2.1 Public Opinion and Foreign Policy

Domestic public opinion plays an important role in shaping the political calculus of foreign policy decision-making. In democracies such as the United States, politicians have direct political incentives to respond to public sentiments. While an early literature asserted that voters tend to focus more on domestic issues and care less about foreign affairs (e.g., Almond Reference Almond1950; cf. Aldrich et al. Reference Aldrich, Gelpi, Feaver, Reifler and Sharp2006) and, consequently, politicians pay little attention to public opinion when formulating foreign policy (e.g., Cohen Reference Cohen1973; Page and Bouton Reference Page and Bouton2006), later scholarship showed that many citizens do, in fact, hold meaningful and coherent foreign policy attitudes. Political psychologists have demonstrated how two dimensions – militant internationalism and cooperative internationalism – can effectively structure individuals’ foreign policy attitudes (e.g., Holsti Reference Holsti2004; Wittkopf Reference Wittkopf1990). These general dimensions are grounded on distinct motivational goals, moral foundations, personal values, and worldviews (Kertzer Reference Kertzer, Leonie Huddy, Sears, Levy and Jerit2023; Rathbun Reference Rathbun2020).

Others point to the sensitivity of public opinion to elite cues. While there is evidence that American citizens’ foreign policy attitudes can be swayed by co-partisan elites (e.g., Berinsky Reference Berinsky2009; Guisinger and Saunders Reference Guisinger and Saunders2017), recent research showed that Americans do not blindly follow their leaders in their policy preferences (Coppock Reference Coppock2023; Fowler and Howell Reference Fowler and Howell2023; Tappin, Berinsky, and Rand Reference Tappin, Berinsky and Rand2023), including their attitudes on foreign issues toward China (Yeung and Xu Reference Yue2025). More generally, Kertzer and Zeitzoff (Reference Kertzer and Zeitzoff2017, 554) offered experimental evidence to show that “rather than simply being shaped from the top down, public opinion is a function both of individuals’ social context and their preexisting attitudes toward the kind of role America should play in the world.”

Recent research suggests that political elites do respond to public opinion on foreign policy. Public opinion influences foreign policy through multiple pathways (Baum and Potter Reference Baum and Potter2015); even foreign public opinion can influence policymakers (Goldsmith and Horiuchi Reference Goldsmith and Horiuchi2012; Goldsmith, Horiuchi, and Matush Reference Goldsmith, Horiuchi and Matush2021). Administering an elite survey experiment to members of the Israeli parliament, Tomz, Weeks, and Yarhi-Milo (Reference Tomz, Weeks and Yarhi-Milo2020, 119) found that “[p]arliamentarians were more willing to use military force when the public was in favor and believed that contravening public opinion would entail heavy political costs.” Building on Tomz and colleagues’ experiment, Lin-Greenberg (Reference Lin-Greenberg2021, 1) replicated their finding with a sample of US military officers, concluding that “public opposition makes military leaders less likely to recommend the use of force.” Leveraging unpublished polling information about British public opinion on UK military presence in the South China Sea, Chu and Recchia (2022, Reference Chu and Recchia1874) conducted an experiment on British Members of Parliament (MPs) and found that “MPs who received the polling information, compared with those who did not, voiced opinions closer to those of the public.” Extending the policy domain from security to economic issues, Peez and Bethke (Reference Peez and Bethke2025, 1) found that US foreign policy practitioners became more supportive of increasing economic sanctions when they “used contemporary public polling highly supportive of increasing sanctions as an information treatment.” These experimental studies highlight how public opinion may causally affect foreign policy decision-making in democracies.

What about nondemocracies? Does the Chinese government, for instance, care about public opinion? Existing research on authoritarian politics shows that even closed autocracies can be responsive to public opinion (Lueders Reference Lueders2022). Chen, Pan, and Xu (Reference Chen, Pan and Yiqing2016) conducted an online field experiment and found that local governments in China were responsive to citizen demands expressed online. Similarly focused on online public demands, Kornreich (Reference Kornreich2019, 547) showed that “public comments have an impact upon policy revisions” (see also Chen and Li Reference Chen and Yiran2024; Distelhorst and Hou Reference Distelhorst and Hou2017), although such responsiveness can be selective based on individuals’ social identities, such as ethnicity (Distelhorst and Hou Reference Distelhorst and Hou2014) and residency (Su and Meng Reference Zheng and Meng2016). Such “authoritarian responsiveness” in China (Heurlin Reference Heurlin2016) – be it substantive (Lueders Reference Lueders2022) or simply performative (Ding Reference Ding2022) – helps the regime enhance the legitimacy that is critical to its survival (Dickson Reference Dickson2016; Truex Reference Truex2017).

Indeed, the Chinese government invests an extraordinary amount of time and money to monitor and manage mass opinion (King, Pan, and Roberts Reference King, Pan and Roberts2013, Reference King, Pan and Roberts2017; Roberts Reference Roberts2018), including under President Xi Jinping (Waight et al. Reference Waight, Yuan, Roberts and Stewart2025). Chinese leaders pay a great deal of attention to social media and read daily briefings on the ongoing activities in Chinese social media (Masterson Reference Masterson2024, 2). Over two million people were reportedly employed as public sentiment analysts and workers, consistent with President Xi Jinping’s instruction to party cadres that “winning or losing public support is an issue that concerns the CPC’s survival or extinction” (Weiss and Dafoe Reference Weiss and Dafoe2019, 963–64). On trade policy in particular, Steinberg and Tan (Reference Steinberg and Tan2023, 147) documented that “China’s central government sought public input in 2018 on how it should respond to Trump’s tariff threats (Ministry of Commerce 2018). China’s leadership subsequently justified its decision to retaliate with tariffs by noting that ‘it had Chinese public opinion on its side’ (Buckley 2018).” More generally, Wang (Reference Wang2024) traced China’s propaganda strategies across different interstate disputes and demonstrated how it deployed media campaigns to mobilize and pacify domestic public opinion. Given the importance of public opinion to the domestic politics that underlie international relations, a large body of research has focused on and investigated American and Chinese citizens’ foreign policy attitudes in recent years.Footnote ¹¹

2.2 Relative Gains and Trade Preferences

While conventional wisdom in international political economy holds that individual trade preferences are predominantly guided by economic self-interest, a sizable body of research also suggests trade preferences are not merely guided by the direct material self-interest of individuals (e.g., Brutger and Rathbun Reference Brutger and Rathbun2021; Guisinger Reference Guisinger2017; Hainmueller and Hiscox Reference Hiscox2006; Lü, Scheve, and Slaughter Reference Lü, Scheve and Slaughter2012; Rho and Tomz Reference Rho and Tomz2017).Footnote ¹²

Particularly, trade preferences can be affected by relative gains considerations – how well the home state fares compared to other states (Powell Reference Powell1991; Snidal Reference Snidal1991a, 389). Under relative gains thinking, a win–win trade policy where domestic gains outweigh the trade partner’s gains would be preferred to a win–win policy where foreign gains outweigh domestic gains. This thinking is particularly likely when people believe that relative economic advantages can be translated into relative military advantages, or that the trade partner is a potential security threat (Snidal Reference Snidal1991b, 703; Waltz Reference Waltz1959; Reference Waltz1979). Analyzing these dynamics at the state (rather than individual) level, Gowa and Mansfield (Reference Gowa and Mansfield1993) found that free trade took place more frequently within, but not across, military alliances, suggesting that states take relative gains into consideration on the issue of trade.Footnote ¹³

Relative gains concerns can affect trade preferences at the individual level through sociotropic considerations. A large body of research has established the importance of sociotropic motivations in policy preferences. Early work by Kinder and Kiewiet (Reference Kinder and Kiewiet1981) found a sociotropic voting pattern in American politics, where voters tend to focus on collective-level instead of individual-level information. Sociotropic influence is relevant to how political perceptions form and may even outweigh egotropic considerations (Hainmueller and Hopkins Reference Hainmueller and Hopkins2015; Hearn Reference Hearn2020; Mutz Reference Mutz1992; Sears and Funk Reference Sears, Funk and Jane J. Mansbridge1990). Mansfield and Mutz (Reference Mansfield and Mutz2009) found that trade preferences depended on how people perceived the economic impact of international trade on the United States as a whole, and that in-group favoritism could affect opinion on trade (see also Mutz and Kim Reference Mutz and Kim2017; Mutz and Lee Reference Mutz and Lee2020). Relative gains considerations operate here insofar as people would prefer a trade policy that brings greater gains to the domestic in-group relative to the foreign out-group. Carnegie and Gaikwad (Reference Carnegie and Gaikwad2022) found that American and Indian citizens both preferred trading with allies to trading with adversaries, underscoring the role of geopolitical concerns – a form of sociotropic considerations – in determining individual trade preferences (see also Bush and Prather Reference Bush and Prather2020; Chen, Pevehouse, and Powers Reference Chen, Pevehouse and Powers2023; Herrmann, Tetlock, and Diascro Reference Herrmann, Tetlock and Diascro2001).

