Global Markets, Corporate Assurances, and the Legitimacy of State Intervention: Perceptions of Distant Labor and Environmental Problems

Collective perceptions of harm and impropriety channel the evolution of capitalism, as shown by research on the moral boundaries of markets. But how are boundaries perceived when harms are distant and observers face competing claims from advocacy organizations and corporations? These conditions are particularly salient in global supply chains, where private voluntary initiatives have been formed to address labor exploitation and environmental degradation. We argue that state intervention is now on the rise and that popular judgments about state intervention carry new insights for the sociology of markets, morality, policy, and globalization. Analyzing data from a conjoint survey experiment, we find that distant labor and environmental problems (e.g., forced labor, natural resource depletion) provoke varied levels of interest in state intervention as well as different justifications for state intervention. We also find an asymmetry of influence by strategic actors: transnational advocacy frames shape judgments to some degree, but they fall flat or backfire among conservatives. Corporate promises of reform reduce the perceived importance of state intervention—across political-ideological divides and regardless of credibility. Moving beyond stylized pro-/anti-trade attitudes, these findings reveal implicit logics of a contested moral field and the legitimacy of state intervention at a formative moment.

The globalization of production has reshaped politics, inequality, health, and family life in the United States (Dorn, Hanson, and Majlesi 2020;Venkataramani et al. 2020) and in societies on the other ends of global supply chains (Meng and Yamauchi 2017;Villarreal and Yu 2007). In many affluent countries, people have grown accustomed to hearing about links between the two ends of the supply chain, especially through exposés of labor exploitation and environmental degradation in the making of consumer products. But it is not clear how individuals decide what counts as objectionable and what can be dismissed as distant, negligible, or the normal working of the market. News about the poisoning of workers making smartphones in China or the burning of forests in Brazil to make leather footwear, for instance, could be perceived as unfortunate but inevitable, as best left to others to solve, or as requiring intervention that places restrictions on global markets. Under what conditions will a particular response predominate? When will "sympathy fatigue" (Hochschild 2016) be high, and when will a sense of global interdependence materialize?
A large literature, mostly outside sociology, has examined the economic and sociocultural bases of "protectionist" attitudes (Mansfield, Mutz, and Brackbill 2019). But this research reads popular perceptions of globalization narrowly as either appreciation or antipathy toward international trade. Sociologists, in contrast, have argued that markets are interpreted through historically-and situationallydefined moral lenses (Abend 2014;Zelizer 2013). Objections to child labor, prostitution, pollution, life insurance, debt, and trade in human tissues have spawned laws to restrict markets or channel their growth (Anderson 2018;Fourcade 2011;Healy and Krawiec 2017;Quinn 2008), although potential objections can be dampened by restructuring or re-framing exchange (Almeling 2007;Chan 2009;Schilke and Rossman 2018). Beyond the primordial question of commodification, perceptions of profit-seeking and inequality depend in part on the moral assumptions activated among observers (DiMaggio and Goldberg 2018;Lamont 2000).
Current architectures of global production raise related but distinct questions-about how the moral boundaries of markets are judged when harms are geographically distant (Bair 2019;Clapp 2015), supply chains are fragmented (Gereffi 2018), and observers face competing claims about severity, responsibility, and benevolence (McDonnell, King, and Soule 2015). We contend that fleshing out situational judgments about global markets within a larger moral field is essential for fully theorizing the "cultural life of market exchange" (Zelizer 2013:371) and making sense of a moment in which global supply chains are increasingly politicized. Sociological research on perceptions of globalization has been strangely stalled since Fiss and Hirsch's (2005) analysis of discourse, and it deserves to be reinvigorated.
We seek to explain the conditions under which individuals in the United States perceive market restriction by the state as important for combatting environmental degradation and labor exploitation in global supply chains. State intervention is a direct and intuitive way of restricting the boundaries of markets, and states have become far more interventionist vis-à-vis global supply chains over the past decade (Kay and Evans 2018). But the legitimacy of state mandates continues to be in question, particularly amid polarized perceptions of evidence (Eyal 2019) and corporate power to resist or reshape government (Hacker et al. 2022). Methodologically, we use a conjoint survey experiment, a useful tool for analyzing multidimensional perceptions of sociopolitical phenomena (Flores and Schachter 2018). By asking respondents to consider both labor and environmental problems, we acknowledge parallels in these controversies (Polanyi 1944) while also asking whether distinctive logics are at play (Latour et al. 2018). In addition, we examine how perceptions of state intervention are shaped by the strategic and competing claims of both corporations (Malhotra, Monin, and Tomz 2019) and advocacy organizations (Kreps and Maxey 2018), rather than examining just one in isolation.
The "repertoire of moral sentiments" (Lukes 2010) and set of "concepts for moral reasoning" (Strand 2015) that predominate in a society at a given point in history can be described as existing in a "moral field." Like other social fields (Bourdieu 1991), moral fields are constituted by power struggles and competing positions that come to define the common sense of an era (Strand 2015). In the current era, struggles between corporations and advocacy groups have heightened the image of the market-and perhaps especially the global market-as a site of moral/immoral action. Companies are named and shamed for their sins and then jockey for redemption by promoting narratives of responsibility (Bartley and Child 2014;McDonnell et al. 2015)-at least in the sense of being responsive without admitting fault (Ricoeur 2000). Individuals are classified-by themselves and others-as good or bad people based on their consumer choices (Johnston and Baumann 2010;Tavory 2011). Our study helps reveal the logic of this moral field and its presumptions about global markets. We find widespread valorization of corporate responsibility-taking, even in the midst of contestation. Moreover, analyzing openended responses, we discover vocabularies of agency, appropriateness, and consequences that justify intervention and attach in different ways to distant labor and environmental problems.

