Introduction
Countries have incrementally employed antidumping trade remedies more frequently throughout the post-war period, with countries employing measures twice as often in present times than they did in 1995. Footnote 1 Previous literature has documented this trend. Footnote 2 Balassa (Reference Balassa1978) commented on this phenomenon, referring to it as the spread of a “new protectionism” encompassing the rapid proliferation of antidumping duties and other non-tariff measures with trade-stifling effects. Footnote 3 Ethier and Fischer (Reference Ethier and Fischer1987) similarly wrote that this trend differs substantially from traditional protectionism in that these measures involve “applications of a limited number of well-defined statutes (notably the antidumping and countervailing-duty laws, safeguard provisions, and unfair trade practices acts)” rather than adding to aggregate national levels of protectionism writ large. Footnote 4 Moore and Zanardi (Reference Moore and Zanardi2011) point to the possibility of substitution effects explaining these trends, with reductions in traditional trade barriers being sustained by the introduction of antidumping trade remedies following multilateral trade agreement rounds or preferential trade agreements. Footnote 5
What factors explain the upsurge in non-traditional trade barriers over the past few decades? Though the dissemination of antidumping laws provides an interesting puzzle for scholars, it may hold implications for trade policy. Antidumping policy has been the subject of debate among trade economists and political scientists. Indeed, antidumping laws have often been argued to promote fair competition and protect domestic economies from unfair trade practices (e.g. dumping products at “less than fair value”) and international market volatility. Footnote 6 Jacob Viner himself argued in 1923 that antidumping laws were a necessity to prevent real economic injuries arising from dumping. Footnote 7 However, other scholars have linked these policies to trade dampening and chilling effects, which carry economic consequences as a corollary. Footnote 8 Many of these scholars have linked antidumping laws to attempts by rent-seeking firms to cartelize industries and engage in monopolistic pricing, resulting in job losses, hobbled downstream production and, ultimately, higher prices charged for consumer goods. Footnote 9 Thus, when abused, antidumping laws may potentially inflict economic costs and losses on economies. From a policy and a scholarly standpoint, it is therefore important to understand what incentivizes firms to utilize these policies.
Previous literature identifies a number of theoretical mechanisms within the data-generating process, including real exchange rates, business cycles, and retaliatory dynamics. I offer an alternative take on this literature, focusing on the role of foreign investment and the demand for antidumping protectionism. I propose that countries that are economically integrated into global markets at national levels are less likely to receive requests for antidumping trade remedies. Through a number of processes, I argue that globalized countries that receive and deploy higher volumes of trade and foreign direct investment (FDI) should be less prone to receive petitions for antidumping investigations. Higher volumes of trade and investment over time should signal the emergence of transnational corporations in domestic economies that are competitive internationally and are thus less likely to require additional antidumping trade remedies and protective measures, ceteris paribus.
This paper contributes to the growing literature within international political economy that studies the relationship between global value chains (GVCs) and trade policy preferences. Footnote 10 The findings indicate that countries with higher degrees of economic integration (in terms of de facto trade and investment flows) experience lower likelihood of observing filed petitions for antidumping trade remedies. The results suggest that the structure of production throughout the global political economy affects trade policy preferences among economic actors. This paper thus simultaneously contributes to the antidumping literature by incorporating systemic variables capturing the structure of the global economic structure to analyze prevailing trends in demand for antidumping trade remedies by economic actors. While economic integration appears to be linked with fewer national petition filings, it is plausible to expect a positive feedback loop between antidumping and globalization, where backlashes against globalization may induce enhanced usage of antidumping procedures for protectionist purposes, causing further retreats from economic integration. The structure of this paper proceeds as follows. I first review the existing literature and data on the topic, examining the empirical puzzles within previous work. I then turn to an exposition of my principal theoretical argument before proceeding to my research design and findings.
Antidumping: Why?
While traditional trade barriers have been on the decline in recent decades, antidumping duties have simultaneously been on the rise during the same period of time. Under the World Trade Organization’s guidelines, antidumping trade remedies are to be reserved for use in the case that imports have been sold below normal market value and that they have caused material injury to petitioning firms. Footnote 11 One unique characteristic differentiating these types of trade barriers from traditional barriers to trade is that they are implemented by procedures carried out by civil servants in accordance with national statutes. Footnote 12 For example, national petition filings for antidumping investigations in the United States are generally filed with the Department of Commerce and the International Trade Commission (ITC). Footnote 13 Companies that claim injury due to unfair competition and pricing practices (e.g. dumping) may file a petition with the Commerce Department and the ITC. If the plaintiff—the petitioner—has legal standing, then the Commerce Department conducts an investigation for evidence that dumping has occurred. Footnote 14 If Commerce determines that dumping has occurred, the department calculates the “dumping margin,” which is comprised of the difference between the export price of the good and the normal value divided by the export price of the good. Footnote 15 The ITC then investigates to determine whether any “material injuries” were sustained by the plaintiff as a result of dumping. If the Commission finds that the plaintiff sustained injuries (alongside an affirmative ruling by Commerce), then the Commerce Department will issue an antidumping trade remedy to offset the damages incurred by dumping (or countervailing duties in the case of foreign subsidies). Footnote 16
Previous literature has investigated the proliferation of antidumping trade remedies around the globe. One prominent explanation for these trends revolves around business cycles and macroeconomic fluctuations. According to this literature, these factors—with attention given to the effects of unemployment, GDP Growth, and import growth—stimulate the demand for protectionism. Footnote 17 Recessions and economic downturns are said to be characterized by higher demands for protection, whereas import growth should signal higher tides of foreign competition, generating incentives for domestic firms and producers to seek insulation from the global economy. Footnote 18 Under these conditions, it is also easier for parties to demonstrate that unfair trade practices have been employed and that material injuries have been sustained as a result. However, these arguments encounter several theoretical and empirical issues. Conceptually, these arguments do not provide an adequate explanation as to why firms would only demand protectionism during recessions and economic downturns when they generally stand to benefit from these policies. Footnote 19 Empirically, the data supporting these arguments is mixed. Footnote 20 Despite their invaluable theoretical and empirical contributions, business cycles and macroeconomic indicators do not fully explain these trends.