Despite the role of nonmaterial factors in trade preferences and the geopolitical implications of the US–China trade war, only a few studies have investigated its public opinion dynamics in the context of great power competition. Schweinberger (Reference Schweinberger2022) found that “promise-keeping trade narratives” – accusations that the other country has a poor track record of upholding trade promises – decreased both American and Chinese respondents’ support for US–China trade cooperation. Steinberg and Tan (Reference Steinberg and Tan2023) primed Chinese respondents about US protectionism amid the trade war and found that reminders of US protectionist measures increased Chinese respondents’ support for raising trade barriers. Focusing instead on the American public, Bulman (Reference Bulman2022) found that informing American respondents about “commercial peace theory” – that countries trading more with each other are less likely to have military conflict – increased support for removing trade war tariffs on China. While these studies show that political rhetoric, negative reciprocity, and security concerns could play an important role in shaping individual preferences for trade protectionism, whether positive reciprocity would hold among American and Chinese citizens in the shadow of a US–China trade war remains an open question.

But before theorizing and testing how different forms and strategies of reciprocity shape the domestic feasibilities of trade war de-escalation, we first draw on our experimental data – collected at the outbreak of the US–China trade war – to evaluate relative gains sentiments among American citizens (Yeung and Quek Reference Yeung and Quek2022). In doing so we offer direct evidence of how relative gains considerations tamp down individual preferences for trade, serving as a domestic political force underlying the ongoing trade war.

2.3 Relative Gains Sentiments in the American Public

Do citizens care about relative gains in trade?Footnote ¹⁴ Are relative gains concerns amplified by geopolitical considerations? To tackle these questions, we administered a survey experiment to 850 American citizens in August 2018 – amidst the early phase of the US–China trade war. We recruited respondents via Amazon Mechanical Turk, a crowdsourcing platform that has been widely used in social science (Berinsky, Huber, and Lenz Reference Berinsky, Huber and Lenz2012; Coppock Reference Coppock2019).Footnote ¹⁵ Although our data draw from a convenience sample, our findings are robust to reweighting the survey data according to national population benchmarks in age, gender, education, and party identification.

Our experiment began with a scenario where the United States proposed to remove import limits on China or Country X. Respondents were told that the removal of the import limits would increase both American citizens’ welfare and Chinese or Country X’s citizens’ welfare (i.e., a win–win trade policy). Half of our respondents were then given an additional vignette stating that the gains enjoyed by Chinese or Country X’s citizens would be significantly greater than the gains enjoyed by American citizens; the others were not given the additional vignette. This additional vignette imputes relative gains to the foreign country. If relative gains affect trade preferences, we should expect respondents to be less supportive of removing import limits on China or Country X when given the additional vignette. The design thus provides a direct test of the effect of relative gains on trade opinion.

After reading the vignettes, respondents were asked how much they supported the given trade policy on a seven-point scale from 0 (“strongly oppose”) to 6 (“strongly favor”). In short, our experiment leveraged a 2 × 2 factorial design, where we randomized the trade partner (China vs. Country X) and relative gains implications (with vs. without direct relative gains concerns against the trade partner). If relative gains are important, support for the policy should decrease when the additional vignette is introduced. The decrease should be particularly substantial when the trade partner is China, if trading with China invokes particularly strong relative gains concerns, given the brewing US–China great power rivalry.Footnote ¹⁶

We found that relative gains considerations strongly affected Americans’ trade preferences, particularly when the trade partner was specified as China. Respondents’ support for the policy decreased by 0.71 on the seven-point scale when they learned that Chinese gains would outweigh American gains (p = 0.0004, n = 435). Support also decreased, but not as much, when the trade partner was Country X: respondents’ support decreased by 0.29 when they learned that Country X’s gains would outweigh American gains (p = 0.0828, n = 415).Footnote ¹⁷

Who were most likely to hold such relative gains sentiments? In our survey, we administered an item that tapped into nationalism. Specifically, we asked whether respondents agreed with the following statement: “Everyone should support their country even when it is wrong.”Footnote ¹⁸ For our exploratory analysis, we classify those who agreed with this statement as nationalists. Because nationalism captures a sense of national dominance over foreign states (de Figueiredo and Elkins Reference de Figueiredo and Elkins2003; Gries et al. Reference Gries, Qingmin Zhang and Cai2011; Johnston Reference Johnston2017; Ko Reference Jiyoung2023; Kosterman and Feshbach Reference Kosterman and Feshbach1989; Yeung, Wang, and Quek Reference Yeung, Wang and Quek2024), there are theoretical reasons to believe that relative gains considerations – particularly against China – are especially salient among American nationalists when it comes to trade.

We found that this is indeed the case. In the control conditions where respondents were not informed about the relative gains implication about the removal of import tariffs, nationalists’ and non-nationalists’ support for the win–win trade policy did not substantially differ regardless of the US trade partner. However, differences in approval emerged when respondents learned about the negative relative gains implications for the United States. Where the trade partner was Country X, nationalists were less supportive of the win–win trade policy than non-nationalists by 0.93 points on a seven-point scale (p = 0.0017, n = 206). Where the trade partner was China, their gap in support was even wider, reaching 1.29 points on a seven-point scale (p = 0.0002, n = 219). Among nationalist respondents, their level of support for removing import tariffs against China was only 2.76 (95% CI [2.17, 3.35], n = 62) in light of relative gains concerns, even when they were explicitly told that the United States would derive absolute gains from the trade policy.

How would these dynamics play out in the current political climate? Although an exact answer to this question would require replicating the experiment, it is likely that the role of relative gains considerations in influencing American trade preferences would be even more pronounced given the growing security concerns about China’s rise (see Copeland Reference Copeland2024; Pomeroy Reference Pomeroy2025). The latest research shows that Americans from across the political spectrum – not only Republicans but also Democrats – overwhelmingly favor hawkish responses to be taken by the US government to counteract the China threat (Yeung and Xu Reference Yeung and Weifang2026). More generally, there has been a surge of nationalism and populist sentiments in the United States, where protectionist foreign policies – and politicians who propose or endorse these policies – have become increasingly popular among the American public. Public support for nationalist foreign policies, particularly against the United States, has also been prevalent in China (Liu and Li Reference Liu and Xiaojun2024; Weiss Reference Weiss2019). While de-escalating the trade war is imperative given its potential to evolve into a new Cold War (Christensen Reference Christensen2023; Lake Reference Lake2018; Rodrik Reference Rodrik2019), the prospect of de-escalation appears dim in light of the domestic forces we found.

3.1 The Nature of Reciprocity

Human beings have innate preferences for reciprocity. Behavioral economists find that individuals exhibit reciprocity preferences in gift-exchange and trust games (e.g., Berg, Dickhaut, and McCabe Reference Berg, Dickhaut and McCabe1995; Fehr, Kirchsteiger, and Riedl Reference Fehr, Kirchsteiger and Riedl1993).Footnote ¹⁹ Neuroscientists provide physiological foundations for these empirical regularities, showing how different regions of the human brain relate to reciprocal behavior (Delgado, Frank, and Phelps Reference Delgado, Frank and Phelps2005; van den Bos et al. Reference van den Bos, Wouter, Westenberg, Rombouts and Crone2009).

Such intrinsically motivated behavior is prosocial rather than egoistic. In fact, reciprocity motives in human beings are so ingrained that even lame-duck legislators, who are some of the less likely individuals to adhere to reciprocity norms and who have no direct incentives to engage in reciprocity, “continue to repay their party leaders’ favors even when they are about to leave politics” (Fong Reference Fong2023, 1036). In international politics, Kertzer and Rathbun (Reference Kertzer and Rathbun2015) showed that reciprocity plays an important role in shaping bargaining behavior (see also Chilton Reference Chilton2015; Chu Reference Chu2019; Kertzer et al. Reference Kertzer, Powers, Rathbun and Iyer2014) and that such preferences for reciprocity are predicated not on selfish motives but on intrinsic human quality.Footnote ²⁰

The experimental literature in International Political Economy (IPE) also suggests that concerns about reciprocity shape individuals’ foreign policy preferences. On foreign direct investment, Chilton, Milner, and Tingley (Reference Chilton, Milner and Tingley2020) found that American and Chinese respondents were “more likely to oppose foreign acquisitions when the foreign firm’s home country did not provide reciprocal market access” (131). On climate change, Bechtel and Scheve (Reference Bechtel and Scheve2013) and Tingley and Tomz (Reference Tingley and Tomz2014) found that institutional arrangements promoting norms of reciprocity were more conducive to amassing public support for global climate arrangements. On trade, Kiratli and Ertan (Reference Kiratli and Ertan2024) and Steinberg and Tan (Reference Steinberg and Tan2023) respectively found that Turkish and Chinese respondents favored retaliation against the United States if it engaged in protectionism, which is consistent with the logic of negative reciprocity.