The Resurgence of State Intervention in Historical Perspective
The U.S. federal government has played a central role in expanding global markets since the mid-1930s, helping forge the supply chains that now structure trade. But U.S. policy has not always been driven by economic orthodoxy on the benefits of market expansion, and there are signs the United States and other countries are now embracing a style of interventionism that rejects or revises neoliberal principles.
From the end of the Civil War to the 1930s, trade loomed large in U.S. politics, with frequent battles between export-oriented and import-competing interests (Hiscox 2002). At times, these battles were tinged with moral reform projects, as with the anti-imperialist "peace progressives" who fought to include a ban on items made with forced labor in the Tariff Act of 1930, "blend[ing] economic protectionism of domestic producers with moral indignation over the poverty, degeneration, and inhumane treatment" of foreign workers (Armstrong 1975:23).
From the 1940s to 1980s, the United States became a champion of trade liberalization, albeit with exceptions to protect some sectors (Chorev 2007). As Cold War anti-communism became a driving force of trade liberalization (Verdier 1994) and rising service-sector employment scrambled earlier coalitions (Guisinger 2017), trade politics lost public salience. U.S. policy helped spread market fundamentalism but rarely intervened to address exploitation or environmental damage (Rosen 2002). Nevertheless, the growth of the domestic U.S. regulatory state had some international implications. The Marine Mammal Protection Act of 1972, for instance, banned imports of fish caught through processes that endangered dolphins-a provision that would later spur formative trade disputes (DeSombre and Barkin 2002).
Several moments in the late 1980s and 1990s increased the public salience of trade and intermingled material and moral concerns about globalization. The Tiananmen Square massacre heightened concerns about human rights and trade with China (Kuk, Seligsohn, and Zhang 2018). Debates about NAFTA generated a resurgence of protectionist sentiment (Guisinger 2017) and a new politics of labor and environmentalism that would cast a long shadow (Kay and Evans 2018). Even before the 1999 "Battle of Seattle" against the World Trade Organization, "globalization" was taking on a negative valence (Fiss and Hirsch 2005). In a few instances, U.S. trade policy shifted. Facing pressure from Black and gay activists and voters, the Clinton administration changed course on global intellectual property protections and AIDS drugs (Chorev 2012:842). Labor and environmental advocates managed to get more stringent standards in 2004's Central America Free Trade Agreement than they had in NAFTA (Kay and Evans 2018).
Yet this kind of policy impact was often overshadowed by the influence of activism targeting the private sector, where labor and environmental campaigns inspired, often unwittingly, the rapid growth of corporate social responsibility and sustainability initiatives in the late 1990s to 2000s. As reputationsensitive companies positioned themselves as responsible actors, advocacy organizations highlighted "greenwashing" or "fairwashing" and pushed companies to demonstrate real improvements. Moralized global markets were forged during this period. In a number of product markets, various enterprises and activists sought to expand, purify, or conventionalize alternative norms about production and consumption (Balsiger 2021). The idea of the global market as a moral arena also became more visible in consumption patterns (Bartley et al. 2015) and public attitudes about trade (Ehrlich 2018). States were never absent, but they refocused on "orchestrating" private forms of regulation rather than directly intervening (Abbott and Snidal 2009).
This moment of state ceding to private orders began to fade in the 2010s. International trade returned to being highly salient in U.S. politics with the 2015 to 2016 Trump campaign and Sanders primary campaign, both of which empowered "political entrepreneurs willing to buck the elite pro-trade consensus" (Guisinger 2017:5). The United States' rejection of the Trans-Pacific Partnership, renegotiation of NAFTA, hobbling of the WTO, and embrace of tariffs are important parts of a rejection or reconfiguration of neoliberalism (Hopewell 2020). But it is important to also recognize a rise in state intervention targeting labor and environmental problems specifically, including crackdowns on illegal logging, illegal fishing, and forced labor (LeBaron and Rühmkorf 2019; Leipold et al. 2016). Figure 1 shows four indicators of this sort of state intervention. Preferential trade agreements around the world, which included few if any labor or environmental conditions through much of the 1980s to 1990s, have become increasingly dense with references to labor rights, environmental protection, and other norms. The density of environmental norms grew dramatically, reaching an average of 50.8 norms per trade agreement in the 2010 to 2018 period (based on data from Morin,
Dür, and Lechner 2018). Labor provisions (coded on a different scale by Raess and Sari 2018) followed a similar pattern-increasing to 7.7 provisions per agreement in the 2012 to 2017 period (compared to 1.6 for 1992 to 2011). Figure 1 also shows a rise in U.S. enforcement actions against international environmental and labor violations, using data we compiled from the U.S. Department of Justice and U.S. Customs and Border Protection. Indictments for international environmental crimes grew from just a handful in the early 2000s (2001 to 2006 mean = 4.67) to a peak of 20 indictments in 2011 (2007 to 2020 mean = 12.36). These indictments most often addressed illegal discharges by ships in international waters or illegal imports of animal or plant species-including illegally-logged timber after its import was banned in 2008 (Leipold et al. 2016).
The U.S. government has also become more active in blocking the import of items made with forced labor, as the 1930 Tariff Act allows. This provision was rarely used until 1991 to 1992, when the United States blocked imports from 22 Chinese firms in the wake of Tiananmen Square. After the Obama administration revised the rules of application in 2016, there was a surge of enforcement in 2019 to 2020, targeting Malaysian rubber glove factories, Taiwanese fishing vessels, Malawian tobacco farms, and cotton from the Xinjiang region of China, among others. The repression of Uyghur people in Xinjiang, which the United States has called genocidal, has heightened attention, but it has not been the sole focus of crackdowns on forced labor during either the Trump or Biden administrations.
Amid the rise of state intervention, corporate responsibility initiatives have also continued to grow. The number of U.S. companies that deposited sustainability reports with the Global Reporting Initiative (n.d.) grew from 213 in 2010 to 564 in 2016. The UN Global Compact (2020) roughly doubled its participants from 2011 to 2020, reaching a total of more than 12,000 companies worldwide. The current configuration of moralized global markets thus includes a rising wave of state intervention and the continued growth of corporate voluntarism.

Why Study Popular Perceptions?
Understanding popular perceptions of state intervention in this context is important for three reasons. First, popular perceptions are resources and constraints for policy entrepreneurs, who "try to take advantage of those elements of public attitudes that support their case in presenting demands or making policy proposals" (Manza and Brooks 2012:92). Although public opinion is weakly related to trade policy in periods of low salience, it makes a discernible difference in periods of high salience (Guisinger 2017), competitive electoral contexts (Hiscox 2002), and moments of collective choice about the institutions of trade policy-making (Verdier 1994). As the United States navigates supply chain disruptions, COVID-era controls, and geopolitical changes, policy entrepreneurs are likely to redouble efforts to tap into popular perceptions of whether, when, and why global markets should be restricted.
Second, there is a growing consensus among scholars-and even some corporate leaders (Whoriskey 2019)-that voluntary responsibility/sustainability initiatives have failed to solve central problems, making greater state involvement necessary (Bartley 2018;Locke 2013). It is unclear, though, if popular audiences share this diagnosis. If they instead perceive corporate voluntarism as a sufficient substitute for the state, or if they reject state intervention for other reasons, then reformers may face an uphill battle in shifting from voluntary to mandatory approaches. More broadly, if private orders have become taken-for-granted, this could add a new sort of friction to Polanyian (1944) movements to re-embed markets through state controls.
Third, popular perceptions are an essential ingredient for theorizing the cultural logics of markets. Economic sociologists have built a strong empirical foundation for theorizing perceptions of status, inequality, and authenticity in markets (e.g., DiMaggio and Goldberg 2018;Hahl, Kim, and Zuckerman Sivan 2018), but perceptions of globalization have been left to other disciplines. The result is a large but paradigmatically narrow literature on perceptions of globalization and a great deal of untapped insight in sociology.