Another prominent argument emphasizes retaliatory, “beggar thy neighbor” trade motives throughout the global political economy. In brief, this literature argues that retaliatory motives influence states to adopt antidumping measures against each other. Footnote 21 Feinberg and Reynolds (Reference Feinberg and Reynolds2006) for example argued that retaliation provided a “significant motive” that has driven the trend toward more frequent utilization of antidumping processes. Footnote 22 The theoretical logical and empirical evidence marshalled in favor of these arguments is indeed compelling. However, it alone cannot explain this phenomenon. Firms in Country B may petition to initiate antidumping investigations against Country A’s imposed antidumping measures, but this does not explain the initial choice of Country A to impose them—nor does it explain why firms in Country A sought to file petitions initially. Conceptually speaking, the “retaliatory motives” thesis omits the first half of the theoretical process and instead focuses solely on the latter—countries retaliating against the measures imposed on other countries. While retaliatory motives are theoretically significant and empirically salient, they do not explain this initial starting point and thus alone do not explain this phenomenon.
Other literature focuses on the role of exchange rates and currency appreciations in generating demand for protectionism. Footnote 23 The exchange rate hypothesis postulates that industries demand protectionism in accordance with appreciations in currencies and real exchange rates. Thus, exchange rate movements alter the quantity and type of firms that demand protection. Footnote 24 As Country A’s currency appreciates and becomes stronger relative to other currencies, firms in Country A have stronger incentives to file petitions for antidumping investigations. These incentives are produced because appreciations in a nation’s currency render its exports more expensive, relative to other competing products on world markets. Firms with products competitive at equilibrium exchange market levels are harmed by appreciations because their products are priced out of foreign markets. Firms further incur harm as foreign imports from abroad generate intense domestic competition. Firms harmed by currency appreciations should therefore be more likely to demand protectionism and file for antidumping petitions, ceteris paribus.
This argument has generally received strong support from previous research. Specifically, work elsewhere in the political economy literature demonstrates this empirical pattern with other types of protectionist trade policy. Appreciations in the U.S. dollar in the 1980s were met with increased protectionist pressures as firms demanded insulation from global market forces, with the fall of the dollar’s value beginning in 1985 relieving some of these protectionist pressures. Footnote 25 Similarly, American firms in the 19th century demanded higher tariff rates as the dollar went through a stringent period of appreciation. Footnote 26 Real exchange rate arguments are therefore useful in explaining why countries adopt antidumping—and protectionism as a whole.
However, recent research reveals some unexplained variation among appreciations in real exchange rates and demand for antidumping protectionism. Footnote 27 While exchange rates during the 1990s in North America and Europe were relatively stable, the quantity of antidumping petitions initiated by firms varied significantly during the period. In contrast, demand for antidumping petitions in East Asian countries remained flat in the 1990s despite comparatively higher levels of exchange rate appreciation. Similar trends were observed in South Asia and the Middle East. Despite having rigorous theoretical logic and robust empirical support, exchange rate fluctuations do not entirely explain these remaining empirical puzzles.
While each of these theoretical literatures provide invaluable contributions to the literature, they alone cannot account for the propagation of antidumping protectionism. As such, it is probable that other additional forces influence this phenomenon. One such factor is economic globalization itself. While economic globalization can be conceptualized at different levels of analysis, Footnote 28 it is frequently analyzed at the country, or national, level. Harris (Reference Harris1993) viewed globalization at the national level as the “increasing internationalization of the production, distribution, and marketing of goods and services”. Footnote 29 Akhter (Reference Akhter2004) broadly defined national economic globalization as a process that “results in increasing integration of a country’s economy with the rest of the world,” with countries gradually integrating into international market processes via FDI and international trade flows. Footnote 30 Much of the literature generally accepts that national economic globalization is a multifaceted process of integration involving interconnectedness in terms of FDI flows and international trade. Footnote 31 National economic globalization, therefore, can be understood as a multidimensional process of “global economic integration” (GEI) comprised of international trade and FDI. Footnote 32
This definition pairs well with contemporary understandings of the modern global economy. A large share of world trade, production, and investment continues to be structured around intrafirm trade and GVCs as countries have gradually integrated within global markets at national levels, steadily embracing de facto flows of trade and foreign investment at national levels. Footnote 33 For example, the U.N. estimated in 1996 that roughly one-third of world trade was conducted within transnational corporations. Footnote 34 The Organization for Economic Co-operation and Development additionally found in 2009 that intrafirm trade accounted for 48 percent of U.S. imports and roughly 30 percent of U.S. exports. Footnote 35 Much of this intrafirm trade consists primarily of intermediate goods within different nodes of GVCs. Footnote 36 A large portion of services trade—about 75 percent—is comprised of intermediate inputs. Footnote 37 De facto trade and capital flows continue to strongly indicate a country’s aggregate level of economic integration, and are in part driven by large, domestic firms that operate internationally.
While speculation has occurred as to how GEI has contributed to demand for antidumping trade remedies, Footnote 38 the growing body of literature has provided comparatively fewer theoretical accounts linking these two phenomena. However, economic integration is clearly a significant mechanism driving outcomes in world politics. Literature elsewhere throughout international relations research emphasizes national economic globalization as a significant force shaping international politics. For example, GEI plays a central role in the development literature, Footnote 39 and is often implicated as a significant force in controversies over domestic regulatory policies and globalization. Footnote 40 Additionally, many of the non-state actors associated with national economic integration—transnational corporations—are generally viewed as prominent examples of non-state actors that exert considerable influence in world politics, along with NGOs and other non-state groups. Footnote 41 Much of the literature elsewhere in international relations scholarship, therefore, emphasizes national economic integration as a crucial mechanism shaping domestic and international political phenomena.
It is likely that de facto national economic integration contributes to the political economy of antidumping protectionism. Excluding these dynamics may therefore misspecify theoretical and empirical models of demand for antidumping protectionism. In this paper, I attempt to integrate these literatures, proposing that countries that are nationally integrated into global markets should experience less demand for protectionism. Specifically, higher levels of trade and financial integration—as indicated by trade and investment flows—should reduce the likelihood of antidumping petitions being filed, ceteris paribus.