These studies are insightful, but they focus largely on negative reciprocity for deterring negative foreign policy actions, rather than positive reciprocity. Among the few exceptions that touch on positive reciprocity, the focus is on how it compares with negative reciprocity (Chilton, Milner, and Tingley Reference Chilton, Milner and Tingley2020) or how it helps to sustain cooperation (Kiratli and Ertan Reference Kiratli and Ertan2024), rather than how it ends a vicious cycle of conflict. Importantly, our theoretical and empirical foci are on how people respond to the reciprocation – rather than the action–in the action–reciprocation sequence. Examining individual responses to reciprocation thus fills in the missing piece, improving our understanding of the domestic political outcomes each country will likely face in different policy pathways that involve different reciprocation strategies or conciliatory gestures.

Studying positive reciprocity is important for both substantive and theoretical reasons. Substantively, it maps onto the processes of interstate rapprochement and conciliation, rather than conflict escalation that is characterized by negative reciprocity. De-escalating hostile relations is not a unilateral effort; it is feasible only if conciliatory gestures are reciprocated (Mattes and Weeks Reference Mattes and Weeks2022). Theoretically, how individuals approach negative and positive reciprocities should differ. Psychologists have found “an inherent asymmetry between reciprocating to giving [positive reciprocity] versus taking [negative reciprocity]” (Keysar et al. Reference Keysar, Converse, Wang and Epley2008, 1280). This asymmetry exists because “in cases of negative social exchange, an action that is perceived to be generous by an actor is likely to be perceived as less generous by a receiver and therefore deserving of a more selfish response” (Keysar et al. Reference Keysar, Converse, Wang and Epley2008, 1285). Yet in the positive domain, the norm of reciprocating in kind – or even more than the initial favor – is less strong.

Theories and findings on sender–receiver gaps and asymmetric inferences in the international relations literature would also lead us to expect different effects, depending on whether the home or foreign country is the sender or receiver. Recent research highlights the sender–receiver gap (e.g., Cho and Chao Reference Cho and Chao2025; Kollner Reference Köllner2025; Quek Reference Quek2016; Quek and Ni Reference Quek and Jiaqian2026; Yarhi-Milo, Kertzer, and Renshon Reference Yarhi-Milo, Kertzer and Renshon2018) – a systematic bias in how the same signal is perceived between the two ends of the signaling relationship, whereby the logic of costly signaling (Fearon Reference Fearon1997; Spence Reference Spence1973) operates significantly more strongly at the sender’s end than at the receiver’s end. The sender systematically treats and weighs the signal more seriously than receivers do, and assumes receivers do the same – but they don’t.Footnote ²¹ Bell and Quek (Reference Bell and Quek2026) and Kertzer, Brutger, and Quek (Reference Kertzer, Brutger and Quek2024) show that people exhibit an asymmetry in the inferences they draw from their own signal or behavior relative to the inferences they draw from the same signal or behavior of others. They see, for example, the same action (e.g., naval deployment or military investment) as offensively and malignly motivated when carried out by the other side, but as defensively and benignly motivated when carried out by their own side. Evidence suggests these asymmetries are sticky and difficult to correct, even when people are made to engage in perspective-taking (Kertzer, Brutger, and Quek Reference Kertzer, Brutger and Quek2024) or to think about the reciprocal nature of their own behavior (Bell and Quek Reference Bell and Quek2026).

3.2 Semi-Reciprocity, Balanced Reciprocity, and Super-Reciprocity

In addition to whether individuals respond favorably to positive reciprocation under great power rivalry, another open question in international relations is how what we call “unbalanced reciprocity” operates between states.

Beyond the concept of balanced reciprocity, unbalanced reciprocity plays an important role in international relations because the equivalence of benefits is not always attainable in interstate exchanges. Perceived unbalanced reciprocity – our theoretical focus – is potentially ubiquitous. Yet, despite “the impossibility of determining exact equivalence” in world politics (Keohane Reference Keohane1986, 8), the existing literature on reciprocity gives little theoretical and empirical guidance on unbalanced reciprocity in international relations (for an exception, see Goldstein Reference Goldstein1995).

We propose that unbalanced reciprocity exists in two logically orthogonal forms: super-reciprocity and semi-reciprocity. The delineation derives from straightforward logic. For two parameters x and y, if x ≠ y, then either x > y or x < y.

In the context of a trade war, super-reciprocity is exemplified when the sender state reduces tariffs on the receiver state by 10 percent, while the receiver state responds by cutting tariffs on the sender state by 15 percent in return. Correspondingly, semi-reciprocity is exemplified when the receiver state decides to reduce tariffs on the sender state by 5 percent in return.Footnote ²³

Behavioral economists, social psychologists, and political scientists have focused predominantly on balanced reciprocity. However, in trade politics where states have significant leeway to decide how much to reciprocate, unbalanced reciprocity can kick in. In the Uruguay Round of General Agreement on Tariffs and Trade (GATT), “some territories, such as the United States, received deeper concessions than they gave; other territories, such as India, South Korea, and Thailand, gave much deeper concessions than they received” (Steinberg Reference Steinberg2002, 366). Several studies have shown that developing countries, in general, made more concessions to developed countries than they received (Goldin, Knudsen, and Mensbrugghe Reference Goldin, Knudsen and van der Mensbrugghe1993; Harrison, Rutherford, and Tarr Reference Harrison, Rutherford, Tarr, Will Martin and Winters1996). Finger, Reincke, and Castro (Reference Finger, Reincke and Castro1999) revealed wide variations between participating countries in terms of how much they gave and took. For example, India accepted a reciprocal tariff reduction of about 6 percent in return for a cut of slightly more than 1 percent by its trading partners on Indian exports; by contrast, Hong Kong, which had no tariffs to reduce, obtained an average reduction of nearly 2.5 percent on its exports (Finger, Reincke, and Castro Reference Finger, Reincke and Castro1999, 16). Many interesting questions spawn from these observations. For example, facing such unbalanced reciprocity, how would domestic constituencies in the receiver state react? And how would people in the sender state respond?

Further interesting theoretical implications emerge when we compare unbalanced reciprocity with balanced reciprocity. While both unbalanced and balanced reciprocity imply the willingness to reciprocate, unbalanced reciprocity activates another social norm: inequality aversion.Footnote ²⁴ Following our parametrization, inequality aversion kicks in if there is a gap between x and y and if an individual incurs a utility loss due to this gap (Fehr, Bernhard, and Rockenbach Reference Fehr, Bernhard and Rockenbach2008, Reference Fehr, Bernhard and Rockenbach1079; Fehr and Schmidt Reference Fehr and Schmidt1999, 822). Just as humans have intrinsic preferences for reciprocity, they also have strong preferences for equality. A large number of replications of the ultimatum game have shown that individuals often reject offers that deviate from an equal distribution of resources (Fehr and Schmidt Reference Fehr and Schmidt1999; Güth, Schmittberger, and Schwarze Reference Güth, Schmittberger and Schwarze1982). Looking directly into the human brain, neural scientists further show that inequality aversion preferences are prevalent among human beings (Tricomi et al. Reference Tricomi, Rangel, Camerer and O’Doherty2010). Conducting controlled experiments on young children and subjecting them to one-shot interactions with anonymous partners (in order to shut down selfish motives as a confounder), Fehr, Bernhard, and Rockenbach (Reference Fehr, Bernhard and Rockenbach2008) found that while most children at age 3–4 reveal selfish tendencies, most children at age 7–8 make choice that avoid advantageous or disadvantageous inequality. Past research, therefore, has shed important light on the prevalence, physiological foundation, and evolutionary roots of egalitarianism.

Preferences for equality are general and can shape foreign policy attitudes (Kertzer et al. Reference Kertzer, Powers, Rathbun and Iyer2014), including in a trade setting. Descriptively, Mutz and Kim (Reference Mutz and Kim2017) found that Americans predisposed to embrace group equality are more supportive of a win–win trade policy that also benefits the trade partner. Powers et al. (Reference Powers, Kertzer, Brooks and Brooks2022) showed that American adults who hold stronger beliefs about equality are more supportive of free trade, even after controlling for standard measures of foreign policy orientations such as militant internationalism, cooperative internationalism, and isolationism. Experimentally, Lü, Scheve, and Slaughter (Reference Lü, Scheve and Slaughter2012) found that American and Chinese citizens are more supportive of protectionism when they believe that it will help equalize domestic distributions of income. Brutger and Rathbun (Reference Brutger and Rathbun2021) showed that Americans are averse to inequality in international trade, especially when the trade agreement will make the United States fare worse than its trade partner.

