Chinese Firms in the US–China Trade War: Decoupling through Reshoring?

Abstract

This paper leverages firm-level data to examine the impact of the US–China trade war on the greenfield investment of Chinese investors. Our research yields a few interesting findings. Importantly, our longitudinal analysis of Chinese greenfield investment projects yields evidence that the tariffs have dampened overall Chinese investment. Further analyses for different world regions indicate that while the tariffs have generally had a chilling effect on Chinese investment in most world regions, they may have prompted Chinese investors to engage in “tariff-jumping” in the United States. Our analysis further shows that the trade war may have incentivized Chinese investors to increasingly invest in countries with good political ties with Beijing, presumably as a buffer against potential downturns in bilateral relations. It may also have contributed, at least in part, to a reduced willingness by Chinese state-owned enterprises to engage in overseas investment. However, we found no evidence that the tariffs have led to a perceptible shift in China's investment flows toward BRI countries. Overall, these findings help to illuminate the extent to which policies of “decoupling” may have induced shifts in Chinese investors’ investment patterns, at least in the short-run.

Este artículo aprovecha datos a nivel empresarial con el fin de estudiar el impacto que ejerce la guerra comercial entre Estados Unidos y China sobre la inversión en nuevas instalaciones en el extranjero (inversiones greenfield) por parte de los inversores chinos. Nuestra investigación arroja algunas conclusiones interesantes. Resulta de importancia destacar el hecho de que nuestro análisis longitudinal de proyectos de inversión chinos en nuevas instalaciones en el extranjero proporciona algunas pruebas de que los aranceles han frenado la inversión china en general. Existen otros análisis, realizados en diferentes regiones del mundo, los cuales indican que, si bien los aranceles han tenido, en general, un efecto disuasorio sobre la inversión china en la mayoría de las regiones del mundo, también pueden haber provocado que los inversores chinos participen en “saltos arancelarios” en los Estados Unidos. Nuestro análisis demuestra, además, que la guerra comercial puede haber incentivado a los inversores chinos a invertir cada vez más en países con buenos vínculos políticos con Pekín, presumiblemente a modo de amortiguador contra posibles desaceleraciones en las relaciones bilaterales. También puede haber contribuido, al menos en parte, a una menor disposición de las empresas estatales chinas a participar en inversiones en el extranjero. Sin embargo, no encontramos pruebas de que los aranceles hayan llevado a un cambio perceptible en los flujos de inversión por parte de China hacia los países de la Iniciativa de la Franja y la Ruta (IFR). En general, estos resultados nos ayudan a arrojar luz sobre hasta qué punto las políticas de “desacoplamiento” pueden haber provocado cambios en los patrones de inversión por parte de los inversores chinos, al menos a corto plazo.

Cet article exploite des données à l’échelle d'entreprises pour analyser les effets de la guerre commerciale entre les États-Unis et la Chine sur les investissements sur site vierge d'investisseurs chinois. Nos travaux de recherche offrent quelques conclusions intéressantes. Notre analyse longitudinale des projets d'investissement sur site vierge chinois nous montre notamment que les droits de douane ont amoindri l'ensemble des investissements de la Chine. En analysant d'autres régions du monde, nous constatons que bien que les droits de douane ralentissent généralement les investissements de la Chine dans la plupart des régions du monde, ils ont également peut-être incité les investisseurs chinois à s'adonner au « tariff-jumping » aux États-Unis. Notre analyse montre en outre que la guerre commerciale pourrait avoir incité les investisseurs chinois à renforcer leurs investissements dans les pays entretenant de bonnes relations politiques avec Pékin. Il s'agirait a priori d'un tampon en cas de diminution des relations bilatérales. Elle pourrait également expliquer, au moins en partie, la réticence des entreprises publiques chinoises à investir à l’étranger. Cependant, nous n'avons trouvé aucun élément indiquant que les droits de douane auraient modifié de façon manifeste les flux d'investissement de la Chine vers les pays BRI. Dans l'ensemble, nos conclusions permettent d’éclaircir dans quelle mesure les politiques de « découplage » pourraient avoir induit une modification des schémas d'investissement des investisseurs chinois, au moins sur le court terme.

Introduction

On July 6, 2018, the US–China trade war officially opened with a set of tariffs imposed by the United States and China. The US imposed a 25 percent tariff on 818 imported Chinese products, valued at US $34 billion, leading the Chinese government to retaliate by imposing 25 percent tariffs on 545 goods originating from the US, also valued at US $34 billion (Bown and Kolb 2022). These actions followed efforts at bilateral negotiations and, immediately before the exchange of tariffs, the conclusion of a Section 301 investigation requested by the US’ Trump administration on “China's Acts, Policies, and Practices Related to Technology Transfer, Intellectual Property, and Innovation.” Subsequent tit-for-tat tariffs, ranging from 10 to 25 percent, affected roughly $350 billion of US imports from China and $100 billion in US exports to China between 2018 and 2019. In January 2020, a phase-one trade agreement was signed, in which China committed to purchasing an additional $200 billion of goods and services in addition to pledging to address some of its alleged unfair trade practices in order to defuse trade tensions with Washington. Despite this temporary truce and more recent discussions by the Biden administration to lift some of these tariffs, most of the Trump-era tariffs have remained in place during the Biden administration, raising concerns that the trade war may further accelerate economic decoupling between the two largest economies in the world and stymie decades of progress towards integrating China into the global economy.

As the largest economic conflict in the contemporary era, the US–China trade war has stimulated growing scholarly interest in its economic impact, including its impact on American and Chinese firms, workers, and consumers; regional or global trade patterns; or global value chains (Amiti et al. 2020, 2019; Bellora and Fontagné 2020; Mao and Görg 2020; Wu et al. 2021). With regard to trade patterns, recent studies have yielded some preliminary evidence that the trade war has dampened US-China trade, at least until the onset of the more recent pandemic (Zeng et al. 2022), although it did not necessarily lead to a slowdown in global trade growth due to the reallocation of exports targeted by the tariffs by producers elsewhere in the world (Fajgelbaum et al. 2024).

This study directs attention to an important gap in our understanding of the US–China trade war: how has the exchange of tariffs affected investment behavior? In examining this underexplored area of the trade conflict, this study offers a firm-level analysis of the impact of the trade war on the determinants and locational patterns of Chinese greenfield investment. Leveraging both the Orbis firm-level data and the Orbis Crossborder Investment dataset, we not only examine the determinants of Chinese outbound greenfield investment at the firm- and country-levels during our sample period (2013–2021) but also track the changes in such investment in the trade war period, which started in 2018, compared to the pre-trade war period.

The analysis yields several findings of interest, both for understanding the consequences of the trade conflict and for the relationship between trade and investment. First, our results yielded some evidence that the Trump administration's tariffs have had a negative impact on Chinese firms’ investment behavior, an effect that is statistically significant in some but not in all the model estimates. Further analyses for different world regions indicate that while the tariffs have generally had a chilling effect on Chinese investment in most world regions, they may have prompted Chinese investors to engage in “tariff-jumping” in the United States.

Second, at the firm-level, our results indicate that state-owned enterprises (SOEs) are important in spearheading China's overseas investment. Our analysis of the trade war period, however, generated some evidence that SOEs have been less active in investing abroad compared to the pre-trade war years. One possible explanation of this pattern may be that, as a major driving force of China's outbound investment, SOEs may have been most directly exposed to the impact of global geopolitical and economic changes. The trade war may also have led the Chinese Communist Party (CCP) to embrace the “Made in China” policy, with its emphasis on “independence” and “self-reliance” (Hirsh 2022) and to scale back its support for Chinese firms’ global expansion, leading Chinese SOEs to increasingly responding to the changes in the policy environment by investing in the domestic market.