I offer a series of theoretical mechanisms that may plausible explain this relationship. Higher levels of de facto trade integration may indicate that transnational corporations are becoming sufficiently “internationalized” over time, becoming more competitive in global markets through outsourcing and fragmenting production, which reduces demand for protectionist shielding. Furthermore, as firms become internationalized, they subsequently become fearful of retaliatory antidumping petitions, and refrain from initiating them. On the other hand, higher levels of financial integration may reflect transnational corporations attempting to internalize production and protect trade secrets. This helps mitigate knowledge and technology spillovers and reduces the emergence of foreign competitors, which dampens demand for antidumping protection. Additionally, financial flows may reflect the deployment of “quid pro quo” FDI to defuse potential antidumping duties by producing the final product within a target country. Thus, countries with higher levels of trade and financial flows should receive fewer petitions for antidumping protectionism overall, ceteris paribus. I expound on these theoretical mechanisms below.
Trade integration and antidumping
Historically, antidumping policy has served as a policy tool to protect domestic producers from unfair trade practices (“dumping”) by foreign producers. Footnote 42 Dumping can take a number of forms. Firms may attempt to expand market shares by discriminating against domestic prices, pricing goods at smaller margins than production costs. Footnote 43 Exporting firms may also engage in “cyclical dumping,” or the process of exporting goods produced in excess capacity at unusually low prices. Footnote 44 It is often superficially difficult to determine whether firms are engaging in price predation or simply competitively pricing products to compete with rival firms. Footnote 45 Regardless, cheap imports resulting from these trade practices undercut the prices of goods sold by domestic firms to consumers, which cuts into their profits and economic livelihood. This may serve to motivate domestic firms to seek out assistance from government channels.
Antidumping trade remedies provide a means of recourse for injured domestic firms to receive protection from severe import competition. These policies allow for firms to receive vital shielding, protecting the competitiveness of domestic producers and offering consumers alternatives to imported goods. Firms will typically seek out these policy devices primarily when the expected benefits outweigh the costs of filing petitions with bureaucratic apparatus. Footnote 46 Domestic firms struggling to compete with foreign producers may thus demand antidumping trade remedies from their respective governments by filing petitions to request an antidumping investigation and a subsequent trade remedy to be imposed.
Over time, however, domestic firms operating within economically integrated countries may become more competitive producers within the global economy, which reduces their need to acquire additional shielding from excess economic volatility and import competition through antidumping trade remedies. Footnote 47 As discussed above, one primary component of GEI is trade integration, which is in part driven by domestic firms that are large, multinational firms operating within integrated countries. Footnote 48 These firms may offshore production to exploit comparative advantages, deriving efficiency gains and cost advantages from doing so. Previous literature indicates that offshoring and outsourcing provide substantial economic boons and cost advantages to firms. Footnote 49 Offshoring allows for firms to realize expanded economies of scale, allowing for lower per-unit production costs and more efficient overall production, which results in cost savings for offshoring firms and lower prices for consumers. Footnote 50
Outsourcing to exploit regional comparative advantages thus helps transnational corporations become more competitive in global markets, which reduces the necessity for these firms to file petitions for antidumping trade remedies to maintain profits and competitiveness. As these firms become more competitive internationally, nationally integrated economies will be less likely to receive petitions for antidumping investigations. This relationship should be reflected via a country’s level of trade; as primary sources of intrafirm trade, transnational corporations tend to import intermediate goods to utilize within the production processes of a final good. As countries begin the process of integrating with global markets, previous literature indicates that trade could signal significant sources of import competition, which may motivate small domestic firms to file petitions for antidumping trade remedies. Footnote 51 However, as firms become internationalized in global markets, trade integration should instead signify the presence of multinational firms that are more competitive internationally. This logic seems to receive support from recent empirical research. Blanchard et al. (Reference Blanchard, Bown and Johnson2016) and Meckling and Hughes (Reference Meckling and Hughes2017) both find that global companies engaged in outsourcing and offshoring preferred open trade. Bown, Erbahar, and Zanardi (Reference Bown, Erbahar and Zanardi2021) found that domestic value-added growth (DVA) embedded within foreign production networks significantly influenced the probability of removing imposed antidumping duties. Because these actors are more competitive owing to exploitation of comparative advantages, transnational companies should be less likely to file petitions for antidumping investigations, which should be reflected in aggregate trade and investment flows.
Additionally, transnational corporations headquartered in trade-integrated countries may refrain from initiating investigations for antidumping trade remedies out of concern for motivating retaliatory antidumping duties against final consumer goods. In contrast to small domestic firms, transnational firms have heavier ties to global supply chains, and generally prefer fewer barriers to trade. Footnote 52 International companies are more likely to produce a variety of final products for export in global consumer markets, which allow for these firms to reap profits across product lines. However, the prospect of antidumping measures being imposed on these products introduces costly risks that globalized firms must account for within their strategic calculus, as they may cut into realized profits if measures are imposed. The linkages that transnational corporations share with the global economy may therefore produce incentive for these firms to refrain from filing for antidumping investigations on other potential products in order to minimize these costly risks. This lowers the costs of doing business by reducing the opportunities for governments to impose additional barriers to international trade, which maximizes economic gains for these companies.
This problem is compounded by the fact that transnational corporations frequently utilize intermediate inputs in the production of final consumer goods within GVCs. International trade is in part constituted by intrafirm trade, and internationalized companies drive a significant portion of this trade. Footnote 53 These international economic processes introduce more potential targets for retaliatory antidumping measures by other countries. These measures may cut into the costs of manufacturing final products for export in consumer markets, which may cause further economic injury to globalized firms. Footnote 54 It is plausible, therefore, that transnational firms may refrain from participating in the antidumping process in order to reduce the likelihood that retaliatory measures may be imposed on intermediate products as well as final consumer goods.
