OUP user menu

International Trade and US Relations with China

Benjamin O. Fordham, Katja B. Kleinberg
DOI: http://dx.doi.org/10.1111/j.1743-8594.2011.00135.x 217-236 First published online: 1 July 2011


US relations with China are critically important for the future of world politics. They are also a useful case in which to test the individual-level implications of the liberal commercial peace argument. A plausible case can be made on both sides of the claim that China poses a security threat to the United States. China's economy is growing far faster than the United States' economy, while the country remains a communist autocracy. At the same time, trade between the United States and China has expanded dramatically in the last three decades. Its dual role as a major trading partner and a growing international rival generates substantial uncertainty about China's future status as friend or foe. Using data from a recent survey by the Chicago Council on Foreign Relations, we find that economic interests help explain individual Americans' assessment of China as a threat and their views concerning hostile policies toward that country. Those who stand to benefit from trade with China hold more positive views of the country and oppose conflictual foreign policies with respect to it. Those whose incomes are likely to decline because of trade with China tend to take the opposite position on these questions.

The nature of relations between the United States and China is critically important for the future of the international system. Both scholars and policy makers often see the two states as potential rivals. Congress requires the Pentagon to issue a public assessment of Chinese military power each year, a report that is generally greeted with concern. 1 Similarly, the fact that China remains a communist autocracy leads some Americans to view its rapid economic growth with trepidation. Power transition scholars predict that a dissatisfied China could become a dangerous rival to the United States as the two approach equality in material capabilities (Lemke and Tammen 2003).

Liberal international relations theory nevertheless suggests one reason for optimism about future Sino–American relations: the enormous volume of trade that flows between two countries. Scholarship in this tradition points out that the exchange of goods and services leads to mutual welfare gains, economic interdependence, and increased communication, all of which arguably make armed conflict between the trading partners a costly and generally unattractive proposition (for example, Viner 1951; Oneal and Russett 1999; Russett and Oneal 2001). Critics of this optimistic view have noted that gains from trade may be distributed unevenly among trading partners, giving rise to concerns about shifts in relative power and influence (for example, Gilpin 1981). Similarly, asymmetric dependence on trade and a growth in the number of potentially contentious issues involving both trading partners may serve to aggravate rather than improve relations between these states (Hirschman 1980 [1945]; see also Barbieri 1996). Arguments in this vein suggest that under certain conditions, commercial relations contribute to the perception of trading partners as a threat.

Should we really expect trade to diminish potential tensions between the United States and China? We will attempt to answer this question by focusing on the impact of trade on individual attitudes. Our focus is thus on a source of foreign policy rather than on policy outcomes. Although previous empirical tests of the liberal claim that trade reduces conflict have focused on dyadic and state-level predictors of interstate conflict, the liberal argument also implies individual-level effects. If the opportunity cost of commerce interrupted by conflict matters, its effects should be apparent in the attitudes of those most directly affected by it. Trade certainly figures in the calculations of state leaders, but it is generally conducted by private individuals who experience the gains and losses from trade and its interruption. These attitudes offer an alternative way of testing liberal claims about trade and conflict. They are also substantively important. Public opinion about other states can influence foreign policy, especially in a democracy. Indeed, some formulations of the liberal argument emphasize the role of political pressure from commercial interests (for example, Russett and Oneal 2001, 130; Simmons 2003, 37–8).

With this in mind, we depart from previous research on the liberal peace by focusing on individual attitudes toward trading partners rather than conflict initiation. A significant amount of scholarship has sought to identify the sources of individual-level attitudes toward international trade. Though some have raised questions about these findings, most research on the topic has found that economic interests, generally represented by the ownership of human capital, shape support for free trade (for example, Scheve and Slaughter 2001; Beaulieu 2002; Mayda and Rodrik 2005; Hainmueller and Hiscox 2006). Here, we draw on insights from this work to test whether the same factors also drive individuals' attitudes toward policies that would influence trade relationships more indirectly, in the manner suggested by the liberal argument about commercial peace.

The fact that China is both a potential security threat and a major trading partner makes American attitudes toward the country a useful setting for testing claims about the influence of trade on threat perception as well as an important case in its own right. One potential source of hostility between the US and China is American support for Taiwan. While the United States does not formally recognize Taiwan as an independent state, it has nevertheless remained committed to preventing China from forcibly reincorporating the island even after normalizing relations in 1979. President Obama announced in January 2010 that the United States would sell $6 billion of arms to Taiwan in order to aid the island's defense against a potential Chinese attack. China objected strenuously to this announcement as it has to previous US assistance to Taiwan. In addition to freezing military relations with the United States, the Chinese government threatened to sanction American firms involved in the arms sale. 2

Threats of this sort are especially serious in light of China's rapidly growing economic power. China boasts the world's third-largest gross domestic product (GDP) in terms of purchasing power parity, and its colossal 8.4% growth rate was the fifth highest in the world in 2009 (Central Intelligence Agency 2010). While precise estimates vary, many scholars agree that Chinese GDP will surpass that of the United States within a generation. As Figure 1 illustrates, China has risen rapidly to become the United States' second largest trading partner. From essentially nothing during the early 1970s, Chinese imports grew to more than 2% of US GDP by 2006. American exports to China have also grown, although less dramatically. Just as Chinese military potential is enough to prompt serious discussion about a Chinese security threat, the volume of Chinese trade is sufficient to have real economic consequences for individual Americans.


American Trade with China, 1970–2007

More broadly, the Organization for Economic Cooperation and Development projects that China will account for approximately one-quarter of world GDP by 2030, greatly increasing the country's influence on the performance of the world economy and its geopolitical leverage (Maddison 2007:93). Already the Chinese are increasingly assertive in their economic as well as their political relations with other countries. For example, The Economist reports that although China had only brought two trade disputes to the World Trade Organization before 2008, it has initiated five since that time, including complaints against the United States and the European Union. 3 Finally, while it is in many ways unique in current world politics, the case of US–Chinese relations may provide a glimpse into the future of US relations with other rising economic powers, such as Brazil and India.