Third, political distance between a host country and China matters more for Chinese firms in the trade war years. Before the onset of the trade war in 2018, the host country's political alignment with China had no definable impact on the likelihood of investment; however, the analysis shows that the imposition of the tariffs has reduced Chinese firms’ willingness to invest in host countries that are politically distant from China, as observed through a lack of alignment in United Nations General Assembly (UNGA) voting records. Fourth, in terms of the geographical distribution of Chinese greenfield investment, there have not been any significant changes in the regional distribution of Chinese greenfield FDI since the beginning of the trade conflict. Further analyses of regional sub-samples yield some evidence of relative decline of Chinese investments in the European Union (EU) and the Association of Southeast Asian Nations (ASEAN) countries in the trade war period.

This paper contributes to a growing body of literature (Blanchard 2019; Sutherland et al. 2020; Nugent and Lu 2021) on the drivers of Chinese outward foreign direct investment (COFDI). Its main findings enrich our understanding of the consequences of the US–China trade war and represent one of the first systematic analysis of how the trade war may have driven changes in COFDI. Our firm-level analysis also yields new evidence that helps shed light on the ongoing debate about whether Chinese investors may have been behaving differently from investors from advanced industrialized countries in their investment decisions.

The paper proceeds as follows. The next section derives our main hypotheses based on a survey of existing literature on the location choice of FDI in general and that of COFDI in particular, which is followed by a descriptive analysis of the pattern of Chinese firms’ greenfield investment. The paper then proceeds to the research design, followed by a report of the main empirical results, robustness checks, and additional analyses. The paper concludes with a summary of the findings, their implications, and directions for future research.

Literature Review and Hypotheses

Investigating the impact of the US–China trade war on Chinese firms’ investments highlights the link between trade protection and foreign direct investment. In this section, we build on previous literature on the factors that influence the location choices of Chinese investors to develop the study's central hypotheses about the impact of the trade war on Chinese firms’ investment patterns.

Motivations and Location Determinants of COFDI

China's rise as a major investor country has captured growing scholarly interest. What motivates this emerging stream of research is the puzzle as to why firms without significant ownership advantages turn out to be successful multinational corporations (MNCs) in their own industries in a way that challenges existing theories of firm internationalization (Child and Rodrigues 2005). Questions have also been raised as to whether firms from emerging market economies such as those from China are more likely to be drawn to host countries with a different set of characteristics than investor firms from developed countries (Buckley et al. 2007; Ramasamy et al. 2012; Shi and Zhu 2018).

Existing theories on firm internationalization provide a useful starting point for understanding the factors that influence the FDI location choice of MNCs from both developed countries and emerging economies. Economists have emphasized the importance of trade costs or firm characteristics in influencing the decision to undertake FDI. Where trade protection in the form of tariffs is regarded as a trade cost, scholars have argued that it depresses investment (Duval and Utoktham 2014). For example, the “proximity concentration” hypothesis holds that firms may be motivated to spread production activities across national borders when trade costs such as those arising from transportation, trade barriers, or other frictions negatively affecting trade flows outweigh the potential gains from scale economies (Brainard 1997). This perspective is consistent with the argument that firms may develop new plants or expand existing operations through the so-called tariff-jumping FDI when trade barriers are sufficiently high. Recent research inspired by new–new trade theory also points to the importance of firm-level characteristics such as size or productivity as determinants of firm internationalization (Antràs and Helpman 2004; Helpman et al. 2004).

While similarly interested in the drivers of multinational production, the international business literature focuses on the liability of foreignness that firms may incur when operating in a foreign country. Foreign firms face additional costs relative to local firms arising from unfamiliarity with the host market, cultural and institutional distance, political and economic differences, and discriminatory policies and practices (Hymer 1976; Zaheer 1995). Firms should find FDI attractive if they possess sufficient organizational and managerial capabilities or firm-specific resources that will help to offset such costs and if they are able to maintain internal control across operations to minimize the information and transaction costs associated with operating in imperfect markets (Buckley and Casson 1976; Hennart 1982; Nelson 1991). Dunning (1977, 1979) builds on the above perspectives to develop the eclectic paradigm, which emphasizes ownership, location, and internalization advantages that could help overcome the liability of foreignness and make it cost-effective for firms to undertake FDI. Ownership advantages refer to a firm's control of ownership rights and proprietary information that may enhance its competitiveness over foreign rivals. Location advantages pertain to host country characteristics such as resources, labor costs, markets, or favorable institutional and political environments that may help lower the costs of production and transportation for foreign firms. Finally, internalization advantages arise when it is more beneficial for the firm to produce the product in-house rather than contracting with a third party. Foreign investment becomes a rational strategy for a firm when all three conditions have been met.

The above theories for explaining when firms decide to become multinational also have implications for understanding their location choices. They highlight the importance of parent firm characteristics, host country characteristics, and the dyadic relationship between home and destination countries for shaping FDI. For outward FDI from emerging markets such as China, studies (Lardy 1998; Buckley 2004; Warner et al. 2013) have argued for a special theory grounded in the above theoretical approaches. For example, firms from emerging economies, in particular SOEs, may be more likely to exploit capital market imperfections such as the availability of capital at below-market rates for an extended period of time and to transform them into ownership advantages. Studies (Wells et al. 1983; Zhan 1995; Lau 2003) have also found that emerging economy investors tend to be attracted to countries with similar cultural backgrounds or those with shared ethnic or familial ties to the home country, which enables them to capitalize on special ownership advantages made possible through location-specific network relations. In addition, studies (Buckley et al. 2007; Morck et al. 2008; Holburn and Zelner 2010) have found that emerging market MNCs are drawn to politically risky destinations, as their experience with weak institutional environments at home makes them less averse to risky business environments abroad.

Building on the above literature, the following section develops hypotheses that capture the influence of a set of factors that have been identified by scholarship on FDI as important determinants of firms’ decision to undertake internationalization. These factors include those specific to the firm (e.g., firm size or productivity), host country (e.g., level of economic development, regime type, or natural resource rents), and the dyad (e.g., bilateral political relations or the existence of international economic agreements such as bilateral investment treaties or free trade agreements). The discussion also addresses the unique role of SOEs in China's outward FDI and that of political risks in shaping Chinese firms’ investment behavior.

Impact of the Trade War On COFDI

Scholarship has not only highlighted the importance of home country characteristics that may influence outward investment, but also the unique features of China's political and social organization, such as state ownership of industry that may affect Chinese investors’ behavior (Wei et al. 2015; Buckley et al. 2018). The discussion below develops the paper's central hypotheses, taking into consideration both sets of factors in shaping Chinese firms’ FDI.

Impact of the Tariffs

The most direct channel through which the trade war may have impacted COFDI is through the imposition of tariffs on Chinese products since July 2018. The trade war has ruptured decades of US policy of engagement with China and accelerated the decoupling of the US–China relations in the spheres of trade, investment, and technology cooperation, a strategy intended to reduce US dependence on China-centered production and supply chain networks. In generating heightened uncertainty and unpredictability in the business environment, the trade war may have undermined business actors’ expectations of the stability of business transactions and reduced their willingness to undertake overseas investments. This effect may have been compounded by the COVID-19 pandemic, which has further aggravated supply chain disruptions and dampened global economic activities. The rising specter of a prolonged commercial conflict amidst the background of growing US–China strategic competition may therefore have undercut Chinese investors’ willingness to invest in not only the US market but also elsewhere in the world. This should especially be the case for the tariffed sectors as they are most directly exposed to rising bilateral tensions and therefore more likely to experience changes in investment patterns. Therefore, Chinese investments may have experienced an overall decline in the sectors subject to the Trump tariffs since the onset of the trade war (Hypothesis 1a).