Previous literature lends some credence to these propositions. While previous literature investigated the possibility of cross-national deterrence against retaliatory antidumping enforcement, Footnote 55 recent literature suggests that transnational corporations linked to global markets may strategically act to avoid retaliatory antidumping filings. Footnote 56 Meckling and Hughes (2017) further suggest that transnational corporations operating in the solar photovoltaic industry may prefer open trade due to fears or threats of retaliation along other product lines, which could damage economic profits (232). Avsar (Reference Avsar2013) similarly found that AD activities in Brazil lead Brazilian exporting firms to increase export prices for the products of named industries to decrease dumping margins and avoid threats of retaliatory measures by other countries. There is reason to believe, therefore, that firms include the possibility of retaliatory antidumping measures within their strategic calculus. With an expanded variety of final products to export to global consumer markets, the risks of retaliatory measures are higher for internationalized firms, which may decrease the likelihood that these firms file additional petitions for antidumping trade remedy investigations.
A brief case comparison could aid in illustrating this logic. Figure 1 depicts the number of antidumping petitions filed by firms within the United States and South Korea between 1978 and 2015. I select these examples as they are representative cases within the sample that allow me to explore the logic of my theoretical argument. Footnote 57 The United States and South Korea experienced differing levels of aggregate demand for antidumping trade remedies during the sample period. Starting with the United States, it is clear that between 1978 and the early 1990s—the hay day of deindustrialization—U.S. firms filed more petitions for antidumping investigations than in recent times, with a high of ninety-four petitions filed during 1992. Footnote 58 During the early 1980s, the U.S. economy experienced a steep recession, which was magnified by fierce import competition from Japanese manufacturers. Many large factory closings in the United States made headlines in the 1980s, including Firestone, Ford, Chrysler, Pabst, and the United States Steel Corporation. Footnote 59 Offshoring and outsourcing may have contributed to these trends via knowledge and technology spillovers, allowing foreign competitors to “leapfrog” up the industrialization ladder and emerge in global markets. Over time, however, United States demand for antidumping protectionism gradually declined. The number of petitions filed sharply decreased following 1992, spiking again to seventy-five petitions filed in 2001 before continuing to decline to comparatively lower amounts of petitions filed within the United States relative to the Cold War era and early 1990s. This may have been the result of U.S. firms becoming sufficiently internationalized. As the United States became integrated into global markets, firms continued to internationalize and became more competitive in global markets, which likely reduced their need for tariff shielding to maintain profits.
Antidumping petitions filed in the United States and South Korea, 1978–2015.
In comparison, South Korea—a newly industrialized Asian country that bucked the deindustrialization trend Footnote 60 —experienced comparatively fewer antidumping petitions filed overall. South Korea and the other Asian Tigers witnessed dramatic growth rates following the implementation of special economic zones and export-oriented industrialization (EOI) growth policies in the early to mid 1960s under President Park Chung-hee. Footnote 61 Liberalization—and EOI growth strategies overall—helped foster massive export booms in South Korea and East Asia, facilitating de facto national trade integration. Footnote 62 In short, South Korea’s export boom and period of industrialization coincided alongside national economic integration. Footnote 63 At the same time, aggregate demand for antidumping petitions in South Korea generally remained lower in comparison to the United States, with a high of eighteen filed in 2003. Demand for antidumping spiked in 1996–97 during the Asian Financial Crisis and again in 2003 while the Korean economy contracted during the same period. Petitions for antidumping petitions again increased in 2007 as financial meltdown—the “Great Recession” —spread across markets worldwide. For the most part, South Korea managed to weather the financial pandemic. However, fears of a financial meltdown may have caused Korean firms to anticipate oncoming economic troubles, triggering a small wave of demand for antidumping protectionism. On the whole, demand for antidumping protectionism tended to remain lower in Korea relative to the United States. This may be a result of high levels of de facto national economic integration and exposure to global markets.
While demand in the United States remained higher compared to demand in South Korea, it gradually tapered off to lower levels following the late 1990s and early 2000s. These variations in demand for antidumping protectionism may be explained by increasing levels of national economic integration and globalization. American firms may have become more efficient and competitive via exploiting shifts in regional comparative advantages to derive cost advantages. By adapting and becoming more competitive in global markets, these firms need not rely on the utilization of domestic antidumping apparatuses to protect profits via protectionist shielding. It seems plausible, therefore, that countries that are sufficiently trade-integrated are less likely to receive petitions for antidumping investigations.
Financial integration and antidumping
Investment and financial flows comprise the other primary element of GEI, which may similarly reduce demand for antidumping trade remedies. I offer two mechanisms that could plausibly explain this relationship. First, while large, domestic corporations that operate internationally frequently outsource to curb production costs, they may also deploy financial capital and internalize production within a wholly owned subsidiary to protect vital intangible assets, which are reflected in a country’s FDI flows. Firms may opt to internalize production via vertical and horizontal integration to protect crucial trade secrets, managerial expertise, and other proprietary assets from exploitation via third parties. Footnote 64 While “contracting out” and outsourcing production provide major economic boons, these strategies also often carry long-term costs: they may enable the emergence of foreign competitors that utilize their trade secrets and production knowledge to leapfrog their way up the industrial ladder. Footnote 65 Firms harboring intangible proprietary assets risk having contractees and outsiders utilizing these assets without compensation. By internalizing production, firms can manage to protect intangible assets from intellectual expropriation by third parties. Footnote 66 . Preventing these spillover effects may help reduce foreign competition that emerges from use of these proprietary assets, which dampens protectionist demand from these firms and reduces the likelihood that nationally integrated economies receive additional antidumping petitions.
Second, I propose that financial flows in part reduce the demand for antidumping protectionism because it serves as a form of “quid pro quo” FDI. The literature on political economy and protectionism has long documented the notion of “quid pro quo FDI” as a means for firms to avoid the prospects of protectionist trade barriers. Footnote 67 This form of FDI differs from traditional tariff-leaping FDI in that firms are concerned about the possibility of protectionist barriers being erected, rather than attempting to dodge existing trade restrictions. Footnote 68 I similarly argue that flows of foreign investment within nationally integrated economies may reflect attempts made by transnational corporations to co-opt protectionist sentiments, which may reduce aggregate demand for additional antidumping trade remedies. While transnational firms tend to produce a variety of intermediate and final products for export to consumer markets, these products serve as a source of steep import competition for other domestic import-competing firms. This provides a potential source of costly risks for prospective exporting firms; additional imposed antidumping measures will push their export prices upward, which will further cut into sales revenue and profits. Footnote 69 Rather than face these risks, firms may instead opt to bear the costs of opening up a subsidiary in a target country to avoid costly risks. This additionally provides incentives for firms to cut back on its manipulation of its terms of trade, cutting back their demand for antidumping trade remedies (Blanchard, Bown, and Johnson, Reference Blanchard, Bown and Johnson2016).
