It is a fact that foreign workers will be harvesting the food we eat in the United States . . . they will either be doing it within our borders with our domestic food supply or they will be doing it outside our borders and shipping us a foreign-grown food supply.
–Tom Nassif, President, Western Growers AssociationFootnote 1Business leaders have long understood that there was a trade-off between trade, the ability to move production overseas, and immigration. Yet, the field of international political economy (IPE) has largely ignored this trade-off and immigration policy more generally.Footnote 2 This is understandable given that, since World War II, policy-makers have treated migration policy as domestic policy. After the war, policy-makers, recognizing the interactions among trade, finance, and security, forged a patchwork of interlocking international regimes to govern global security, trade, and finance in hopes of recreating the nineteenth-century liberal international order. Yet apart from provisions to accommodate refugees, provisions governing international economic migration are conspicuously absent from the Anglo-American postwar order. Scholars studying migration have similarly conceptualized migration as domestic concern, focusing on three domestic-level variables to explain changes in policy: the power of organized labor, the importance of immigrant groups, and the rise of nativism.
This domestic perspective, however, fails to explain US postwar immigration policy. Briggs, for example, argues that immigration policy has been driven by variation in the power of organized labor.Footnote 3 Yet immigration was reopened somewhat after World War II when labor was relatively strong and was closed in the 1990s when labor was relatively weak. Other scholars focus on immigrants as an important lobbying group.Footnote 4 Nonetheless, the foreign-born have never been more than 14 percent of the population and naturalized foreign born, those who can vote, have never been more than 7 percent of the population, limiting their ability to affect politics.Footnote 5
Finally, nativist backlash, which is often thought of as the conventional wisdom, has been argued to explain changes in policy.Footnote 6 Yet, nativist backlash has occurred several times in US history without leading to a change in policy. In the 1840s and 1850s, there was a major outcry against Irish and German immigration, which led to the creation of nativist parties, but not restrictions. Backlash against southern and eastern European immigrants in the 1890s again led to little action on immigration. Most recently, nativist backlash has led to some action on the state level, for example, Arizona's SB1070 or Alabama's Self-Deportation law, but not at the federal level. Nativism is too ubiquitous a phenomenon to be the full explanation for immigration policy. Instead, I argue that scholars must examine when firms serve as a bulwark against nativism and when they choose to stay on the sidelines of the low-skill immigration debate.
This article argues that immigration policy—particularly low-skill immigration policy (LSIP)—is largely driven by the economy's need for low-skill labor that, in turn, is affected by the country's trade policy and the ability of firms to move production overseas (what I term, firm mobility.) It is generally accepted that trade closure leads to an increase in low-skill-intensive production in low-skill labor-scarce states, such as the United States, and a concomitant increase in wages. Without an increase in the labor supply, any advantage that firms gain from trade protection may be erased because of increasing wages. One expects, then, that firms lobby for liberalizing LSIP when trade is restricted. As firms tend to be powerful, LSIP should be relatively open. Similarly, when firms are immobile across international borders, because they are legally or technologically unable to move capital or because there are few safe places for investment, their need for low-skill labor at home increases as does their support for LSIP.
In contrast, trade openness leads to a decrease in low-skill labor-intensive production, reducing the need for labor and, in many cases, forcing businesses to close. Businesses that close no longer lobby policy-makers, and businesses that remain open also have less incentive to lobby policy-makers for open LSIP as wages for low-skill workers have decreased. Similarly, when firms are mobile, because of open capital policies, new technologies, or greater investor protections from foreign governments, their support for LSIP decreases because of their outside option. Given the existence of groups who oppose LSIP, one expects that policy-makers will respond to less support for open LSIP by restricting it.
In this article, I show how openness to trade and other states' openness to foreign capital affect LSIP through the lens of US senators' voting behavior on immigration after 1950. This case was chosen because, empirically, examining Senate voting allows one to establish causality. Although trade and capital policy are likely to be endogenous to firms' and policy-makers' preferences, I argue in this study that there are two measures—the level of tariff barriers and average world openness to capital flows—that US senators have had little ability to control since World War II.Footnote 7 With the Reciprocal Trade Agreements Act, Congress tied its own hands on tariff policy. Tariff rates could now stay the same or be cut; they could not be increased. Trade was also opened using international institutions such as the General Agreement on Tariffs and Trade/World Trade Organization (GATT/WTO), which help perpetuate trade openness far removed from the influence of US senators. The ability to move production overseas during this time period was largely driven by the decision of other countries to open their markets to foreign direct investment (FDI); again, something that was beyond the influence of a single senator.
Foreshadowing the results, voting on immigration in the post-1950 period can largely be explained by trade openness and the ability of firms to move overseas. There is little difference in voting behavior based on the explanations in the literature, including the party or ideology of the senators, welfare spending, unemployment, gross domestic product (GDP) growth, and the percent of foreign-born in the senators' states.
For the immigration literature, this article returns the focus to firms. Firms have often been conceptualized as having static preferences for openness.Footnote 8 This article shows that firms' preferences can change based on their production strategies, competitiveness, and locational choices. In a world of increasingly internationalized firm operations, understanding how endogenous locational choices by firms affect where and how they lobby is becoming increasingly important. This article sheds light on this problem and suggests a theory for how the internationalization of firms may affect other policy areas, such as labor or environmental policy.
Moreover, this article helps return focus to immigration policy as part of IPE. One of the key questions of IPE is why do countries open their borders to the free movement of goods and services, capital, and people.Footnote 9 IPE scholars have long examined the determinants of trade policyFootnote 10 and have increasingly examined the determinants of policies toward capital in all its forms,Footnote 11 but, except for a nascent literature on public opinion on immigration,Footnote 12 the third flow has been largely ignored.Footnote 13 This inattention is somewhat understandable given the youth of the IPE field: migration, especially low-skill migration, has played a lesser role in the current era of globalization than it did in the nineteenth century.Footnote 14 But, this smaller role has been attributable to policy choices, especially those made by the largest immigrant-receiving state, the United States. To better understand globalization, then, one needs a better understanding of why low-skill immigration has not been liberalized in the post–World War II era whereas trade from and capital movements to less-developed states have been.
Further, this article is part of an important trend to bring the different areas of IPE together.Footnote 15 It highlights, then, that the choice of openness policies matter. Although the economists are correct that any combination of openness of the three factors—people, money, and goods—will have similar effects on the size of the economy and the return to factors, they miss the political effects of the changing composition of industry that arises. Opening trade and capital will lead those firms most reliant on low-skill labor to become more productive, move overseas, or close their doors. As such, these firms will no longer lobby for LSIP and low-skill immigration will be restricted. The choice to open trade and capital, therefore, changes the political landscape, leading to changes in immigration policy and, likely, other areas of domestic policy as well.