Trade and threat perception

The reason international commerce could reduce hostile attitudes among trading partners is clear enough in liberal arguments about its influence on interstate conflict. The conflict-reducing effects of trade depend on the benefits that flow from the trading relationship and the opportunity cost of policies that might disrupt it. National leaders care about the benefits of trade because they contribute both to national wealth and to domestic political support. The societal actors who gain from trade should oppose conflictual foreign policies that jeopardize these gains.

The distributive effects of trade suggest one important modification to the liberal argument at the individual level. Although international trade provides aggregate benefits to both states, it does not follow that all individuals within the two trading states also benefit. Trade creates domestic winners and losers. The implications of the liberal argument are reversed when one considers the effects of trade on the attitudes of the losers. Those who are harmed by international trade have reason to see trading partners as threats, and to support correspondingly hostile foreign policies. To the extent that hostile policies disrupt trade, they may be expected to produce results similar to a protective tariff. If economic interests indeed influence attitudes on security matters, the domestic political battle lines on these issues will resemble those in debates over trade protection.

Calculations of economic interest are not the only way in which trade might influence individual attitudes. A second pathway is affect. Psychological accounts of attitude formation stress the role of personal experiences in particular. It is plausible that individual experiences with the consequences of free trade, such as employment lost due to outsourcing or financial gain through international investment, may contribute to a person's view of the trading partner. To the extent that people generalize from such experiences, they are more likely to believe that those countries with which one has had positive encounters tend to do things one will like, such as support goals one favors and oppose things one would also oppose. Conversely, individuals tend to believe that countries with which one has negative experiences tend to “make proposals that would harm us, work against the interests of our friends, and aid our opponents” (Jervis 1976:118). Put differently, negative economic experiences can contribute to an “enemy image” of the trading partner.

Perceptions of economic self-interest may be based not only on one's own fortunes but also on the fortunes of the local community. Some regions contain more industries that gain from international trade and investment than others. The economic well-being of a region affects the individual materially, for example through its impact on taxation, social spending, and property values. It is possible that the effects of trade on their local community could influence individuals' foreign policy attitudes. Additionally, people who are not directly affected by the consequences of international trade may adopt positions as a result of frequent interactions with those who are. A region's economic interests are likely to be reflected in the local political climate for this reason. Opinion leaders such as local politicians and local media have incentives to advocate foreign policy positions that benefit the community. By doing so, they may generate additional support for these policies among the local population. We argue that economic self-interest and attitudes that stem from personal economic experience should influence individual threat perceptions by giving rise to such attitudes directly and also by making individuals more receptive to arguments about a Chinese threat which are already present in the public discourse.

If our line of argument is correct, then those who lose from trade will be more likely to perceive the trading partner not only as a detriment to their own economic welfare but also as a political-military threat. In past research, perceived threats of this kind have been shown to increase support among Americans for the use of armed force against the source of the threat (Jentleson 1992; Jentleson and Britton 1998). Using an experimental survey design, Herrmann, Tetlock, and Visser (1999) also find that certain situational factors are related to increased support among Americans for military intervention, most likely because they cause the opponent to be perceived as a threat. More broadly, their study lends support to the argument that enemy images play a significant role in mass foreign policy attitudes. To the extent that economic interests have a systematic effect on how individuals view the trading partner, they may also influence whether a person favors hostile rather than cooperative policies.

It is still possible that economic self-interest does not influence foreign policy attitudes in the way the liberal argument about the conflict-reducing effects of trade indicates. People may be unaware of the economic reasons they should expect to gain or lose from international trade. Moreover, there is a substantial body of research raising doubts about whether individual economic self-interest influences political attitudes at all (for example, Mansfield and Mutz 2009; Mutz and Mondak 1997; Sears and Funk 1990). On the other hand, individuals might acquire attitudes that reflect their self-interest through their identification with groups or opinion leaders who share their perspective. Individuals are likely to interact more often with people who share their interests, such as their family, coworkers, union officers, and others. This process will shape their perceptions in ways that line up with their actual economic interests. Ultimately, the influence of economic interest on individual foreign policy attitudes is an empirical question. In the remainder of this paper, we will use survey data to test whether the winners and losers in this expanding commercial relationship characterize Chinese power differently.

Research design

We will use the 2006 American Public Opinion and US Foreign Policy survey conducted by the Chicago Council on Foreign Relations (CCFR) to test hypotheses about the effect of trade on threat perception (CCFR 2006). 4 The CCFR has conducted similar surveys roughly every 4 years since 1974, and they have been widely used in research on public opinion and foreign policy (for example, Page and Bouton 2006). The survey we use was administered in July 2006 to a random sample of Americans aged 18 and over. Of the 1,570 contacted for the survey, 1,227 completed it, generating an enviable response rate of 78%. In addition to a wide range of questions on American foreign policy, the survey gathered demographic information essential for testing our argument, including respondents' income, education, and state of residence.

Measuring the distributive effects of trade is important for testing the individual-level implications of the liberal argument. The well-known Stolper–Samuelson theorem implies that owners of the relatively scarce factor of production in their country should see their incomes decline under free trade. At the same time, owners of the relatively abundant factor of production should realize income gains. Most survey research on the subject has found that the Stolper–Samuelson argument best accounts for attitudes toward free trade in the United States and we therefore adopt it here.

Our argument about the effect of trade on threat perception suggests several hypotheses about American attitudes toward China. First, Americans with relatively greater income and education should be less likely to view China as a threat. Compared with China, the United States is relatively capital-abundant and labor-scarce. The Stolper–Samuelson theorem indicates that Americans with more human and financial capital should see their incomes rise as a result of trade with a nation like China. Those with less human and financial capital, whose principal asset is labor, should see their incomes fall. These changes in factor prices should affect the real income of all holders of capital and labor, even if they are not directly involved in trade with China. The Stolper–Samuelson theorem has proven useful in predicting individual attitudes toward international trade (for example, Scheve and Slaughter 2001). The question to be answered here is whether the economic stakes it identifies also help explain threat perception.