Under unbalanced reciprocity in trade, inequality aversion comes into play in two ways. First, consider super-reciprocity. Here, the receiver state makes a greater trade concession than it receives from the sender state. Such reciprocity not only generates relative gains to the sender state but also produces unequal outcomes between the two states. For the receiver state, it triggers disadvantageous inequality aversion, “the loss individuals incur because others have better outcomes than they do” (Lü, Scheve, and Slaughter Reference Lü, Scheve and Slaughter2012, 639). For the sender state, it provokes advantageous inequality aversion, “the loss individuals incur because others have worse material outcomes than they do” (Lü, Scheve, and Slaughter Reference Lü, Scheve and Slaughter2012, 639). A recent meta-analysis (Nunnari and Pozzi Reference Nunnari and Pozzi2022) suggested that inequality aversion in both forms have been routinely documented across different disciplines (economics, neuroscience, psychology), experimental tasks, and subject populations.

Compared to balanced reciprocity, super-reciprocity generates additional relative gains and inequality aversion considerations. To preempt confusion, here we will refer to sender as “initiator” and receiver as “responder.” Under super-reciprocity, the responder state should suffer public backlash because both egoistic and other-regarding preferences would point to the same direction: Relative losses and disadvantageous inequality aversion would make individuals less supportive of super-reciprocity than balanced reciprocity. The theoretical prediction for the initiator state, however, is less clear. While super-reciprocity generates relative gains to the initiator state, it also produces advantageous inequality aversion for domestic citizens through the same process. Public preferences for super-reciprocity versus balanced reciprocity, therefore, will depend on the relative forces of egoistic and other-regarding preferences among individuals in the initiator state.

Now consider semi-reciprocity. Here, the responder state makes a smaller trade concession than it receives from the initiator state. Thus, the roles of the responder (receiver) and initiator (sender) states are reversed. Citizens in the responder state enjoy relative gains and confront advantageous inequality aversion, whereas citizens in the initiator state incur relative losses and face disadvantageous inequality aversion. Egoistic and other-regarding preferences would point to the same direction for the initiator state: Citizens should prefer balanced reciprocity to semi-reciprocity. For the responder state, however, the theoretical prediction is ambiguous: It depends on the relative influence of relative gains and inequality aversion preferences.

When unbalanced reciprocity triggers inequality aversion, would individuals prefer super-reciprocity, semi-reciprocity, or no reciprocity at all? Research suggests that inequality aversion and reciprocity are both intrinsic in human nature, but when they collide with one another, how would people react?

The literature offers little guidance, as it has neither theorized nor tested how individuals respond to super-reciprocity and semi-reciprocity. These forms of unbalanced reciprocity, however, are not simply thought experiments; they are prevalent in the real world, including in the context of international trade.Footnote ²⁵ The Uruguay Round, as described above, is a case in point. The history of US trade liberalization, dating back to the passage of the Reciprocal Trade Agreement Act in 1934, provides another illustration: “Congress did not stipulate the degree to which reductions in the US tariff should be matched by foreign tariff reductions. This issue was left to the executive, and indeed greater access to the US market via lower tariffs was not initially matched by equal access to European markets, which remained more restricted due to quotas and exchange controls” (Irwin Reference Irwin and Jagdish Bhagwati2002, 81).

Crucially, how people reciprocate – and how they respond to how the other side reciprocates – is at the heart of strategies for conflict de-escalation. De-escalation often requires one side to send a de-escalatory signal and the other side to reciprocate. While specific strategies of reciprocation differ in their empirical details, they fall within three general theoretical forms – balanced reciprocation, semi-reciprocation, and super-reciprocation – which may elicit different perceptions and reactions on the sender and receiver sides. To understand how the dynamics of de-escalation play out in de-escalating a trade war, it is important to delineate the different strategies of reciprocity available to each side of a trade war and assess the reactions they elicit.

3.3 Past Experimental Approaches

Despite a large literature on how reciprocity shapes preferences and decision-making, the question of how people respond to unbalanced reciprocity in international relations remains an empirical puzzle.

Ultimatum games from behavioral economics offer experimental evidence that norms of reciprocity exist among humans. In an ultimatum game, one player proposes how to divide a sum of money and the second player either accepts or rejects the proposal; if rejected, both players receive nothing. A large number of experimental replications have found that “unfair” offers are often rejected by the second player, even at a cost to themselves, to punish the proposer for perceived unfairness (see Fehr and Schmidt Reference Fehr and Schmidt1999; Güth, Schmittberger, and Schwarze Reference Güth, Schmittberger and Schwarze1982). Modifying the game and making it “analogous to how the first wave of bargaining theorists thought about war,” Kertzer and Rathbun (Reference Kertzer and Rathbun2015, 622) found a pattern of “behavioral assimilation” from a sample of US college students, such that prosocials (i.e., those with high social preferences pretreatment) also stopped cooperating and behaved like proselfs when faced with a proself partner. This strand of literature focuses on a negative response to unfairness – responding negatively to an “unfair” action – rather than the positive reciprocity that we study here.

Bechtel, Genovese, and Scheve (Reference Bechtel, Genovese and Scheve2019) fielded conjoint experiments in four advanced industrial democracies (including the United States) to investigate the role of social norms in shaping public support for climate policy. They showed each respondent pairs of international climate agreements, randomized the number of participating countries and the share of global emissions represented by participating countries (which tapped into levels of reciprocal cooperation), and asked respondents to choose between each hypothetical pair. They found that higher levels of reciprocal cooperation increased support for a climate treaty.

Chilton, Milner, and Tingley (Reference Chilton, Milner and Tingley2020) ran a survey experiment on US and Chinese samples that randomized the extent to which an unnamed foreign country recently made acquisitions of its companies harder. They found that lower reciprocal market access increased American and Chinese respondents’ support for making it harder for companies from that foreign country to acquire domestic companies. In a follow-up study, they presented American respondents the following information: “On a scale of 0 to 10, where 0 is no restrictions and 10 is an absolute ban on foreign ownership, in the past, Country A has had a score of [0 / 3 / 6] for the ability of US companies to buy companies in Country A. Today this country is now a [0 / 3 / 6].” After randomizing the foreign scores, they “told the respondents that the United States is currently a 3 on this scale and asked the respondents whether the United States should make it easier or harder for companies from Country A to buy American companies” (Chilton, Milner, and Tingley Reference Chilton, Milner and Tingley2020, 146). Using this design, they found that respondents appeared more willing to engage in negative reciprocity compared to positive reciprocity.

Schweinberger and Sattler (Reference Schweinberger, Sattler, Henry Gao, Raess and Zeng2023) administered cross-national survey experiments to American, Australian, and German respondents to study whether and how they “value reciprocity as a key cooperative principle in international trade towards different trading partners” (334). Using a factorial design, they randomized not only whether the foreign trade partner “[strongly decreases / does not change / strongly increases] its tariffs on goods produced in the US” (343) but also who the trade partner is, including ally countries such as the UK and Japan as well as rival countries, namely China and Russia. They found that respondents were generally willing to support reducing tariffs on the trade partner’s goods when the trade partner was an ally initiating a tariff reduction on US goods, but such preferences for positive reciprocity decreased when the trade partner was an adversary.

Brutger and Rathbun (Reference Brutger and Rathbun2021) conducted a survey experiment to examine how unequal tariff concessions made by the United States and another country affect Americans’ support for trade agreements. For a hypothetical trade agreement that the United States is considering negotiating with an unnamed trade partner, they “randomly varied the concessions each side made, such that each side could make a 30 percent, 60 percent, or 90 percent cut to their tariffs” (Brutger and Rathbun Reference Brutger and Rathbun2021, 886). They found that support for a trade agreement was equally high when the United States had a balanced or favorable agreement, although respondents felt that the balanced agreement was the fairer one. In addition, support and perceptions of fairness were the lowest when the United States had an unfavorable agreement.

Together, these experiments elucidate how norms of reciprocity influence individual support for international cooperation. Unpacking the properties of different forms of positive reciprocity, however, was neither the theoretical nor the empirical focus of these studies. Crucially, these studies do not conceptualize positive reciprocity as a factor for ending a vicious cycle of conflict. They focus on how individuals respond to their country’s action (e.g., in a potential climate or trade agreement), instead of their country’s reciprocation, in the action–reciprocation sequence.

Motivated by the policy relevance of strategies of reciprocation in conflict de-escalation; the scarcity of research on unbalanced reciprocity in international relations; and the lack of scholarly attention to how people respond to reciprocation in the action–reciprocation sequence, we theorize and test experimentally how balanced reciprocity, semi-reciprocity, and super-reciprocity influence domestic public dynamics in a trade war between the two largest economies in the world – the United States and China.