Because the antidumping process is largely bureaucratic in nature, it is difficult for firms to utilize information derived from a target country’s domestic politics to anticipate the likelihood of tariff formation. Footnote 70 Nonetheless, my theoretical argument focuses upon the demand side of the equation—the possibility of firms initiating antidumping petitions. The risk of motivating antidumping protectionism through exports to consumer markets may effectively incentivize firms to purchase a subsidiary or otherwise directly invest in a target market to produce final goods for sale. Firms take these risks into account and instead directly produce final goods within a target consumer market, importing intermediate components from abroad into domestic economies for final production to avoid the prospects of antidumping protectionism being imposed on these goods. This is more likely to be true of large, internationalized firms than of less-experienced firms; owing to more experience and resources available, internationalized firms are more capable of successfully navigating international markets. Footnote 71 It therefore follows that foreign investment flows may reflect the efforts internationalized firms to co-opt protectionism, reducing the aggregate number of antidumping petitions filed. From this theoretical argumentation, I derive my core hypothesis:
• H1: Countries with higher levels of trade and financial globalization should receive fewer antidumping petitions.
Research design
I employ quantitative statistical analysis to empirically test my argument. Because the dependent variable is count distributed, standard ordinary least squares (OLS) techniques will produce biased and inefficient statistical estimates. The dependent variable also exhibits signs of overdispersion. I therefore employ count modeling techniques, utilizing negative binomial regression to model the overdispersion present within the dependent variable. I opt for fixed effects to account for unit-specific effects in the data. I analyze all available data from the World Bank’s Temporary Trade Barriers Database. The TTBD hosts data collected from thirty-three countries between 1978 and 2020. To extend these data to 2022, I utilize aggregate petition filing data available for these countries from the World Trade Organization. In total, the analyses encompass data collected from 33 countries between 1978 and 2022. This sample comprises the entirety of the data available for antidumping petition filings within the Temporary Trade Barriers Database. Footnote 72
Many of the explanatory and control variables suffer from missing data. One standard approach to issues with missing data involves case-wide deletion, which outright removes data with missing values. However, this technique can induce bias in the estimates produced by model estimation by excluding valuable information about any relationships between variables that are present in the observations deleted. Footnote 73 Thus, case-wide deletion suffers from unfortunate drawbacks. I therefore employ multiple imputation techniques to control for a wider variety of variables within the empirical model to include a greater variety of control variables within the models. Additionally, these techniques have been utilized within previous research in comparative and international political economy. Footnote 74 I employ these techniques with the “Amelia” package in R. I employ twenty imputations of the data. A summary of missing data is included in Table 1. I include all summary statistics in the Appendix as well in Table A8.
Missing data summary
Autocorrelation is present within the antidumping petitions data. Footnote 75 However, technical difficulties emerge with modeling both autocorrelation and overdispersion simultaneously. This is particularly prevalent when utilizing software to apply multiple imputation techniques; available options for modeling poisson and negative binomial autoregressive models with time series cross-sectional imputed data are nonexistent. Footnote 76 I therefore opt for trade-offs. I employ an autoregressive poison model in the Appendix to check for robustness in key variables across model specifications that account for autocorrelation. Due to software limitations in terms of performing imputation as well as estimating multiple imputation autoregressive-count models, this model utilizes the base model within the unimputed dataset.
Dependent variable
I measure aggregate antidumping petition initiations filed using data collected from the Global Antidumping Database (GAD), which is hosted under the World Bank’s Temporary Trade Barriers Database (TTBD). Footnote 77 This data includes information on all 33 countries included within the GAD. Table A5 in the Appendix provides details on the countries included within the sample. The sample series for each country begins with the first year they appear in the database. Footnote 78 I extend the data series to 2022 with aggregate petition filings data from the World Trade Organization. I opt to focus primarily on aggregate antidumping petition initiations, as they provide a clear and direct measure of firm demand for antidumping trade remedies. I produce this measurement by manually aggregating the quantity of petitions filed by firms within countries per country-year. Footnote 79 This measurement is preferred to other measurements (e.g. imposed duties) as it offers a more direct measure of firm demand for antidumping trade remedies. I additionally check for robustness in the findings across model specifications by relying on a measure of imposed trade remedies, which is also hosted within the World Bank’s Temporary Trade Barriers Database.
Independent variables
• Trade Globalization—To construct my first set of primary explanatory variables, I first collect trade globalization data from the KOF Index of Globalization. KOF dichotomizes between de jure and de facto measures of trade globalization. I opt to collect the de facto measure of trade globalization, as it best empirically captures the substance of my theoretical argument. These values are further specified as lagged variables within the imputation algorithm. KOF’s de facto trade indicator measures trade globalization as the “exchange of goods and services over long distances” and is computed via exports and imports of goods and services as a share of GDP. Footnote 80 KOF’s measure relies primarily on principle component analysis to measure each weight utilized in the process of constructing the variable. This measurement is preferred to other measurements of de facto trade, as it accounts for geographical distributions of linkages in real trade in goods by measuring trade partner diversity. This provides an added benefit to utilizing this data in that it measures the extent to which countries are globally oriented toward trade in international markets as opposed to trade in regional markets. Footnote 81 This choice of measurement is also supported by previous literature. Footnote 82 “Trade Globalization” is measured on an interval from 0 to 100. Scores closer to 100 indicate higher levels of de facto trade globalization, whereas lower scores indicate lower levels of de facto trade globalization. Footnote 83
• Financial Globalization—I measure aggregate financial integration by similarly collecting financial globalization data from the KOF Index of Globalization. Footnote 84 This variable is a quantity-based measure and is constructed by measuring “capital flows and stocks of foreign assets and liabilities”. Footnote 85 This variable further includes measurements of FDI, portfolio investments, international debt, and international reserves, which are calculated as “the sum of stocks of assets and liabilities and normalized by GDP”. Footnote 86 Furthermore, this measurement includes information about relative positions within the international financial system for a large swath of countries. Footnote 87 Similar to the trade globalization measurement, KOF’s measure primarily utilizes principle component analysis to measure each weight employed in the process of developing the indicator. As such, I prefer this indicator to other measures of financial globalization as it is the most comprehensive measure of real capital stocks and assets, reflecting the extent to which economies are financially integrated within the global economy. This modeling choice is also supported by previous literature. Footnote 88 “Financial Globalization” is measured on an interval from 0 to 100. Scores closer to 100 indicate higher levels of de facto financial globalization, whereas lower scores indicate lower levels of de facto trade globalization.