Although our expectations about the effects of human and financial capital are clear enough in theory, our decision to use income and education to capture these effects requires some explanation. While income is not equivalent to wealth, individuals with higher incomes are more likely to have money available for investment. Education is commonly used to represent an individual's skill level. Highly skilled persons should enjoy benefits from trade with China similar to those that accrue to holders of financial capital. While education creates human capital, it may also affect individual attitudes through what might be called its “information effect.” In addition to providing people with economically valuable skills, it introduces them to ideas with important policy implications. Hainmueller and Hiscox (2006) found that the effect of education on support for free trade was more consistent with exposure to arguments in favor of free trade in college economics classes than with increasing human capital. Increasing education below the college level had relatively little effect, but there was a large increase in support for free trade among individuals with at least some college education. Although both aspects of education may influence attitudes toward China, the human capital effect is of primary interest here.

The implications of the information effect of education for American perceptions of a Chinese threat are not as clear as they were in Hainmueller and Hiscox's analysis of support for free trade. Economists are nearly unanimous about the potential value of free trade, so the likely effect of college economics classes on attitudes toward free trade is unambiguous. The information effect of education on assessments of a potential Chinese security threat is not so obvious. Basic information about China arguably makes claims of a Chinese threat more plausible. Many of the IR theories students encounter in college classes, including both realist accounts of the balance of power and liberal arguments about the democratic peace, suggest that China's rapidly growing economy and nondemocratic government are danger signs. Moreover, an increasing proportion of American elites view growing Chinese power with concern. A 2002 elite survey conducted by the CCFR found that 47% viewed China as a “critical threat” to American vital interests in the next decade, up from 16% in 1990. Only 6% said China posed no threat at all, down from 25% in 1990 (CCFR 2005). These attitudes among highly educated and influential people are bound to influence commentary on China in the US media.

If the human capital and the information effects of education run in opposite directions, then education might appear to have no effect. In order to distinguish between these two effects, we include a control for individual interest in news coverage of foreign policy. The CCFR survey includes the following question: When you follow the news these days, how interested are you in news about the relations of the United States with other countries? 5

We expect that greater exposure to media content should make individuals more hostile toward China, as news reports often focus on human rights violations, Chinese economic prowess, and its communist regime. While it does not fully overlap with the informational story suggested by Hainmueller and Hiscox, this variable provides a useful way to control for the effect of individuals' exposure to information about China and US–Chinese relations beyond the college classroom.

One potential concern about our account of individual interests in trade with China is that our hypotheses consider only the effects of trade on income through factor ownership. As Baker (2005) has pointed out, attitudes toward trade may be driven by one's role as a consumer as well as a producer. Baker argues that income is correlated with a preference for consuming high-skill goods. Trade between a skill-abundant country like the United States and a relatively skill-scarce country like China should increase the price of these skill-intensive goods. If this line of argument is correct, then high-income individuals' loss of real income from the consumption side of trade with China should lead them to express greater political hostility toward the country. As the empirical analysis in the remainder of this paper will make clear, we find little support for this claim. It is possible that the production side of trade is more important in shaping political attitudes. Hiscox (2006) found that question frames discussing job losses due to trade were more consequential than those emphasizing consumer prices. It is also possible income does not really capture the winners and losers from the consumption side of trade. Unfortunately, the CCFR survey does not include a variable that could better represent this aspect of the trading relationship. The consumption side of trade is potentially important but it does not invalidate the evidence presented here. To the extent that the consumption of traded goods mitigates differences in economic interests based on wealth and education, it should make statistical support for our hypotheses less likely.

A second set of hypotheses concerns regional economic interests. As we noted in the last section, individual self-interest is not the only potential pathway through which trade could influence threat perception. The aggregate economic impact of Chinese trade on their community could also influence individuals' opinions about a potential Chinese security threat. Even if they are relatively well endowed with human and financial capital, those who live in areas where local industries are threatened by Chinese competition could develop hostile attitudes toward China. These individuals are more likely to see the negative consequences of trade on others, even if they do not experience them firsthand. To the extent that proponents of trade protection use arguments about national security to make their case, those living in regions threatened by Chinese competition are probably also more likely to hear about these concerns in the local media. Conversely, those living in areas that benefit from Chinese trade are more likely to see and hear about these positive effects, and perhaps to adjust their views accordingly.

In order to capture the regional effect of trade, we computed indices of the export-orientation and import-sensitivity of the survey respondent's home state to Chinese trade. States are an imperfect representation of each survey respondent's home area, but more precise information about their location was not available. These indices were constructed using state-level data on sectoral employment and national data on sectoral output and trade with China. We first grouped exports and imports into North American Industrial Classification System (NAICS) sectors comparable with those used to record output and employment. We then created an export-orientation index for each of these sectors by dividing the value of exports to China by the total output of the sector. Similarly, we created an import-sensitivity index by dividing the value of imports by the sum of domestic output and imports in that sector. For each state, we then summed these indices across all the sectors for which we had data using each sector's share of overall employment in the state as a weight.

Although it is our primary interest, trade is clearly not the only systematic influence on threat perception. The models estimated below will also include an indicator of political ideology. Previous research indicates that conservatives place a higher priority on anti-communism and on national security more generally, so they should also be more likely to see China as a security threat (Hurwitz and Peffley 1987; Kaltenthaler, Gelleny, and Ceccoli 2004). Including this variable is useful because income and education are likely to be related to liberal–conservative ideology. If ideology has important effects on threat perception but is omitted from the model, income and education may proxy its effects, possibly creating spurious support for our hypotheses about the effect of economic interests. The estimated effect of ideology also offers a point of comparison for assessing the substantive significance of education and income.

Gender is another individual-level factor consistently found to influence both threat perception and attitudes toward free trade. A number of studies indicate that women are more likely than men to support protectionist trade policies. 6 Women may be more likely to become unemployed as a result of trade liberalization and subsequently face greater challenges finding new jobs at their previous income levels (Cardero, Barron, and Gomez 2000; Ozler 2000). At the same time, studies of threat perceptions related to war and terrorism indicate that women perceive greater security risks (Arian and Gordon 1993; Huddy, Feldman, Taber, and Lahav 2005). Like political ideology, gender is likely to be related to education and income. Omitting it from our model might thus create spurious support for our hypotheses about economic interests.