On the other hand, increased protectionist barriers on Chinese products may have prompted Chinese firms to evade the tariffs on their exports to the United States by establishing production facilities in the US market. Known as “tariff-jumping FDI,” such moves are often considered to be motivated by cost-saving considerations and tend to occur when the costs of exporting exceed those of direct investment (Horstmann and Markusen 1992; Hwang and Mai 2002). For example, Linton Technologies Group, which is a US company with Chinese ownership, decided to move production back to the United States as the “geopolitical issues have been difficult to navigate.”¹ Scholarship (Riker and Schreiber 2019) has further suggested that whether a tariff increase will affect FDI flows will depend on the magnitude of the tariff change, with more significant tariff increases more likely to induce shifts in FDI flows. Following this logic, Chinese investments in the US market may have increased in the sectors subject to the tariffs since the onset of the trade war (Hypothesis 1b).

Yet another possibility is that the tariffs have led Chinese investors to search for alternative production sites, to adjust suppliers, or to reallocate resources, thus adopting a strategy of shifting away from the heightened risks and uncertainty associated with investing in the United States. Such diversion of investment may take the form of increased Chinese investments outside the United States and thus Chinese investments may have increased in third countries in the sectors subject to the tariffs since the onset of the trade war (Hypothesis 1c). Chinese solar panel giant Trina Solar, for example, announced in September 2023 its third factory in Vietnam, strategically placed to boost exports to the United States due to anticipated US tariffs.²

In the next set of hypotheses, the discussion further expands the pathways through which the trade war may have encouraged Chinese investments in third countries.

Ally-shoring

According to a recent Biden administration report on US supply chains, strategies such as ally-shoring or friend-shoring emphasize the importance of cooperating with “allies and partners to foster and promote collective supply chain resilience” (Kollewe 2022). The concept was initially developed by democracies to reduce the vulnerabilities associated with dependence on foreign supplies of essential materials, goods, and services by strengthening supply chain relationships with countries with shared values and ideology. Beijing has also sought to deepen relations with its regional neighbors and countries along the BRI route in an effort to expand China's international influence and undercut Washington's perceived encirclement strategy of China. Beijing has emphasized the shared priority attached to socioeconomic development through the adoption of initiatives such as the China-ASEAN agenda for 2021 (Petri and Plummer 2020).

In this context, political alignment with China captures the diplomatic environment for China's investor firms. The hypothesis of interest is that Chinese firms are likely to locate investments in countries that are closely aligned with China, as observed through United Nations (UN) voting patterns (Bailey et al. 2017). Chinese firms, in an effort to minimize risk, will prefer investment destinations with favorable diplomatic climates for China's economic actors. Recent scholarship has found that Chinese investments are increasingly subject to strict screening in host countries, whose governments may cite national security concerns as grounds for rejecting investor applications (Canes-Wrone et al. 2020; Bauerle Danzman and Meunier 2021; Chan and Meunier 2022). In the United States, the government undertook a reform of the Committee on Foreign Investment in the United States to tighten screening procedures. In the European Union, the first EU-level investment screening mechanism was adopted in 2019. Chan and Meunier (2022) finds that investment screening in EU member states is stricter in high-tech sectors and notes that overall, investment screening is likely to become more restrictive over time. In the United States, scholarship has noted increasing public backlash against Chinese investment activities in recent years (Canes-Wrone et al. 2020). With animosity towards Chinese investors rising in many recipient countries, Chinese firms are likely to invest in countries that have good political relations with China. The trade war may have reinforced this pattern as Chinese investors seek to cushion the impact of tariff exposures. The analysis thus tests the following hypothesis on political relations between China and the destination country:

 

H2:

Chinese investment is less likely to flow to countries that are politically distant from China with the onset of the trade war.

A BRI Effect?

The US–China trade war is first and foremost a political war, pitting the world's leading economy against the second largest and challenging the US’ dominant position in the international economic order (Liu and Woo 2018; Chong and Li 2019; Kim 2019). Scholarship advancing China's position has gone so far as to claim that the Trump administration's initiation of the trade war is actually an attempt to obstruct the rise of China (Lai 2019). As part of its economic statecraft, the central government has sought to promote export-related overseas investment, securing the supply of natural resources, enhancing Chinese firms’ competitiveness, as well as tending to the political side by maintaining strong and positive political ties with host countries (Wei 2010).

Notable among these was Beijing's renewed emphasis on the Belt and Road Initiative (BRI), an ambitious infrastructure and investment project launched in 2013 to increase China's political and economic influence by offering incentives for infrastructure development and financial investment in the region. The BRI comprises two components, namely the Silk Road Economic Belt on land and the twenty-first-Century Maritime Silk Road in the seas. The BRI has been considered as China's grand strategy designed to forge international cooperation on Beijing's terms, to wield influence regionally and globally, and to counter the political and economic influence of the United States, especially in Asia (Callahan 2016; Cheng 2016; Huang 2016; Pencea 2017; Chaisse 2018; Kim 2022).³

Even before the onset of the trade war in 2018, Chinese academics had come to view the initiative as an important means of offsetting the potentially damaging effects of trade tensions with the United States by fostering international economic connectivity with partner countries in the BRI. In the early years of the BRI, as “One Belt, One Road,” partner countries were located in the broader Eurasian continent. Since 2017, BRI partners include countries from Africa, Latin America, Oceania, and the Artic Ocean. The trade war has given additional credence to the view that it may not only represent an offensive tool to enhance China's global standing but may also serve a defensive purpose by providing alternative sources of trade and investment to help mitigate the impact of the trade war. The trade war may thus have led Chinese firms to increasingly shift investment towards BRI partner countries.

 

H3:

Chinese firms are more likely to invest in BRI partner countries during the trade war compared to the pre-trade war period.

Expectations concerning BRI partner countries should be tempered by the understanding that shifting investment towards the BRI countries also entails a sectoral change, moving away from those goods subject to tariff exposures towards sectors such as construction or infrastructure development that are more compatible with the objectives of the BRI. Such a geographical shift is likely to take place in the medium- to long-term, as more years pass since the outbreak of the trade war. Thus, this hypothesis is largely investigative and exploratory, with the expectation of a positive impact of the trade war on investment in BRI countries but noting that for many firms, shifting investments away from established destination markets and towards countries along the BRI route would be a long-term strategy.

Firm Ownership

Scholarship on COFDI has found that firm ownership may play an important role in influencing the investment decisions of Chinese firms, with SOEs more likely to undertake FDI than firms in the non-state sector. This is because SOEs’ historically strong ties to the government provides them with access to financing, foreign exchange, and other forms of support such as heavily subsidized raw materials, energy, input, and monopoly production rights needed to engage in international operations (Dollar and Wei 2007; Haley and Haley 2013; Stone et al. 2022). SOEs’ privileged access to natural and financial resources may therefore have made them more likely to venture into international markets compared to non-state firms. Firms under the direct supervision of the State-Owned Asset Supervision and Administration Commission, in particular, are more likely to respond to the government's calls by conducting international activities aimed at acquiring natural resources or other strategic assets (Ramasamy et al. 2012).

Despite growing evidence that SOEs and state-run banks have gained flexibility and autonomy in the implementation of Beijing's main initiatives designed to expand Chinese firms’ global footprints, such as the Go Out Policy and the BRI (Ye 2020), SOEs may have remained the principal agents for China's global investment expansion (Zhao and Lee 2021). Scholarship on the BRI, for example, has found that SOEs are leading economic actors in investor activities in partner countries (Wildau and Ma 2017), carrying out key responsibilities such as negotiating contracts and establishing partnerships. SOEs are reported to have undertaken more than 3,100 projects in BRI partner countries, and participating SOEs include not only national-level SOEs but also local SOEs, semi-SOEs, as well as their subsidiaries (Rolland 2017; Suokas 2018). SOEs invest in “hard” infrastructure, covering a wide range of activities in transportation, energy supply, and telecommunications.

The trade war may have enhanced the incentives for Chinese SOEs to invest more aggressively in overseas markets to offset the impact of the trade war. SOEs, which played a pivotal role in the government's efforts to promote mutually beneficial economic initiatives in partner countries and faced strong imperatives to meet political mandates, may therefore have made a more forceful foray into international markets in order to compensate for the reduction in market access to the US market.

