The growth of world trade slowed sharply after 2011, both in absolute terms and relative to the growth of world output. Rather than growing faster than world gross domestic product (GDP), it only grew at the same speed; indeed, in 2016 it grew less rapidly.Footnote 1 These facts prompted concern on the part of organizations such as the World Trade Organization (WTO) and International Monetary Fund, and led journalists to ask whether we were witnessing the end of globalization. In 2015, for example, the Washington Post suggested that “the phenomenon … may be coming to an end.”Footnote 2 During the following year, journalistic speculation regarding the imminent demise of globalization increased still further, following the U.K.’s Brexit referendum in June and the election of Donald Trump to the U.S. presidency in November. Commentators and politicians have debated the causes of both political upsets, as well as their potential consequences, and now academic economists and political scientists are entering the fray (e.g., Allcott and Gentzkow Reference Allcott2017; Becker, Fetzer, and Novy Reference Becker, Thiemo and Dennis2017; Dhingra et al. Reference Dhingra, Hanwei and Gianmarco2017; Inglehart and Norris Reference Inglehart and Pippa2016). What, if anything, can economic history contribute to the discussion?
During the 1990s, as the integration of the international economy was deepening at an accelerating rate, economic historians responded to the commentary of the time by pointing out that globalization was not a new phenomenon, as did other scholars such as Paul Hirst and Grahame Thompson (Reference Hirst and Grahame1996). They also argued that it was not irreversible, since it had been reversed in the past. Paul Bairoch (Reference Bairoch, Peter and Sidney1989) had already written about the protectionist trade policy U-turn experienced in Europe from the late 1870s onwards. At the very end of his Presidential Address to the Economic History Association, Jeffrey Williamson suggested that late nineteenth century globalization might have “seeded its own destruction,” by creating losers as well as winners (1996, p. 302). That suggestion was followed up by authors such as Kevin O’Rourke (Reference O’Rourke1997), Ashley Timmer and Williamson (Reference Hatton and Jeffrey1998), Williamson (Reference Taylor and Jeffrey1997, Reference Hatton and Jeffrey1998), and Timothy Hatton and Williamson (Reference Hatton and Jeffrey1998), and was one of the main themes of O’Rourke and Williamson (Reference O’Rourke and Jeffrey1999). Harold James (Reference James2001) continued the story into the interwar period. The events of 2016 may have come as less of a surprise to authors such as these.
But just as economic historians’ first instinct was to try to temper some of the “globaloney” of the 1990s (Strange Reference Strange, Hirst, Thompson, Ruigrok, van Tulder, Doremus, Keller, Pauly and Reich1998), so their first task today should perhaps be to calm talk of the imminent demise of globalization. Economic history has something to say about what deglobalization looks like, and about whether what we are experiencing today matches that description. Economic historians, and historians more generally, can speak to the issue of whether the causes of Brexit and Trump were economic or not. And finally, economic history has something to say about the political stresses to which globalization may be subjected in the years ahead.
After a brief discussion of what I mean by “globalization” and “deglobalization,” and of whether we are already experiencing the latter, I consider three economic history literatures that speak to the state of the world today: the 1990s literature on the anti-globalization backlash of the late nineteenth century, focused in particular on trade and migration; the literature on interwar protectionism; and the literature on the links between trade and warfare.
Globalization has many facets, several of which are noneconomic (e.g., international criminal justice or the globalization of culture). Economic historians have, not surprisingly, focused on the economic dimensions of the phenomenon. In this essay, therefore, when speaking of “globalization” I primarily mean the integration of international markets for commodities, labor, and capital. By the integration of markets, I mean reductions in the costs of doing business internationally—of moving goods, people, or capital between countries or continents. Those costs could fall because of better technology—such as better navigation techniques, the discovery of new ocean routes, the construction of canals, more efficient forms of overland or maritime transport—or because of political factors promoting international economic integration, such as the emergence of geopolitical stability or more liberal domestic trade policies. Technological transfer is another important economic dimension of globalization that is crucial to the long-run dynamics of economic growth. I will occasionally allude to it, as well.
As O’Rourke and Williamson (Reference O’Rourke and Jeffrey2002) emphasize, globalization thus defined will be reflected in declining price gaps between markets. Ceteris paribus falling costs will lead to more trade (or migration, or capital flows, as the case may be), which is why it makes sense to look at the size of flows as well as price gaps. It seems reasonable on a priori grounds to expect that price gaps between markets that were closer together fell earlier than price gaps between markets that were further apart, and that ceteris paribus price gaps fell earlier for more valuable commodities than for cheaper and bulkier ones. O’Rourke and Williamson argued, based on the evidence available to them at the time, that intercontinental price gaps only started to fall systematically in the nineteenth century. The claim gave rise to a large literature (for a recent survey see Federico (Reference Federico, Claude and Michael2018)). We now know that while the volume of trade and the speed of price convergence were indeed much more impressive in the nineteenth century than before, evidence of price convergence can also be found during the early modern period (e.g., Pim de Zwart (Reference de Zwart2016) on the trade between the Netherlands and Asia). O’Rourke and Williamson (Reference O’Rourke and Jeffrey2009) themselves argue that Vasco da Gama played an important role in integrating Eurasian spice markets.