• FDI Stock Outflows—To construct my second set of primary explanatory variables, I collect FDI stock outflow data from UNCTAD’s Foreign Direct Investment Database. This measure captures the percentage of FDI stock in a country’s GDP and is “the value of capital and reserves attributable to a non-resident parent enterprise”. Footnote 89 This measure is preferred as it measures the accumulation of capital deployed by investors over time. As such, this measure should capture the entrenchment of foreign investment within host economies as well as its level of importance overall, as it measures the role foreign capital investment has played within a host country’s economy over time. Additionally, this measurement has been employed by previous scholarship. Footnote 90 I expect a negative coefficient for this variable. I additionally employ a natural log transformation of this variable to check for robustness, which can be found in the Appendix. Footnote 91
FDI Stock Inflows—I similarly construct stock inflows using data from UNCTAD’s Foreign Direct Investment Database. This measure also captures the percentage of FDI stock in a country’s GDP. Footnote 92 As with stock outflows, this measure is similarly preferred to other measures as it measures the accumulation of capital invested over time. Footnote 93 I expect a negative coefficient for this variable. As with FDI outflows, I employ an additional model estimating antidumping petitions utilizing a natural log transformation of this variable to check for robustness across model specifications in the Appendix. I additionally specify these variables as lags within the imputation analyses.
• Trade—To construct my third set of primary explanatory variables, I collect trade data from the World Bank’s World Development Indicators Database. “Exports” captures a country’s annual exports of goods and services as a percentage of its GDP, whereas “Imports” measures a country’s yearly imports of goods and services as a percentage of its GDP. I employ both of these variables as alternative specifications of a country’s degree of integration within global markets. Footnote 94 I expect a negative coefficient for both of these variables. Footnote 95
Other control variables
• Real Exchange Rates—Previous literature implicated real exchange rate movements as a theoretically significant determinant of the quantity of aggregate antidumping petitions filed. Omitting exchange rate movements may risk misspecifying theoretical and empirical models of antidumping protectionism, increasing the risks of wrongly attributing statistical estimates to incorrect causal factors. I therefore control for movements in real exchange rates. This data is collected from Bruegel’s real effective exchange rate (REER) database. Footnote 96 Bruegel’s data contains CPI-adjusted exchange rate data for 178 countries for the period 1960–2018. I expect this variable to be positively associated with higher likelihood of petition initiations being filed during a given country-year.
• Retaliatory Motives—The retaliatory motives literature implicates tit-for-tat trade dynamics as a primary determinant of gross antidumping petitions and measures imposed (Feinberg and Reynolds, Reference Feinberg and Reynolds2018). To avoid mistakenly attributing changes in antidumping petitions filed to my primary explanatory variables, I include “Retaliation” to control for retaliatory dynamics. This measure is generated by aggregating the number of petitions filed against each country by every other country collected in the GAD across the sample series. Additionally, this measurement has been utilized by previous research. Footnote 97 I expect a positive coefficient for this variable.
• Regime Type and Institutional Quality— IPE scholarship traditionally emphasizes the role of regime type in affecting the extent to which countries adopt freer trade or protectionist policies. Footnote 98 Regime type may exhibit a multiplicity of effects on the aggregate number of antidumping petitions filed during a given country-year. Specifically, democracies may be more strongly associated with lower tariffs because they take a larger range of interests into account when crafting trade policy. Additionally, democratic regimes may signal presence of stronger domestic institutions. I measure “Democracy” using the data collected from the Polity IV Project’s database. Footnote 99 I specifically employ the database’s “polity2” variable. The polity2 variable measures a country’s “Polity Score,” which ranges on an interval between −10 and 10. Scores between −10 and −6 are considered autocratic regimes; scores between 6 and 10 are considered democracies. I expect negative coefficients for this variable.
I also employ a measure of private property rights protection to further assess institutional quality across model specifications. “Contract-Intensive Money” measures the level of property rights enforcement within a country. Contract-intensive money refers to “the ratio of non-currency money to the total money supply, or (M
${{\rm{\;}}_2}$
− 2)/M
${{\rm{\;}}_2}$
, where M
${{\rm{\;}}_2}$
is a broad definition of the money supply and
$C$
is currency held outside banks”.
Footnote 100
This variable is measured on an interval between 0 and 1, with scores closer to 1 indicating stronger enforcement of private property rights.
Footnote 101
I similarly expect negative coefficients for these variables.
• Import Growth—“Import Growth” is defined simply as the annual percent change in imports during a given country-year. Higher percentages of foreign import growth may trigger additional demand for trade remedies during a given country-year due to short-term economic distress. Omitting this variable runs the risk of wrongly attributing changes in antidumping petitions filed to my primary explanatory variables when they are instead more significantly associated with import competition. Footnote 102 . I collect this data from the World Bank’s “World Development Indicators.” Pursuant to previous literature, I expect a positive coefficient for this variable.
• GDP Growth—Some scholarship emphasizes the role of business cycles in affecting the demands for protectionism. Theoretically, measuring fluctuations in GDP growth could capture cyclical economic downturns within an economy. I thus include “GDP Growth” to account for mechanisms related to business cycles. Declining GDP growth may signal the onset of an economic recession during a given country-year. I collect this data from the World Bank’s “World Development Indicators.” It is specifically operationalized as the percent change in annual GDP (current international dollars). I expect a negative coefficient for this variable.