 

H4:

SOEs are more likely to engage in outward foreign direct investment relative to non-state-owned enterprises during the trade war compared to the pre-trade war period.

Chinese Greenfield FDI Patterns

China's phenomenal economic ascent and the country's emergence as a major trading nation and center of global manufacturing activities have facilitated its transition from a capital-importing to a capital-exporting country. By 2015, China had overtaken Japan to become the country with the second largest FDI flows in the world. While China's outward FDI stock of US$2.4 trillion in 2020 is still a relatively small fraction (6.4 percent) of the world total, China's share of global FDI stock has experienced a five-fold increase during the past decade (Molnar et al. 2021). Chinese greenfield investment in the sample of analysis, illustrated in figure 1, shows Chinese firms completed a total of 4,217 investment projects between 2013 and 2021. The total number of projects gradually increased from 310 in 2013 to a high of 573 in 2018, followed by a decline to 161 projects in 2021.⁴ The capital expenditures associated with these projects demonstrated a somewhat different trend. While peaking at $53.27 billion in 2014, total expenditures dropped to $8.04 billion in 2016 before rebounding to $20.08 billion in 2018. This was followed by another decline to $3.94 billion by 2021.

Figure 1.

Chinese Greenfield FDI, 2013–2021.

Figure 2 presents the top 10 destinations of Chinese greenfield investment by the total number of projects. The top three destination countries of Chinese investment prior to 2018 were Germany, the United States, and India. The United States has surpassed Germany as the number one destination market for Chinese investors since 2018. Other top destinations in both periods include the United Kingdom, France, and Hong Kong. Spain and Mexico have replaced Singapore and Russia as the other two major destinations in the 2018–2021 period.

Figure 2.

Top Destinations of Chinese Greenfield Investment by Number of Projects, 2013–2021.

Figure 3 reports changes in the sectoral composition of Chinese greenfield investment by the total number of projects in the primary investing sector. Sectors such as retail, business services, and biotechnology have become increasingly popular sectors of investment post-2018. Within the manufacturing industry, computer software, transport manufacturing/OEM manufacturing, and industrial, electric, and electronic manufacturing remain relatively important sectors during both periods.

Figure 3.

Sectoral Composition of Chinese Greenfield Investment by Number of Projects in Primary Investing Sector, 2013–2021.

Figure 4 presents the geographical distribution of Chinese greenfield investment by world region between 2013 and 2021. Chinese investment in the European Union far outpaced investment in other world regions, with North America trailing behind at least until 2016. The data additionally point to the negative impact of the trade war on Chinese outward investment. COFDI peaked in 2018–2019, saw a clear drop in 2019–2020, and only started to experience a gradual rebound in 2020–2021 for most world regions.

Figure 4.

Geographical Distribution of Chinese Greenfield Investment by Capital Expenditure, 2013–2021.

Data and Variables

To test our hypotheses, we construct an estimation sample by matching data on Chinese firms obtained from the Orbis firm-level dataset with those from the Orbis Crossborder Investment dataset. Somewhat similar to other databases that monitor global greenfield FDI flows, such as fDi Markets, released by fDI Intelligence, the Orbis Crossborder Investment database tracks greenfield investment projects, including the capital expenditures and job creation associated with these projects, in real time. The database provides comprehensive country and sectoral coverage and includes both listed and private companies. Both databases offer researchers valuable tools for assessing globalization trends and their key drivers. One advantage of this database compared to the fDi Markets database is that it captures not only greenfield FDI projects but also cross-border mergers and acquisitions deals. However, one drawback is that it provides more limited time coverage, containing data dating from 2013 instead of 2003 as in the case of fDi Markets.⁵

The analysis covers Chinese greenfield investments. The resulting sample includes 4,217 greenfield investment projects made by 1,627 unique Chinese firms in 129 destination countries in the 2013–2021 period. We run cross-sectional, time-series models with the firm-destination market as the cross-section.

The dependent variable, Invest, is a dummy variable that equals 1 if firm i has invested in country j in year t. Given that very little data are available in Orbis Crossborder investment dataset on the actual value, or capital expenditures of the greenfield FDI projects, the analysis includes the recorded incidence(s) of greenfield FDI projects. The dependent variable thus indicates whether a Chinese firm has undertaken at least one greenfield FDI project in a country in a given year. The analysis includes the following independent variables of interest:

Tariff Exposure. We include a dummy variable indicating whether a given firm has made an investment in a sector that is subject to Trump tariffs in a given year to capture the effects of the trade war and to test the different expectations under Hypothesis 1 (1a, 1b, and 1c). Data on the Trump tariffs were obtained from the United States Trade Representative's Office.⁶ The tariff data, which are recorded for products classified in the 8-digit subheadings of the Harmonized Tariff Schedule of the United States, are matched to the firm's 6-digit North American Industry Classification System (NAICS) industry code to facilitate the following empirical analysis. Tariff Exposure is coded as 1 if the firm's 6-digit NAICS industry includes a product that is on the Trump administration's tariff list, and 0 otherwise.

Political Distance. The analysis employs the ideal point distance between China and its partner country in United Nations General Assembly (UNGA) voting as a proxy for political relations (Bailey et al. 2017).⁷ A larger value on this variable indicates greater Political Distance between China and the destination country, which is expected to have a negative effect on the likelihood of investment (Invest).

BRI partner. This variable takes a value of 1 when a country has joined the BRI by signing either a memorandum of understanding (MOU) or a cooperation agreement with China in a given year; and 0 otherwise. Data were obtained from the Council on Foreign Relations.⁸ 139 countries joined the BRI by 2021 (Sacks 2021).

SOE. The variable takes on a value of 1 if a given firm is a Chinese state-owned enterprise and 0 otherwise. Data were obtained from the Orbis Crossborder Investment dataset. SOEs include firms classified as the following legal entities as defined by the Orbis database: State-owned and collective company (China), State-owned associates (China), State-owned company (China), and State-owned limited liability company (China).

The analysis employs interaction terms between Tariff Exposure and the above main variables of interest—SOE*Tariff Exposure; Political Distance*Tariff Exposure; and BRI Partner*Tariff Exposure—to assess changes in the effects of these variables since the imposition of the tariffs in the trade war.

The analysis also controls for factors at the firm, industry, country, dyad, and region levels that may influence a firm's investment decisions. At the firm level, it takes into account the potential influence of a firm's size and productivity on its FDI patterns. Theories of firm heterogeneity (Blomström and Lipsey 1991; Helpman et al. 2004) would lead us to expect that larger and more productive firms should be more likely to engage in multinational production. We use the firm's Number of Employees and Profit Margin as proxies for size and productivity, respectively.

At the industry level, the analysis considers Chinese firms in industries that are more dependent on exports to the United States. Such firms may be more vulnerable to tariff exposure as protectionist trade barriers not only increased the costs of exports but also generated heightened uncertainty in the business environment more broadly, which may reduce the likelihood of investment. The analysis thus controls for exports to the United States of the industry to which the firm belongs as a share of the industry's total exports (Export Share to the United States).⁹

At the country level, the analysis controls for the possibility that Chinese investment is more likely to flow to countries with a large market and a high level of economic development, as measured by the logged values of its GDP and GDP per capita, respectively.¹⁰ A host country's natural resource rents are expected to have a positive association with the likelihood of investment, as Chinese firms may be drawn to host countries with an abundant supply of mineral deposits, raw materials, and other natural resources. China's unprecedented economic growth may have further accentuated the demand for natural resources, reinforcing the resource-seeking motivations behind COFDI (Buckley et al. 2007; Kolstad and Wiig 2012; Wang and Yu 2014).