There is, of course, a vast literature on international trade during the early modern, medieval, and even classical periods: long-distance markets may have been characterized by large price gaps, but flows of commodities, and perhaps more importantly of slaves, technology, germs, and specie were sufficient at times to have transformative effects on economies (see inter alia Abu-Lughod Reference Abu-Lughod1989; Flynn and Giráldez Reference Flynn and Arturo2004; Findlay and O’Rourke Reference Findlay and Kevin2007; Wilson and Bowman Reference Wilson and Alan2017). And just as you can find evidence of globalization before the nineteenth century, depending upon your definition, so also can you find evidence of deglobalization, often linked to geopolitical upheaval such as the collapse of the Roman Empire, the end of the pax Mongolica, the disintegration of the Timurid Empire, or political crises in Persia or China (Lopez Reference Lopez, Cynthia, Edward and Postan1987, pp. 385–89; Rossabi Reference Rossabi and James1990; Ward-Perkins Reference Ward-Perkins2005). But as we will see, just as the globalization of the nineteenth century and later was different from what had gone before, so were the backlashes produced by it.
HAS GLOBALIZATION ENDED?
The 1930s remain the canonical example of deglobalization. That decade saw a worldwide increase in tariffs, as well as widespread resort to quotas, exchange controls, multiple exchange rates biased against imports, and a variety of clearing arrangements. Discrimination was rife. In contrast, average tariffs barely rose between 2008 and 2017, although 7.4 percent of world exports were facing higher tariffs in the latter year than in the former (O’Rourke Reference O’Rourke2018a; personal communication from Simon Evenett). It is true that 2018 saw the introduction of a wide range of tariffs by the U.S. administration, as well as retaliatory measures by several of its trading partners. However, despite this fact, as well as the widespread resort to non-tariff barriers of various kinds and policies artificially promoting exports, documented by the Global Trade Alert database, most economic historians would agree that interwar protection was far more ubiquitous and intrusive than what we have seen since 2008.Footnote 3 Figure 1 shows that as of 2017 U.S. tariffs remained at historically low levels.Footnote 4 At the time of writing (November 2018), it seems entirely possible that Brexit will eventually lead to a serious disintegration of markets linking the United Kingdom and European Union (EU), and the implications of President Trump for U.S. trade policy and the WTO appear increasingly alarming. But to date there has been no return to protectionism in the aftermath of 2008 comparable to what was experienced after 1929.
World trade growth accelerated in 2017. But in any event, using trade volumes alone as evidence of deglobalization is mistaken. These can rise or fall because of underlying shifts in supply or demand, unrelated to the costs of doing business across borders. If those costs do not change, and companies decide to slow or even reverse their outsourcing activities because of changing relative wages, for example, this hardly counts as “deglobalization.” That is why, as noted earlier, many economic historians have used price evidence to assess the integration or disintegration of international markets. Comparatively little work has been done on the twentieth century, but what there is finds evidence of deglobalization in the aftermath of the Great Depression. William Hynes, David Jacks, and O’Rourke (Reference Hynes, David and Kevin2012) use prices of agricultural commodities taken from the International Yearbook of Agricultural Statistics; Chris Hajzler and James MacGee (Reference Hajzler and James2015) use retail food prices in Canada and the United States. Both papers find evidence of increasing international price gaps during the 1930s.Footnote 5 I am unaware of any similar evidence for the period since 2008.
In summary, while the events of 2016 were dramatic, it is inaccurate to suggest that the world has, to date, experienced extensive deglobalization. That could still change in the future however. Even if it does not, no one could deny that the successful votes for Brexit and Trump, as well as the unsuccessful 2017 vote for the French National Front, the March 2018 election results in Italy, and support for populist parties elsewhere in Europe, represent a revolt against the established order in general and, at least in part, against globalization in particular. To what extent does economic history speak to the causes of this revolt, or help us to understand how to respond to it?
THE LATE NINETEENTH CENTURY
The argument that the populist revolts of 2016 represented, at least in part, a backlash against globalization rests on three assertions. First, the world economy became significantly more integrated after 1990, and particularly after 2001 when China entered the WTO. Second, this globalization had significant distributional effects. And third, those distributional effects ended up having political consequences. While there is little dispute about the first of these assertions, the second and third are hotly contested.
During the 1990s, economic historians tried to make precisely the same three-step argument to explain the anti-globalization backlash of the late nineteenth century.Footnote 6 First, they argued that the late nineteenth century international economy was integrating at an impressive rate: that globalization, viewed as a process, was dramatic during this period. Technological change was largely responsible, notably steamships, railroads, and the telegraph (e.g., Garbade and Silber Reference Garbade and William1978; Harley Reference Harley1988; O’Rourke and Williamson Reference O’Rourke and Jeffrey1999; Hoag Reference Hoag2006; Pascali Reference Pascali2017; Juhász and Steinwender Reference Juhász and Claudia2018).Footnote 7 Political and geopolitical factors such as European imperialism which opened colonies to trade, the imposition of close to free trade on countries such as China and Japan which remained independent, the pax Britannica which ensured a relatively peaceful century in Europe (Schroeder Reference Schroeder1994, p. vii), and (until the 1870s) trade liberalization in Europe also played an important role (Findlay and O’Rourke Reference Findlay and Kevin2007, Chapter 7).
Following scholars such as Knick Harley (Reference Harley1980), O’Rourke and Williamson (Reference Boyer, Timothy, Kevin, Timothy and Jeffrey1994) used price evidence to make the case for trans-Atlantic commodity markets, and they drew on work by other authors, such as A.J.H. Latham and Larry Neal (Reference Latham and Larry1983) and William Collins (Reference Collins1996), to make the case for commodity markets more generally. Taken in conjunction with the evidence on transport costs (Harley Reference Harley1988) and trade volumes, it is clear that the nineteenth century saw a dramatic integration of commodity markets. It is also clear that international markets for labor and capital became much better integrated. By the end of the nineteenth century, more than one million Europeans were leaving the continent every year, and there were substantial outflows from China and India, as well. Foreign investments as a share of world GDP rose from 7 percent in 1870 to almost 20 percent on the eve of WWI (Hatton and Williamson Reference Hatton and Jeffrey1998; Obstfeld and Taylor Reference Obstfeld and Alan2004).