PErCEPTIonS oF STATE InTErvEnTIon: ThEory AnD hyPoThESES
If moral judgments about markets are situational and multidimensional, then what situations and dimensions are likely to influence perceptions of market restriction in the context of global supply chains? We contend that the desire for state intervention is likely to be activated to different degrees depending on the controversies encountered, the principles they evoke, and various actors' strategic attempts to promote or avoid market restrictions. In the contentious fields that constitute contemporary moralized markets, the strategic and competing claims of advocacy organizations and corporations are especially central (Balsiger 2021). We therefore turn to research on transnational advocacy and corporate responsibility to theorize these claims and their currencies.

Advocacy Frames and the Structuring of Concern
Theories of transnational advocacy suggest perceptions of distant problems are malleable if confronted with the right information. In line with constructivist premises, many scholars of transnational advocacy take a Durkheimian approach to moral authority (Durkheim [1898(Durkheim [ ] 1995, emphasizing the power of broad norms that are both externally legitimate and personally cherished, such as those pertaining to individual rights, rationality, and fairness in the "western cultural account" (Boli and Thomas 1997). The power of principles ultimately depends, though, on the strategic activities of advocacy organizations to select, name, frame, and translate distant problems into pressing local concerns (Keck and Sikkink 1998;Krause 2014).
Specifically, advocacy organizations often draw on the legitimating power of science, which gains authority from its perceived disinterestedness and universalism (Drori et al. 2003), even as its autonomy has become more complicated. Scientific frames are most commonly noted in research on environmentalism (Longhofer and Schofer 2010) and public health (Best 2012), but scientific legitimation has also proven powerful in advocacy focused on global poverty (de Souza Leão and Eyal 2019), occupational health and safety (Hilgert 2013), and corporate exploitation of vulnerable populations (Quark 2016). A second approach is to mobilize personal testimony from affected individuals. This can help build sympathy or solidarity (Keck and Sikkink 1998), demonstrate the deservingness of victims (Best 2012), and tap into underlying beliefs about the sanctity of individual rights. Testimony might be conveyed in person through visits and speaking tours or in a more mediated fashion through documentary films or press quotations.
For some problems, transnational advocates emphasize corporate irresponsibility and greed. This can heighten publicity, appeal to anti-corporate sentiment in various parts of the political spectrum, and activate powerful feelings of indignation (Boltanski and Chiapello 2006). It can be particularly potent to counterpose the power, profits, and greed of corporations with the desperation of workers and communities in their supply chains (Bair 2019;Bartley and Child 2014).
Taken together, these lines of research portray perceptions of global problems as open to influence from advocacy groups, and they specify particular frames that should be potent. If market observers are as sensitive to advocacy frames as constructivist scholarship suggests, then one would expect the following: Advocacy hypothesis: Statements from advocacy groups that mobilize science, testimony, or anti-corporate critique increase the perceived importance of state intervention in global supply chains.
Delving further into anti-corporate critique, some scholars worry that spotlighting corporations fuels the "private politics" of corporate social responsibility rather than strengthening state capacities (Seidman 2007). This might be rectified if advocacy groups explicitly pushed for state intervention as a solution to corporate malfeasance. This approach has contributed to a move away from private, voluntary initiatives to state regulation in several instances (Evans 2020;Leipold et al. 2016), and it leads to a supplemental expectation: Anti-corporate critique has a larger positive effect on the perceived importance of state intervention when it explicitly calls for state action rather than merely criticizing corporations.

Corporate Promises and the De-legitimation of the State
A growing body of research suggests perceptions of state regulation are open to a different type of influence, based on companies' promises of voluntary reform (Malhotra et al. 2019). Neo-Marxist and power elite theories have long argued that corporations can derail or defang democratic controls by inserting their own preferred solutions (Domhoff 1990), exercising "moral leadership" to establish their hegemony (Gramsci 1971), and more recently, making voluntary "soft law" seem like the obvious answer to injustices (Rodríguez Garavito 2005). From this perspective, corporate assurances (i.e., confidence-inspiring pledges) of responsibility and private regulation should tend to preempt state intervention.
Corporate assurances come in different forms, as emphasized by neo-Weberian theories of credibility in private governance (Boström 2006) and organizational reform more generally (e.g., Dobbin, Schrage, and Kalev 2015). When confronted with problems in their supply chains, consumer-facing corporations often develop new standards for their suppliers and pledge to audit compliance (Thorlakson, de Zegher, and Lambin 2018). This kind of "first party" private regulation leaves companies open to charges of untrustworthy greenwashing or fairwashing (Balsiger 2021), because it fails to specify "who watches the watchers" (Shapiro 1987). To increase the credibility of their assurances, companies often work with NGOs in multistakeholder initiatives that oversee compliance (Boström 2006;Cashore 2002). After two decades of growth, multi-stakeholder standards were used by nearly 20 percent of publicly listed food, clothing, and forest products companies (Thorlakson et al. 2018). A third response is for corporations to cut off recalcitrant suppliers; this response occurs more rarely, and typically after unsuccessful remediation attempts (Amengual, Distelhorst, and Tobin 2020).
Prior research paints a mixed picture of whether and how these forms of private regulation actually derail state intervention. Casebased research portrays private regulation as an influential but imperfect shield that does not necessarily block pro-regulatory efforts (Boghossian and Marques 2019; Evans 2020). Looking at public opinion data in Europe, Burgoon and Fransen (2018) find at best contingent effects of private regulation on attitudes about redistributive policy. Yet some studies show that private regulation can affect popular perceptions, at least in particular policy domains and when highlighted for respondents. In Switzerland, respondents became less supportive of government regulation of companies' foreign activities when told that Swiss commodity producers had "voluntarily committed themselves to protect people and the environment" abroad, although only when this included oversight by "independent non-profit organizations" as a sign of credibility (Kolcava, Rudolph, and Bernauer 2021:A5). In the United States, support for environmental regulation (e.g., banning genetically modified foods or the sale of bluefin tuna) declined when respondents were told of firms' voluntary efforts, especially when more firms participated but regardless of the depth of the voluntary action (Malhotra et al. 2019).
It remains to be seen whether corporate assurances of private regulation resonate in a crowded discursive arena that also includes advocacy organizations and a variety of labor and environmental problems for individuals to consider. If corporate voices come through as loudly as elite and neo-Marxist theorists of corporate power suggest, then we would expect support for an overarching hypothesis: Private regulation hypothesis: Corporate assurances that suppliers are being privately regulated reduce the perceived importance of state intervention in global supply chains.
A supplemental hypothesis with a more limited scope is suggested by theories of organizational credibility and some prior research: Corporate assurances of private regulation reduce the perceived importance of state intervention more when corporations partner with nonprofit organizations than when corporations take unilateral actions. Symbolic statements from companies, even if they do not amount to assurances of private regulation, may also undermine more stringent interventions. Theories of impression management (Goffman 1959) suggest actors can gain deference from their audiences by altering their language and demeanor. Applying this perspective to corporations, McDonnell and King (2013:388) find that firms in the midst of controversy often make prosocial (i.e., socially benevolent) claims that "show a firm's commitment to socially acceptable norms, beliefs, and values and protect its image by diluting, rather than refuting, the negative claims made by activists." Carlos and Lewis (2018) argue that prosocial claims especially buffer firms from further criticism if they avoid the specific issue under contention, because this reduces charges of hypocrisy. On the other hand, diluting language can also be perceived as cheap talk or empty symbolism (Darnall, Ji, and Vázquez-Brust 2018). Therefore, we test the following claims: Prosocial hypotheses: (a) Corporations' prosocial statements reduce the perceived importance of state intervention, and (b) the effect of prosocial statements is smaller than assurances of private regulation.
By some accounts, assurances of private regulation could have the opposite effectthat is, to legitimate state intervention. Private regulation may help demonstrate that alternatives are feasible and create reformminded business coalitions (Tzankova 2021). Although these processes are unlikely to be visible or salient to average citizens, Dana and Nadler (2019) find evidence that firms' voluntary actions can sometimes legitimate state regulation among conservatives, who would otherwise be opposed. This research reinforces the need to examine political heterogeneity in the effects of corporate promises.