The article proceeds as follows. I begin by theorizing how trade openness and increased firm mobility can lead to restrictions on immigration because of their effects on firms. Next, I discuss the ways in which tariff policy and the ability to invest overseas were out of the control of US senators. Third, I provide evidence that trade policy and firm mobility help explain this variation, and I examine the evidence for the alternative explanations. Fourth, I provide evidence showing how firm lobbying on immigration has changed with trade openness and increases in firm mobility. Finally, I conclude by discussing the implications of this article for research on other economic policies.
Trade Policy, Firm Mobility, and Support for Immigration
In this section, I examine how changes in trade policy and firm mobility affect firms' support for openness to low-skill immigrants and how this, in turn, affects policymakers' support for open LSIP. I argue that immigration policy toward low-skill workers and the policy toward high-skill workers should be studied separately; although, they often get bundled together conceptually and in legislation.Footnote 16 These policies target two different populations that are not interchangeable. Low-skill immigrants do not have the skills to take the place of high-skill immigrants and high-skill immigrants are unlikely to give up the wage premium that their skills provide them by taking low-skill jobs. Policy similarly is often used to target these populations; many countries have policies specifically targeting high-skill or low-skill workers. In general, when states are open to low-skill migrants they are also open to high-skill migrants but not vice-versa.Footnote 17
Further, the flow of low-skill migrants is more politicized than the flow of high-skill migrants. Natives tend to have much more favorable views of high-skill immigrants than low-skill immigration.Footnote 18 Nor do these preferences seem to be a new feature of politics. In the United States, for example, most of the anti-immigrant sentiment has been targeted toward the least-skilled migrant group: the Irish in the 1840s, the Chinese in the mid-1800s, the southern and eastern Europeans at the turn of the last century, and Hispanics today. It is, therefore, politically easier for a politician to support a more open policy toward high-skill immigrants than to low-skill immigration.
Although beyond the scope of this research, the politics of high-skill immigration may also be affected by trade openness and firm-mobility as many high-skill occupations can be easily outsourced overseas.Footnote 19 However, there is less opposition to high-skill immigration and governments may favor high-skill immigration given these immigrants' contributions to the tax base.Footnote 20 Therefore, even if there is less support for high-skill immigration from firms, high-skill immigration may remain open.
Finally, whereas high-skill immigration gains more attention from the media, low-skill immigration is, arguably, more important. The vast majority of all potential immigrants have been low-skill workers.Footnote 21 Moreover, studies have shown that the migrants' remittances lead to better outcomes for their families and communities and that increased low-skill migration would greatly increase both developing nations' and world income.Footnote 22 Given the benefits of low-skill migration, IPE scholars should be interested in why low-skill immigration is so restricted in most wealthy states, including the United States, today.
Although I examine US policy toward low-skill immigration in the empirical section, my argument applies to any low-skill labor scarce state; although, these states' policies may not “look” like US LSIP.Footnote 23 First, there are different ways to restrict LSIP. For example, the United States uses numerical quotas to control the flow of low-skill immigrants whereas Canada and Australia, among others, use a point system. The United States, Canada, and Australia adopted these regulations in the 1960s and 1970s with the same goal in mind; namely, they wanted to replace their racist, national origin restrictions but still restrict low-skill, nonwhite immigration.Footnote 24 Although these states chose different regulations, they have all been relatively successful at restricting low-skill immigration.Footnote 25 Nonetheless, these policies have had different effects on the number of high-skill immigrants entering these countries, something that is beyond the scope of this article. Further, as discussed next, the level of openness will depend on states' openness to trade and the mobility of their firms. Thus, it may not be surprising that these states have different LSIPs even though these policies are driven by the same factors.
Finally, the argument in this article is, like most, an “all else equal” argument. There are many groups in the polity that could affect LSIP: firms, labor, nativists, taxpayers, and immigrants themselves. In classic economic models, native labor dislikes immigrants because low-skill immigrants compete for jobs and push wages down (which is exactly why firms support immigration). Since recent survey data do not support these economic models, some scholars have turned to cultural reasons for opposition to immigration.Footnote 26 Nativists dislike immigrants for the obvious reason: immigrants are different from natives. Other scholars have examined the fiscal costs of immigrants as the source of anti-immigrant sentiment.Footnote 27 Finally, immigrants typically are pro-immigration to protect their position in society and/or to bring in friends and family; however, immigrants tend not to be a powerful group. Until they gain the rights of citizens (and, sometimes, not even then), immigrants can be expelled from the country, which limits their political power. Nonetheless, the relative power of these groups is likely to affect the level of openness to immigration and these groups are likely to have different amounts of power in different countries, which would affect their openness to immigration. In the empirical analysis below, I control for these other factors.
I focus on firms' support for immigration because whereas these other groups likely play a role in LSIP formation, firms are the most powerful group that could be pro-immigration. Given firms' important role in LSIP, I examine how their preferences change because of changes in productivity, trade openness, and firm mobility and how policy-makers respond to these changes. In the United States, a low-skill labor-scarce economy, trade openness, and increases in firm mobility should affect the preferences of firms that are low-skill labor intensive and/or less productive the most, which also are the firms that use the most low-skill immigrant labor. Openness to foreign goods and increasing firm mobility, therefore, may not need to affect the entire economy for them to have an effect on LSIP; instead, they need to affect only low-skill-intensive firms.
Low-Skill Immigration Policy Under Autarky
To contrast the effects of trade openness and firm mobility, I begin by examining support for LSIP by firms under trade protection and limited opportunities to move abroad (autarky). The goal of firms when they interact with policy-makers is to convince policy-makers to pass policies that decrease their costs and increase their profitability. Firms do not necessarily have a preference over which policy policy-makers choose so they are likely indifferent between LSIP that lowers their labor costs and some other policy that lowers other costs or increases profits.
Firms across and within industries differ in their need for low-skill labor; some industries are more capital and/or high-skill-intensive and some firms within an industry are more productive. The terms low-skill labor intensive and low-productivity are used somewhat interchangeably. In economics, these two aspects of the firms are treated as separate and the effect of trade is modeled differently: differences in the factor intensity of production is modeled by the Ricardo-Viner model and productivity differences are modeled by the Melitz and other similar models.Footnote 28 In the real world, firms differ on both dimensions—some firms in low-skill labor-intensive industries are more productive than others and some industries are more capital/ high-skill-intensive than others.