• Unemployment—Alternatively, fluctuations in unemployment rates may also signal the onset of business cycles. “Unemployment” is therefore included to check for robustness across model specifications. Unemployment rates may capture this factor by measuring drops in employment, which may result from periodic economic recessions. This is operationalized specifically as the percent of the civilian labor force actively seeking employment. I collect this data from the World Bank’s “World Development Indicators.” Theoretically, I expect higher rates of unemployment to be positively associated with stronger demand for antidumping protectionism.
• Economic Shocks—I include a measure to capture economic and financial crises within the model, which may influence national filing patterns. Short-term financial crises may influence decisions to file for antidumping investigations, which may influence national filing patterns. “Economic Crisis” is a dichotomous variable that captures systemic economic and financial shocks occurring within a given country. This variable receives a value of 1 if a jarring financial shock occurred within a country during a given year, and 0 if otherwise. I construct this variable with data collected from the “Behavioral Finance and Financial Stability” dataset, which is hosted by the Harvard Business School. Footnote 103
Results
Table 2 displays the parameter estimates derived from Models 1 to 5. I derive exponentiated coefficients to report substantive effects within the data. Model 1 estimates the effects of de facto trade globalization and financial globalization on aggregate petitions filed, controlling for retaliatory motives, real exchange rate appreciations, regime type, and a host of macroeconomic control variables. Models 2 and 3 estimate the effects of trade flows on aggregate petitions filed, controlling again for retaliatory motives, real exchange rate appreciations, regime type, and macroeconomic controls. Finally, Models 4 and 5 incorporate FDI stock outflows and inflows while including the previous controls within the model specification.
Models 1–5: Negative binomial models, fixed effects—antidumping petitions, 1978–2022—KOF indices, FDI stock flows, trade (percent of GDP)
Standard errors in parentheses.
Parameter estimates reported in the table.
${{\rm{\;}}^{\rm{*}}}$
$p \lt 0.05$
.
Overall, the models indicate a series of consistent statistical relationships. First, across the models, national economic integration has a consistent negative and statistically significant effect on the aggregate number of antidumping petitions filed, holding all else equal. Model 1 provides one exception to these statistical relationships, with “Trade Globalization” receiving stronger empirical support than “Financial Globalization” in the results. This may suggest that de facto trade integration impacts petition filing patterns significantly more than de facto financial integration. However, a post-estimation test for joint significance indicates that both variables are jointly significant in the model (
$p \lt .001$
). Second, the “retaliatory motives” argument receives significant and robust support across all five models, with the effects of retaliation being consistently positive and statistically significant. Third, the “real exchange rates” hypothesis receives strong empirical support in the models, which is consistent with existing literature that relies on smaller country samples. The effects of exchange rate appreciations are consistently positive and statistically significant across each model. Fourth, a country’s regime type does not seem to be significantly associated with the likelihood of petitions being filed at traditional levels of alpha. Finally, the macroeconomic controls are generally not significantly associated with petition filings at traditional levels of alpha.
Turning to Model 1, we can see that “Trade Globalization” is negatively associated with the number of petitions filed, and is statistically significant (
$p \lt 0.05$
). Countries that were trade-integrated experienced about a 1.7 percent decrease in the incidence rate of a petition being filed. Countries with higher levels of de facto trade were less likely to receive petitions for antidumping investigations during a given country-year. As mentioned above, “Financial Globalization” does not meet traditional levels of statistical significance in the model, but a post-estimation test for joint significance indicates that it is jointly significant with “Trade Globalization” (
$p \lt .001$
) in the model, lending credence to the theoretical hypotheses developed here. This may provide evidence that trade integration impacts petition filing processes more significantly than financial integration. The model returns positive coefficients for both “Retaliation” and “REER,” which are statistically significant (
$p \lt 0.05)$
. Countries that experienced retaliatory motives and currency appreciations were more likely observe demand for additional antidumping protectionism, lending support to hypotheses developed in previous scholarship.
Models 2 and 3 illustrate a similar empirical story, estimating the effects of a country’s exports and imports (as a share of GDP) on national petition filings. The models indicate that “Exports” and “Imports” are negatively associated with the number of petitions filed, and are significant at traditional levels of alpha (
$p \lt 0.05)$
. The incidence rate of an antidumping petition decreases in both models by about 1.2 percent and 1.1 percent, respectively. Countries experiencing higher volumes of exports and imports were less likely to observe additional antidumping petitions filed during the sample period. Similar to Model 1, “Retaliation” and “REER” are positively associated with the number of antidumping petitions filed, and are statistically significant at traditional levels of alpha (
$p \lt 0.05)$
.
Models 4 and 5 incorporate variables measuring a country’s inflows and outflows of FDI capital stock. “Stock Outflows” and “Stock Inflows” are both negatively associated with national petition filings, and are statistically significant at traditional levels of alpha (
$p \lt 0.05)$
. The models again return positive coefficients for “Retaliation” and “REER,” which are statistically significant at traditional levels (
$p \lt 0.05)$
, providing robust support for these hypotheses. Models 4 and 5 thus indicate that countries experiencing higher levels of FDI stock flows also observed fewer petitions filed throughout the sample period. Finally, Table A2—located in the Appendix—displays the findings from the models employing natural logs and squared terms of FDI stock flows and trade flows, respectively. Intriguingly, the results are divergent across both de facto trade and financial integration. While the models employing squared terms for de facto trade integration do not indicate the presence of significant non-linear effects, the models employing natural log transformations of both inward and outward FDI stock flows seem to be capturing these effects (
$p \lt 0.05)$
. While further investigation of this empirical puzzle falls outside the current scope of this paper, the jarring nature of these findings warrants further inquiry in future research.
Table 3 displays the results from Models 6 to 10, replacing petition initiations with final petitions imposed as the outcome of interest. Model 6 estimates final petition impositions utilizing both “Trade Globalization” and “Financial Globalization,” controlling for retaliatory motives, real exchange rates, regime type, and the macroeconomic measurements. Models 7 and 8 include “Exports” and “Imports,” including the previous control variables. Models 9 and 10 estimate granted trade remedies utilizing “Stock Inflows” and “Stock Outflows,” retaining the previous specification of control variables.
Models 6–10: Negative binomial models, fixed effects—antidumping petitions imposed, 1978–2022
Standard errors in parentheses.
Parameter estimates reported in the table.