Political Ideologies and Divergent Judgments
While emphasizing widespread objections and approvals, scholars of moralized markets also recognize population heterogeneity in perceptions of market processes and state controls (DiMaggio and Goldberg 2018). The political polarization of attitudes (Della Posta 2020) and knowledge claims (Eyal 2019) in U.S. society means political orientations should play an especially important role in evaluating both advocacy frames and corporate promises-and thus moderate their effects. In particular, to the extent that political orientations imply different levels of trust in the claims-makers and principles that animate moralized markets, we should see divergent judgments.
Public opinion research suggests conservatives should be less responsive to predominant advocacy frames in global markets. Scientific expertise, testimony, and corporate critique are not necessarily anathema to conservative advocacy (Gross, Medvetz, and Russell 2011), but conservatives, on average, have lower trust in science (Gauchat 2012), environmental organizations (Brewer and Ley 2013), and labor unions (Newman and Kane 2017) than do individuals with moderate or liberal views. At the same time, conservatives tend to have more trust in business (Leibrecht and Pitlik 2019) and should therefore be more responsive to corporate statements. 1 Principles of small government and voluntary provision of public goods may also make private regulation especially attractive to those on the right (Vandenbergh and Gilligan 2017). In broad strokes, we thus expect that among conservatives, statements from advocacy groups have smaller effects and statements from companies have larger effects than they do among moderates and liberals.
Attitudes about government regulation should also affect the degree to which people see private action by corporations as a reasonable substitute. The rise of private regulation has been fueled by skepticism about "command and control" regulation (Short 2011) and by projects to "reinvent government" in search of efficiency (Abbott and Snidal 2009). At the individual level, people with ambivalent or negative views of government regulation should be especially keen to let the private sector substitute for the state. In contrast, "true believers" in government regulation should be less swayed by companies' promises of private regulation and prosocial activities. In other contexts, scholars have similarly argued that prior beliefs about states and markets anchor responses to new information (Brooks and Manza 2013). We expect statements from companies to have smaller effects among people who believe government regulation is generally beneficial than among those with more ambivalent or negative attitudes. Other, more fine-grained distinctions are also possible, but these claims provide a grounded framework for asking whether central axes of disagreement in U.S. society also structure responses to a contentious field of moralized markets.

Labor and Environment in Global Supply Chains
The perceived importance of state intervention may also depend on the problem being considered. Research and advocacy have made it clear that supply chains for consumer products frequently rely on, produce, or exacerbate varied forms of labor exploitation and environmental degradation. This includes child labor, forced labor, dangerous work, withheld wages, and repression of unions in supply chains for food, clothing, and electronics (Chan, Ngai, and Selden 2020;Kerrissey and Schuhrke 2016); as well as deforestation, land degradation, water pollution, toxic waste, and violence against communities proximate to sites of natural resource extraction (Jorgenson 2006;Smith, Sonnenfeld, and Pellow 2006).
Perceptions of these problems are filtered through distance, as global supply chains shift the hazards of production to distant locations, with ambiguous connections to quality of life at the point of consumption. Some scholars have begun to theorize this distancing (Clapp 2015), but empirical research on perceptions of globalization remains focused elsewherewhether on how individual characteristics (occupation, education, ethnocentrism) shape attitudes about free trade (Mansfield et al. 2019) or on consumer interest in Fair Trade or eco-labeling (Bartley et al. 2015), rarely with cross-domain comparisons.
Although it is difficult to predict precisely which problems will elicit the most (or least) interest in state intervention, several problems stand out due to their relationship to widespread principles of moral order. Forms of labor exploitation that violate sacred categories of childhood and personhood (Zelizer 2013) should strike observers as highly objectionable. This suggests a high degree of support for intervention against child labor and forced labor-with the latter commonly referred to as "modern slavery" (LeBaron and Rühmkorf 2019). Yet other labor problems, such as dangerous work, are also capable of spurring widespread outrage (Kerrissey and Schuhrke 2016), as are environmental problems with diffuse but emotional valences (Farrell 2014), which may gain added salience through a sense of interdependence. The relative ranking of these problems must be seen as an empirical question.
A further rationale suggests support for state intervention will be heightened when problems are described as criminal violations. Theorists of crime and markets have argued that criminality is unique as a domain in which there is widespread support for strong state intervention even amid market fundamentalism (Harcourt 2010). This is clearest in the discourse of global environmental crime, which fuses environmental and penal rationales (Pellow and Brehm 2013). As documented above, crackdowns on illegal logging, illegal fishing, and other environmental crimes have escalated. We expect that when a problem is defined as illegal or criminal, support for state intervention will be higher than when a similar problem is described in other terms. Although rooted in environmental research, this tendency may also extend to labor issues, where activists have begun to use criminal framings, such as using "wage theft" to describe unpaid wages (Bobo 2011).
Other justifications for state intervention could come to the fore as observers consider distant environmental and labor concerns. Examining open-ended rationales for state intervention will allow us to move beyond the measurement of how much different problems matter to engage with theories of how different problems activate assumptions about interdependence, spillover, agency, justice, and moral responsibility (Latour et al. 2018;Lukes 2010).