Trade and Income Distribution in the Late Nineteenth Century and Today
Second, the 1990s literature argued that commodity market integration and mass migration had significant distributional effects, although these varied across countries. Let me start with trade. Ceteris paribus, narrowing commodity price gaps raised New World agricultural prices and European manufactured goods prices, while they lowered European agricultural prices and New World manufactured goods prices. O’Rourke and Williamson (Reference Boyer, Timothy, Kevin, Timothy and Jeffrey1994) used computable general equilibrium (CGE) techniques to argue that these price shocks benefited abundant factors and hurt scarce factors, in classic Heckscher-Ohlin fashion. In particular, falling transport costs hurt British landowners and benefited British workers, while they boosted land rents in the United States.Footnote 8 The impact was significantly bigger in Britain, reflecting its greater openness to international trade. O’Rourke, Alan Taylor, and Williamson (Reference O’Rourke, Alan and Jeffrey1996) provided econometric evidence for a small panel of seven “Atlantic economy” countries that relative goods prices were related to relative factor prices in line with Heckscher-Ohlin theory. Williamson (Reference O’Rourke and Jeffrey2002) broadened the argument to include what in the twentieth century came to be called the Third World.
Regarding the econometric approach of O’Rourke, Taylor, and Williamson (Reference O’Rourke, Alan and Jeffrey1996): a panel regression with seven countries and eight time periods might not strike people nowadays as providing particularly compelling evidence. On the other hand, it was at least a theory-consistent econometric test of the Heckscher-Ohlin model, which is rare enough. Since this is a general equilibrium theory, which predicts a relationship between a vector of economy-wide goods prices and a vector of economy-wide factor prices, the appropriate unit of observation is the economy, rather than regions or industries. As Pinelopi Goldberg and Nina Pavcnik (Reference Goldberg and Nina2007, p. 58) say, “The general equilibrium nature of the Heckscher-Ohlin model makes it extremely hard to bring it to the data.
Given that the model’s predictions refer to economy-wide factor returns, one has only one observation per year to work with.” Not surprisingly, therefore, later researchers such as Kris Mitchener and Se Yan (Reference Mitchener and Se2014), when revisiting these issues, have established a link between trade (i.e., traded goods prices) and income distribution using model-based simulation methods.
The most influential work linking trade and income distribution today is, deservedly, the work of David Autor, David Dorn, and Gordon Hanson. Autor, Dorn, and Hanson (Reference Autor, David and Gordon2013) show that U.S. regions which were more heavily exposed to rising Chinese import competition between 1990 and 2007 experienced higher unemployment, lower labor force participation, and lower wages. Since they “circumvent the degrees-off-reedom problem endemic to estimating the labor-market consequences of trade” (p. 2124) by focusing on regions, they are not testing long-run Heckscher-Ohlin theory. On the other hand, their research suggests that the “short run” regional disequilibria arising from major trade shocks are much more persistent and severe than previously thought, and that we need to worry about these more than we once did. That, in turn, suggests an agenda for future historical research, assuming that the required regional data are available.
Migration, Wages, and Employment in the Late Nineteenth Century and Today
When it came to the impact of mass migration on income distribution, the literature of the 1990s used a similar mixture of techniques. Since the scholars involved were interested in economy-wide effects, there was the same degrees of freedom problem as that described earlier, and similar responses to the problem. George Boyer, Hatton, and O’Rourke (Reference Boyer, Timothy, Kevin, Timothy and Jeffrey1994), O’Rourke, Williamson, and Hatton (Reference Boyer, Timothy, Kevin, Timothy and Jeffrey1994) and O’Rourke and Williamson (Reference O’Rourke and Jeffrey1995) used CGE models. Taylor and Williamson (Reference Taylor and Jeffrey1997) assumed a three-factor (land, labor, and capital) production function, and estimated the elasticity of substitution econometrically using data for a panel of 14 countries. This elasticity, together with information on factor shares, yielded aggregate labor demand elasticities, which allowed them to estimate the wage impact of migration-induced labor supply shocks. For example, the 24 percent increase in the U.S. labor force between 1870 and 1910 was estimated to have lowered U.S. wages by 8 percent (O’Rourke and Williamson Reference O’Rourke and Jeffrey1999, p. 155).
Other research tried to circumvent the degrees of freedom problem by using variation across cities or states, in particular in the United States. As in the case of trade, this approach cannot deal with the issue of whether or not there were economy-wide effects of immigration on wages. On the other hand, such studies may pick up a local effect if intra-regional labor market adjustment is sluggish, as the work of Autor, Dorn, and Hanson (Reference Autor, David and Gordon2013) suggest is the case today.
By and large, the literature suggests that nineteenth-century immigration put pressure on local U.S. labor markets. Hatton and Williamson (Reference Hatton and Jeffrey1998, Chapter 8) found that higher rates of immigration were correlated across U.S. states with higher rates of native-born out-migration, controlling for a variety of other factors, while Collins (Reference Collins1997) found a negative correlation between foreign-born and African-American net migration to a sample of northern U.S. states and cities. Claudia Goldin (Reference Goldin, Claudia and Gary1994) found a negative city-level relationship between wages and the proportion of the population that was foreign-born: a 1 percent increase in the latter lowered wages between 1 and 1.5 percent, an effect big enough that it cannot plausibly be attributed to composition effects (as noted by Hatton and Zachary Ward 2018), and notable in the context of the evidence on intra-regional labor market adjustment provided by Collins, Hatton, and Williamson.