3.4 Research Design

To investigate how different forms of reciprocation impact domestic reactions amid a trade war, we fielded simultaneous surveys in the United States and China from April to May 2023 and embedded parallel experiments within the surveys. We worked with our survey partners to construct national samples that targeted the American and Chinese population benchmarks in sex, age, ethnicity, and geography (see Online Appendices §A and §B). We follow our pre-analysis plan to exclude speeders, defined as those spending less than 40 percent of the median response time in the full survey, from our analysis.Footnote ²⁶ The remaining samples consist of 1,913 American citizens and 1,601 Chinese citizens.

3.4.1 Experimental Procedures

In our experiment, respondents considered a future scenario in which the United States and China are in a trade war. While a trade war was ongoing at the time of the experiment (2023), a scenario set at the present would introduce confounds from the politics of the day. Such confounds would be better mitigated if respondents were to consider the scenario beyond the tenure of the then-current Biden administration.

All respondents read this preamble:

• The year is 2025.

• The United States and China are engaged in a trade war.

• Each side has imposed import tariffs (i.e., taxes) on the other, and the trade war is costly to both countries.

Having established this context, our experiment randomized two elements: First, whether the United States or China was the sender state that initiated the de-escalatory signal by reducing its tariffs on the other side; second, how much the other country (i.e., the receiver state) reciprocated by reducing tariffs on the sender state. In four experimental conditions, the United States was the initiator state that removed tariffs on US$100 billion worth of Chinese goods. China then responded by removing tariffs on [US$150 billion / US$100 billion / US$50 billion] worth of US goods, or by not removing any tariffs on US goods. Specifically, respondents randomly assigned to these conditions read:

In the latest development:

• The US calls on China to work together toward ending the trade war. As a start, the United States will remove tariffs on US$100 billion worth of Chinese goods.

• [If assigned to the balanced-reciprocity condition] China says that it welcomes the removal of tariffs on Chinese goods. In return, China will remove tariffs on US$100 billion worth of US goods.

• [If assigned to the super-reciprocity condition] China says that it welcomes the removal of tariffs on Chinese goods. In return, China will remove tariffs on US$150 billion worth of US goods.

• [If assigned to the semi-reciprocity condition] China says that it welcomes the removal of tariffs on Chinese goods. In return, China will remove tariffs on US$50 billion worth of US goods.

• [If assigned to the no-reciprocity condition] China responds that it will not change its policy toward the United States. All existing tariffs imposed on the United States will remain in force.

In four other experimental conditions, China was the initiator state that removed tariffs on US$100 billion worth of US goods. The United States then responded by removing tariffs on [US$150 billion / US$100 billion / US$50 billion] worth of Chinese goods, or by not removing any tariffs on Chinese goods. Because we simply reversed the sender and receiver states, the vignettes for these experimental conditions were identical to the above vignettes, except that China’s action (reaction) became US action (reaction), and vice versa. Thus, each experiment had a 2 × 4 factorial design (Figure 1).

Figure 1

Parallel dyadic experimental design

Note: In our dyadic experimental design, China’s action was the treatment for the US, and the US’s response was the treatment for China, and vice versa.

Across all eight conditions, respondents were asked, through two-stage branching, whether they approved or disapproved of the way their country handled the situation, and how much they approved or disapproved – resulting in a seven-point scale ranging from 0 (strongly disapprove) to 6 (strongly approve).Footnote ²⁷ We avoided asking respondents whether they approve or disapprove of their leader or the government, due to the political environment in China and the need to mitigate political sensitivity biases. Because substantive conclusions do not change if we instead use binary approval as the dependent variable (see Online Appendix §D), we report the main results based on percentage approval for ease of interpretation.

Online Appendix §C presents our survey instruments in English and in Chinese. Because our experiments were parallel and dyadic, linguistic precision was important. We followed a rigorous protocol to ensure our translations were accurate. After we developed the English version of the survey instrument, two native speakers – bilingual in English and Chinese – independently translated it into Simplified Chinese (the written form of the language used in Mainland China). A third person – also bilingual in both languages – then proofread and reconciled the independent translations.

Thereafter, through the connections of a Chinese contact, we tested the translated survey instrument with a small sample of native speakers from diverse backgrounds in China and refined the translations according to feedback. This iterative process ensured that our language was accurate and easy to understand for our target local population. The pretesting also helped to ensure that we had crafted our survey in a way that was contextually appropriate.

4.1 Balanced Reciprocity versus No Reciprocity

We begin by looking at balanced reciprocity. Where China was the initiator removing $100 billion of tariffs on US goods, 72 percent of American respondents approved when the United States reciprocated by reducing the same amount of US tariffs on Chinese goods. Approval dropped sharply to 45 percent when the United States did not reciprocate (p < 0.0001, n = 479). In the same scenarios, 76 percent of Chinese respondents approved of their country’s policy when they saw the United States respond with balanced reciprocity, while approval dropped to 63 percent when the United States did not reciprocate (p = 0.004, n = 408).

The empirical patterns were similar when it was the United States that initiated the trade concession. Where China engaged in balanced reciprocity by reducing $100 billion worth of Chinese tariffs on US goods in return, 79 percent of American respondents approved of their country’s policy. However, approval dropped to 60 percent when China did not reciprocate (p < 0.0001, n = 479). At the same time, 77 percent of Chinese respondents approved when China engaged in balanced reciprocity, while approval was at 70 percent when China did not (p = 0.12, n = 403).Footnote ³³ These results indicate general preferences for positive reciprocity among the American and Chinese publics – even in the shadow of great power rivalry that amplified relative gains and international status concerns.

4.2 Super-Reciprocity versus Balanced Reciprocity

Now we turn to super-reciprocity. Recall that if relative gains considerations prevail over inequality aversion preferences, citizens in the initiator state (i.e., the “sender” state which expects to be reciprocated) should express stronger approval in the super-reciprocity condition than in the balanced-reciprocity condition. If inequality aversion preferences trump relative gains considerations, however, approval in the super-reciprocity condition should be lower instead. Citizens in the responder state (i.e., the “receiver” state which decides how to reciprocate), on the other hand, should express lower approval in the super-reciprocity condition because both relative gains and inequality aversion preferences point in the same negative direction.

What do the data tell us? When respondents’ own country initiated the de-escalation and was super-reciprocated by the foreign country, they – despite deriving relative gains from super-reciprocity – did not approve of their own country more. Among American respondents, 77 percent approved when China engaged in super-reciprocity by removing $150 billion worth of tariffs on US goods, compared to 79 percent when China engaged in balanced reciprocity by removing an equal amount of tariffs (p = 0.56, n = 492). In parallel scenarios, 76 percent of Chinese respondents approved when the United States engaged in super-reciprocity by removing $150 billion worth of tariffs on Chinese goods, while approval remained at 76 percent when the United States engaged in balanced reciprocity (p = 0.85, n = 492). Super-reciprocity was not perceived more favorably than balanced reciprocity by citizens in the initiator state, even if it generated relative gains in a trade war.

Shifting our focus to the responder country, we also do not find strong evidence that engaging in super-reciprocity reduced domestic approval. Where China initiated the trade concession, American respondents’ approval remained relatively stable when their country decided to reduce $150 billion, rather than $100 billion, worth of tariffs on Chinese goods (p = 0.18, n = 466).Footnote ³⁴ Where the United States initiated the trade concession, Chinese respondents maintained the same level of approval of their country, regardless of whether it decided to super-reciprocate or reciprocate with the same amount (difference-in-proportions = +0.02, p = 0.63, n = 411).Footnote ³⁵ Public approval thus appeared indifferent to super-reciprocity and balanced reciprocity when the domestic country was the one reciprocating.

4.3 Semi-Reciprocity versus Balanced Reciprocity

The dynamics changed when we turn to semi-reciprocity. Public support decreased decisively when the foreign country engaged in semi-reciprocity rather than balanced reciprocity. Where the United States was the initiator removing $100 billion worth of tariffs on Chinese goods, only 63 percent of American respondents approved of the policy when they saw China semi-reciprocating by removing only $50 billion worth of tariffs on US goods in return, a drop of 16 percentage points of support against the scenario in which China removed an equal amount of tariffs (p < 0.0001, n = 491). On the other hand, where China was the initiator, only 67 percent of Chinese respondents approved of their country’s policy when the United States engaged in semi-reciprocity, which is 9 percentage points lower than the Chinese approval when the United States engaged in balanced reciprocity (p = 0.04, n = 397).