rESEArCh DESIGn AnD DATA
Survey experiments help uncover shared logics of judgment that can be obscured in standard survey research (Wallander 2009). Conjoint survey experiments, in which a variety of elements are randomized simultaneously, are commonly used to study multidimensional perceptions of sociopolitical phenomena, including immigration (Flores and Schachter 2018), terrorism (Huff and Kertzer 2018), and political candidacy (Hainmueller, Hopkins, and Yamamoto 2014).
In a conjoint design, respondents select or rate profiles that contain a randomized array of information (Hainmueller et al. 2014). A conjoint approach allows us to construct scenarios that approximate the noisy way information about problems, advocacy frames, and corporate promises is ordinarily conveyed in news reports. For example, a 2019 Reuters article on labor abuses in the Ethiopian garment industry included both a statement from the corporation H&M ("We take seriously any allegations of violations of labor standards and will . . . implement our programs addressing . . . workers' rights") and a statement from an NGO, the Worker Rights Consortium ("Unfortunately . . . there is a yawning gap between the brands' ethical pretentions and the workplace reality for the people sewing their clothes") (Woodyatt 2019). Our design approximates the experience of reading such news stories in a condensed form.

Outcome
To measure the perceived importance of state intervention in global supply chains, we asked respondents when and to what degree it would be important for the U.S. government to "ban items from being sold in the U.S. if they are made in especially harsh or damaging ways." A ban is a straightforward way of restricting the market to prevent morally objectionable practices, as illustrated by research in economic sociology (Healy and Krawiec 2017;Zelizer 2013). It is also an intuitively meaningful-and increasingly common-way for the U.S. government to intervene in global supply chains.
After showing respondents a pair of scenarios containing our treatments (described in the next subsection), we asked for two types of judgments. The first asks respondents to make a prioritizing choice about when it would be most important to impose a ban, choosing between the two scenarios shown. Comparing a pair of scenarios increases respondents' engagement (Hainmueller, Hangartner, and Yamamoto 2015) and allows us to examine judgments in a constrained decision environment, in which respondents cannot revert to overarching defaults (e.g., toward a crackdown in every case). Because we view public perceptions as resources for policy entrepreneurs, measuring prioritization helps identify messages that nudge public choices down one path or another. A second measure focuses on respondents' degree of support or opposition to a ban in each scenario. Here, respondents registered their opinion on whether a ban should be imposed on a seven-point scale from "definitely should not" to "definitely should," with a central value of "not sure." This allows preferences for a ban in both or neither of the two scenarios, as well as gradations of support or opposition. Measuring what "moves the needle" of support/opposition is especially apt if policy entrepreneurship depends on tapping into strong feelings. We also gathered qualitative evidence by asking respondents an open-ended question about the reasons for their choice.

Treatment Dimensions
Each scenario consisted of information on (1) a product made through global supply chains and consumed in the United States, (2) a labor or environmental problem, (3) a statement from an advocacy group, (4) a statement from companies in the industry, and (5) a description of which companies made this statement (see Table 1). Research suggests five dimensions is well within the range respondents can process (Bansak et al. 2019). We designed our scenarios such that all elements could be fully randomized.
Products. To make the scenarios realistic, they refer to the production of clothing, electronics, or seafood. Labor and environmental abuses have been documented in each product's supply chain. For instance, fishing boats may use forced labor, pollute waters with leftover plastic nets, and over-harvest threatened species (Bush and Oosterveer 2019). Apparel and electronics factories may have unsafe working conditions, emit toxic chemicals into air and water, and utilize inputs that have been harvested or mined in dangerous or environmentally damaging ways (Kerrissey and Schuhrke 2016; Locke 2013). To ensure our findings are not driven by how or how often products are consumed, we chose products that allow full randomization.
Problems. We focus on nine problems that are well-documented and applicable to these products. The problems include pollution of rivers and oceans, emission of toxic chemicals, dangerous work, child labor, and modern slavery. For two sets of problems, we varied whether the problem was framed in terms of illegality (i.e., "overuse of natural resources" versus "illegal use of natural resources"; and "unpaid wages" versus "wage theft by employers") to probe whether an illegality/criminality frame increases support for state intervention.
Advocacy frames. We developed four statements to be conveyed by "a major advocacy group." Following the theoretical guidance above, these used science, personal testimony, corporate critique, or an explicit call for state action to underscore the problem and legitimate intervention. We also developed a baseline statement, which merely restated the commonness of the problem, to serve as a reference category while maintaining narrative consistency.
Corporate promises. We developed four statements that companies could make in response to the problem in their industry, as well as a baseline statement that the company was simply "weighing the evidence." Our prosocial statement makes references to ethical conduct, community, and charity, which are common elements in prosocial corporate discourse (McDonnell and King 2013). The other statements capture the three assurances of private regulation described above: (1) a unilateral response in which companies impose "strict rules" for their suppliers and audit compliance, (2) partnership with a nonprofit organization to certify suppliers' compliance, and (3) a decision to cut ties to problematic suppliers. Common problem "This is a common problem in this industry. We are looking into how it can best be addressed." Corporate coverup "Corporations might say that they are taking responsibility. But they only want to cover up their relentless pursuit of profits, which is the root of the problem." Testimony "Listen to the voices of affected people like J.N., who told us: 'This has caused great suffering for me and my community.
We need a solution now.'" Science "The scientific evidence is clear: This is a pressing problem that severely harms many people around the world. We need to take action now." Do not trust; need government "We should not trust corporations to solve a problem as important as this. It is time for the government to act."

Corporate Promise
No comment "We are weighing the evidence about this issue, and we cannot comment further at this time." Prosocial "We believe in ethical conduct and good corporate citizenship. We will continue to promote these values through our charitable and community engagement initiatives." Strict rules "We have adopted strict new rules to prevent this problem and are closely monitoring our suppliers' compliance." Partnership to certify "We are partnering with independent non-profit organizations to certify that our suppliers meet strict standards for responsible production." Cut supplier "We have decided to stop doing business with any suppliers that exhibit this problem."

Which Firms
A large retailer Target Walmart A group representing half of the companies in this industry A group representing nearly all the companies in this industry Which companies made the statement. We also included information about which companies voluntarily chose to respond. This serves several purposes. First, excluding this information might lead respondents to assume from the statement which type of firm made it-a problem referred to as masking (Hainmueller et al. 2014). Second, this allows us to see whether the presence of particular firms or larger/smaller groupings alters perceptions of state intervention, as it might if perceptions of firms are highly politicized or sensitive to industry unity. Third, this allows us to perform supplemental tests of whether corporate promises carry more weight when made by larger groups of companies (Malhotra et al. 2019) or by well-known, highstatus companies (McDonnell and King 2018). In addition to an unnamed "large retailer," we named Walmart and Target, because these are the largest and most well-known multiproduct retailers. Both are high-status companies, as measured by McDonnell and King (2018), and one (Walmart) is perceived as politically conservative, which could allow seemingly liberal issues to speak more clearly to conservatives (Dana and Nadler 2019).