There is a vigorous debate regarding whether present-day immigration lowers wages or employment levels. To take some recent examples, Gianmarco Ottaviano and Giovanni Peri (Reference Ottaviano and Giovanni2012) find that immigration into the United States between 1990 and 2006 had a small positive effect on native wages, but lowered the wages of previous immigrants; Marco Manacorda, Alan Manning, and Jonathan Wadsworth (Reference Manacorda, Alan and Jonathan2012) find broadly similar evidence for immigration into the United Kingdom from the 1970s to the 2000s; Christian Dustmann, Tommaso Frattini, and Ian Preston (Reference Dustmann, Tommaso and Ian2013) find that immigration between 1997 and 2005 lowered the wages of native-born U.K. workers below the 20th percentile of the wage distribution, but slightly increased the wages of well-paid workers; and Dustmann, Uta Schönberg, and Jan Stuhler (Reference Dustmann, Uta and Jan2017) find that immigration in 1991 had a moderate negative impact on local native-born German wages, and a significant negative impact on native employment.
Such mixed evidence stands in contrast to the unambiguous conclusions of the 1990s literature on late nineteenth-century migration. Is this contrast due to different types of data and research methodologies, or to different historical contexts? One obvious historical difference concerns the proportional scale of migration flows: Taylor and Williamson (Reference Taylor and Jeffrey1997) estimate that immigration raised labor supply by 24 percent in the United States, 44 percent in Canada, and 86 percent in Argentina. It would be odd if such enormous flows had not had an impact, even if only counterfactual, on wages. More speculatively, the size of the flows may have made it easier to identify wage elasticities by shifting labor supply curves to a greater extent.
Ran Abramitzky and Leah Platt Boustan (Reference Abramitzky and Leah2017) list several reasons why the elasticity of wages with respect to immigration might have been larger in the United States a hundred years ago than today: greater similarity in the labor market characteristics of immigrants and natives; less employment in the service sector, where the impact on wages might be muted; or a similar regulatory framework (involving not much regulation) for both categories of workers, rather than segmentation between legal and illegal workers. Bin Xie (Reference Xie2017) further suggests that manufacturing jobs may have been more homogenous then than now, implying less scope for natives to specialize so as to avoid competition from immigrants; and (citing Rosenbloom (Reference Rosenbloom1996)) that inter-regional mobility within the United States may have been less strong in the nineteenth century, implying greater local impacts for econometricians to find. Modern minimum wage laws are another reason why wage effects might be smaller today than a hundred years ago.
Having said that, Abramitzky and Boustan’s (Reference Abramitzky and Leah2017) judgment that immigration “creates winners and losers in the native population and among existing immigrant workers, reducing the wages of low-skilled natives, encouraging some native born to move away from immigrant gateway cities, and spurring capital investment” seems reasonable. So does the frequent finding that the greatest negative wage impacts of immigration may be felt by previous immigrants. We might expect the short- and long-run impacts of immigration to be different, with accumulation and technological responses to immigration potentially muting its wage impact (for a survey, see Lewis (Reference Lewis2013)). We might also expect the impact of immigration to vary depending on the nature of the flows and the institutional and economic context.
Abramitzky and Boustan are surely right to call for more research, using modern techniques, on the labor market impact of migration during the late nineteenth century. Recent studies for the interwar and early postwar periods show the way. Consistent with the 1990s research on the nineteenth century, Xie (Reference Xie2017) finds that U.S. immigration restrictions raised manufacturing wages and promoted South to North migration by African Americans during the 1920s; less consistent are Philipp Ager and Casper Worm Hansen (Reference Ager and Casper2017), who also use quotas as an identification strategy, but rely on occupational data, finding that while quotas boosted the status of African Americans, they pushed native whites into lower wage occupations; even less consistent are Alan Green and David Green (Reference Accominotti and Barry2016), who find only small effects of immigration on Canadian wages during the 1920s, once general equilibrium adjustments are taken into account; Marco Tabellini (Reference Tabellini2019), who finds only very small (and not statistically significant) effects on manufacturing wages of immigration in a panel of U.S. cities between 1910 and 1930; and Michael A. Clemens, Ethan Lewis, and Hannah Postel (Reference Clemens, Ethan and Hannah2018), who find that the exclusion of almost half a million Mexican farm workers from the United States at the end of 1964 failed to substantially raise the wages of native-born workers. The evidence is thus mixed, as it is for the modern period.
In summary, the literature on the late nineteenth century found that globalization created both winners and losers, with the most prominent losers being European landowners and native-born New World workers. What was the political impact of this?
Backlash in the Late Nineteenth Century and Today
The third step in the argument that late nineteenth-century globalization undermined itself was that the distributional effects of migration and trade led to an anti-globalization backlash. When using regional data to test whether trade or migration are linked to voting behavior at the local level, there is a problem analogous to the “degrees of freedom” issue discussed earlier. Such exercises will miss any national-level political effects of globalization operating equally across districts. Despite this, when Goldin (Reference Goldin, Claudia and Gary1994) studied 1915 Congressional voting on the U.S. Literacy Test, she found that higher local wage growth lowered the likelihood of an anti-immigration vote. Timmer and Williamson (Reference Timmer and Jeffrey1998) developed a cross-country index of immigration policy for a panel of countries, and found that policy became less restrictive the higher were unskilled wages relative to average incomes. Put these findings together with those on the impact of immigration on wages, and you have a clear case of a dimension of globalization (in this case migration) undermining itself.