Age is yet another potential influence on threat perception that may also be related to education and income. Americans whose attitudes toward China were formed before the normalization of relations with the United States in the 1970s may have substantially darker views of Chinese intentions than those who have little or no memory of Cold War hostility between the two states. In order to capture this cohort effect, the models that follow will include variables indicating both age and age squared. This specification allows us to estimate the age at which hostility to the Chinese peaks rather than imposing an assumption about where the Cold War cohort begins and ends.

The extant literature suggests at least two potential determinants of threat perception that cannot be captured here due to the limitations of the survey instrument. Research indicates that cultural values such as national and ethnic loyalties as well as attitudes toward social justice have an independent and significant impact on public opinion regarding issues such as immigration and trade liberalization (for example, Eichenberg and Dalton 1993; Wolfe and Mendelsohn 2005). Despite its partial adoption of capitalism, China remains a communist regime whose history and culture differ dramatically from that of the United States. These differences may play a role in Americans' perceptions of Chinese dispositions and intentions. To some extent, ideology may capture their effects. The CCFR data do not permit more convincing controls for them.

Secondly, and more importantly, negative views of China among Americans may result in part from a broader economic anxiety due to stagnant wages, rising income inequality, and shrinking benefits such as pensions and healthcare. 7 The condition of the macro-economy has been shown to influence foreign policy attitudes; economic downturns tend to make voters more cautious regarding trade liberalization (Bowler and Donovan 1998) and in their support for European integration (Jenssen, Pesonen, and Gilljam 1998). As one of the newest and largest US trading partners, China may simply serve as a scapegoat for Americans' concerns about the state of the economy. The available data currently does not allow us to control for this possibility. Future research should include an analysis of the correlation between respondents' assessment of the US economy as a whole and threat perceptions.

Empirical results

The first survey item we will model taps assessments of whether China is a threat to the United States quite directly. Respondents were asked about the Chinese threat as part of a list of other potential concerns. The survey item reads as follows: Below is a list of possible threats to the vital interest of the United States in the next 10 years. For each one, please select whether you see this as a critical threat, an important but not critical threat, or not an important threat at all … . The development of China as a world power.

Aggregate responses to this question show substantial concern about a Chinese threat, though less than that indicated in the recent elite surveys discussed earlier. Overall, 36% said China posed a critical threat, 54% said the threat was important but not critical, 8% said China was not a threat at all, and 2% declined to answer. Table 1 presents the results of an ordered logit model of the individual responses to this item including the independent variables discussed above. For each of the categorical independent variables, we included a dummy for each category except the lowest. The table reports a Wald test for the joint significance of these dummies.

View this table:

Ordered Logit Model of Economic Interests and Threat Perception

Regional economic interests
Home state sensitivity to Chinese Imports24.34 (49.60)9.10 (51.65)
Home state export orientation to China256.88 (183.44)290.64 (188.08)
High school graduate−0.14 (0.21)−0.12 (0.21)
Some college−0.04 (0.23)−0.02 (0.23)
College graduate−0.27 (0.22)−0.21 (0.23)
Graduate degree−0.34 (0.30)−0.29 (0.29)
Wald test of joint significance χ 2 (4 d.f.) = 4.00 χ 2 (4 d.f.) = 2.80
Second-lowest quintile0.04 (0.18)0.05 (0.18)
Middle quintile−0.23 (0.17)−0.21 (0.17)
Second-highest quintile−0.13 (0.15)−0.12 (0.15)
Highest quintile−0.51 (0.18)*−0.49 (0.17)*
Wald test of joint significance χ 2 (4 d.f.) = 12.49* χ 2 (4 d.f.) = 12.31*
Slightly liberal0.49 (0.45)0.50 (0.45)
Moderate, middle of the road0.52 (0.41)0.53 (0.40)
Slightly conservative0.54 (0.36)0.53 (0.37)
Conservative0.99 (0.41)*0.99 (0.41)*
Extremely conservative0.31 (0.50)0.28 (0.51)
Wald test of joint significance χ 2 (5 d.f.) = 12.02* χ 2 (5 d.f.) = 11.52*
Interest in media coverage of American foreign relations
Hardly interested−0.03 (0.38)−0.03 (0.37)
Somewhat interested0.55 (0.30)*0.57 (0.29)*
Very interested0.90 (0.34)*0.89 (0.34)*
Wald test of joint significance χ 2 (3 d.f.) = 24.40* χ 2 (3 d.f.) = 23.32*
Female0.20 (0.11)0.22 (0.12)
Age0.08 (0.02)*0.07 (0.02)*
Age squared−0.0007 (0.0001)*−0.0006 (0.0002)*
Favorable attitude toward globalization−0.30 (0.14)*
Cut point 10.96 (0.82)0.78 (0.86)
Cut point 24.06 (0.83)3.88 (0.87)
Brant test of parallel regression assumption: H0: assumption is met χ 2 (21 d.f.) = 17.21 χ 2 (22 d.f.) = 17.05
  • (Notes. The dependent variable is the response to the survey question concerning the seriousness of the Chinese threat to American vital interests. Responses were coded as follows: 1–not an important threat at all, 2 –important but not critical threat, 3 –critical threat. Robust standard errors clustered on the state are reported in parentheses.

  • *Statistical significance at the p < .05 level in a two-tailed test.)


Ordered Logit Model of Economic Interests and Threat Perception

The results support our hypotheses concerning individual self-interest but not those concerning regional economic interests. Respondents with relatively higher incomes tended to hold less dire views of the Chinese security threat. Although the education variables are not jointly significant, this outcome is sensitive to the categories into which the variable is divided. Respondents with at least a college degree perceived significantly less threat than those without one. The effect of attentiveness to media coverage of foreign affairs was the opposite of education. The fact that exposure to more information about China tended to increase threat perception makes our claim that education embodies the effect of human capital rather than a proxy for economic ideas more plausible. On the other hand, the state export-orientation and import-sensitivity scores were not statistically significant. Because these variables could take on only 50 values, this null result is not especially surprising. It would be a mistake to conclude that the local impact of trade with China had no effect on threat perception based on the failure to reject the null hypothesis here. Nevertheless, our data offer no support for this claim.