Data Collection and Analysis
We conducted the survey experiment online from mid-February to early-March of 2020 with a sample of 4,489 respondents provided by Dynata. This is a national sample of adults, intended to approximate the U.S. population in terms of region, age, gender, income, and education. Although not a probability sample, it provides a varied set of respondents with which to test our hypotheses. In a pre-treatment question, 33.8 percent of respondents described themselves as slightly to extremely conservative and 34.1 percent described themselves as slightly to extremely liberal. Respondents were 48.5 years old on average (sd = 17.3), and 52 percent were women. Although people with some college experience were over-represented, 25 percent of the sample had only a high school diploma or less. The location of respondents is similar to each state's share of the population. Section 1 of the online supplement provides more information on the sample and its representativeness. Before fielding the survey, we registered our analysis plan with the Open Science Framework. 2 To maximize comprehension and engagement, we first presented each scenario in a narrative, bullet-point style, approximating a news story (see Figure 2). To maintain narrative flow, we started with the problem and product and then moved to the business and advocacy statements; we randomized the order of these statements by respondent, to ensure any differences in these theoretically A group representing half the companies in this industry chose to issue a statement saying: "We have decided to stop doing business with any suppliers that exhibit this problem." What is a major advocacy group saying about illegal use of natural resources in the production of seafood?
"This is a common problem in this industry. We are looking into how it can best be addressed." important treatments are not driven by presentation order. Respondents who read the narrative descriptions very quickly were reminded that most people take longer and were asked if they wanted to go back to review (Zhang and Conrad 2016). Respondents had earlier been asked a simple multiple-choice question on the instructions, and those who failed were reminded of the correct answer, similar to Hahl and colleagues (2018).
After seeing two scenarios in narrative form, respondents were shown a comparison table (see Figure 3). This allowed respondents to recall and contrast details before registering their judgments about (1) the scenario they would prioritize for a ban by the U.S. government and (2) the degree to which they support or oppose a ban in each scenario. Attention checks were included after each pair of scenarios (see Section 3 of the online supplement).
As is standard in conjoint designs, each respondent was shown two pairs of scenarios, generating 17,956 observations in total (Bansak et al. 2018). Following established practice, we cluster standard errors by respondent. We follow recommendations to be minimally exclusionary of low-attention respondents and use attention checks in robustness tests. 3 Our analyses focus on the Average Marginal Component Effect (AMCE) of each element on the perceived importance of state intervention. The AMCE can be interpreted as the causal effect of an attribute, conditional on the joint distribution of all other attributes (Hainmueller et al. 2014). We present the results in graphical form, with 95 percent confidence intervals. Supplemental models using alternative specifications produce consistent findings.

rESuLTS
The average level of support for state intervention across all scenarios (mean = 5.05, sd = 1.6, median = 5, 1-7 scale) suggests a fairly high degree of interest in imposing bans to address distant labor and environmental problems (the mean of prioritization is a probability of .5 by design). We proceed by examining the effects of advocacy frames and corporate promises, followed by the effects of labor and environmental problems. We then turn to the question of whether people with different political-ideological orientations responded differently to the advocacy and corporate statements, as hypothesized. In a later section, we examine the justifications  Table of Scenarios Here is a table summarizing the two scenarios you just read about. Remember, these refer to products that are made abroad and sold by many companies in the U.S.

Scenario 1 Scenario 2
Problem: illegal use of natural resources modern slave labor Product: seafood sold by many companies clothing sold by many companies Business statement: "We have decided to stop doing business with any suppliers that exhibit this problem." "We have adopted strict new rules to prevent this problem and are closely monitoring our suppliers' compliance."

Which businesses chose to make this statement:
A group representing half of the companies in this industry A large retailer

Advocates' statement:
"This is a common problem in this industry. We are looking into how it can best be addressed." "Corporations might say that they are taking responsibility. But they only want to cover up their relentless pursuit of profits, which is the root of the problem." for state intervention provided in response to our open-ended question. Figure 4 shows the AMCEs of advocacy frames and corporate promises on the probability of a scenario being prioritized for state intervention (left panel) and the degree of support for state intervention (right panel).

Advocacy Frames
(Results are based on a model with all treatments but are isolated here for ease of presentation; see Table S3 in the online supplement.) Consistent with the advocacy hypothesis, statements by advocacy groups shaped the perceived importance of state intervention, although some more reliably than others. Anti-corporate critique-that is, accusing companies of wanting to cover up their relentless pursuit of profit-increased the likelihood of a scenario being prioritized by 2.8 percentage points, on average (p = .016, 95 percent CI[. 5 -5.1]), compared to the baseline statement, which only observed that the problem is common. It also influenced the degree of support for state intervention, generating an increase of .11 points on our seven-point scale (p = .004, CI[.04 -.18]). The scientific frame similarly increased the likelihood of prioritization by 4.3 percentage points (p < .001, CI[2.0 -6.5]) and support for state intervention by .078 points (p = .043, CI[.002 -.15]). Although these effects do not represent dramatic shifts, recall that the advocacy statements were part of a mix of information to consider.
Personal testimony and the explicit call for government intervention increased the likelihood of a scenario being prioritized (by 2.4 percentage points, p = .040, CI[. 1 -4.7] and 3.2 percentage points, p = .007, CI[. 9 -5.5], respectively, compared to the baseline), but they did not have discernible effects on support for state intervention. Contrary to the claim that foregrounding the state should provide a powerful alternative to private politics, we find no evidence that an explicit call for government intervention had a larger effect than the simple anti-corporate critique. In fact, the anti-corporate critique had a larger effect on the degree of support (F(1,4488) = 12.96, p < .001).