The 1990s evidence on the determinants of nineteenth-century trade policy was more qualitative, although it was able to draw on an extensive political science literature, such as Peter Gourevitch (Reference Gourevitch1977) and Ronald Rogowski (Reference Rogowski1989). O’Rourke (Reference O’Rourke1997) argued that the more liberal trade policy responses of the United Kingdom and Denmark to the grain invasion could be explained by the different distributional effects it had in those two countries based on CGE evidence. Another approach is to study election results in settings where trade was the dominant issue to see whether votes on trade policy are consistent with economic interests. The typical finding of authors such as Douglas Irwin (Reference Irwin1994) is that economic interests matter. He finds a role for sectoral interests in explaining why the United Kingdom did not revert to protectionism in 1906. Sibylle Lehmann (Reference Lehmann2010) uses ecological inference methods and concludes that while small farmers favored free trade in the 1878 German election, large estate owners favored protection, which is in line with Heckscher-Ohlin theory. On the other hand, the fact that smaller farmers shifted towards protectionist candidates in Sweden in 1887 seems puzzling. Lehmann and Oliver Volckart (Reference Lehmann and Oliver2011) interpret that shift as representing a loss of trust in the incumbent government, a reminder that even in elections where trade is a major issue votes are cast for many reasons.Footnote 9
The question of whether the electoral upsets of 2016 were due to economic factors, and in particular whether they represented an antiglobalization backlash, is fiercely disputed. Some have argued that the causes were cultural, but if the causes were purely cultural one has to ask what it was that led working class voters to become more culturally conservative in the United Kingdom, United States, France, Italy, and elsewhere over the course of the last 10 or 15 years. As historians, economic historians are presumably comfortable with complexity. Think of the debate about why Robert Peel repealed the Corn Laws: Was this due to ideas or interests? A long literature, expertly summarized by Cheryl Schonhardt-Bailey (Reference Schonhardt-Bailey2006), has shown that it was much more complicated than that. Yes, there is evidence that Members of Parliament voted in their own economic interests, and in the economic interests of their constituents. The lobbying of interest groups mattered, as did the diversification of landlords’ portfolios. But Peel’s ideas about the determinants of wages also mattered (Irwin Reference Irwin1989), as did his sense of ethics, and the ability of the Duke of Wellington to deliver the House of Lords, and the constitutional reasons why the Duke decided to do so, and so on. Imagine our successors looking back at our own period in 50 years’ time, and asking students whether the turmoil of 2016 was due more to economics or to culture. Is it not likely that the answer to that question will be the same as it usually is, when students are given the choice between two pedagogically useful but overly simplistic alternatives— that it was, in fact, a bit of both?
As indicated previously, my view is that economic historians have done a reasonable job of showing that late nineteenth-century globalization undermined itself by creating both winners and losers, with the latter successfully influencing trade or migration policies in several countries. But economics was not the only factor at work even then.
Yes, immigration restrictions in the United States were linked to the impact of immigration on wages, but even a cursory examination of the relevant legislation shows an obsessive concern with Asian immigration in particular. No-one reading Avner Offer’s (Reference Offer1989) account of how anti-Asian racism was ubiquitous throughout the late nineteenth-century Anglosphere could be in any doubt that “culture” was a powerful factor underlying late nineteenth-century immigration restrictions. But economics mattered also.
Can you fully understand Brexit without taking account of English nationalism, or the obsessive hostility towards Obama without reference to the color of his skin (on the former, see O’Rourke (Reference O’Rourke2018b, Chapter 9))? Surely not, but at the same time, evidence suggests that economic factors were also important during 2016. This does not automatically mean that voters were being “rational,” in the sense that they were voting in their own economic self-interest. Take, for example, the evidence presented by Sascha O. Becker, Thiemo Fetzer, and Dennis Novy (Reference Becker, Thiemo and Dennis2017) and Fetzer (Reference Fetzer2018) showing a clear link between austerity policies and the pro-Brexit vote in Britain. This is a good example of economically motivated voting, but it is not a good example of economically rational voting for the simple reason that Conservative austerity policies had nothing to do with the EU.
However, other evidence on twenty-first century voting behavior is more consistent with the late nineteenth-century anti-globalization backlash narrative. Becker, Fetzer, and Novy find that migrant inflows from Eastern Europe were correlated with the pro-Brexit vote, though the effect is small. Italo Colantone and Piero Stanig (Reference Colantone and Piero2018) find that the Brexit vote was significantly higher in regions that were more exposed to Chinese import competition. Autor et al. (Reference Autor, David and Gordon2016) find that increased imports from China were associated with political polarization at the Congressional district level in the United States from 2000 onwards. Taken in conjunction with their evidence, cited earlier, on the impact of Chinese imports on local labor market outcomes, and you have a compelling argument linking globalization to both economic outcomes and political change. The precise mechanism is unclear, however. Some Trump supporters were presumably voting in favor of protection, in the manner of nineteenth-century European landowners. However, as Autor et al. note, populist voters could also have been voting for politicians promising a greater slice of the public service pie for white voters at a time of increased competition for government services. Alternatively, populist politicians could have opportunistically responded to economic hard times by supplying extremist policies designed to increase voter turnout and campaign contributions (Autor et al. Reference Autor, David and Gordon2016, p. 4). The literature on the link between economic distress and populism is too vast to be summarized here, but a reasonable conclusion is that trade and immigration have contributed to the surge of populism in Europe and America.