What about the responder state deciding whether and how to reciprocate? Recall that if the responder state engages in semi-reciprocity, compared to the balanced-reciprocity condition, it will generate relative gains to itself and advantageous inequality at the same time. Thus, if individuals’ inequality aversion offsets their relative gains concerns, they should exhibit similar – or even weaker – approval in the semi-reciprocity condition compared to the balanced-reciprocity condition. This is indeed the case among our American respondents. When their country decided to reciprocate by reducing only $50 billion – rather than $100 billion – worth of tariffs on Chinese goods, public approval did not increase but in fact fell in absolute terms by 3 percentage points (p = 0.49, n = 474).Footnote ³⁶ In parallel conditions, when China engaged in semi-reciprocity rather than balanced reciprocity in response to US trade concessions, Chinese public approval also did not increase (difference-in-proportions = +0.07, p = 0.07, n = 399).Footnote ³⁷ While the specific results differ in the two countries, they both illustrate how advantageous inequality aversion may countervail and offset relative gains preferences among both American and Chinese citizens.

4.4 Unbalanced Reciprocity versus No Reciprocity

How did unbalanced reciprocity fare compared to no reciprocity at all? We find that people preferred super-reciprocity to no reciprocity. In both scenarios of super-reciprocity, and at both sender’s and receiver’s ends, approval was consistently higher in the super-reciprocity condition than in the no-reciprocity condition (all two-tailed p-values were less than 0.05; see Figure 2 for the substantive differences).

Preferences, however, were more nuanced for semi-reciprocity; it depended on the sequence of moves. Where one’s country was the first mover that initiated the concession, public approval remained low regardless of whether the foreign country engaged in semi-reciprocity or zero reciprocity (all two-tailed p-values were greater than 0.20; see Figure 2 for the negligible substantive differences). Where one’s country was the second mover that received the initial concession, however, approval increased by 24 percentage points in the United States (p < 0.0001, n = 483) and by 14 percentage points in China (p = 0.001, n = 400) when it reciprocated to the foreign country with $50 billion worth of tariff reductions than when it refused to remove any tariffs.

What might explain this puzzling difference in approval when one’s country was the second mover, but not when it was the first mover? We may tap on behavioral theories and evidence on sender–receiver gaps and asymmetric inferences to shed light on this puzzle. For citizens in the second-mover country, they saw their own country giving a $50 billion benefit (tariff reduction) to the foreign country, which they approved significantly more than not giving at all. What they gave or sent (sender side) would be salient to them, if they were indeed biased by the sender–receiver gap whereby senders treat what they send more seriously than receivers do (see Section 3.1). For citizens in the first-mover country, however, they saw their own country receiving $50 billion after giving $100 billion to the foreign country. If they were systematically biased by the sender–receiver gap, the $100 billion concession they initially gave (as sender) should be more salient than the $50 billion reciprocation they subsequently received (as receiver). The evidence shows their approval in this situation sank as low as in the nonreciprocity condition.

The puzzle can also be interpreted as a manifestation of asymmetric inferences (Bell and Quek Reference Bell and Quek2026; Kertzer, Brutger, and Quek Reference Kertzer, Brutger and Quek2024): If we give a $50 billion benefit (even after receiving $100 billion), our giving of $50 billion is benign and well-intended – our country is doing “good,” and we approve. But if others give us $50 billion, their giving of the same $50 billion (after receiving our $100 billion) is malign and ill-intentioned – our country has been exploited as the “sucker,” and we disapprove. Such is the asymmetry to be expected in the inferences we draw from our own giving/sending behavior, relative to the inferences we draw from the same giving/sending behavior of others.

What is clear from these results is that people were averse to seeing no reciprocity at all. Moreover, the majority supported de-escalating the trade war in all scenarios, except where the United States refused to engage in any form of reciprocation following China’s reduction of tariffs – an outcome approved by 63 percent (95% CI [56%, 70%], n = 204) of Chinese respondents but by only 45 percent (95% CI [39%, 51%], n = 244) of American respondents.

4.5 Discussion

Given the abundance of results harvested through our dyadic experiments, we take a step back to highlight three general patterns.

First, public approval was consistently higher in unbalanced-reciprocity conditions than in the no-reciprocity condition. This finding holds regardless of the country sample, the sequence of moves, and the form of unbalanced reciprocity (semi or super). Even when unbalanced reciprocity triggered relative gains concerns and/or inequality aversion, both American and Chinese publics strictly preferred reciprocity over nonreciprocity – even in the context of great power competition and animosity.

Second, domestic public support was lowered by semi-reciprocation from the foreign country, compared to when balanced or super-reciprocity was the response strategy of the other side. Americans’ approval of their country’s policy was 63 percent when China semi-reciprocated, compared to 77 percent or 79 percent when China super-reciprocated or reciprocated with equal terms. Similarly, China’s domestic approval was 67 percent when the United States semi-reciprocated, compared to 76 percent when the United States responded with super- or balanced reciprocity. Interestingly, however, respondents did not appear to draw a clear distinction between semi-reciprocation and nonreciprocation from the foreign country. This observation is consistent with individuals making qualitative, heuristic judgments based on categorical thinking. When their country was the first mover, both US and Chinese respondents appeared to only care about whether the other side reciprocated at least as much ($100 billion or $150 billion) or not ($0 or $50 billion).

Third, balanced, semi-, and super-reciprocity as response strategies garnered majority support in both the United States and China. The breadth of domestic support for these response strategies ranged from 67 percent to 72 percent in the United States, and from 77 percent to 84 percent in China (see Figure 2).

In Online Appendix §E, we offer suggestive evidence for the behavioral foundation of these findings, showing that retaliation concerns were unlikely to be the driving force behind these results. In particular, risk-takers, who should be less anxious about retaliation following nonreciprocation, did not respond differently than non-risk-takers (Figure A2).

Unpacking further, we shift the focus from risk preference to nationalism following our preregistered analysis.Footnote ³⁸ Nationalism reflects the belief about national superiority, the desire for ingroup favoritism, and the tendency of outgroup animosity (Kosterman and Feshbach Reference Kosterman and Feshbach1989) – conditions conducive to relative gains-seeking behavior. The empirical literature shows that nationalism strongly predicts hawkish foreign policy preferences among American and Chinese citizens (Gries et al. Reference Gries, Qingmin Zhang and Cai2011; Herrmann, Isernia, and Segatti Reference Herrmann, Isernia and Segatti2009). It turns out that the public opinion premium of positive reciprocity over nonreciprocity as a response strategy was mainly driven by non-nationalists. Figure 3 shows that while nationalists in the United States and China were generally supportive of their country’s policy regardless of their country’s response, non-nationalists’ approval was especially low when their country decided not to reciprocate. This finding suggests that norms of reciprocity are not universally held. Individuals more concerned about relative gains and international status may be more aggressive and hawkish, and less other-regarding and cooperative, in their policy preferences.

Figure 3

Public approval across experimental conditions, by nationalism

Note: While nationalists in both the US and China generally supported their country’s policy regardless of its response strategy, non-nationalists showed particularly low approval when their country chose not to reciprocate. To identify nationalists, we asked: “How much do you believe [the United States / China] is superior to and better than other countries? (1 = not at all; 5 = very much)” Those who rated their country 4 or 5 were coded as nationalists; those who rated lower were non-nationalists. Error bars represent 95 percent confidence intervals.

Overall, our findings open a middle ground between the multidisciplinary literature on reciprocity, which suggests norms of reciprocity widely held by individuals, and the neorealist literature on international cooperation, which paints a pessimistic picture for reciprocity due to relative gains concerns in the shadow of great power rivalry.

4.6 Additional Analysis of Heterogeneous Treatment Effects

To further examine heterogeneity in respondents’ attitudes, we report a set of exploratory subgroup analyses in Online Appendix §E. We summarize the findings here.

First, we test if political knowledge moderates our experimental findings. To the extent that political knowledge serves as a reasonable proxy for tariff misperceptions, examining heterogeneity by political knowledge helps to probe whether our findings may change in a more naturalistic setting where respondents’ knowledge about tariffs varies. We find that the response patterns among the politically knowledgeable and the less knowledgeable were generally consistent across both American and Chinese samples (Figure A4). This analysis partly assuages concerns about external validity that center on tariff misperceptions surrounding the US–China trade war.

Second, we explore if education may play a moderating role. The IPE literature argues that education strongly influences trade preferences, especially by fostering a more cosmopolitan worldview (Hainmueller and Hiscox Reference Hiscox2006; Solodoch Reference Solodoch2024). We find that when the home country was the first mover, respondents with and without college education disapproved of under-reciprocation (semi- or nonreciprocation) from the foreign country at similar levels; yet when the foreign country engaged in balanced or super-reciprocity, the college educated significantly increased their level of approval, compared to the noncollege educated whose approval only increased mildly (Figure A5). This empirical pattern points to the possibility that the educated care more about the relative gains implications of trade.

Third, we examine heterogeneity by partisanship. In foreign policymaking, the ruling elite may pay more attention to a certain subset of the domestic constituency. In the United States, Donald Trump may be more sensitive to Republicans’ opinions. In China, the government may be more attentive to Chinese Communist Party (CCP) members’ sentiments. Our subgroup analysis shows that Republicans’ response pattern was not substantively different from Democrats’. They had lower approval rates than Democrats across all experimental conditions, but maintained their preference for engaging in any form of positive reciprocity in response to China’s tariff reduction over no reciprocation (Panel A of Figure A6). In our Chinese sample, the picture is more nuanced. There was a lack of variation in approval rates, such that CCP members maintained their strong support regardless of the reciprocation strategy at play (Panel B of Figure A6).