Corporate Promises
All three corporate assurances of private regulation negatively affected the probability of a scenario being prioritized for state intervention and the degree of support for state intervention, strongly supporting the overarching private regulation hypothesis. When companies promised to set strict rules for their suppliers and audit compliance, it reduced the likelihood of a scenario being prioritized by 6.0 percentage points (p < .001, CI [-8.4 --3.7]) compared to the baseline statement, in which companies simply say they are weighing the evidence. It also decreased support by .12 points (p = .002, CI[-.19 --.04]). A more credible form of private regulation, in which companies partner with nonprofit organizations to certify suppliers' compliance, affected judgments to a similar degree, reducing the likelihood of prioritization by 6.7 percentage points (p < .001, CI [-9.2 --4.5]) and decreasing support by .17 points (p < .001, CI[-.24 --.09]). When companies promised to cut ties to problematic suppliers, it reduced the likelihood of prioritization by 5.1 percentage points (p < .001, CI [-7.4 --2.3]) and reduced support by .11 points (p = .002, CI[-.19 --.04]). Note that these effects, which are highly significant but modest in magnitude, are conditional on the uniform distribution of all other treatments, including two advocacy statements that explicitly cast doubt on corporate promises.
Contrary to the expectation that partnering with nonprofit organizations provides more credibility than unilateral corporate action, we do not find any significant differences among private regulatory assurances (see Table S12 in the online supplement). Scholars have extensively theorized credibility and legitimacy in global standard-setting, but a key distinction in this literature (between corporate-led and multi-stakeholder initiatives) did not register in the court of public opinion.
Observers did differentiate, though, between assurances of private regulation and prosocial statements. The prosocial statement, in which companies espouse ethical conduct, corporate citizenship, charity, and community, did not have a discernible effect on either the likelihood of prioritization (p = .156) or the degree of support for state intervention (p = .304), compared to the baseline statement. In addition, compared with this prosocial statement, each of the private regulatory actions had significantly greater negative effects on prioritization and support (see Table S12). This pattern is consistent with the claim that assurances of private regulation are more influential than prosocial statements (part b of the prosocial hypothesis) but not with the claim that prosocial statements matter (part a). Figure 4 also shows that which corporations made a statement generally did not affect judgments. The one exception is that prioritization was more likely when half of firms in the industry spoke, compared to a single generic large retailer (b = 2.5, p = .029, CI[.3 -4.8]) or "nearly all" businesses (F(4,488) = 3.86, p = .049). To probe further, we examined interactions between these elements (estimating Average Component Interaction Effects [ACIEs]) (Table S13 in the online supplement). Given the large number of combinations, these estimates are less precise and should be taken with caution. In general, we do not find evidence that corporate assurances have larger effects when made by larger groups of firms. Nor do we find clear evidence that statements are more influential when made by a well-known firm-Target or Walmart-than by an unnamed large retailer. The single exception comes with Walmart's prosocial statements, which reduced prioritization (ACIE = -.074, p = .047) and degree of support (ACIE = -.244, p = .035) more than those of an unnamed large retailer. This raises intriguing questions about how firms' status and sociopolitical valence might combine to shape the appeal of prosocial language. We also examined whether corporate assurances might be less influential when accompanied by advocates' critiques of corporations. These interactions are imprecise and do not reveal any clear evidence of this kind of countervailing effect. 4 Advocacy frames and corporate promises both resonated, but their combination was not so salient as to drive perceptions amid the array of information respondents were considering.

Problems and Products
Specific labor and environmental problems clearly affected perceptions of when and to what degree state intervention is needed (see Figure 5). Compared to the reference category (dangerous working conditions), four problems significantly increased both prioritization and support for state intervention.
Child labor, modern slave labor, emission of toxic chemicals, and water pollution generated the greatest interest in state intervention. When the problem was child labor, the likelihood of the scenario being prioritized was 18.1 percentage points higher (p < .001, ), and the degree of support was .40 points higher (p < .001, CI[.30 -.50]), than when the problem was dangerous working conditions. Modern slave labor had a similarly large effect on the degree of support (b = .33, p < .001, CI[.23 -.43]) and a slightly smaller effect on the likelihood of prioritization (b = 11.3, p < .001, CI [8.2 -14.4]). Even though it does not violate sacred norms about personhood and individual rights, emissions of toxic chemicals inspired a similar likelihood of prioritization (b = 11.0, p < .001, CI [7.8 -14.1]) and increased support to a slightly lesser extent (b = .21,p < .001,). In turn, pollution of rivers and oceans had a similar effect on support (b = .19, p < .001, CI[.09 -.29]) and increased prioritization to a slightly lesser extent (b = 8.2, p < .001, CI [5.1 -11.3]). 5 Note that the problem effects are larger than those of private regulation and advocacy frames, underscoring questions about how respondents perceived distant problems (which we address in a later section).

Figure 5. Effects of Problems and Products on the Perceived Importance of State Intervention
Other problems generated much less interest in state intervention. Most notably, overuse of natural resources reduced the likelihood of prioritization by 7.6 percentage points (p < .001, CI [-10.8 --4.5]), and reduced support by .20 points (p < .001, CI [-.39 --.19]), compared to the reference category. Interestingly, when this was defined as illegal use of natural resources, the probability of prioritization and degree of support was significantly higher (F(1,4488) = 8.6, p = .003 and F(1,4488) = 22.9, p < .001, respectively), although still comparatively low. This supports the claim that an illegality frame increases the perceived importance of state intervention. Note, however, that the contrast between unpaid wages and its illegal analogy-wage theft by employers-worked in the opposite direction (F(1,4488) = 21.1, p < .001 for prioritization and F(1,4488) = 6.58, p = .010 for support). This is an unexpected and intriguing finding, which raises questions about how well criminal framings of workplace problems resonate in broad public audiences.
A handful of open-ended responses implied that wage theft was comparatively minor/partial or attributed it to employees, but we suspect respondents mostly assumed theft was sporadic or found the recentlyrising discourse of wage theft unfamiliar (and thus unconcerning) or equivocal (see Levin 2021).
Finally, the product in question did not shape when or to what degree state intervention was seen as important. We estimate precise null effects for electronics and seafood compared to clothing, suggesting few respondents judged scenarios based on their taste or perception of different products.