${{\rm{\;}}^{\rm{*}}}$
$p \lt 0.05$
.
Similar to the findings reported in Table 2, the model results reported in Table 3 illustrate a series of consistent relationships. Throughout the models, the employed indicators of de facto economic integration appear to be negatively correlated at traditional levels of statistical significance (
$p \lt 0.05$
), with the exception of “Financial Globalization.” Model 6 therefore may suggest that de facto trade integration impacts trade remedy production along with initial filing patterns more significantly than financial integration. I similarly employ a post-estimation test for joint significance, finding that both trade and financial integration are jointly significant in Model 6 as well (
$p \lt .001$
). The “retaliatory motives” and real exchange rate hypotheses receive significantly robust support across each model in Table 3, with both measurements being consistently positively and significantly correlated with the likelihood of trade remedies being granted. For the most part, regime type explanations do not receive much empirical support in Models 6–10, with the exception of Model 10 (
$p \lt 0.05$
). The macroeconomic controls included in each model also receive a strong amount of robust empirical support from the results reported in Table 3. “GDP Growth” is positively and significant correlated with the likelihood of a trade remedy being granted by a government (
$p \lt 0.05$
) across each model, indicating that the onset of business cycles play a role in the production of finalized antidumping policies. The overall findings from these models are striking, given that this measure of antidumping trade policy focuses on supplied policies, rather than aggregate petition filings per say. Globalization—as measured in terms of de facto trade and financial integration—appear to significantly affect the process of trade remedy production within domestic institutions. However, to provide a full explanation of these findings, this would require developing a theory of bureaucratic decision-making, which lies outside the scope of this paper. I thus leave these questions for future research to address.
Conclusion
What factors drive demand for antidumping trade remedies? In this paper, I proposed that countries that are nationally integrated into global markets should be associated with less demand for antidumping trade remedies. In particular, countries with multinational firms deploying higher volumes of trade and FDI will be less likely to observe demand for trade remedy policies overall. Trade and investment flows possibly reflect the preferences of emergent competitive transnational companies that require fewer antidumping services from governments, thus being less likely to file for antidumping investigations. The results demonstrate that the structure of economic production and integration within global economic processes affect the composition of trade policy preferences among economic actors. In doing so, this paper also contributes to the antidumping literature by incorporating systemic variables capturing the structure of the global economic structure to analyze prevailing trends in demand for antidumping trade remedies by economic actors. I investigated this argument with an analysis of antidumping petition data collected from over thirty countries between 1978 and 2022, finding evidence in support of this argument. I also find additional evidence in support of exchange rate arguments and retaliatory motives, along with limited evidence for macroeconomic determinants.
While economic integration has thus far been shown to be associated with dampened demand for antidumping protectionism, these globalization mechanisms may induce further retreats from economic integration in the future. With emergent backlashes against globalization, it is possible that economic integration could serve to generate feedback dynamics, which produces further demand for antidumping trade remedies and further retreats from global markets. The ongoing trade disputes between the United States and the European Union resulting from green industrial policies may serve as a direct example of these possibilities. Footnote 104 While national economies have gradually become more integrated over time, the possibility of a retreat from economic globalization remains plausible.
Future research can improve upon these findings in a number of ways. First, analyzing the determinants of antidumping protectionism via a dyadic framework would be instrumental for a clearer inquiry into the matter. For example, this would allow for a clearer inquiry as to precisely which firms are motivated to demand additional antidumping protectionism due to exchange rate appreciations. Additionally, an analysis of dyadic trade and financial stock data would allow for future scholars to check for specific effects across pairs of countries within the data sample.
Finally, future research should attempt to formulate the role of bureaucratic politics in driving the “supply-side” portion of the antidumping policy process. This paper primarily provides assesses the “demand-side” element, focusing on what factors drive demand for additional antidumping protectionism. While I find little evidence here to suggest that domestic political institutions have any significant effects on antidumping petition initiations, they may play a role in the bureaucratic apparatus that supplies the final imposed measures.
Despite the limitations arising from the data in the study, the models employed demonstrated a clear relationship between indices of national de facto economic integration and antidumping petition filings. Namely, that higher levels of FDI stock outflows appear to be negatively correlated with the propensity for firms to initiate antidumping investigations. I also replicated the findings produced by existing literature on other prominent explanations for the proliferation of antidumping laws around the globe. The question of whether antidumping laws are a desirable policy or not lies outside the scope of this paper, and remains a pressing question for future research.
Competing interests
The author declares none.
Appendix A: Models and data summaries
Models 1–5: Negative binomial models, fixed effects—antidumping petitions, 1978–2022—KOF indices, FDI stock flows, trade (percent of GDP), substituting GDP for unemployment rates
Standard errors in parentheses.
Parameter estimates reported in the table.
${{\rm{\;}}^{\rm{*}}}$
$p \lt 0.05$
.
Models 1–5: Negative binomial models, fixed effects—antidumping petitions, 1978–2022, replacing polity with contract-intensive money
Standard errors in parentheses.
Parameter estimates reported in the table.
${{\rm{\;}}^{\rm{*}}}$
$p \lt 0.05$
.
Negative binomial models, fixed effects—1978–2022—antidumping petitions, 1978–2022—squared trade and logged FDI
Standard errors in parentheses. Parameter estimates reported in the table.
${{\rm{\;}}^{{\rm{**}}}}$
$p \lt 0.05$
,
${{\rm{\;}}^{\rm{*}}}$
$p \lt 0.1$
.
Antidumping petitions filed, 1978–2022—negative binomial models, fixed effects: exports, imports, and FDI indicators
Standard errors in parentheses.
Parameter estimates reported in the table.
${{\rm{\;}}^{{\rm{**}}}}$
$p \lt 0.05$
,
${{\rm{\;}}^{\rm{*}}}$
$p \lt 0.1$
.
Note: high levels of collinearity in the model.
Autoregressive Poisson model: 1978–2022
Standard errors in parentheses.
Parameter estimates reported in the table.
${{\rm{\;}}^{\rm{*}}}$
$p \lt 0.05$
.
Countries and years included in regression models
Correlation matrix—trade globalization
Correlation matrix—financial globalization
Summary statistics