Considering the disruptions that the COVID-19 pandemic has generated for global trade and investment activities, the analysis further controls for a country's Stringency of COVID-19 Measures, obtained from the Oxford COVID-19 Government Response Trackers (OxCGRT) database (Hale et al. 2020).¹¹ The stringency index is a composite measure based on nine response indicators, including school closures, workplace closures, cancellation of public events, restrictions on public gatherings, closures of public transport, stay-at-home requirements, public information campaign, restrictions on international movements, and international travel controls. The index ranges from 0 to 100, with 100 indicating the strictest measures. As this measure reflects the strictness of government policies in response to the severity of the pandemic, it should capture the restrictions on capital mobility due to the pandemic.

To control for regulatory barriers in the host country that may deter foreign investment, the analysis utilizes the FDI Regulatory Restrictiveness Index (FDI Restrictiveness), which measures “statutory restrictions on foreign direct investment in 22 economic sectors across 69 countries, including all OECD and G20 countries.“¹² Since data for this variable are available only for a subset of the countries in our sample, models including this variable necessarily have a smaller number of observations, and inference is limited to the OECD and G20 countries.

Whether Chinese firms are more likely to be attracted to countries with a higher level of political risk has also been the subject of considerable scholarly debate. Buckley et al. (2007) found that as most Chinese investments were led by the government, amicable political relations between China and host countries in the developing world may have made Chinese firms less sensitive to political risks in the destination country. Kolstad and Wiig (2012) and Ramasamy et al. (2012) yielded similar conclusions, showing that SOEs are less risk averse than firms in the non-state sector. Studies have also found that investors from emerging market economies such as China may be more likely to be attracted to autocracies as their greater familiarity with systems that are not fully transparent may offset the liability of foreignness they face in weak institutional environments (Child and Rodrigues 2005; Morck et al. 2008; Holburn and Zelner 2010).

Recent scholarship has challenged the view that Chinese investments favor countries with poor institutions. Shi and Zhu (2018), for example, argue that as MNCs embedded in autocracies cannot easily transfer the political connections and influence they have developed at home to other contexts, this may limit the use of similar strategies to cope with political risks abroad. Biglaiser and Lu (2021) find that, except for natural resource investment, Chinese state investment tends to flow to low-risk developing countries in a way that is not too dissimilar to the pattern observed for US investors. The global expansion of Chinese investment may have increased Chinese investors’ sensitivity to host country political risks. Other studies (e.g., Zeng 2019; GaoYan 2020) have yielded limited to no evidence that Chinese investors favor politically risky destinations.

Given that many of the indicators of political risks, such as rule of law, control of corruption, and regulatory quality, are highly correlated with one another as well as with a country's regime type, the analysis includes a control for regime type. Regime Type, which is measured as the sum of a country's political rights and civil liberties ratings provided by the Freedom House score, is a continuous variable ranging from 0 for a highly authoritarian country to 14 for a highly democratic one. The analysis does not place any expected sign on the relationship between political risk and Chinese greenfield investment due to the contrasting theoretical expectations and mixed empirical evidence described above.

At the dyadic level, the analysis controls for geographic distance (distance), which may inhibit FDI, and the presence of a bilateral investment treaty (BIT) between China and the partner country that may promote investment. BITs’ liberal investment protection and liberalization provisions appear to boost COFDI (Lu et al. 2021), and this effect is especially pronounced for SOEs (Zhao and Lee 2021). BIT is a dummy variable that equals 1 if China and the host country have a BIT in effect in a given year and 0 otherwise. Data for this variable are drawn from the International Investment Agreements Navigator maintained by UNCTAD. Appendix 1 presents the descriptive statistics of the main variables. All of the above variables are lagged by one year to take account of the potential for reverse causality.

Finally, the analysis includes controls for the following regions, measured as dichotomous variables with a value of 1 if a destination market is within the region. The regions include the Association of Southeast Asian Nations (ASEAN) comprising the 10 countries in Southeast Asia, the European Union (EU), North America (NA), South Asia (SA), Latin America and the Caribbean (LAC), and the Middle East and North Africa (MENA). Regions are defined in terms of World Bank country classifications, and the variables test for any region-specific effects associated with Chinese greenfield investments.¹³ The reference category consists of all countries not included in the above regional classifications. The analysis further examines the interaction terms between each of these region variables and the Tariff Exposure variable to evaluate whether the adoption of tariffs brought about any changes to the geographical distribution of Chinese investment since the onset of the trade war.

Findings

We estimate population-averaged logit models for panel data recorded at the firm-destination-year level during the 2013–2021 period. The population-averaged logit model (Neuhaus et al. 1991; Neuhaus 1992), in contrast to the random effects model, highlights cross-sectional differences in a panel context. The estimates, therefore, provide a comparison of firms across destination countries rather than within-unit comparisons of firms in the same host country. Table 1 reports the results of the analysis. Model (1) provides a baseline set of results, and it reports estimates for our major independent variables of interest—Tariff Exposure, SOE, Political Distance, and BRI Partner. The analyses control for the share of exports to the United States in the industry (Share of Exports to the United States) in which a greenfield investment project is located and national level characteristics such as GDP, GDP per capita, Distance, the existence of a bilateral investment treaty (BIT) with China, Stringency of Covid-19 Measures, and regions in which the investment was made. The models also include the interaction terms between Tariff Exposure and the main variables of interest—SOE, Political Distance, and BRI Partner—to examine how sectors affected by the imposition of tariffs have affected the investment decisions of Chinese firms. Model (2) builds on Model (1) to account for interaction effects between the regions and Tariff Exposure. Models (3) and (4) include controls for Regime Type, FDI Restrictiveness, and Natural Resource Rents. As the limited data available for these latter three variables significantly reduces the sample size, estimates including these variables are reported separately.

Table 1.

Cross-Sectional Time-Series Logit Models of Greenfield FDI by Chinese Firms

	Dependent variable: likelihood of investment
Independent Variable	(1)	(2)	(3)	(4)
Tariff Exposure	−0.868∗∗∗	0.513	−0.934∗∗∗	3.392
	(−7.35)	(0.56)	(−6.73)	(1.55)
Political Distance	−0.119∗	−0.0761	−0.0352	−0.0574
	(−1.68)	(−0.91)	(−0.24)	(−0.32)
BRI Partner	−0.0835	−0.343	−0.0387	−0.153
	(−0.48)	(−1.51)	(−0.17)	(−0.56)
SOE	0.0968	0.211∗∗	0.0980	0.201∗
	(1.02)	(1.97)	(0.93)	(1.75)
Political Distance*Tariff Exposure		−0.196		−0.625∗
		(−1.33)		(−1.88)
BRI Partner*Tariff Exposure		0.00174		−0.341
		(0.01)		(−0.62)
SOE*Tariff Exposure		−0.507∗∗		−0.610∗∗
		(−2.28)		(−2.20)
Number of Employees	0.000662	0.00228	0.0126	0.0142
	(0.02)	(0.05)	(0.26)	(0.30)
Profit Margin	−0.0531	−0.0554	−0.0637	−0.0661
	(−1.26)	(−1.31)	(−1.33)	(−1.38)
Share of Exports to US	−0.00787∗∗	−0.00806∗∗	−0.0105∗∗	−0.0106∗∗
	(−2.03)	(−2.07)	(−2.41)	(−2.42)
GDP	1.881∗∗∗	1.888∗∗∗	1.769∗∗∗	1.833∗∗∗
	(9.34)	(9.33)	(6.69)	(6.59)
GDP per capita	0.455∗∗∗	0.420∗∗∗	0.582∗∗∗	0.544∗∗∗
	(6.95)	(6.65)	(5.15)	(4.87)
Stringency of COVID-19 Measures	−0.354∗∗∗	−0.351∗∗∗	−0.369∗∗∗	−0.349∗∗∗
	(−3.83)	(−3.86)	(−3.62)	(−3.49)
Regime Type			0.00547	0.00426
			(0.12)	(0.09)
FDI Restrictiveness			−0.723	−0.609
			(−0.64)	(−0.54)
National Resource Rents			0.0585	0.102∗∗
			(1.26)	(2.05)
BIT	1.007∗∗∗	0.978∗∗∗	1.260∗∗∗	1.224∗∗∗
	(3.90)	(3.80)	(3.47)	(3.45)
Distance	0.709∗∗∗	0.726∗∗∗	0.543∗∗	0.594∗∗
	(3.19)	(3.25)	(2.23)	(2.38)
ASEAN	2.167∗∗∗	2.145∗∗∗	2.245∗∗∗	1.973∗∗∗
	(7.99)	(6.99)	(4.51)	(3.76)
South Asia	1.958∗∗∗	1.903∗∗∗	2.406∗∗∗	2.113∗∗∗
	(7.17)	(6.28)	(5.67)	(4.48)
European Union	1.376∗∗∗	1.384∗∗∗	1.481∗∗∗	1.535∗∗∗
	(8.89)	(7.93)	(6.99)	(6.93)
North America	−12.31∗∗∗	−12.12∗∗∗	−11.18∗∗∗	−11.48∗∗∗
	(−6.51)	(−6.48)	(−4.64)	(−4.67)
Latin America and Caribbean	0.521∗	0.248	1.374∗∗∗	0.864∗
	(1.81)	(0.80)	(2.84)	(1.68)
Middle East and North Africa	1.327∗∗∗	1.082∗∗∗	1.319∗∗∗	1.209∗∗
	(5.50)	(3.84)	(2.65)	(2.26)
ASEAN*Tariff Exposure		0.395		0.339
		(0.91)		(0.53)
South Asia*Tariff Exposure		0.0903		−0.430
		(0.20)		(−0.66)
European Union*Tariff Exposure		−0.0796		−0.111
		(−0.25)		(−0.28)
North America*Tariff Exposure		−0.747∗		−0.701
		(−1.83)		(−1.60)
Latin America and Caribbean*Tariff	Exposure	0.733		0.873
		(1.61)		(1.60)
Middle East and North Africa*Tariff	Exposure	0.900∗∗		−0.484
		(2.27)		(−0.39)
Constant	−14.54∗∗∗	−14.92∗∗∗	−13.76∗∗∗	−14.00∗∗∗
	(−7.48)	(−7.55)	(−7.03)	(−6.80)
N	658,375	658,375	274,557	274,557