What Is to Be Done?
One shortcoming of O’Rourke and Williamson (Reference O’Rourke and Jeffrey1999), in my opinion, is that its view of the political process was excessively simplistic. Politics was effectively reduced to a binary choice: governments could choose to stay open, or not. They remained open where the configuration of interests permitted it, and they erected barriers to trade or immigration otherwise. A practicing politician might find suggestions regarding how to maintain openness in the face of protectionist pressures more useful.
This is where the work of Michael Huberman and co-authors comes in. In a series of articles and a subsequent book (Huberman Reference Huberman2012), Huberman describes the gradual introduction, during the late nineteenth century, of labor market regulations and social insurance programs designed to protect workers. In some cases, notably Belgium, workers’ support for trade liberalization was conditional on the introduction of this “labor compact.” Importantly, the labor compact was more advanced in countries more open to trade: there was no sign of globalization leading to a race to the bottom during this period. This mirrors the finding of Dani Rodrik (Reference Rodrik1998) that more open economies had bigger governments in the late 1980s and early 1990s. Rodrik interprets this correlation as reflecting a higher demand for the insurance services provided by the government sector in countries more exposed to trade shocks.
In this view, states and markets are complements rather than substitutes, in that state action can help to maintain political support for openness. It is possible that financial globalization and greater competition for mobile direct investment may today be placing constraints on the ability of states to finance such policies (Rodrik Reference Rodrik2017). On the other hand, there has surely never been a more tightly integrated group of economies in human history than the 28—soon to be 27—member states of the EU. Despite the hyper-globalization linking its members, the EU seems to be able to accommodate wildly differing preferences regarding the share of GDP going to government. Setting aside Ireland, whose GDP statistics are not comparable with those in other countries, general government expenditure as a share of GDP ranged in 2016 from a low of 34 percent in Romania to a high of 56.4 percent in France. The ratio was less than 40 percent in eight member states, and higher than 50 percent in five.Footnote 10 The late nineteenth-century lesson that globalization and pro-worker state intervention can go hand in hand may not be out of date yet.
Economic history also suggests that states can directly cooperate with each other to avoid mutually harmful regulatory races to the bottom that might undermine support for openness. Huberman and Christopher Meissner (Reference Huberman and Christopher2010) discuss how countries traded market access in return for promises to strengthen regulatory standards during the late nineteenth century. The 1904 agreement between France and Italy also linked regulatory standards and migration: France allowed Italian immigrants to benefit from the French labor compact, on condition that Italy bring its own regulatory standards up to international levels. Alan Milward (Reference Milward2000) emphasized that European integration was an essential part of that continent’s response to the traumas of the interwar period, since it boosted the capacity of states to deal with the problems facing them. It was important for countries to be able to reap the benefits of trade, but this could not come at the expense of broader social and political objectives. “The problem genuinely was how to construct a commercial framework which would not endanger the levels of social welfare which had been reached . . . The Treaties of Rome had to be also an external buttress to the welfare state” (Milward Reference Milward2000, p. 216). And so the Treaty of Rome (1957) envisaged not just a customs union, but inter alia a Common Agricultural Policy, regional policy, and the harmonization of social policies. The supranational elements of the European project were a logical consequence. These always bothered the British, but they were (and still are) seen as essential in maintaining popular support for European economic integration (O’Rourke Reference O’Rourke2018b).
THE GREAT DEPRESSION
O’Rourke and Williamson suggest that the deglobalization of the interwar period was not solely the result of an exogenous shock, namely the Great War, but had its roots in the anti-globalization backlash of the late nineteenth century (1999, p. 286). We did not discuss in detail the extent to which this was really the case, but it merits reflection. I think that we were on fairly solid ground, for example, when it comes to U.S. immigration policies. There is a clear continuity between the gradually tightening restrictions over the course of the late nineteenth century, the introduction of the U.S. Immigration Act of 1917, and the immigration quotas of the 1920s. The European agricultural protection introduced before 1914 survived the war intact, and remains with us today. The late nineteenth century also saw a shift towards protecting heavy industry in particular, and manufacturing more generally, from British competition. This phenomenon was so ubiquitous that Robert Allen (Reference Allen2011) describes it as being part of a developmental “standard model.” It was adopted by countries across the globe as they regained tariff autonomy in the twentieth century.
On the other hand, it would be wrong to suggest that the outbreak of WWI and the onset of the Great Depression did not represent important discontinuities that had serious negative consequences for the openness of the world economy. To what extent could either of these be regarded as examples of globalization undermining itself, as opposed to exogenous shocks imposed upon an otherwise robust and open international system?