Finally, we consider heterogeneity by income. This analysis carries substantive relevance because governments can be disproportionately responsive to more well-off citizens, not only in the United States (Gilens Reference Gilens2012) but also in China (Fu, Li, and Yan 2025). Overall, we do not find meaningful variation in how the higher and lower income groups judged the different reciprocity scenarios (Figure A7). The only potential nuance is that the higher income group in our Chinese sample appeared to favor balanced reciprocation from the United States more than the lower income group did, although we caution against overextrapolation.

Overall, the lack of heterogeneous treatment effects uncovered in our data – apart from nationalism, the theoretically motivated moderator we preregistered to examine – suggests that the results should be generalizable to wider populations whose demographic and dispositional traits may vary (Bassan-Nygate et al. Reference Bassan-Nygate, Renshon, Weeks and Weiss2025; Coppock, Leeper, and Mullinix Reference Coppock, Leeper and Mullinix2018; Druckman Reference Druckman2022, 70–83).

4.7 Analysis of Domestic Political Feasibilities

Tapping on the data from our parallel dyadic experiments in both countries, Figure 4 fills in the domestic outcomes for both the United States (in blue) and China (in red) at the terminal nodes of each action–reaction sequence.

Figure 4

Domestic public support across experimental conditions

The worst domestic public outcomes are found in the action–reaction sequence where China made the first move by removing tariffs on American goods, and the United States did not reciprocate. This action–reaction sequence produced the lowest level of domestic approval in both countries, with approval falling to below majority support (45 percent) in the United States.

Domestic sweet spots are found in the action–reaction sequences where the United States made the first move by removing tariffs on Chinese goods, and China responded with either balanced reciprocity or super-reciprocity. In both cases, their aggregate domestic support was maximized, with approval within each country pitched at the range of 77 percent to 79 percent, representing a remarkable win–win public outcome for both. If China were to respond with semi-reciprocity, its domestic approval would see a small nominal bump, but it is statistically insignificant or only marginally significant compared to its approval under super-reciprocity (p = 0.17, n = 408) and balanced reciprocity (p = 0.07, n = 399). It would, however, create a drastic and significant fall in domestic approval in the United States compared to super-reciprocity (−14 percentage points, p = 0.001, n = 485) and balanced reciprocity (−16 percentage points, p < 0.0001, n = 492), jeopardizing the political feasibility for future cooperation with China.

4.8 Caveats and Insights from Open-Ended Responses

It is important to place these findings and analyses in perspective. Firstly, it is useful to note that our analyses operate at the aggregate level. Across different times and situations, there may be specific political and institutional cleavages that justify specific disaggregations. For instance, the political and institutional context could be such that the ruling elite only cares about – or overweighs by some factor – a specific subset of the domestic constituency (e.g., partisan voters in American politics or urban citizens in Chinese politics). The aggregate analyses can be disaggregated accordingly to speak to these specific situations. Researchers may also design their experiments to oversample specific subpopulations instead of the national population if the purpose is primarily to inform policymakers of the potential consequences of their choices and if there are indeed strong reasons to believe that some subpopulations are politically important but not others.

Secondly, while our dyadic experimental design has produced a richness of findings, it is not designed to target particular mechanisms that drive particular findings in each particular scenario. This task is not trivial – given the large number of contingent scenarios generated through our interactive design – and requires future research in a series of carefully designed mechanism tests. But while decisive tests of mechanisms are not within our scope, we designed an open-response segment, as per our preregistration, to gather first-cut evidence on why citizens responded the way they did in different scenarios.

Specifically, we provided an opportunity for respondents who disapproved of the way their country handled the situation to explain their disapproval in an open-ended response. Here, we zero in on respondents who expressed disapproval in the following treatment groups:

1. (1) Condition 1 – super-reciprocity from the foreign country: The home country was the first mover that removed tariffs on $100 billion worth of goods from the foreign country. The foreign country engaged in super-reciprocity by removing tariffs on $150 billion worth of goods from the home country in return.

2. (2) Condition 2 – semi-reciprocity toward the foreign country: The foreign country was the first mover that removed tariffs on $100 billion worth of goods from the home country. The home country engaged in semi-reciprocity by removing tariffs on $50 billion worth of goods from the foreign country in return.

Because the disapproval rates in these conditions across both American and Chinese samples were low in the first place, and providing an open-ended response was voluntary, our qualitative data are limited:

1. (1) American sample – Among thirty-three respondents who disapproved in Condition 1, there are seventeen open-ended responses that are nonmissing and interpretable. Among fifty-eight respondents who disapproved in Condition 2, there are fifty open-ended responses that are nonmissing and interpretable.

2. (2) Chinese sample – Among sixteen respondents who disapproved in Condition 1, there are eight open-ended responses that are nonmissing and interpretable. Among the fourteen respondents who disapproved in Condition 2, there are ten open-ended responses that are nonmissing and interpretable.

In Condition 1, there is some, albeit weak, evidence of advantageous inequality aversion being the core concern among respondents who expressed disapproval. Only one American respondent explicitly invoked the language of inequality or unfairness (“[The] amount should be paid equally”), while many other respondents justified their disapproval by saying that any trade with China is bad and therefore the US should not remove tariffs on China in the first place (e.g., “I don’t think the U.S. should do any trades with China. We can depend solely on our own.” “China is very sly and they will hurt the United states,” “I disapprove of this because something America has always done and we should stick to is ‘we do not negotiate.’”). On the other hand, no Chinese respondents invoked the language of inequality or unfairness when explaining their disapproval. Among the few respondents who gave classifiable responses, two of them justified their disapproval by saying that China should not be the first mover (“Our country shouldn’t remove tariffs on the U.S.” and “China should let the U.S. remove [tariffs on Chinese goods] first and make its next move by gauging the U.S.’s sincerity”). One believed that all tariffs should be removed (“[We] should remove all tariffs”).

In Condition 2, however, there is strong evidence of advantageous inequality aversion being the core concern among American respondents who expressed disapproval. Most responses (thirty-two) invoked the language of inequality or unfairness (Table 1). The other responses reflected principled opposition to trading with China. The evidence of advantageous inequality aversion is less strong, but still existent, among Chinese respondents who expressed disapproval. Of the eight open-ended responses that are nonmissing and interpretable, two touched on the language of inequality or unfairness (“[It/We] should be equal” and “China should respond more proactively by removing $100 billion worth of tariffs or even more”). Most remaining responses are not sensible explanations of disapproval (e.g., “Because China’s solution is correct every time. Every time it’s resolved peacefully,” “Because I think this trade war is hard to fight”).

Table 1Open-ended responses from American respondents that invoked the language of inequality or unfairness in Condition 2

Table 1 Long description

Many respondents invoked the language of inequality or unfairness in Condition 2, where the home country engaged in semi-reciprocity by removing tariffs on $50 billion worth of goods from the foreign country. Examples of these open-ended responses include: “The amounts being waved should be equal,” “The US should remove Tariffs on an equal amount as China,” and “It doesn’t seem like the United States is matching the effort of the Chinese to reach common ground.” These responses were written when equality or fairness concerns were not prompted by the question and providing explanations for policy preferences was voluntary.

Note: Original quotes are shown without edits.

Taken together, the open-ended responses provide some, but admittedly limited, evidence that Conditions 1 and 2 evoked advantageous inequality aversion – which might offset relative gains preferences – among respondents who disapproved of their government’s action. The qualitative data should nevertheless be taken as suggestive rather than conclusive, and it would be useful for future research to tease out more precisely the theoretical mechanisms and how they might interact with one another.

5.1 Implications for Policy

The policy implications of these findings are sharp. In recent years, US foreign policy has taken a hawkish turn, with many policymakers from across the political spectrum now advocating for decoupling with China. At the time of this writing, the second Trump administration imposed further tariffs on China, which continued to intensify the US–China trade war by provoking retaliatory tariffs from the Chinese government. While scholars are concerned that such economic nationalism will pose a threat to the liberal international order (Lake, Martin, and Risse Reference Lake, Martin and Risse2021) and will further escalate the political tensions between the United States and China (e.g., Christensen Reference Christensen2023; Lake Reference Lake2018; Rodrik Reference Rodrik2019), some politicians and foreign policy analysts argue against active de-escalation on the US side because they assert that China will not reciprocate.Footnote ³⁹

At the same time, however, the Chinese government would also be concerned that the United States would not reciprocate. The lack of evidence that either side would enter into reciprocation constitutes a barrier to initiating de-escalation.Footnote ⁴⁰ Our book offers the evidence. Our results show clearly the willingness to reciprocate on both sides from the standpoint of their domestic publics. To the extent that politicians are sensitive to domestic sentiments (see Section 2.1), our findings lay an empirical foundation for the prospect of de-escalation, if and when the politicians are ready to try.