What is important for this analysis is that these two dimensions collapse when one examines preferences on immigration; henceforth, less-productive and low-skill-intensive firms and capital/high-skill and more productive firms will be referred to as low-skill- and high-skill-intensive, respectively. Low-skill-intensive firms benefit from open LSIP more than firms that are high-skill-intensive.Footnote 29 Assuming that each firm's political capital is limited, low-skill-intensive firms should be willing to spend more political capital on LSIP than high-skill-intensive firms. It is not necessarily the case that high-skill-intensive firms do not want more low-skill immigration; it is simply that they prefer to spend their political capital elsewhere, including on high-skill immigration policy. As the proportion of high-skill-intensive firms increases, policy-makers will receive less political capital for the same level of LSIP and given the existence of groups that oppose immigration, one should expect that, all else equal, senators from states with higher high-skill intensity will support more restrictive LSIP. If one allows for oligopoly, increasing immigration conveys an advantage to firms that use more labor because it lowers these firms' costs to a greater extent, allowing them to capture a greater share of the market. High-skill-intensive firms may want to keep their competitive edge by giving contributions for restrictions.
Trade closure, all else equal, should have an effect on the composition of firms in the United States; it should increase the number and/or size of low-skill-intensive firms, as, under the Ricardo-Viner model, domestic production becomes more competitive or, under the Melitz model, less-productive firms are able to stay in business.Footnote 30 Thus, trade restrictions increase the demand for low-skill labor, increase the wages firms pay and, without an increase in the labor force, may erase gains from trade protection.Footnote 31 Further, they increase the wage for low-skill labor in the nontradable or export sector as well. Therefore, one should expect that higher trade barriers should lead to increased support by firms and senators for open LSIP. Similarly, the inability to move production overseas (or low firm mobility) will also increase the production of low-skill labor-intensive goods, as firms have no choice but to produce at home, and increase support by firms and by senators for more open LSIP.Footnote 32
Low-Skill Immigration Policy Under Open Trade and Low Firm Mobility
Under autarky, firm preferences—and their willingness to spend political capital—are driven by their skill intensity. Opening trade does not affect these preferences; instead, it increases the proportion and/or size of firms that are high-skill-intensive. I begin by assuming that trade openness is exogenous to both the firm and the policymaker, which I relax later. The effect of trade openness does not depend on whether the trade is opened to countries with different endowments (Ricardo-Viner model) or the same endowments (Melitz model). Under the Ricardo-Viner model, trade openness affects firms by increasing the price of high-skill-intensive goods and decreasing the price of low-skill-intensive goods. As prices for low-skill-intensive goods decrease, firms that produce these goods (the threatened firms) have to decrease costs or close. Under models of intraindustry trade (the Melitz model), only the most productive firms can export. As the highly productive firms export, they increase the amount of labor they need and bid up the real wages, forcing the least productive firms to close.Footnote 33 Under either model, low-skill-intensive firms face increased competition because of trade openness and are likely to close.Footnote 34 If low-skill-intensive firms close, they will not provide any support for open LSIP and senators should vote for restrictions more often. Further, the closure of some firms may lead others to spend less political capital on LSIP as well. When firms close, they lay off their workers which leads to lower wages, negating the need for as much low-skill immigration, even for nontradable industries. Given that firms have many issues on which they may want to spend their political capital, a decrease in the wage because of other firms' layoffs should make them less likely to spend political capital on LSIP.
Because closing is an undesirable outcome for the firm, firms likely will pursue strategies to stay in business, including increasing their use of high-skill labor. As I have shown, if firms increase their use of high-skill labor, they will be less supportive of open LSIP and senators will vote for restrictions more often. Firms have long understood that increasing their use of technology will decrease their need for low-skill labor and make them more competitive. For example, in his address to the New England Cotton Manufacturers' Association in 1896, Mayor Josiah Quincy of Boston argued that “As other sections, nearer to the sources of the supply of fuel, enabled to command cheaper labor in some respects, coming to the market and taking up manufacturing industries as the South is largely doing, it seems to me that those who are engaged in similar industries in New England must recognize the fact that in order to hold their supremacy they must put more brains, more skill and more education in the carrying on of the manufacturing business.”Footnote 35 In this case, the competition was not only from overseas but also domestic competition from the South, from which it was impossible to gain any trade protection. New England producers heeded this advice and increased the production of high-skill-intensive textile manufacturing; the increased use of skilled labor meant that by 1908, the association could argue that “immigration is, however, no longer as necessary to this country as it was in pioneer times.”Footnote 36
More recently, the Western Growers Association (WGA), which represents farmers in California and Arizona, has made a similar argument in response to foreign competition: “It is time for those anti-immigration reform legislators in Washington DC to realize that the higher the use of technology and innovation, the lower the need for foreign labor.”Footnote 37 Unlike the textile industry, there have been few advances in technology that would decrease the need for agricultural labor and, as I show in the next section, agriculture therefore has continued to lobby for immigration.
Threatened firms may also lobby for increased low-skill immigration or subsidies to stay in business. For example, Tom Nassif, the president of the WGA, argued that “You cannot say we're going to take every illegal alien out of this country without at least factoring in the need for some foreign workers in this country. China is coming at us like a freight train. We can't compete with them.”Footnote 38 Even if firms do not lobby, policy-makers may increase immigration or give subsidies because they do not want threatened firms to close.
The policy-maker will open LSIP or subsidize the firm if the costs of doing so are lower than the costs of losing these firms. On the one hand, if the firms close, the policy-maker loses the tax revenue and jobs it provided along with any political capital. On the other hand, the policy-maker can restrict LSIP, which makes taxpayers and nativists happier, without making the surviving firms worse off because they can hire the laid-off native labor. Additionally, opening LSIP or giving subsidies is not costless; increasing low-skill immigration will increase nativism and the fiscal costs of immigrants and giving subsidies reduces the income that can be spent on other constituents.Footnote 39
At moderate levels of trade openness, it is unclear whether policy-makers will open LSIP, subsidize firms, or allow them to close. However, the amount of subsidies or low-skill immigration needed to keep firms in business increases as trade opens because greater openness reduces prices further; reduces the price of more goods; or allows the export sector to expand, further driving up costs for threatened firms. Further, recent trade negotiations have chipped away at the ability to use subsidies to protect firms, making it less likely that senators can provide firms with subsidies. As trade openness continues, more firms will be allowed to close and senators should vote for restrictions more often.