Notes: Estimates generated using xtlogit in Stata 18.

t statistics in parentheses.

^∗  p < 0.1, ^∗∗ p < 0.05, ^∗∗∗ p < 0.01.

The results in table 1 show findings for the four hypotheses developed in this paper. On the first hypothesis, the estimates for Tariff Exposure are negative and statistically significant in Models (1) and (3). The variable has a positive sign in Models (2) and (4) but did not achieve statistical significance. These results provide some support to Hypothesis 1a, indicating that the Trump tariffs may have had a dampening effect on Chinese investors’ worldwide greenfield investments.

Table 1 shows that before the trade war, Political Distance from China, observed through UNGA voting records, did not have any significant effect on Chinese investment, even though it had the expected negative sign. The trade war appears at least to have reinforced the importance of political relations for Chinese investors’ locational decisions. Political Distance*Tariff Exposure is negatively signed in both models (2) and (4) and is statistically significant in the latter.¹⁴ This result provides some measure of empirical support for the “ally-shoring” argument (Hypothesis 2).

The third hypothesis in this study is that Chinese firms would be more likely to invest in BRI partner countries in the post-trade war compared to the pre-trade war period. The results of the analysis do not strongly support this geostrategic dimension to US-China relations. The main effects for BRI Partner are generally not statistically significant. For the trade war period, the estimates for BRI partner diverge for Models (2) and (4) and do not achieve statistical significance, indicating that the trade war has not greatly altered Chinese investment flows toward BRI partners.¹⁵ The negative estimate for BRI partner in the main models is contrary to expectations. The expansion of BRI partners and the consequent lack of variation may explain the lack of statistical significance. The negative estimate may reflect the public backlash against the BRI project as detailed in recent studies (Shi and Zhu 2018; Wang et al. 2022), which may have limited the BRI's potential to boost Chinese outward foreign direct investment.

Finally, the expectation in the fourth hypothesis is that SOEs should be more likely to engage in outward foreign direct investment compared to non-SOEs. The results reported in table 1 suggest that before the onset of the trade war, SOEs were more likely to invest overseas. The coefficient estimates are positive and statistically significant in Models (2) and (4). However, the results also show that, following the onset of the trade war, SOEs were significantly less likely to locate investments overseas. The estimates for SOE*Tariff Exposure are consistently negative and statistically significant in both models in which it was included. The overall decline in investment flows in the post-2018 period due to the heightened uncertainties in the business environment and changes in the Chinese government's general orientation and approach towards outward investments may have more disproportionately affected SOEs, which were the main actors involved in China's overseas investment activities in the previous era. A follow-on analysis of this effect can examine whether SOEs have shifted their production to concentrate more on the domestic market during the trade war years.

The estimates can also be interpreted in terms of marginal effects for the independent variables of interest—Tariff Exposure, Political Distance, BRI Partner, and SOE—while holding the values of all other variables at their mean.¹⁶ For Models (1) and (3), where a sector has been subject to US tariffs during the trade war, the probability of greenfield investment by a Chinese firm drops by 0.08 percent (Model (1)) and 0.17 percent (Model (3)). The magnitude of these changes can be interpreted in the context of the average probability of investment for the whole sample, which is very low, at 0.09 percent. A reduction of 0.08 percent (Model (1)) or 0.17 percent (Model (3)) in the probability of investment, relative to the sample average of 0.09 percent, is a significant drop. The marginal effect (0.607 percent) for Tariff Exposure in Model (4) indicates that the probability of investment increases more than six-fold over that of the sample mean (0.09 percent).

In terms of the effects of the control variables, the results yielded no evidence that firms with higher Number of Employees or those that are more profitable (Profit Margin) should be more likely to invest abroad. As for characteristics of the destination country, the estimates for GDP and GDP per capita show evidence that Chinese investors favor countries that have a larger market and that are more developed. Estimates for both variables are positive and statistically significant across model specifications.

The analysis also yielded evidence that geographical distance is a major determinant of Chinese firms’ investment patterns. Distance is positively signed and has achieved statistical significance in all model specifications. The positive sign for Distance indicates that the probability of Chinese greenfield investment is higher in distant lands. Unlike trade, where the distance to the destination market entails higher transport costs, investments are more likely in distant lands, where production may take place directly if the intention is to get closer to the destination market. Finally, the stringency of COVID-19 measures in the host country also has a negative sign and is statistically significant across model specifications, suggesting that the pandemic has had a major chilling effect on investment flows.

Among the institutional and political variables, the estimates for BIT with China are positive and statistically significant across model specifications. This result is in line with the findings of previous studies (Lu et al. 2021), which indicate that bilateral investment treaties boost Chinese foreign direct investment. Regime Type, measured in terms of a country's political rights and civil liberties as provided by the Freedom House database, is not statistically significant. This is consistent with expectations, given contrasting arguments and findings in the existing scholarship showing that, at best, Chinese investment is not affected by regime type. Estimates for Natural Resource Rents are positive and statistically significant in Model (4), a result that lends some support to the resource-seeking motivations behind COFDI. Furthermore, the analysis yielded no evidence that the regulatory barriers to investment in a destination country, as measured by FDI Restrictiveness, had any impact on the likelihood of investment, as the estimate is not statistically significant.

Finally, the results of the analysis also show regional patterns in Chinese FDI before and after the onset of the trade war. The estimates for the main effects show that Chinese investment was more likely in countries in ASEAN, South Asia, European Union, Latin America and Caribbean, and the Middle East and North African regions, relative to those in the reference group.¹⁷ North America, which includes the United States, Canada, and Bermuda, was less likely to receive Chinese FDI, again relative to the same reference group. These patterns were little changed during the trade war period. Except for North America, where we observe a negative and statistically significant effect, and the Middle East and North Africa where we observe a positive and statistically significant effect (Model (1)), the interaction effects for the regions are largely not statistically significant, indicating only weak changes in regional patterns of Chinese greenfield FDI. These results suggest that Chinese investors are not necessarily more likely to invest abroad except for the two regions mentioned above compared to the reference category during the trade war period.