While our findings assuage pessimism about the practical prospects of de-escalating the US–China trade war, we caution against overoptimistic interpretations. A large segment of the US public, especially Republicans, appeared to support a certain level of trade war. Such attitudes could stem from tariff misperceptions (Rho and Tomz Reference Rho and Tomz2017; Stantcheva Reference Stantcheva2022), protectionist ideology (Jedinger and Burger Reference Jedinger and Burger2020; Mutz Reference Mutz2021), and political framing (Ardanaz, Murillo, and Pinto Reference Ardanaz, Murillo and Pinto2013; Hiscox Reference Hiscox2006). For instance, Donald Trump not only used the terms “reciprocal tariff” and “Liberation Day” to justify his trade war escalation, but also framed it as necessary for the United States to strengthen manufacturing.

To the extent that elites are willing to de-escalate the US–China trade war, political frames can also be harnessed for de-escalation (Koo and Quek Reference Koo and Quek2024; Wong Reference Wong2024). Politicians, for example, may frame trade war de-escalation – through the engagement of positive reciprocity – as reinforcing US strategic competition with China. If delivered by Republican elites, such framing – albeit potentially counterintuitive – can be particularly effective on Republicans, as recent work on American political behavior indicates (Barber and Pope Reference Barber and Pope2019). State media in China, on the other side, can also induce citizens to take a more dovish foreign policy stance through their framing of the issues (Pan, Shao, and Xu Reference Pan, Shao and Yiqing2022). Indeed, the success of de-escalation hinges on multiple factors, with elites’ ability to sell rapprochement at home being one critical component of the peace process (Mattes and Weeks Reference Mattes and Weeks2024). Our discussion and empirical evidence thus offer a measured optimism about the prospects of reciprocity today, insofar as elites have the political will to temper the spiral of conflict.

At the time of writing, Trump and Xi reached a decision to de-escalate the US–China trade war with a ninety-day pause of the “reciprocal” or retaliatory tariffs. While this mutual de-escalation effort was likely attributable to the alignment of multiple domestic factors (such as interest group influence and elite ideology), circumstantial evidence points to the importance of US public opinion. Since Trump initiated the trade war with China (as well as Canada and Mexico), an overwhelming majority – including Republicans – thermostatically responded to his tariff escalation, viewing trade as an opportunity for economic growth rather than a threat to the economy (Sides Reference Sides2025). Such public opinion dynamics coincided with Trump’s tariff pause against Canada and Mexico on April 10, 2025. But anti-escalation sentiment continued to brew among the American public amid Trump’s refusal to de-escalate the trade war against China. National polls indicated that most Americans believed the trade war would be bad not only for the United States but also for them personally.Footnote ⁴¹ At the same time, a majority of Americans believed the United States should pursue a policy of global free trade,Footnote ⁴² and even Republicans began to lose confidence with Trump’s economic stewardship and were especially dissatisfied with Trump’s tariff policy.Footnote ⁴³ In this context, Trump’s approval rating continued to drop,Footnote ⁴⁴ with many voters considering his tariff policy the “biggest” mistake of his second term.Footnote ⁴⁵ Republican elites became anxious about a midterms wipeout due to Trump’s unpopular tariff policy, exerting further pressure on Trump to de-escalate the US–China trade war.Footnote ⁴⁶ Although we cannot put precise relative weights on the role that public opinion played in contributing to the recent de-escalatory efforts, these observations are consistent with our claim that analyzing the domestic feasibility of conflict de-escalation would benefit from deeper understandings of how domestic publics would respond (see also Milner Reference Milner1997; Putnam Reference Putnam1988).

5.2 Looking Forward

One major empirical contribution of this book is to show that conditional on conciliatory gestures to end a trade war, both American and Chinese publics are likely to favor reciprocation. Because we did not study the condition in which both the United States and China maintained the import tariffs they imposed on each other, one limitation is that we were not able to shed light empirically on the level of domestic approval when neither side extended an olive branch. Comparing public support with this condition would help us better understand leaders’ incentives, in the first place, to initiate rapprochement and de-escalate the trade war. Considering recent events, we speculate that it is unlikely that Americans would have a high(er) approval rate in this condition. In a Pew Research Center survey conducted on March 24–30, 2025 (after Trump implemented tariffs on China in February and again in early March), most US respondents said increased tariffs on China would be bad for their country and for them personally.Footnote ⁴⁷

Of course, it remains a speculation because our sample and survey timing were different. Given the stakes, future work can follow up on the present study and investigate the domestic consequences leaders face when neither side makes a conciliatory gesture. Despite the limitation, our experiments speak directly to much debate about the prospect of US–China de-escalation, which is framed around the suspicion or assertion that neither country would be willing to reciprocate (see Footnote footnotes 39 and Footnote 40). In democracies such as the United States, “hostile public opinion might initially constrain leaders from approaching an adversary” (Mattes and Weeks Reference Mattes and Weeks2024, 195; see also Bayer Reference Bayer2010; Lind Reference Lind2020). This argument may also extend to China, as our literature review in Section 2.1 contends. Our study sheds light on the prospect of de-escalating the trade war insofar as it provides systematic evidence that public opinion would not be hostile to reciprocation, which should give both sides more confidence to extend an olive branch. The Joint Statement on US–China Economic and Trade Meeting in Geneva – released on May 12, 2025, in an effort to de-escalate trade tensions – should serve as a powerful precedent in addition to our experimental findings. In this political episode, both national publics appeared to favor the tariff reductions and de-escalating (even if not ending) the US–China trade war.

While the experimental findings generate practical insights, our theorization of unbalanced reciprocity is fresh to the literature and opens avenues for further research. Firstly, future work could investigate the implications of super-reciprocity and semi-reciprocity in other contexts. While we focused on the US–China trade war – a hard case for positive reciprocity due to the salience of national security and superpower rivalry concerns – future work may explore the generalizability of our findings by experimenting with different issue domains and other bilateral conflicts.

Positive semi- and super-reciprocity in the IPE domain was our focus. Moving forward, the dynamics of both positive and negative semi- and super-reciprocity in the security domain should also be explored. For instance, do arms races in the real world reflect spirals of balanced or unbalanced reciprocities? Would the strategic and psychological aspects of the spiral model (Jervis Reference Jervis1976) change significantly in super-reciprocity or semi-reciprocity – and if so, how? Strategically, for example, one may have to consider its own – and the other’s – reputation for reciprocity. One might have incentives to cultivate a reputation for negative super-reciprocity to deter further and future challenges, yet super-reciprocation at the first instance might be overly provocative. Psychologically, a spiral of semi-reciprocations (super-reciprocations) should reduce (increase) fear and insecurity over time. Semi-reciprocation at the first instance might help to calm the security dilemma and reduce the other’s fears, tapering off the spiral.

Furthermore, because public perceptions of whether a reciprocal policy action is “balanced” are important but potentially pliable, hawkish and dovish leaders might have different sets of political incentives to frame what counts as “balanced” and therefore “reciprocal” in international interactions. We leave the question of how political framing can manipulate perceptions of balanced and unbalanced reciprocity, as well as the implications of the framing strategies, for future research. Future work may also explore how citizens from a third-party country assess the United States and China under different action–reciprocation pathways. Such analysis would bring new insights to the study of international relations, insofar as the findings have implications for the global influence and reputation of the two superpowers.

Finally, future studies may extend our experiments with further-stage interactions between states to study how to end the trade war. Even though de-escalation is a necessary step, de-escalating a trade war does not necessarily mean ending it. While our experiments captured the initiation-reciprocation dynamic in de-escalation, real-world conflicts are likely to involve more than the first round of action and reciprocation. And given the role of time horizons in the decision-making process (Krebs and Rapport Reference Krebs and Rapport2012), it would be interesting to investigate how the shadow of the future shapes the dynamics of reciprocation in interactions extending further into the future.

At a more general level, our work demonstrates how we may experiment with different policy pathways and their domestic public implications through interactive survey experiments that cross two countries in a conflict. In the real world, only the outcome from one policy pathway is observable. The fundamental problem of causal inference is that the counterfactual is never observable (Holland Reference Holland1986). In the experimental worlds we constructed, the different outcomes from different potential policy pathways become discernible – not only the domestic consequences faced by the United States and by China when the United States acts and China reacts, but also the consequences produced when China acts and the United States reacts. As the domestic consequences of international action–reaction sequences crystallize from unobservable counterfactuals into observable evidence, governments will be in a better position to assess their policy options for de-escalating a trade war that has impacted the welfare of their own people and that of the world.