Low-Skill Immigration Policy Under High Firm Mobility
What happens as firms become more mobile? Firm mobility is affected by several factors: the ability to control overseas agents, the legal ability to move, the fear of expropriation, and whether it is profitable to move, either to gain market access or to exploit lower costs. Regardless of why firms move, the option to move production makes the firm less willing to spend political capital on LSIP. Moreover, once firms have moved, they may support restricting LSIP at home if they move to a state that sends many emigrants to the home state. By restricting LSIP at home, the wage in the country of production will decrease as fewer people emigrate. Firms' support for open LSIP, therefore, should decrease as firm mobility increases, leading to restrictions in LSIP.
To the policy-maker, moving production overseas is the same as if the firm went out of business; both the jobs and the tax revenue associated with the firm are lost.Footnote 40 As above, the policy-maker may want to open LSIP or subsidize these firms through tax breaks. As more firms find moving overseas profitable, however, the policy-maker would have to offer more subsidies to more firms. This is likely to be unsustainable at high levels of firm mobility and senators should be more likely to vote for restricting LSIP.
As well, trade openness and firm mobility may have an interactive effect. Moving overseas is a better option for the owners of threatened firms than closing because they continue to earn profits rather than extracting what capital they can from selling the firm's assets. Under trade openness and firm mobility, one expects more firms that are low-skill-intensive to move overseas. This also decreases support for open LSIP and will lead senators to vote for restrictions more often.
This trend of firms moving in the face of trade competition has even affected agriculture, an industry previously thought to be relatively immobile. For example, the WGA argues that “For years many people have been speaking out about the threat of competition from other countries … What is not being spoken about is the lure of relocating our own operations to foreign countries or face extinction.”Footnote 41 To keep farms in the United States, “it is time for politicians to understand that we are serious when we say the survival of the specialty crop industry in America is at stake. We need a guest worker program. We need specialty crop support in the new Farm Bill.”Footnote 42 Unfortunately, for the threatened farmers of California and Arizona, members of Congress have not granted them more access to labor; a bill that would create an agricultural guest worker program has failed in every congressional session for the past fifteen years. In this case, it seems that it is too costly, politically, to protect these firms.
In the discussion here, trade openness and firm mobility were exogenous forces that affected firms and policy-makers. Yet policy-makers control trade and policies that affect firm mobility, especially capital policy, and firms often lobby over trade and firm mobility policies in hopes of increasing their profits. Opening trade will increase the profits for high-skill-intensive firms by increasing the size of the market they can sell to (assuming that trade is open reciprocally) but it decreases the profits for low-skill-intensive firms. High-skill-intensive firms are likely to lobby for open trade but not on LSIP and low-skill-intensive firms are likely to lobby for closed trade and open LSIP. If policy-makers choose to open trade, it is because they have already privileged the demands of high-skill-intensive over low-skill-intensive firms either because as benevolent social planners, opening trade will be better for the country or as partisan senators, opening trade will be better for their (partisan) constituents. The policy-maker or senator may then try to subsidize the low-skill-intensive firms with low-skill immigration and subsidies, but as I have discussed previously, these are difficult to maintain as trade opens further.
Open trade and capital is a better policy combination for low-skill-intensive firms than open trade alone as they can move production to where they can operate profitably. Under open trade and low firm mobility, low-skill-intensive firms are likely to lobby either for increased low-skill immigration/subsidies to remain profitable at home or lobby for open capital policies so they can leave. Again, policy-makers will weigh the costs of subsidies against the costs of allowing firms to exit the economy.
Finally, one can consider how trade and capital policy respond to LSIP. As I have shown in the case of subsidizing firms with immigration, open LSIP may increase support for open trade because it makes threatened firms more competitive. However, LSIP would have to continue opening as trade openness increased to maintain this support and this is unsustainable at high levels of trade openness. Open immigration may also reduce pressure for open capital, again because threatened firms would be more competitive at home. In contrast, a restrictive LSIP may make open trade harder to achieve, as labor costs remain high and may increase pressure for open capital so that threatened firms can move overseas. Thus, even when policy-makers control trade, firm mobility, and immigration, one expects that if policy-makers choose to restrict trade and keep firms immobile, they open LSIP; if they choose to open and allow firms to be mobile, they will restrict LSIP.
In sum, firm preferences over immigration vary along two dimensions: low-skill labor intensity and the ability to move production overseas. Firms that use a low-skill-intensive production technology will need more low-skill labor than other firms. One expects that low-skill-intensive firms will support low-skill immigration whereas high-skill-intensive firms prefer to spend their limited political capital elsewhere. The second dimension captures the ability of the firm to move production overseas. Firms that are relatively immobile will be more likely to need low-skill labor at home, especially when trade is open and, therefore, support LSIP more than firms that can move production. Once firms move overseas, they will be indifferent to or potentially against low-skill immigration at home. Table 1 summarizes these two dimensions.
Table 1. Firm preferences for immigration along two dimensions
Support for immigration is driven, then, by the low-skill-intensive immobile firms. With trade openness, the size of the low-skill-intensive firms shrinks as firms become high-skill-intensive or close their doors. Similarly, with high firm mobility, firms decrease their support for open LSIP as they have an outside option. At moderate levels of trade openness and firm mobility, however, one thinks that policy-makers may respond to firms' threats to close or to move overseas by offering them incentives, including increased low-skill immigration and tax breaks, to stay open. Yet one expects that when trade is very open and firms are highly mobile, policy-makers will respond to the decreased size of the pro-immigration coalition by voting for restrictions.
Voting on Immigration in the US Senate
There are several obstacles to overcome when testing the argument. First, I discuss how using voting on immigration by the US Congress after 1950 mitigates the endogeneity problem.Footnote 43 Second, I argue that using all votes in the Senate reduces the effects of the strategic roll-call vote-generating process. Third, I discuss how to measure senatorial preferences on immigration and how to overcome the potential spurious correlations between trade, firm mobility, and immigration. Finally, I test the causal mechanisms by examining whether firm lobbying on immigration conforms to the expectations of the argument.
The Sequencing of Trade, Capital, and Low-Skill Immigration Policy After World War II
As I have argued previously, it is less likely that increasing immigration restrictions lead to increasing trade openness. LSIP restrictions and trade openness doubly hurt low-skill-intensive firms by decreasing the price they receive for their goods whereas increasing the price they pay for labor. If anything, one should expect that firms lobby for increased low-skill immigration and increased trade restrictions. Nonetheless, examining voting on immigration in the US Senate mitigates concerns about reverse causality.
