inequality
My analysis so far has focused on trade, but it would be incomplete without some reference to the distributional effects of the international mobility of corporations and of financial globalization. There is evidence that greater capital mobility produces strong inequality effects. In particular, that capital liberalization leads to long-lasting declines in the labor share of income along with corresponding increases in income inequality and in the shares of the top 1, 5 and 10 percent of the income distribution.
It’s worth noting, though, there has been much greater cooperation and information-sharing among advanced economies in recent years, with a view to preventing a race to the bottom on taxing capital
There is no analogue to trade theory’s Stolper- Samuelson theorem in international macroeconomics implying that capital mobility creates absolute losers. But there is a more obvious explanation: as long as wages are determined in part by bargaining between employees and employers, capital mobility gives employers a credible threat — accept lower wages or we move abroad. This bargaining explanation is also consistent with the finding that foreign direct investment in particular is associated with the rise in inequality.
Another problematic consideration is that capital mobility increases volatility of labor earnings and, in particular, shifts the burden of economic shocks to workers. This, too, follows from the differences in cross-border mobility of labor and capital. The factor that is stuck within borders has to absorb the costs of miscellaneous shocks. Workers with the lowest skills and qualification — those least able to move across borders — are typically the most affected by this risk-shifting.
Another distributional shift has to do with the burden of taxation. As capital becomes globally mobile, it becomes harder to tax. Indeed, corporate tax rates have come down sharply in advanced economies since the late 1980s, sometimes by half or more. Such trends have been linked explicitly to tax competition in countries with free capital-mobility regimes. And since corporate taxes are generally progressive, downward pressure on tax rates is generally regressive. Meanwhile, the tax burden on wages (much of it social security charges) has remained roughly constant and rates of value-added taxes, which are usually regressive, have generally increased.
It’s worth noting, though, there has been much greater cooperation and informationsharing among advanced economies in recent years, with a view to preventing a race to the bottom on taxing capital income. Agreement has been reached recently among leading economies to establish a floor on corporate income taxation. Whether this will produce a significant shift remains to be seen.
• • •
The thrust of my analysis is that freer trade remains an unquestioned goal of mainstream economic policy, even though the direct efficiency gains have shrunk and the costs in terms of income inequality and household financial instability have grown. This is not a time for revisionist history in which we blame all (or even most) of Americans’ economic woes on globalization. But in an era in which the rising tide has plainly left many boats stranded and right-wing populists have been willing and able to exploit the resulting discontent, a more nuanced view is long overdue.
Freer trade is good — except when it isn’t.