Taken together, income and education had substantively large effects on assessments of the Chinese threat. Figure 2 depicts the predicted responses of several hypothetical respondents. Compared with respondents in the median category on education and income, those in the highest categories on both these variables were roughly twice as likely to say that China was not a threat at all, and about half as likely to say that it posed a “critical threat.” The effects of education and income were just as large as those of political ideology or interest in media reports on foreign affairs. In assessing the Chinese threat, a hypothetical respondent who said he or she was “very interested” in news about foreign affairs, or who said he or she was “conservative,” was roughly comparable with a respondent who fell into the lowest categories on income and education. This comparability across the three possible responses to the question is easy to see in Figure 2.


Predicted Assessments of the Chinese Threat by Hypothetical Respondents

Age also had substantial effects on perceptions of a Chinese threat. Our results indicate that threat perception was highest among older respondents who had longer memories of Cold War conflict between the United States and China. An 18-year-old respondent had 0.20 probability of seeing China as a “critical threat.” The effect of age on threat perception was roughly constant for respondents between the ages of 48 and 68, all of whom had a roughly 0.40 probability of seeing China as a “critical threat.” Most of these individuals would have been old enough to have their attitudes influenced by the hostility between the two states before Nixon's opening to China in 1971.

One possible objection to these results is that they reflect general attitudes toward international trade and globalization rather than considerations that are specific to China. This possibility is not necessarily a problem for the liberal claim that trade should promote peaceful attitudes toward trading partners. Those who benefit from international trade generally should also have more friendly attitudes toward most trading partners. However, in order to sort out the possibility that China simply benefits from generalized support for free trade, we estimated an additional model that included a variable indicating support for economic globalization in general. The CCFR survey included the following question: Turning to something else, do you believe that globalization, especially the increasing connections of our economy with others around the world, is mostly good or mostly bad for the US?

Our “favorable attitude toward globalization” variable was coded “1” if respondents said that globalization was “mostly good.” As the results in the second column of Table 1 indicate, this variable had the expected relationship with attitudes toward China, but it was not especially strong. The probability that a typical respondent like the one presented in Figure 2 would see China as a “critical threat” fell by roughly 0.02. The probability that this respondent would see China as “an important but not critical threat” increased by the same amount. The probability of seeing China as “not an important threat” remained unchanged. The other estimates were also essentially the same. Overall, the influence of trade interests on attitudes toward China was not simply a byproduct of support for globalization in general.

Does threat perception translate into support for more aggressive foreign policies? Two additional items from the 2006 CCFR survey help answer this question. The first item is fairly general: In dealing with the rise of China's power, do you think the US should:

  1. Undertake friendly cooperation and engagement with China

  2. Actively work to limit the growth of China's power.

In the aggregate, support for an aggressive policy toward China was less than threat perception. Overall, 67% preferred the more cooperative option, 27% selected the more aggressive policy, and 6% declined to answer.

The second item concerns the use of force to defend Taiwan, which is arguably the most likely scenario for a military confrontation between the United States and China. There has been some discussion about the circumstances that might justify using US troops in other parts of the world. I would like to ask your opinion about some situations. Would you favor or oppose using US troops: If China invaded Taiwan?

Most Americans declined to endorse this hypothetical use of force. Of those surveyed, 62% opposed it, 32 favored it, and 6% declined to answer.

Table 2 presents the results of two models of individual responses to these questions. In both models, the dependent variable was coded “1” if the respondent endorsed the more aggressive policy: option 2 in the first question, and the use of American troops to defend Taiwan in the second.

View this table:

Logit Model of Support for Policies toward China

Limit Chinese PowerDefend TaiwanDefend Taiwan
Regional economic interests
Home state sensitivity to Chinese Imports89.47 (47.39)−30.67 (71.18)−16.27 (68.79)
Home state export orientation to China10.19 (203.18)185.71 (201.67)178.62 (203.06)
High school graduate−0.61 (0.18)*0.08 (0.25)0.13 (0.26)
Some college−0.88 (0.20)*0.37 (0.23)0.38 (0.23)
College graduate−1.28 (0.21)*0.45 (0.26)0.37 (0.26)
Graduate degree−0.94 (0.29)*0.52 (0.30)0.45 (0.29)
Wald test of joint significance χ 2 (4 d.f.) = 43.58* χ 2 (4 d.f.) = 10.00* χ 2 (4 d.f.) = 6.07
Second-lowest quintile−0.05 (0.16)−0.08 (0.19)−0.11 (0.20)
Middle quintile−0.39 (0.23)0.04 (0.22)0.04 (0.22)
Second-highest quintile−0.26 (0.21)−0.42 (0.23)−0.45 (0.24)
Highest quintile−0.60 (0.20)*−0.50 (0.15)*−0.52 (0.16)*
Wald test of joint significance χ 2 (4 d.f.) = 10.64* χ 2 (4 d.f.) = 17.58* χ 2 (4 d.f.) = 15.48*
Slightly liberal0.13 (0.40)0.18 (0.30)0.14 (0.30)
Moderate, middle of the road0.33 (0.41)0.48 (0.26)0.50 (0.28)
Slightly conservative0.22 (0.43)0.21 (0.29)0.18 (0.30)
Conservative1.12 (0.40)*0.91 (0.31)*0.89 (0.34)*
Extremely conservative1.48 (0.57)*1.24 (0.37)*1.28 (0.40)*
Wald test of joint significance χ 2 (5 d.f.) = 52.58* χ 2 (5 d.f.) = 20.27* χ 2 (5 d.f.) = 19.56*
Interest in media coverage of American foreign relations
Hardly interested0.23 (0.47)0.39 (0.42)0.29 (0.44)
Somewhat interested0.11 (0.33)0.66 (0.39)0.49 (0.41)
Very interested0.22 (0.35)0.91 (0.39)*0.72 (0.42)
Wald test of joint significance χ 2 (3 d.f.) = 0.91 χ 2 (3 d.f.) = 8.53* χ 2 (3 d.f.) = 5.13
Female−0.15 (0.17)−0.23 (0.11)*−0.26 (0.12)*
Age0.03 (0.02)0.04 (0.02)*0.02 (0.02)
Age squared−0.0004 (0.0002)−0.0004 (0.0002)−0.0002 (0.0002)
Isolationism−0.75 (0.18)*
Constant−1.29 (0.75)−2.91 (0.69)*−2.14 (0.68)*
  • (Notes. The dependent variables are the responses to survey questions concerning support for efforts to limit the growth of Chinese power, and to deploy American troops to defend Taiwan. See the text for exact wording. Responses were coded “1” if the respondent supported these policies and “0” otherwise. Robust standard errors clustered on the state are reported in parentheses.