In 1934 with the Reciprocal Trade Agreements Act (RTAA) Congress effectively tied its hands on tariff policy.Footnote 44 The RTAA meant that on tariffs, and after 1974 on nontariff barriers (NTBs), the choice for US senators when approving negotiating authority or a treaty—in the first version of the RTAA, congressional approval was not needed—was the status quo rate (which for a long time was the historically high rate set by the Smoot-Hawley Tariff Act in 1930) or a new lower rate. Further, Peril Point legislation and threats not to extend the RTAA were about keeping tariffs at their current rate, not about increasing them as Congress did in Smoot-Hawley.Footnote 45 Throughout this time period, then, a senator could not give increased tariff protection, the outcome one would expect if senators were responding to firm lobbying in the face of increasing LSIP restrictions.Footnote 46
Choosing the status quo was, in fact, a choice for openness in many cases because specific tariffs were inflated away. Irwin estimates that only about 29 percent of the drop in tariffs from 1932 to 1954 was attributable to tariff cuts; the rest was attributable to inflation.Footnote 47 Congress could have corrected the tariffs to adjust for inflation, but this would have violated the reciprocal agreements that the United States had signed.Footnote 48 Nonetheless, it is telling that Congress has not set tariff rates as it did before the RTAA in large omnibus tariff bills. Bailey, Goldstein, and Weingast argue that the durability of the RTAA is attributable to its endogenous effects on the composition of exporters and import-competing firms, which are the effects that my argument is built upon.Footnote 49
It is more likely, however, that restricting LSIP led firms to lobby for policies that would increase firm mobility. Firms that are hurt by increasing trade competition and immigration restrictions can increase their profits by moving production overseas. Yet for most US firms the obstacle for successful off-shoring was not the US government, but foreign governments. Throughout the postwar era, the United States has had few capital controls; therefore, to produce overseas, firms need locations where they legally could invest and there is a low risk of expropriation.Footnote 50 To measure the ability to move capital into other countries, I use two measures: the Quinn-Toyoda and the Chinn-Ito measures of average world capital openness.Footnote 51 Both measure the intensity of capital controls and may also capture the expropriation risk. Von Stein has argued that states sign the International Monetary Fund (IMF) Article VIII, which encourages the openness that these variables measures, as a Good Housekeeping seal of approval signaling their friendliness to foreign investment.Footnote 52
These measures are also exogenous to US senators. During the Bretton Woods era, the United States was leery of interfering with other countries' capital controls because it had helped provoke the Sterling Crisis in 1947 when it pushed Great Britain to reduce its capital controls.Footnote 53 Since the end of Bretton Woods, there has been a push for removal of capital controls through IMF Conditionality Agreements. Although the United States has influence over these agreements, individual senators and the Senate as a whole does not. Further, foreign intervention in other countries' capital policies does not affect all countries in a given year; most countries choose their capital policies for their own reasons, exogenous to the preferences of the United States. Thus, the measures of trade openness and firm mobility during this time period were out of the control of the US Senate.
Neither the passage of the RTAA nor the open US capital policy was linked to immigration. The Democrats passed the RTAA in 1934 because they did not think they could pass unilateral tariff cuts given the economic depression; they were concerned that the tariff reductions would have little effect given other countries' trade barriers; and they wanted a durable tariff reduction.Footnote 54 At the same time, there was little congressional action on immigration: Congress passed only two minor immigration laws from the beginning of Franklin D. Roosevelt's first term to the start of World War II.Footnote 55 In fact, Congress considered and passed few laws, and none of them major, on immigration from the passage of the Quota Act in 1924 until the McCarren-Walter Act in 1952.Footnote 56
US trade and capital policy continued to open after World War II because leaders in the United States, especially Secretary of State Cordell Hull, believed that the trade wars of the Great Depression helped foment the conflicts that led to World War II and that trade openness would provide economic growth as a bulwark against communism.Footnote 57 For instance, Hull argued: “Though realizing that many other factors were involved, I reasoned that, if we could get a freer flow of trade—freer in the sense of fewer discriminations and obstructions—so that one country would not be deadly jealous of another and the living standards of all countries might rise, thereby eliminating the economic dissatisfaction that breeds war, we might have a reasonable chance for lasting peace.”Footnote 58 Open capital became part of the Bretton Woods system, in part to allow US firms to invest in Europe and Japan to help them rebuild.
By contrast, there was little planning either by the US government or the Western powers on migration. Roosevelt and, later, President Harry S. Truman were worried that the large numbers of displaced persons in Europe after the war were targets for communist infiltrators and would lead to an employment crisis, which would also increase support for communism. These fears led to the creation of the Intergovernmental Committee for European Migration (ICEM), the UN High Commission for Refugees, and the Displaced Persons Act.Footnote 59 Tellingly, however, migration was not part of the Bretton Woods agreement; the Displaced Persons Act was temporary and while the United States financially supported the ICEM, it did not take any migrants under its auspices.Footnote 60 Instead, the status quo on immigration was largely maintained in the first major immigration bill after World War II, the McCarren-Walter Act, which passed four years after the GATT was signed. Thus, at the end of World War II, the United States decided to open trade and allow firms to move but to maintain its LSIP.