Overall, the above results do not provide strong evidence that the trade war has induced a shift in China's investment flows toward BRI countries. They also suggest that the pandemic may have been a more proximate source of the overall decline in Chinese greenfield investment in recent years. However, they do provide some evidence indicating that the trade war has reduced the willingness of Chinese SOEs to engage in greenfield investment. It may also have led Chinese investors to increase investments in countries with good political relations with China, although the results are not overwhelming.

To test our hypotheses regarding the behavior of Chinese investors in the United States versus other world markets, we run regression models for these two different subsamples and report the results in tables 2a and 2b.¹⁸ In table 2a, Tariff Exposure is positively signed across model specifications, a result that lends some weight to the “tariff-jumping” argument. In table 2b, Tariff Exposure is negative and statistically significant in Models (1) and (3), suggesting that the trade war has had a negative effect on Chinese investment in non-US markets. However, it has a positive and statistically significant effect in Model (4). The results for the other main variables of interest are broadly consistent with those reported in table 1. Taken together, these results lent some tentative support to Hypothesis 1b regarding the impact of the tariffs on Chinese greenfield investments in the US market. They also yielded little support to Hypothesis 1c regarding the investment diversion effects of the tariffs on Chinese investments in third markets.

Table 2a.

US Market

	Dependent variable: likelihood of investment
Independent Variable	(1)	(2)
Tariff Exposure	0.220	0.314
	(0.64)	(0.64)
SOE	0.457∗∗	0.614∗
	(2.29)	(1.84)
SOE*US Tariff	−0.313	−0.650
	(−0.57)	(−0.96)
Number of Employees	−0.0285	−0.00574
	(−0.33)	(−0.05)
Profit Margin	−0.110	−0.367∗∗∗
	(−1.22)	(−3.15)
GDP	−1.713∗∗∗	−1.607∗∗∗
	(−5.94)	(−2.92)
Share of Exports to US		−0.000255
		(−0.02)
Constant	9.849∗∗∗	8.791∗
	(4.13)	(1.95)
N	13,198	5,725

Notes: Estimates generated using xtlogit in Stata 18.

t statistics in parentheses.

∗

p < 0.1, ^∗∗ p < 0.05, ^∗∗∗ p < 0.01.

Table 2b.

Non-US Markets

	Dependent variable: likelihood of investment
	(1)	(2)	(3)	(4)
Tariff Exposure	−0.869∗∗∗	0.304	−0.750∗∗∗	4.263∗
	(−7.66)	(0.33)	(−5.41)	(1.95)
Political Distance	−0.170∗∗	−0.135∗	−0.172	−0.173
	(−2.52)	(−1.72)	(−1.21)	(−1.00)
BRI Partner	0.224	−0.0418	0.354	0.295
	(1.36)	(−0.18)	(1.39)	(0.92)
SOE	0.0491	0.155	0.0362	0.135
	(0.49)	(1.36)	(0.32)	(1.10)
Political Distance*Tariff Exposure		−0.182		−0.787∗∗
		(−1.24)		(−2.36)
BRI Partner*Tariff Exposure		0.127		−0.510
		(0.38)		(−0.79)
SOE		−0.488∗∗		−0.632∗∗
		(−2.06)		(−2.07)
Number of Employees	0.00178	0.00315	0.0153	0.0172
	(0.04)	(0.07)	(0.29)	(0.32)
Profit Margin	−0.0131	−0.0161	−0.0151	−0.0176
	(−0.29)	(−0.36)	(−0.29)	(−0.34)
Share of Exports to US	−0.00809∗∗	−0.00814∗∗	−0.0113∗∗	−0.0113∗∗
	(−1.98)	(−1.98)	(−2.42)	(−2.41)
GDP	2.572∗∗∗	2.556∗∗∗	2.790∗∗∗	2.802∗∗∗
	(23.47)	(23.24)	(18.95)	(18.95)
GDP per capita	0.479∗∗∗	0.446∗∗∗	0.871∗∗∗	0.809∗∗∗
	(6.41)	(6.14)	(6.32)	(5.96)
Stringency of COVID-19 Measures	−0.315∗∗∗	−0.308∗∗∗	−0.400∗∗∗	−0.344∗∗∗
	(−3.37)	(−3.34)	(−3.74)	(−3.18)
Regime Type			0.0940∗	0.0786
			(1.68)	(1.38)
FDI Restrictiveness			−1.346	−1.165
			(−1.03)	(−0.87)
Natural Resource Rents			0.183∗∗∗	0.221∗∗∗
			(3.22)	(4.00)
BIT	0.488∗∗	0.473∗∗	0.712∗∗	0.687∗∗
	(2.54)	(2.48)	(2.57)	(2.47)
Distance	1.211∗∗∗	1.208∗∗∗	0.986∗∗∗	1.002∗∗∗
	(6.20)	(6.17)	(4.01)	(4.01)
ASEAN	2.838∗∗∗	2.850∗∗∗	2.889∗∗∗	2.597∗∗∗
	(11.63)	(10.18)	(4.78)	(4.21)
South Asia	2.322∗∗∗	2.267∗∗∗	2.681∗∗∗	2.415∗∗∗
	(8.35)	(7.33)	(5.57)	(4.62)
European Union	1.298∗∗∗	1.320∗∗∗	1.790∗∗∗	1.758∗∗∗
	(8.31)	(7.49)	(7.94)	(7.82)
North America	0.241	0.196	0.487	0.417
	(0.61)	(0.43)	(1.13)	(0.87)
Latin America and Caribbean	0.390	0.128	1.386∗∗	0.872
	(1.27)	(0.38)	(2.34)	(1.40)
Middle East and North Africa	1.830∗∗∗	1.618∗∗∗	1.597∗∗	1.447∗
	(7.97)	(6.03)	(2.06)	(1.83)
ASEAN*Tariff Exposure		0.147		0.367
		(0.32)		(0.55)
South Asia*Tariff Exposure		0.0478		−0.574
		(0.10)		(−0.87)
European Union*Tariff Exposure		−0.0926		0.284
		(−0.29)		(0.71)
North America*Tariff Exposure		0.161		0.114
		(0.19)		(0.13)
Latin America and Caribbean*Tariff	Exposure	0.717		0.867
		(1.53)		(1.50)
Middle East and North Africa*Tariff	Exposure	0.667		−0.545
		(1.60)		(−0.43)
Constant	−18.65∗∗∗	−18.78∗∗∗	−17.52∗∗∗	−17.40∗∗∗
	(−10.51)	(−10.39)	(−8.55)	(−8.19)
N	652,650	652,650	268,832	268,832

Notes: Estimates generated using xtlogit in Stata 18.

t statistics in parentheses.

^∗  p < 0.1, ^∗∗ p < 0.05, ^∗∗∗ p < 0.01.

Robustness Checks

Results from a series of additional analyses corroborate our main findings. First, additional analyses employed an alternative measure of a firm's exposure to the Trump tariffs, calculated as the average simple tariffs of the destination country on the Chinese firm's industry in a given year. Test results are consistent with those reported in table 1.¹⁹

Second, analyses also included country- and year-fixed effects to account for unobserved country or yearly effects which may have been caused by fluctuations in world markets or changes in the Chinese government's regulation of outward FDI. Test results do not alter the interpretation of our main findings.