Let me start with the Depression. We know from Peter Temin (Reference Temin1989) and Barry Eichengreen (Reference Eichengreen1992) that the gold standard and Great Depression were inextricably linked. The gold standard was part of the institutional infrastructure underpinning global capital markets during the late nineteenth century. Going back on gold in the 1920s was an important part of the attempt to recreate the highly globalized pre-war world, and was seen by politicians as signaling a commitment to a generally open international system. To that extent, you could regard the Great Depression that followed as being caused by 1920s globalization. Perhaps that is a slightly forced argument. A more compelling reason to view the Depression as having been caused, at least to some extent, by the globalization that preceded it, is the role played by international capital flows in transmitting the crisis around the world. This was presumably always self-evident to scholars of countries such as Germany, and was a central focus of Charles Kindleberger (Reference Kindleberger1973), but I think that Anglo-Saxon economic historians have tended, since the work of Eichengreen and Temin, to think about the Depression more in Mundell-Fleming terms, with exchange rates, interest rates, and monetary and fiscal policy being at the heart of the analysis. This tendency fits well with Keynes’s repeated denunciations of interwar macroeconomic policy-making. Since 2008, however, and in particular since the Eurozone crisis of 2010, it has become more natural to once again think of the Depression in Europe in terms of a sudden stop of capital flows, a perspective recently exemplified by Olivier Accominotti and Eichengreen (Reference Accominotti and Barry2016). Seen in this perspective, the Great Depression in Europe appears more clearly as a negative consequence of a particular dimension of globalization, namely capital mobility: When the lending stopped, countries like Germany found themselves obliged to cut expenditure, while their financial systems came under strain as foreign depositors withdrew money from banks (Eichengreen Reference Eichengreen2015, pp. 137–44). And perhaps the German historical literature debating whether Chancellor Brüning, widely criticized for his economically and politically destructive austerity policies, could have behaved in a less dangerous manner resonates more strongly today than it once did to observers of the Eurozone periphery.
To the extent that the Depression was caused by malfunctioning international capital markets, or the gold standard that was thought at the time to underpin them, we have yet another case of globalization undermining itself, since the Depression was the major cause of interwar protectionism (Smoot-Hawley, which had its roots earlier, notwithstanding). Eichengreen and Irwin (Reference Eichengreen and Douglas2010) show that protectionism during the 1930s was associated with adherence to the gold standard, a point that had earlier been noted by, among others, the League of Nations (1943, p. 31). Staying on gold longer meant balance of payments problems and a more protracted depression, and doing so deprived policymakers of alternative ways of addressing these problems.
Even worse, the Depression promoted political extremism, notably but not only in Germany. Gregori Galofré-Vilà et al. (Reference Galofré-Vilà, Christopher and Martin2017) find that austerity was positively correlated with the Nazi vote in Germany between 1930 and 1933, but there is econometric evidence linking economic crises with right-wing extremism more generally. Alan de Bromhead, Eichengreen, and O’Rourke (Reference de Bromhead, Barry and Kevin2013) find evidence that protracted periods of poor economic performance were associated with higher votes for right wing political extremists in the interwar period, while Manuel Funke, Moritz Schularik, and Christoph Trebesch (Reference Funke, Moritz and Christoph2016) find, in a sample of 20 advanced economies between 1870 and 2014, that far-right parties increased their vote share by 30 percent on average following financial crises. Nationalist extremism fueled even more dramatic deglobalization, as the Nazis developed a tightly controlled war economy characterized by the search for strategic self-sufficiency and bilateral trading arrangements (Tooze Reference Tooze2006).
What could have prevented these disasters? Superior macroeconomic policy-making is the obvious answer: countries could have chosen to go off gold earlier, preferably in a coordinated manner, and used the resultant policy flexibility to reflate their economies. To that extent, the global policy response after 2008 was far superior. The experience of the Eurozone periphery after 2010, when countries found themselves unable to devalue, obliged to engage in procyclical austerity, and (until Mario Draghi took over the reins of the European Central Bank) locked into a monetary union pursuing excessively conservative policies, shows that these basic macroeconomic lessons of the Great Depression still have relevance (O’Rourke and Taylor Reference O’Rourke and Alan2013). That experience surely helps to explain the election of a populist Italian government in 2017.
WORLD WAR I
Finally, what about WWI? Adam Tooze and Ted Fertik (Reference Tooze and Ted2014) argue that we should not regard it as having ended the globalization of the late nineteenth century, since it was itself a globalised conflict, involving high levels of international borrowing and lending and the large-scale transfer of military resources between continents. The point is well made. On the other hand, the available price evidence suggests an impressive disintegration of international commodity markets during the war (Findlay and O’Rourke Reference Findlay and Kevin2007, p. 434), while the conflict had a host of longer-run consequences, ranging from the creation of new nation states in eastern Europe, to structural imbalances affecting industries and agricultural sectors worldwide, to the Russian Revolution, all of which helped to ensure that it would be difficult to recreate the cosmopolitan international economic order that had, by and large, obtained before 1914 (ibid., pp. 435–43). Some interwar restrictions on trade and migration had their roots in the late nineteenth century, but the Great War had an independent disintegrating impact on the international economy.
To what extent should we think of the war as an exogenous shock hitting the world economy in 1914? Or did it represent, at least to some extent, an even more dramatic form of anti-globalization backlash? Tooze and Fertik (Reference Tooze and Ted2014) make what is perhaps the most direct connection between globalization and war: capital flows and technology transfer helped Russia converge on the established powers. This disrupted preexisting geopolitical equilibria and led Germany in particular to devise military strategies to counter the threat. They could have added that the spread of the Industrial Revolution to Germany had earlier undermined the geopolitical equilibrium in Europe, perhaps to an even more fundamental extent. Similarly, the spread of modern growth to Japan undermined Asian stability in the first half of the twentieth century. If there is an analogy with today it is both obvious and worrying: few would dispute that Chinese growth has been intimately linked to globalization or that this will have consequences for the international balance of power.
David Rowe (Reference Rowe2005) suggests a more subtle mechanism linking pre-WWI globalization with the outbreak of that conflict. Globalization raised wages throughout labor-abundant Europe, raising the financial cost of recruiting soldiers or the political cost of conscripting them. This made states feel more insecure militarily and less able to credibly threaten military retaliation. It also made them worry about the reliability of their military allies. All these factors, according to Rowe, help to explain the catastrophic series of political decisions that were made in the run-up to war.