Roll-Call Votes on Immigration and Agenda Control
Given that studying the US Congress provides empirical tractability, one has to account for the roll-call vote-generating process. As Lee argues, parties not only want to enact their preferred policies but they also want to make political gains for their party.Footnote 61 As such, parties use whatever institutional powers they have to control the agenda, giving votes to only those propositions that will create divisions between, rather than within, the parties, which results in roll-call votes that are more partisan than standard theories of ideology would predict. Agenda control is limited in the Senate because there are fewer restrictions on senators' ability to offer amendments and there are procedural votes—such as the filibuster—that give the minority party greater control over the agenda, which means the effect of party will be overestimated in an analysis of House votes to a greater degree than in Senate votes.Footnote 62 For example, Irwin and Kroszner found that analysis of House roll-call votes on the Smoot-Hawley tariff overestimated the effect of partisanship because of strong agenda control by the ranking members of the House Ways and Means Committee.Footnote 63 In contrast, Senate roll-call votes showed less partisanship and more influence of the senators' constituency.Footnote 64
To examine how senators vote, I collected every roll-call vote on immigration in Vote View.Footnote 65 All votes that affect the number of immigrants in the United States are included, instead of including only final passage or “important” bills as other scholars have done.Footnote 66 Examining only final passage and important votes overestimates the effect of party as these votes gain greater attention from the press, interest groups, and constituents, leading the parties to enforce party discipline.Footnote 67 If one wants to mitigate the selection bias on final passage and important votes by including more votes, it is not clear a priori which votes one should include. First, most of the immigration legislation has taken the form of omnibus bills; therefore, votes on the amendments may affect the final passage of the bill and should be included even if they do not directly affect the number of low-skill immigrants. Additionally, businesses often have preferences over refugee and asylum legislation because refugees become workers. For example, according to congressional lobbying reports, the American Farm Bureau Federation lobbied for the Displaced Persons Act in 1947 and 1948 in hopes of increasing farm labor. Finally, it is important to include procedural votes as they can be used to kill bills. Out of the fifty-nine years examined, then, there were votes in forty-eight of those years, a total of 313 votes on 117 bills and 530 senators voted on those bills. On average there were 6.52 votes on 2.44 bills per year.Footnote 68
The substance of each vote, what the senators were actually voting on, was coded as restrictive or expansive. Votes that seek to restrict immigration are given a score of 0 and votes that seek to open immigration are given a score of 1. Each vote by each senator was given a score of 0 or 1. A zero indicates that the senator voted in the restrictive direction—either by voting for a restrictive bill or voting against an expansive bill. A one indicates that the senator voted in the expansive direction—either by voting for an expansive bill or against a restrictive bill.Footnote 69
Figure 1 shows that both voting on immigration in the US Senate and US LSIP have become relatively more restrictive since the early 1950s.Footnote 70 The United States slightly opened its door to immigrants directly after World War II. Before this period, the United States had restricted immigration to a great degree in the early 1920s and had kept immigration restricted throughout the Great Depression and World War II. With the postwar recovery, there was greater support for immigration by businesses needing workers. The postwar openness continued until the late 1970s and early 1980s as can be seen in the relatively small changes in LSIP in Figure 1. Yet even during this time there was an undercurrent of anti-immigrant sentiment in the Senate, as can be seen in the large plurality of votes (40 to 50 percent) for restrictions. As immigration was restricted in the 1980s and early 1990s, the Senate as a whole was more open to immigration (support was slightly above 50 percent) but even in the Senate, support for immigration was falling. By the late 1990s and early 2000s, the Senate had become, on average, anti-immigration.
Figure 1. United States low-skill immigration policy and voting behavior of the US Senate, 1950–2010
Testing the Argument
I now examine whether trade and firm mobility have affected the policy-makers' preferences. Table 2 provides a test of the argument of this study as well as alternative explanations on the voting behavior of senators from 1950 to 2008. The dependent variable is the year-over-year change in the proportion of votes for open immigration. The year-over-year change is used because of concerns about spurious correlation. Both the measures of trade openness (one minus the tariff level) and average world capital openness generally increase over this time period while, as Figure 1 shows, senators' votes on average became more restrictive, raising concerns of spurious correlation.
Table 2. Voting behavior regressed on trade openness, world capital openness and controls, weighted least squares, and standard errors clustered by congress
Notes: Robust standard errors clustered by congress are in parentheses. Trade openness is 1 minus the ad valorum tariff rate. Firm mobility is the average world capital openness. CI = Chinn and Ito. QT = Quinn and Toyoda. Lagged foreign born is not included in Model 5 and welfare × foreign born and state gdp are not included in Model 6 as the estimates could not be computed because the matrices including those variables are not positive definite. Senators did not vote on immigration every year and not all senators voted, leading to fewer than 5,900 observations.+ p < .10, * p < .05, ** p < .01, *** p < .001.
Each model is a weighted least squares (WLS) regression with robust standard errors clustered by congress. The dependent variable is censored—it could take values from -1 to 1—however, the vast majority of observations lay in the interior of the interval making WLS appropriate.Footnote 71 I weight each observation by the total number of votes to adjust the results for the different number of votes in each year and the results are similar if one uses ordinary least squares (OLS). The standard errors are clustered by congress because votes are rarely independent within congresses.
Models 1 and 2 are the base models. Both models include the change in the percentage of salaried workers as a fraction of all workers in manufacturing at the state level to measure skill intensity.Footnote 72 Goldin and Katz have found that the percentage of salaried workers in an industry is correlated with the average years of education of blue-collar workers in that industry.Footnote 73 Thus, producers who use more salaried workers should be less likely to demand the relatively low-skilled immigrant.Footnote 74 Next, I include the change in trade openness, measured as one minus the ad valorem tariff rate, and firm mobility, measured using the average world capital mobility from Quinn and Toyoda and Chinn and Ito.Footnote 75 In the above discussion of firm preferences, I argued that there was an interactive effect of trade and firm mobility; therefore, I also include an interaction term between trade openness and the two measures of firm mobility.
In Models 3 and 4, I include an interaction term of trade policy and firm mobility with party to examine the effect of preferences over taxation on immigration. As I have argued previously, at moderate levels of trade openness and firm mobility, policymakers may be willing to subsidize firms to keep them at home through increased immigration or tax breaks. As Republicans typically are viewed as more willing to cut taxes, one might expect that there would be a different effect of trade openness and firm mobility on Republicans than on Democrats.
Models 1 and 2 show support for the argument that senators' voting behavior is affected by changes in firm preferences because of changes in trade openness. Comparing Models 1 and 2 with Models 3 and 4, I show that the effect of firm mobility is conditioned by party. To help interpret the effect of the coefficients, Figure 2 plots the predicted change in voting behavior on the change in trade policy and firm mobility for both the Democrats and Republicans.Footnote 76 In all models, the effect of trade openness is negative and significant—increased trade openness leads to more votes for immigration restrictions—and this effect is somewhat more pronounced for Republicans than Democrats. Additionally, the trade effect seems to be stronger than the firm mobility effect. This is not surprising given estimates of the job losses because of trade have been larger than those lost to off-shoring, which suggests that the effect on overall firm demand for LSIP should be greater for trade than firm mobility.Footnote 77 Model 3 shows that there is a negative and statistically significant effect of increasing firm mobility on the voting behavior of Democrats (the excluded category). In contrast, there is little effect of increasing firm mobility on the voting behavior of Republicans. From these two coefficients, it appears that Democrats are more willing to let firms exit in the face of rising firm mobility whereas Republicans are more willing to subsidize firms through tax breaks.
Figure 2. Predicted effects of the change in trade policy and firm mobility
The results in Models 2 and 4 on firm mobility are slightly different than in Models 1 and 3. Trade openness still has a negative and statistically significant effect on voting for immigration openness for members of both parties. Firm mobility on its own has little effect; for Democrats, there is a slightly negative effect of firm mobility on immigration. For Republicans there is a slightly positive effect of firm mobility on immigration, which shows their willingness to subsidize some firms—based on what was voted on during this time period these were agricultural firms—with immigrant workers to keep them in business.Footnote 78
The differences between Models 1 and 3 and Models 2 and 4 are likely driven by the time periods that they cover. Earlier in the postwar era, many countries had relatively closed capital. The likelihood that firms would move overseas was less of a concern than it became in the late twentieth and early twenty-first centuries. Because firm mobility became a larger issue as the twentieth century went on, it is not surprising that it had more of an effect. Nonetheless, in both models, the effect of trade dominates the effect of firm mobility, leading to greater immigration restrictions.