  • *Statistical significance at the p < .05 level in a two-tailed test.)


Logit Model of Support for Policies toward China

The results on the policy of limiting the growth of Chinese power indicate that respondents who benefited from trade with China were much less likely to endorse this course of action. As was the case with the threat perception question, education and income had substantial negative effects on respondents' support for this hostile policy. Figure 3 displays the most important results of the model graphically. 8 The panel on the right shows the results concerning the question on limiting Chinese power. Respondents with higher incomes and more education were substantially less likely to support such a foreign policy. The differences associated with slightly smaller than the difference between self-identified liberals and conservatives on the issue.


Effect of Education, Income, and Ideology on China Policy Attitudes

The results concerning the deployment of American troops to defend Taiwan are somewhat more complicated. As expected, high-income respondents were less willing to endorse such a military action. Contrary to our expectations, however, more educated respondents were actually more likely to endorse troop deployment. One likely explanation for this anomaly concerns the fact that the policy in question concerns not just generalized hostility to China, but the deployment of American “troops.” Less-educated Americans tend to take more isolationist positions on foreign policy questions in general, another pattern that can be linked to the consequences of international trade (Fordham 2008). A question that explicitly mentions the use of American troops, as this one does, is likely evoke these sentiments. The addition of a variable indicating support for isolationism strongly suggests that this is the case. 9 The results are presented in the third column of Table 2. The probability that an isolationist would support the deployment of troops to defend Taiwan was 0.30, substantially lower than the 0.48 probability associated with an otherwise identical non-isolationist respondent. The effect of education in this specification is not statistically significant.

The results concerning income, education, and ideology from the third model in Table 2 are depicted graphically in the left pane of Figure 3. As the graph indicates, the patterns associated with income and education we found in responses to other questions were weaker in shaping attitudes toward the use of American troops in Taiwan. Political ideology appears to have played a more important role in this case. Even so, high-income respondents were substantially less likely to support this policy option than were their low-income counterparts.

One possible problem with our inference that education and income show the effects of trade on political attitudes toward China is that these variables could indicate other things. We have already discussed the possibility that education may influence individual attitudes by exposing people to arguments and information that make China appear more threatening, as well as by providing them with human capital. This is not the only alternative effect of wealth or education, however. People with high incomes and more education may simply be more cosmopolitan and less prone to seeing other states as threats without direct provocation. Another possibility is that wealthy and educated people are less sensitive to international threats in general because wealth and education themselves may offer some protection.

There is probably no way to be absolutely certain that the effects of wealth and education on attitudes toward China stem from access to human and physical capital, but we can test the possibility that they stem from other considerations by examining attitudes toward China before trade with the country became a serious consideration. What all the arguments mentioned in the last paragraph have in common is that they should apply even in the absence of international trade. Finding that the wealth and education effects estimated here also existed before it was clear that China would become a major force in international trade would bolster the claim that these relationships represent something other than trade effects.

The CCFR has been surveying Americans about their foreign policy attitudes since 1975, asking many of the same questions in each such study. The 2006 survey used here included an unusually large number of questions about China, many of which were new. The question about the growth of China as a world power was first asked in 1990. By this time, trade with China was already growing rapidly. The Chinese began liberalizing their economy in 1978, a process that accelerated during the 1980s. It is necessary to examine data from before this time in order to test hypotheses about American attitudes toward China before trade was a major concern. Two questions from the 1975 and 1979 CCFR surveys recur on the 2006 survey. The first, which was introduced in the preceding section, concerns whether the United States should send troops in response to a Chinese invasion of Taiwan. The results in the preceding section indicate that wealthy respondents were substantially less likely to endorse this policy choice. This is consistent with our argument that these respondents' better access to physical and financial capital influenced their views about policies that would bring the United States into conflict with China. An identical question was asked in 1975. Support for sending troops was higher in the more recent survey. Only 16% of respondents endorsed this policy option in 1975, while 32% did so in 2006.

In the second recurring question, respondents were asked to rate the warmth or coolness of their feelings toward a series of states on a “feeling thermometer” scale of 0–100, with 50 degrees serving as a neutral point. The CCFR survey contained a similar question in 1979. 10 In spite of the slightly different lead-in to the question in the two surveys, as well as many changes in US–Chinese relations in the intervening 27 years, aggregate responses to this question were quite similar in 1979 and 2006. The mean temperature assigned to China in 1979 was 44. In 2006, it was 41. There was substantial (and similar) variance in both surveys, with a standard deviation of 21 in 1979 and 24 in 2006.

Table 3 presents the results of comparable models of responses to the 2006, 1979, and 1975 CCFR surveys. In order to facilitate comparisons between the older data and the 2006 study, some of the variables were coded somewhat differently from those used in the models presented in Table 2. The regional economic interest variables also could not be included here because comparable data from the 1970s were unavailable. Given the insignificance of these variables in the models presented in Tables 1 and 2, this omission is not likely to make much difference. The models of responses to the question about the deployment of American troops to defend Taiwan were estimated using logistic regression. The dependent variable was coded “1” if respondents said they favored the deployment of troops. The models of the feeling thermometer score for China were estimated using a tobit model to account for the fact that respondents could rate China no higher than 100 and no less than 0.