Another mechanism linking late nineteenth-century globalization to geopolitical instability is suggested by Offer (1989). Structural change, magnified by international trade which allowed industrializing countries to reap the benefits of comparative advantage, led first the United Kingdom, and later Germany, to become increasingly dependent on imported supplies of food and raw materials. This dependence made countries feel militarily vulnerable, and made naval superiority—or, in the case of Germany, as close to naval parity as was feasible—a matter of vital national strategic interest. The Anglo-German naval rivalry of the decades prior to 1914 was a logical consequence. Offer shows how naval strategists were motivated first by purely defensive considerations, but later began to explore the offensive possibilities of naval warfare (see also Lambert Reference Lambert2012). These considerations were of course not responsible for the outbreak of WWI, but they were at least in part responsible for the U.K.’s decision to enter the war. If Germany defeated France, it would be able to project continental-scale power directly into the Atlantic, safe from the threat of British blockade. As Sir Edward Grey said in 1911, if a European power achieved continental hegemony Britain would permanently lose its control of the sea, which would in turn mean its separation from the Dominions and the end of Empire (Howard Reference Howard1972, pp. 51–52).
Tooze (Reference Tooze2006) and Michael Barnhart (Reference Barnhart1987) have shown how a concern to achieve strategic self-sufficiency was both a cause and an effect of the drive to war during the interwar period in Germany and Japan respectively. Roberto Bonfatti and O’Rourke (Reference Bonfatti and Kevin2018) show more generally how a follower country’s strategic dependence on imported raw materials can, in conjunction with the naval hegemony of an established power, give the follower country an incentive to launch pre-emptive wars in the hope of gaining strategic self-sufficiency. Higher costs of war in high-wage globalized economies can also give politicians an incentive to gamble on rapid offensives, in the hope of striking a knockout blow (Eloranta and Harrison Reference Eloranta, Mark, Kevin and Stephen2010, p. 137).
Again, if there is a contemporary parallel, it is with China. Like Britain and Germany in the nineteenth century, it has been undergoing rapid industrialization and structural change, and it has become far more dependent on international trade. On the import side, it relies heavily on foreign oil, much of which passes through the Straits of Malacca; on the export side, it has toyed with using its near monopoly of rare earths for strategic purposes. The international state system has a poor record of peacefully accommodating the arrival of newcomers to the top table, and there is a fascinating theoretical and historical literature on the topic that is too vast to be summarized here (e.g., Gilpin Reference Gilpin1981; Powell Reference Powell2006). If China started to doubt that it could rely on the market to provide it with the food and raw materials it requires, the world could become a much more dangerous place.
One of the features of interwar trade that struck contemporary observers as particularly dangerous was its decreasingly multilateral nature, with countries trading more and more with their colonial possessions or spheres of influence. As Folke Hilgerdt (Reference Hilgerdt1935, p. 188) put it, “As bilateralism particularly renders the supply of raw materials to certain countries difficult, it threatens to lead to an intensified fight for influence upon (or the domination of) the undeveloped countries, and thereby to political controversies, which may adversely affect all forms of peaceful collaboration between nations.” de Bromhead et al. (Reference de Bromhead, Alan and Markus2019) show that discriminatory U.K. trade policies can explain more than 70 percent of the increase in the British Empire’s share of U.K. imports after 1930. And so it is no surprise that Article 1 of the GATT prohibited discriminatory trade policies, subject to a number of well-known exceptions. The interwar period suggests that the greatest costs of aggressive and discriminatory trade policies can be political rather than economic. It is in this context that upholding the international rule of law is so important, and that the implications of Brexit and Trump become potentially alarming. In particular, the U.S. refusal to reappoint judges to the WTO’s Appellate Body threatens to render the work of that body impossible.
The world has not yet experienced extensive deglobalization, but it could in the future, with dangerous geopolitical consequences. What should policy makers do?
Fortunately, the lessons of the Great Depression influenced macroeconomic policy after 2008, but they are not the only relevant lessons of history. The political upsets of 2016 occurred in the United States and the United Kingdom, where post-crisis macroeconomic management was far superior to that in the Eurozone, and where recoveries came correspondingly sooner. This suggests that longer-run forces, linked to trade and migration, may be having a greater impact on voting than the post-2008 macroeconomic and financial crisis. Is it a coincidence that the upsets of 2016 occurred in countries where the twenty-first century equivalent of Huberman’s labor compact has been eroded the most? My reading of history suggests that if governments push anything too far, including markets in general, and globalization in particular, a reaction is to be expected. Rather than promoting ever-closer integration in an already highly globalized world, maybe advocates of openness should declare victory and focus on how to protect individuals and regions from the risks that markets, both domestic and international, inevitably give rise to.
There is a final contribution that economic historians, like all historians, can bring to present-day debates about the political upsets of 2016, the causes of populism more generally, and the potential implications— some of which are very worrying—for the international order. As economists, we are trained to look for patterns and to seek general explanations for entire classes of phenomena rather than specific explanations for individual historical events. But as historians we are trained to recognize the uniqueness of individual events and acknowledge the roles played in them by context, contingency, and the choices made by individual actors. This is the creative tension that lies at the heart of our discipline and makes it so fascinating. And the perspective of the historian is important today, since it reminds us that we and our leaders enjoy free will, which we can exercise for better or for worse. Nothing is inevitable. In dangerous times that is a useful thought to bear in mind.