There is little effect of increasing use of technology as measured by percentage of salaried workers in any of the models.Footnote 79 Technology was often adopted in response to trade openness during this time period; therefore, the effect of increasing technology may be captured by the coefficient on trade openness. This result is also somewhat counterintuitive given the public opinion literature on immigration that has found high-skill individuals are more supportive of LSIP.Footnote 80
Importantly, there is no independent effect of party in either model. Although Republicans and Democrats react to changes in trade policy and firm mobility differently, they do not have different preferences simply based on ideology. This result differs from previous research that found an independent effect of party.Footnote 81 That research, however, examines only important votes. As discussed above, there is greater partisanship on important votes than on all votes, which leads to an overestimation of partisanship. Moreover, neither of these studies interacted party with trade policy or firm mobility, which might be driving their results.
Models 5 and 6 account for other alternative explanations from the immigration literature. I use multiple imputation on the missing variables where necessary using Amelia II.Footnote 82 These models show that the implications of the argument hold even when accounting for other factors; the coefficients on trade openness and firm mobility are stable to the inclusion of these variables.
To measure the power of unions and low-skill workers, I include the change in percentage of the labor force in each US state represented by a union.Footnote 83 In contrast to those who argue that unions have had an important effect on LSIP, there is no effect of changing unionization rates.Footnote 84 The statistical insignificance of unions is not surprising given the rise of public-sector unions and the change in the composition of union membership. Public-sector employees often do not compete with immigrants and may serve low-skill immigrants; open LSIP, therefore, may lead to increased work in the public sector. Additionally, unions in the private sector have increasingly organized low-skill immigrants; therefore, these unions, such as the Service Employees International Union, may act as immigrant rights organizations. Thus, some unions may oppose immigration whereas others favor immigration, leading to null results.
Next, I include the change in unemployment in each US state as a second test of the political power of workers.Footnote 85 If workers are powerful, then senators' support for immigration should vary inversely with unemployment. Model 6 reveals a statistically significant effect of unemployment, but it is the opposite direction of what scholars would hypothesize and is not robust to the model specification in Model 5. Therefore, it is less likely that workers—whether unionized or not—affect voting on immigration.
To examine the power of immigrant groups and nativism, I include the change in the percentage of foreign-born in each US state as a rough test of nativism, as in Goldin and Timmer and Williamson.Footnote 86 As the percentage of foreign-born in a state increases, there are more opportunities for the foreign-born to interact with the native-born, which may lead to a nativist reaction. Support for immigration should vary inversely with the change in the percent foreign-born if the nativist hypothesis is correct. The change in lagged foreign-born, in contrast, is used to test the power of immigrants argument.Footnote 87 If immigrants form powerful lobbies, as the number of immigrants who can vote (that is, have been here long enough to have citizenship) increases, so too should support for immigration by the state's senators. The models show, however, that neither the change in percentage of foreign-born or the lagged change have an effect on support for immigration by the senators, giving one less confidence that immigrant groups or nativist have an effect on policymakers.Footnote 88 I have also included the estimated number of undocumented immigrants and similarly found little effect.Footnote 89
I include the change in welfare spending per capita in each US state and interact it with the change in foreign-born to capture the costs of immigrants.Footnote 90 Senators from states facing increased costs of fiscal immigration should support open immigration less.Footnote 91 Again, there is no effect of the size of the welfare system on votes on immigration, showing less support for this hypothesis. As a robustness check, I have also included the change in state spending as a percentage of all government spending, as a measure of burden sharing.Footnote 92 When burden sharing is higher, taxpayers within each state should be less concerned about low-skill immigration. I find similar, nonsignificant results. Finally, I include the change in state GDP to control for economic effects of the overall economy and find little effect.
These results are robust to many different specifications of the model. First, because of concerns that lobbying by high-skill firms may affect the passage of bills that include language on high-skill immigration or omnibus bills, I drop the votes on those bills and the results are similar.Footnote 93 I examined characteristics of the senators, such as their race/ethnicity, and tenure in the Senate and found little effect.Footnote 94 Instead of interacting trade openness and firm mobility with party, I interacted them with Republicans and southern Democrats as one party, since southern Democrats held similar views to Republicans and, after the passage of civil rights legislation, many southern Democrats switched parties; tax preferences of senators, and the first dimension of the DW Nominate scores, which measure how liberal or conservative a senator is.Footnote 95 The results are quite stable across these different specifications. I also repeat the above analysis with standard errors clustered by state instead of by congress as one might believe that senators from a single state do not vote independently and the results are similar.Footnote 96 Further, instead of examining the year-over-year change in support for open immigration to control for potential spuriousness of the relationship, I repeat the above analysis using a panel logit model on each vote.Footnote 97 When one clusters the standard errors by state, the results are substantially and statistically similar to the results in the year-over-year change regressions. Finally, I similarly examined the relationship using a subset of only final passage votes and a subset of all votes on important immigration bills and found similar results.Footnote 98 Thus, one can have greater confidence in the results.
In sum, the data support the argument that the increasing integration of world markets has an unintended effect on support for open LSIP. Increasing trade openness leads to less support for open LSIP. In most cases, increasing firms mobility also leads policy-makers to support restricting LSIP. Yet sometimes policy-makers, especially Republicans, have been willing to subsidize firms, especially in agriculture, with increased low-skill immigration to keep them in the country. Finally, the data provide little support for alternative explanations; voting behavior does not seem to be driven by partisan differences, the influence of unions, nativist or the foreign-born, the fiscal costs of immigrants, or the state of the state's economy.
Evidence of Firm Lobbying
Although the data I have presented so far show a robust relationship between trade openness and firm mobility and voting on LSIP, I have not shown directly that this relationship is caused by changes in firm lobbying. In this section, I provide evidence that firms are an important lobbying group and that industries that have been affected by trade openness and firm mobility lobby less on LSIP than industries that have been less affected. Data on firm lobbying are available from 1998 to 2011 from the Center for Responsive Politics.Footnote 99 The data are on lobbying, not on campaign contributions, but they should still provide a sense of how firms have tried to influence LSIP.Footnote 100 The filings include data on what issues the firms or organizations lobbied on, but not if they lobbied for or against, and the total amount they spent on all lobbying, but not how much was spent on any issue. The lobbying data show that on average about 579 firms or business associations and about 269 nonbusiness groups lobbied on immigration each year. Thus, firms are an important lobbying group in the LSIP debate.