View this table:

Sources of Attitudes toward China in the 1970s and in 2006

Send troops to defend TaiwanChina feeling thermometer
High school graduate0.22 (0.27)0.07 (0.22)−5.54 (3.58)1.04 (2.66)
Some college0.34 (0.29)0.30 (0.23)0.73 (3.94)3.85 (2.78)
College graduate−0.45 (0.38)0.28 (0.25)−3.79 (4.21)8.32 (2.92)*
Graduate degree−1.02 (0.51)*0.37 (0.29)
Test of joint significance χ 2 (4 d.f.) = 15.01* χ 2 (4 d.f.) = 3.11 F (3, 583) = 2.93* F (3, 1061) = 4.20*
Second-lowest quintile0.19 (0.31)−0.13 (0.21)−5.48 (3.89)1.42 (2.53)
Middle quintile0.20 (0.24)0.03 (0.21)−4.42 (3.24)5.92 (2.60)*
Second-highest quintile0.11 (0.27)−0.44 (0.22)*−3.01 (3.52)3.34 (2.58)
Highest quintile0.64 (0.34)−0.53 (0.23)*−2.29 (5.14)7.35 (2.71)*
Test of joint significance χ 2 (4 d.f.) = 4.17 χ 2 (4 d.f.) = 10.57* F (4, 583) = 0.67 F (4, 1061) = 2.58*
Somewhat liberal−2.31 (5.21)−8.70 (4.03)*
Moderate, middle of the road0.25 (0.20)0.40* (0.19)−5.59 (4.94)−5.85 (3.62)
Somewhat conservative−7.16 (5.04)−7.86 (3.68)*
Very Conservative0.46 (0.21)*0.58 (0.19)*−14.59 (5.45)*−22.74 (5.14)
Test of joint significance χ 2 (2 d.f.) = 4.78 χ 2 (2 d.f.) = 9.06* F (4, 583) = 3.80* F (4, 1061) = 5.65*
Interest in media coverage of American foreign relations
Somewhat interested0.04 (0.23)0.26 (0.22)4.82 (2.94)3.83 (2.49)
Very interested0.16 (0.22)0.52 (0.23)*8.91 (2.87)*0.30 (2.61)
Test of joint significance χ 2 (2 d.f.) = 0.84 χ 2 (2 d.f.) = 6.52* F (2, 583) = 5.47* F (2, 1061) = 2.62
Female−0.27 (0.15)−0.28 (0.14)*−0.61 (1.83)3.31 (1.63)*
Age−0.01 (0.03)0.02 (0.02)0.12 (0.24)−0.57 (0.29)*
Age squared−0.0002 (0.0004)−0.0002 (0.0002)−0.01 (0.002)0.005 (0.003)
Isolationism−0.39 (0.19)*−0.74 (0.16)*
Constant−1.42 (0.69)*−1.56 (0.67)*48.65 (7.67)*48.94 (8.29)*
  • (Notes. The dependent variables are the responses to survey questions concerning support for deploying American troops to defend Taiwan, and for feelings toward China on a “feeling thermometer” scale of 0–100. See the text for exact wording. The dependent variable in the model of support for troop deployment to defend Taiwan was coded “1” if respondents supported this choice, and “0” otherwise. The model was estimated using logistic regress. The model of the feeling thermometer score was estimated using tobit regression to model the fact that responses were limited at 0 and 100.

  • *Statistical significance at the p < .05 level in a two-tailed test.)


Sources of Attitudes toward China in the 1970s and in 2006

The results concerning the defense of Taiwan indicate that the patterns found in 2006 were not evident in 1975. Neither income nor education had the same effect on attitudes toward military conflict with China in the years before China became a major trading partner. Income was unrelated to respondents' answers to the question about the defense of Taiwan. Education was associated with opposition to the use of American troops in 1975, but not consistently. Only respondents with graduate degrees showed this effect. Those with only high school diplomas or some college were, if anything, more hostile to China than those with less than a high school education. These results offer little support for claims that income and education proxy cosmopolitanism rather than economic interests because these effects should have been evident before trade with China became a major consideration.

The results for the China feeling thermometer point to the same conclusion. Not surprisingly, the patterns found in 2006 are very similar to those evident in responses to the question about whether the growth of Chinese power posed a threat to American interests. Respondents with more education and income tended to hold “warmer” views of China. Respondents in the highest categories on these variables rated China roughly 15 points higher than respondents in the lowest categories. This pattern is not evident in the data from the 1979 survey. In responses to that question, income was not statistically significant. The education variables were jointly significant, but had inconsistent effects not unlike those found in responses to the 1975 question on the defense of Taiwan. Respondents in the highest category on the education variable appeared to have somewhat colder views of China than those in the lowest category, though individuals in the second lowest category had the most negative views.

Overall, the CCFR data from the 1970s offer little evidence that the relationship between education, income, and attitudes toward China predates the growth of trade between the two countries. If considerations like cosmopolitanism or relative insulation from international threats drove these relationships, they should have been apparent before China became a major player in the international trading system. Instead, there appears to have been little or no relationship between education or income and attitudes toward China in the 1970s, an outcome that is consistent with our claim that these relationships are primarily due to international trade.


Our analysis of American attitudes toward China, the country's second largest trading partner, suggests that international trade can have a substantial effect on individual threat perception and support for hostile foreign policies. Evidence from the 2006 CCFR survey indicates that Americans who benefit from trade with China are less likely to view China as a threat, even though there may be plausible grounds for doing so. The effects are not uniformly positive, however. International trade holds aggregate benefits for both nations in the relationship, but it still creates sub-national winners and losers. The losers hold substantially darker views of the trading partner than do the winners. We found that Americans with relatively low levels of education and income were more likely to see China as a threat, and to support conflictual foreign policies for containing Chinese power, than were their wealthier and more highly educated fellow citizens.

These findings suggest that the liberal claim about the pacifying effects of trade must be qualified. Trade can indeed influence political relations between trading partners like the United States and China, but its effects depend on whether the winners or the losers from trade control policy choice. So long as beneficiaries of trade are at the helm, relations between trading states may remain cordial. If those who are harmed by trade come to power, there are no grounds for expecting this benefit. The evidence reviewed here concerns only the US side of the relationship but the theoretical argument applies to China as well. Whether trade will foster cooperative relations between the two countries depends in part on domestic political outcomes in each one.