Third, to take account of the effect of new restrictive barriers on Chinese FDI, such as the Foreign Investment Risk Review Modernization Act of 2018 (FIRMMA) in the United States, we re-ran the main models using two alternative measures from the Politics and Regulation of Investment Screening Mechanisms (PRISM) dataset (Danzman and Meunier 2023). Specifically, these measures include a dummy variable capturing whether the screening mechanism applies to greenfield investment Greenfield Covered and the interaction term between this variable and the number of mechanisms related to screening Number of Mechanisms × Greenfield Covered. In this set of tests, Greenfield Covered is negatively signed and statistically significant across model specifications, while Number of Mechanisms × Greenfield Covered is insignificant. The use of these alternative measures once again does not alter the interpretation of our main findings. However, since the PRISM data only covers 38 advanced industrialized countries, mostly OECD countries, we chose to present the results using the PRISM data as robustness checks (see Appendix) instead of in the main models.

Fourth, models in robustness checks took into account additional factors that may affect Chinese greenfield investment, such as the level of education, measured by the share of the population with post-secondary education, and infrastructure development, proxied by the length of railway lines in the host country. This set of analyses yielded a positive relationship between educational level and the main dependent variable. However, there is no evidence that infrastructure development in the host country is a main determinant of COFDI.²⁰ Importantly, our main findings have once again sustained the inclusion of these variables.

Fourth, the preceding analysis considers the possibility that Chinese greenfield investment may be more likely to flow to countries with good political relations with China since the beginning of the trade war. Another possibility that is worth considering is whether Chinese investors may have been less likely to seek out countries with good political relations with the US. To address this possibility, we re-estimated the main models adding the ideal point distance between the United States and the host country. Estimation results show that this variable is largely insignificant. The addition of this variable also does not affect the sign and significance of the main variables of interest.

The Appendix provides additional analyses (Tables 3a–6) comparing the effect of US tariffs on FDI in different regions of the world, labor-intensive versus capital-intensive industries, and how destination markets’ participation in global value chains affects Chinese greenfield FDI. Results remain largely consistent with our main findings.

Conclusion

This paper has analyzed the impact of the US–China trade war on the outward investment of Chinese firms and provided insights into the politics and economics of their foreign investment activities. The findings lent some support to the “tariff-jumping” argument, providing some preliminary evidence that the tariffs generated incentives for Chinese firms to increase investments in the US in order to evade the protectionist trade barriers. Firm-level analyses showed that the trade war has incentivized Chinese investors to invest in countries with good political ties with Beijing. While the analyses have found little evidence that the tariffs have shifted Chinese firms’ investment toward BRI countries in the short term, continued observation may yield a long-term trend of shifting investment toward BRI activities such as physical and digital infrastructure. Still another interesting finding is that the trade war is associated with a decline in the willingness of Chinese SOEs to engage in FDI. SOEs may have been most directly impacted by the downturn in bilateral political and economic relations, or the shifting pattern may reflect the government's recent re-emphasis of self-sufficiency and autonomous domestic development. A more reticent stance and stringent regulation of outward investments on the part of the central government may have contributed to this phenomenon. The findings concerning SOEs should thus be interpreted within the broader policy context in China. Future studies could engage in a more detailed examination of the extent to which the CCP's recent “dual-circulation strategy” may have galvanized Chinese investors to direct investment toward the domestic mfarket through the so-called backshoring, or the relocation of production toward the home market.

The results of the analyses point to the importance of international economic agreements such as BITs and host countries’ natural resource endowments as major determinants of COFDI (Li and Liang 2012; Shi and Zhu 2018; Lu et al. 2021). There is also strong evidence that the COVID-19 pandemic has had a major dampening effect on Chinese investment flows. The analyses do not provide strong evidence that the host country's regime type influences the locational choice of Chinese investors, a finding that presents some challenge to arguments emphasizing how MNCs from emerging market autocracies may be more likely to invest in other autocracies due to institutional proximity (Morck et al. 2008).

Moving forward, mergers and acquisitions (M&As) present a potentially fruitful area of comparison in future research. Tariff-jumping behavior with M&As, for example, may be more substantial than greenfield investment due to the lower demand for capital expenditures and other forms of investment for M&As, though they are more opportunistic as they tend to be more sensitive to temporary shocks such as currency crises (Davies, Desbordes and Ray 2018). To the extent that this study has found evidence of changes in firms’ greenfield investment behavior, it suggests that the tensions generated by the trade war may have had even broader implications for Chinese outward investments.

Footnotes

1

Linton Crystal Technologies Announces Groundbreaking US Manufacturing Investment, February 9,  2023.

2

Trina Solar Plans Third Factory in Vietnam to Boost Exports Amid US Tariff Measures, September 27, 2023 (Guarascio 2023).

3

Recent studies have explored the relative role of the state versus market actors in the implementation of the BRI as well as public responses to the initiative in partner countries. See, for example, Shi and Ye 2021; Wang et al. 2022.

4

For further reference, the Appendix includes a list of the top 20 Chinese investor firms by project count and value. The authors thank one of the reviewers for this suggestion.

5

Orbis Company Database and Orbis Cross-Border Investment dataset. Chinese firms are defined as per Orbis’ classification of global ultimate ownership (GUO). The global ultimate owner of a firm is the individual or entity found at the top of the corporate ownership structure with minimum controlling share of 50.01 percent of a given subject firm.

6

Available at https://ustr.gov/issue-areas/enforcement/section-301-investigations/tariff-actions.

7

United Nations General Assembly Votes and Ideal Points dataset. Accessed July 1, 2022.

8

Countries in China's Belt and Road Initiative: Who's In and Who's Out. Accessed July 22, 2022.

9

Orbis’ Cross-Border Investments database provides sectoral information using the NAICS classification. These were matched with product-level export data using the HS classification from the UNComtrade database. We employed the concordance package in R to match product codes.

10

Data for the above country-level variables were obtained from the World Development Indicators. Accessed July 15, 2022.

11

Oxford COVID-19 Government Response Trackers (OxCGRT). Accessed November15, 2023.

12

OECD, FDI Regulatory Restrictiveness Index (accessed August 20, 2022).

13

World Bank classification of countries and income groups by region.

14

The Joint test of Political Distance and Political Distance*Tariff Exposure in Model (2) is statistically significant at p < 0.10: chi2(2) = 5.08, Prob > chi2 = 0.0790. For Model (4), joint test is also statistically significant at p < 0.10: chi2(2) = 4.94, Prob > chi2 = 0.0848.

15

Adopting alternative measures of the BRI, such as coding it as 1 if two countries have signed an MOU or a similar agreement regarding the BRI in the previous two years and 0, otherwise does not change the above finding.

16

See Appendix for a table of marginal effects as well as interaction plots for the major independent variables of interest: Political Distance, BRI Partner, and SOE.

17

The reference group includes non-EU countries and countries in Central Asia and Sub-Saharan Africa.

18

The US models necessarily leave out dyadic variables such as BRI Partner and BIT due to the lack of variation in such variables.

19

Tariff data for the pre-trade war period are drawn from the World Integrated Trade Solution (WITS) database maintained by the World Bank

20

Test results are available from the authors upon request.

Author Biography

Ka Zeng is Professor of Political Science at the University of Massachusetts Amherst.. Her research focuses on China's role in the global economy. She is the author or co-author of Trade Threats, Trade Wars, Greening China, and Fragmenting Globalization.

Soo Yeon Kim is Associate Professor of Political Science at the National University of Singapore. Her research areas include the politics of trade, rising powers and global economic governance, and regional integration in Asia.

Notes

Authors’ notes: This article has benefited from feedback at multiple conferences and workshops, including the International Studies Association, the American Political Association, and International Political Economy Society annual meetings. Ka Zeng acknowledges the generous funding support provided by the Charles Koch Foundation Trade Policy Research Grant. Soo Yeon Kim acknowledges funding from the Singapore Ministry of Education. The authors are grateful to the ISQ editorial team and two anonymous reviewers for their insightful comments and guidance. All remaining errors are our own. The data underlying this article are available on the ISQ Dataverse at https://dataverse.harvard.edu/dataverse/isq.