Although firms are important, not all industries lobby on LSIP equally. A measure of intensity of lobbying on immigration examines what percent of issues that firms/organizations in an industry lobby on are on LSIP.Footnote 101 Immigration is not the only issue that firms lobby on; firms that lobby on immigration also lobby on average on 7.15 other issues in a given year. Even firms lobbying on agricultural or seasonal low-skill immigration lobby on average on 7.63 other issues a year.
Lobbying on immigration conforms to the expectation of the argument; firms that have less trade protection and are more mobile lobby less on LSIP than firms with more trade protection and/or firms that are immobile.Footnote 102 If one examines a difference of means, 14 percent of issues lobbied on by nontradable/immobile firms are on LSIP compared with only 4 percent of issues lobbied on by tradable/mobile firms (p < 0.000).Footnote 103 Similarly, the correlation coefficient between the percent of issues on immigration and the tariff level lagged five years is 0.34 (p < 0.02).Footnote 104 As tariffs decrease, firms are less likely to lobby on LSIP. Similarly, the correlation coefficient between the percent of issues on LSIP and log direct investment lagged one year is smaller, −0.23 (p < 0.007), but still in the hypothesized direction, and if one lags log direct investment two years, it is -0.25 (p < 0.003).Footnote 105 As firms increase foreign direct investment they are also less likely to lobby.
Additionally, examining which industries do not lobby on immigration provides confirmatory evidence for the argument. Sectors that did not lobby on LSIP, but did lobby on other issues during this time period include the retail sector, hospitality and restaurants, wholesale, earthenware and glass, wood products, sugar, tobacco, beverages, paper, and sundries. Yet one has reason to believe that some of these industries lobbied on immigration in the past, as many of these industries were once top employers of immigrant labor. Using data from the Integrated Public Use Microdata Series, Table 3 shows the top ten immigrant-employing industries by percentage of immigrants employed for every other census year from 1900 to 2010.Footnote 106 Several of the top employers in the early twentieth century did not lobby on immigration at all in 1998 to 2005, including beverages, sugar, earthenware and glass, and sundries, and many of these industries no longer employ a large percentage of immigrants. The data show that this decrease in immigrant employment is correlated with the tariff level (correlation coefficient 0.54, p < 0.000) and the amount of FDI the industry engages in (-0.11, p < 0.08 with total direct investment; -0.17 (p < 0.01) with percent of total US FDI). Increases in trade openness and firm mobility for these industries have led them to employ fewer immigrant workers, and in many cases fewer workers as a percent of total US employment, likely leading these firms to spend their lobbying dollars on issues besides immigration.
Table 3. Top immigrant employers 1900–2010
Conclusion
After World War II, the United States sought to rebuild the nineteenth-century liberal international order in trade and capital; yet, little attention was paid to immigration policy. Instead, immigration policy was treated as domestic policy and low-skill immigration was increasingly restricted over the past sixty years. In this article, I argue that it was, in fact, the building of the open trade and capital regime that led to the restrictions in low-skill immigration policy.
As an unintended consequence of increased globalization of trade and capital, firm preferences over low-skill immigration changed. Trade openness subjected firms, especially low-skill-intensive and less-productive firms that use immigrant labor, to increased competition. As a result, some firms increased their capital intensity, high-skill intensity, or productivity and some firms closed their doors. Either way, these firms no longer provided political capital for open immigration for low-skill workers. The increase in firm mobility compounded this effect; firms hurt by trade competition or seeking market access moved to where production costs were lower. Again, these firms also stopped supporting open immigration for low-skill immigrants. With less support for low-skill immigration, policy-makers could please other constituencies by restricting it.
As theorized, voting behavior in the US Senate supports the argument; changes in voting behavior on immigration were driven by changes in trade openness and firm mobility. Immigration does not simply fall along the left-right dimension. Nativism, measured as the percent foreign-born, does not affect the way that senators vote and neither does a constituency with a large percentage of potential immigrant voters. There is little evidence unions or welfare spending in the state affect voting either. Moreover, data on industry lobbying on low-skill immigration provide confirmatory evidence of the causal arguments; industries decrease their lobbying on low-skill immigration when their trade protection is reduced or when they engage in more direct investment overseas.
Although this article examines the United States, the argument may apply to other low-skill labor-scarce states. Many of these states have also lowered their trade barriers and capital controls while simultaneously restricting low-skill immigration; however, unlike the United States, these states have set up explicit barriers to low-skill immigration through point systems or other skill restrictions. It is for future research to see if this argument explains other countries' immigration policies as well.Footnote 107
This article sheds light on both immigration policy and on IPE. First, it is one of the few studies quantitatively testing different theories of immigration policy against each other. From the results, few of the existing theories seem to have much explanatory power. Instead, changing firm preferences seem to be a major driver of immigration policy.
The immigration literature, like many, has also assumed that firm preferences are static; I have shown that they can change with different production technology, trade openness, and the ability to move production overseas. This line of research could also help increase scholars' understanding of other domestic policies that affect firms including labor, welfare, and environmental policy by examining how firms endogenously choose where to produce and, therefore, where they lobby. Perhaps, for instance, environmental policy in wealthy countries follows a similar trajectory—it may be easier for policy-makers to pass more restrictive, pro-environment policies as “dirty” firms adopt cleaner technology, close because of increased competitive pressures, or move their production overseas. However, this same process may mean that it is harder for developing countries to pass environmental laws, as the dirty firms that locate in those states are likely to lobby against these laws.
Further, this article increases scholarly understanding of IPE and globalization by bringing immigration back into the picture. Although IPE scholars have studied the movement of goods and money, they have largely ignored the third aspect of international flows, the movement of people.Footnote 108 This article brings immigration back into focus by examining international, and not simply domestic, explanations.
Finally, this article increases scholars' understanding of policy formation by refocusing attention on the interaction of trade, capital, and migration policy. Economists have long argued that the movement of goods, people, and money are substitutes; yet political analysts have ignored this argument when they have studied immigration policy. Instead, scholars have focused on domestic factors and have therefore missed the role that trade and firm mobility have played in the formation of immigration policy. This article brought the substitutability of these international economic policies back into focus by arguing that trade and firm mobility affect firm preferences over immigration and affect the way that policy-makers respond to firms. Moreover, this article shows the importance of examining policy substitutes. How policy-makers create immigration policy cannot be understood without examining trade and firm mobility. Similarly, these other policy areas should be examined in light of immigration policy and each other as well.
Supplementary material
Online appendices are available at http://dx.doi.org/10.1017/S0020818314000150.