While the evidence presented here points to some previously overlooked complications in the effect of trade on foreign policy attitudes, it also raises at least three questions that we cannot answer with the data we have used here. First, do trade-induced changes in individual attitudes actually influence foreign policy outcomes? Policy makers are usually a step removed from the immediate economic effects of international trade. The process through which the winners and losers from trade influence policy makers' positions is thus somewhat different from the one examined here. Other research, such as Bartels' examination of the impact of public opinion on the Reagan military budget (Bartels 1991), indicates that public opinion has influenced both congressional consideration of security issues and actual policy outcomes. However, representatives might not always reflect the aggregate interests of voters in their district on every issue. The wealthier and better educated segment of the population favoring better relations with China might be smaller than the poorer and less-educated group that holds more hostile attitudes yet possess other political resources with which to make their voices heard. Representatives might also ignore relatively narrow interests when building their reelection constituency, especially if their district is fairly diverse (Bailey and Brady 1998). Further research is needed to test whether trade-induced patterns in public opinion have comparable effects.

Second, do the domestic losers from the international trading relationship actually become more hostile toward the trading partner? We know that the winners and losers differ in their attitudes toward the trading partner, but we do not know how either group would view that country in the absence of trade. It seems likely that the winners would be less positive, and the losers less negative toward the other country in this case, but it is also possible that the divergence in threat perception between these groups results primarily from the emergence of more positive or more negative attitudes by one group. We have tested our theoretical argument using a 1-year snapshot of the relationship between two prominent trading states, the United States and China. Future research may be able to answer these questions by incorporating additional trading partners or by examining a single dyad over time.

Third, does trade influence threat perception in every case, or only when other features of the relationship make the claim that the trading partner is a threat plausible? Our data indicate that income gains and losses from trade have substantively large effects on American attitudes toward China, but this is only one case. The same considerations might play a role in future US relations with other emerging market economies such as India and Brazil. On the other hand, it is possible that other factors might prevent a state from being seen as a threat at all, regardless of the income effects of trade. Chief among them, as noted earlier, are affinities based on shared political culture and normative values (for example, Russett 1993). Perhaps for this reason, there is little serious discussion of a European threat in the United States, even though American trade with the European Union also generates domestic winners and losers. Future research that includes additional trading partners could reveal whether trade influences threat perception in the absence of conditions that make it plausible to view another state as a security threat.

The fact that trade has aggregate benefits for the nation as a whole suggests that the winners should be able to compensate the losers while still leaving themselves better off than they would be under autarky. Such transfers of wealth might mitigate the patterns in individual threat perception found here. Unfortunately, these transfers are quite limited in many countries, including the United States. Scheve and Slaughter (2007) have warned that the failure to spread the gains from trade widely within American society threatens the liberal trading order, which depends heavily upon American participation. Our results suggest that this failure might also put at risk the prospects for a cooperative relationship between the United States and China.


We are grateful to the Chicago Council on Global Affairs for providing us with the survey data used in this paper. They bear no responsibility for our interpretations and conclusions. For their helpful comments and suggestions on this paper, the authors thank Joanne Gowa, Ole Holsti, Kenneth Scheve, and participants at the Conference on Domestic Preferences and Foreign Economic Policy at Princeton University, the world politics workshop at Binghamton University, and the 2008 annual meeting of the International Studies Association. Any remaining errors and omissions are our responsibility.


  • 1 The latest such assessment suggests that Chinese advances in military technology may negate the existing American military edge (Thom Shanker, “US Sees Chinese Military Rise,”The New York Times, March 26, 2009).

  • 2 The Economist, “By Fits and Starts,” 6 February 2010, pp. 25-28.

  • 3 The Economist, “When Partners Attack,” 13 February 2010, p. 77.

  • 4 The CCFR has since changed its name to the Chicago Council on Global Affairs. We will use the older name in order to avoid confusion. The older name was current at the time of the survey we use and appears in the documentation we cite.

  • 5 As with the other survey items in our analysis, we will include dummies for all but one of the response categories: (i) very interested; (ii) somewhat interested; (iii) hardly interested; and (iv) don't follow the news. We will treat the fourth of these responses as our base category.

  • 6 Examples include O'Rourke and Sinnott 2001; Kaltenthaler et al. 2004; Mayda and Rodrik 2005. For a focused analysis, see Burgoon and Hiscox 2004.

  • 7 The Economist, May 17, 2007.

  • 8 Except as noted in the figure, the predicted probabilities assume a respondent in the median category on the categorical variables and the mean value on the continuous variables. This individual would be a 47-year-old male in the third income quartile, with some college but no degree, and a moderate political ideology.

  • 9 The question used is similar to that used by Fordham (2008) from the American National Election Study. Respondents were asked the following: “Do you think it will be best for the future of the country if we take an active part in world affairs or if we stay out of world affairs?” Those who responded that the United States should “stay out” were coded as isolationists.

  • 10 The exact wording of the 1979 and 2006 questions varied. The 2006 study was conducted on a computer, where respondents read the following question and were presented with a list of countries in random order: “Please rate your feelings toward some countries and peoples, with one hundred meaning a very warm, favorable feeling, zero meaning a very cold, unfavorable feeling, and fifty meaning not particularly warm or cold. You can use any number from zero to one hundred, the higher the number the more favorable your feelings are toward that country or those people. If you have no opinion or have never heard of that country or those people, leave the box blank and move on to the next question.” The 1979 study was conducted through face-to-face interviews. Respondents were handed a card with a drawing of a thermometer ranging from 0–100 and asked the following: “Next I'd like you to rate the same countries on this feeling thermometer. If you feel neutral toward a county, give it a temperature of 50 degrees. If you have a warm feeling toward a country give it a temperature higher than 50 degrees. If you have a cool feeling toward a country, give it a temperature lower than 50 degrees.”


View Abstract