Heterogeneous Treatment Effects
To assess whether political orientations moderate the effects of corporate and advocacy statements, we utilize pre-treatment, preinstructions questions about political ideology (using a scale of 1 for extremely liberal to 7 for extremely conservative) and attitudes about government regulation in general (based on a seven-point scale for agreement/ disagreement with the statement that "government regulation of business is generally good for society," mean = 4.57). To capture stark differences, we contrast strong conservatives (6 or 7 on our scale, 24 percent of the sample) with all others; and "true believers" in government regulation (6 or 7 on our scale, 27 percent of the sample) with all others, as specified in our pre-analysis plan. Figure 6 shows the effects of advocacy frames and corporate promises in these split samples (for full models and omnibus tests, see Section 4 of the online supplement). Starting with advocacy frames, we find some evidence that conservatives are distinctive, as hypothesized. In the analysis of prioritization, the liberal/moderate group is responsive to all advocacy frames, but the conservative group is not, although formal tests show the interaction to be non-significant. Analyzing the degree of support, we see stark and significant differences between conservatives and others. Here, there are negative ACIEs for the combination of conservative and the explicit call for state intervention (p < .001), testimony (p = .012), and the corporate coverup message (p = .043) (see Table S7 in the online supplement). Moreover, the explicit call for state regulation had a negative effect on conservatives' degree of support for state intervention-that is, leading them toward opposition more than when advocacy groups made only a neutral statement. Testimony had a negative although less discernible effect (p = .058) among conservatives. These results may indicate a backlash to being told that state intervention is crucial. Note that conservatives were moderately supportive of state intervention (mean = 4.87) when the advocacy statement was neutral, perhaps because this cued a sense of "making up your own mind." Turning to corporate promises, we find more similarity than difference, contrary to expectations. For instance, both conservatives and liberals/moderates became less likely to prioritize state intervention when companies promised to enforce strict rules for their suppliers or partner with nonprofits to certify compliance. In addition, support for state intervention decreased to a similar magnitude among both groups when companies partnered with nonprofits to certify compliance, despite conservatives' skeptical views of advocacy organizations, as reported above. In general, we find no significant interactions between political ideology and corporate promises. 6 This result suggests, surprisingly, that even groups that usually have comparatively low levels of trust in business give credence to corporate promises to rectify labor and environmental problems in the global economy.
Similarly, individuals who are "true believers" in government regulation turn out not to be significantly different from others in their responses to private regulation and prosocial statements. Although proregulation individuals were generally supportive of state intervention in global supply chains (see Table S4b in the online supplement), they perceived it as less important (across both outcome measures) when companies promised to partner with nonprofits or cut problematic suppliers; and they were less likely to prioritize state intervention even when companies merely promised to set strict rules for their suppliers. As detailed in the online supplement, the lack of heterogeneous treatments is confirmed with non-significant ACIEs and omnibus tests and persists when defining "true believers" with lower thresholds of our scale. Thus, although scholars and proponents often portray private regulation as a remedy for those with little faith in the state, we find no evidence of this. The appeal of private regulation in the global economy is much broader and not conditioned by people's views about government regulation in general.
Overall, these results reveal an important and unexpected asymmetry in responses to the different actors who are competing to define the meanings of responsibility, regulation, and morality in global markets. Only liberals and moderates respond to advocacy frames by becoming more supportive of state intervention. But all groups respond similarly to corporate promises. Although we cannot empirically assess the mechanisms behind the widespread credence given to corporate promises, we discuss two theoretical explanations in the conclusion.

Robustness Checks
We utilized open-ended attention checks (asking respondents to name one of the products/ problems just encountered) to run supplemental analyses with only respondents who passed at least one check (90 percent of the full sample). The estimates are substantively identical to those above (see Figure S4a in the online supplement), indicating our findings are not driven by noise from low-attention respondents. Similarly, the results were substantively identical when we excluded speeders (in the lowest decile of completion time) ( Figure S4b).
Although our sample approximates the U.S. population more closely than many experimental studies (particularly in terms of geography, gender, and age, with some over-representation of White individuals, see Tables S1 and S2 in the online supplement), lack of representativeness could bias our estimates if under/over-represented characteristics interact with treatments (Mutz 2011). In the online supplement, we report analyses that weight the sample to match key characteristics of the U.S. population (age, education, gender, and race/ethnicity). The estimates are somewhat less precise but lead to identical conclusions about problems, products, private regulation, and heterogeneous treatments (see Figure S2 and replication archive). Some effects of anti-corporate critique (p = .099 for prioritization), testimony (p = .274 for prioritization), and science (p = .161 for support) are less discernible using weighted estimates, suggesting anti-corporate critique most clearly moves the needle of support, whereas science and the explicit call for state intervention nudge decisions within a given choice set. The prosocial statement becomes harder to differentiate from strict rules and cutting suppliers in the analysis of support but remains distinct in other respects.
We probed the assumptions of the conjoint analysis, confirming balance on observables and finding no evidence of carryover effects across rounds or order effects for business and advocacy statements (see the online supplement). In addition, models with controls for respondent characteristics (Table S4a) show substantively identical treatment effects and that support for state intervention varies by gender (with men rating intervention .24 points lower on average), political ideology, and general attitudes about regulation (Table S4a). We found similar responses in additional split-sample models by age and education subgroups, which may differ in their modal form/amount of news consumption (see Section 6 of the online supplement).

JuSTIFICATIonS For STATE InTErvEnTIon
Further insight comes from our open-ended question, which asked respondents to explain why they prioritized a particular scenario. Discursive justifications can reveal socially salient vocabularies of motive and worth (Boltanski and Thévenot 2006;Tilly 2012).
Consider two examples from our respondents. One justified state intervention by contrasting the weakness of companies' prosocial statement in the first scenario to concrete corporate action in the second: "While both are important concerns, scenario 2 shows industry is taking steps to address the problem. Charity and 'community engagements' are nice-but again, it doesn't address the problem." Another respondent focused on the problem, articulating a rationale for why one problem (pollution of rivers and oceans) was more deserving of U.S. government intervention than another (dangerous working conditions): "To me pollution is something that affects everybody.
[T]hat is why I chose Scenario 2.
[T]he other is more concentrated into the industries." We developed a thematic coding scheme, starting inductively, and hand-coded a random sample of 1,800 open-ended responses (20 percent of total) (see Section 7 of the online supplement). 7 As shown in Table 2, 14 percent of codable justifications argued for state intervention by noting the insufficiency of the voluntary business response. For example, respondents were concerned that corporations' prosocial statement "is completely without personal responsibility" or "needs government intervention because the companies do not use an independent group to verify the standards." 8 This lends further credence to the quantitative finding that state intervention was perceived as more important when companies did not provide concrete assurances of private regulation. 9 Several respondents described corporate actions as legitimating government intervention (Dana and Nadler 2019), but this was a tiny slice.
In most cases, respondents discussed the labor or environmental problems at hand, and in doing so articulated a variety of rationales for state intervention. One rationale (present in 7.3 percent of cases) focused on how the problem harmed people at distant points of production (e.g., "chemicals released into the air seem more problematic to me because it would be impossible for people living in the area to avoid breathing") (see also Table 2). Less commonly, 4.2 percent of justifications described how the problem could spill over to harm people within the United Stateswhether as workers (e.g., "wage theft . . . has the effect of lowering wages in the U.S.A. for American workers") or as consumers (e.g., "toxic chemicals can also harm consumers as well as production workers"). Given the centrality of domestic material interests in theories of trade protectionism, it is striking that these themes appeared so rarely (see also Kreps and Maxey 2018).
In 8.3 percent of justifications, respondents suggested the problem could spill over to harm larger collectivities (e.g., "all of us," "everyone around the globe"), essentially justifying state intervention on the basis of cross-border collective goods/bads. This included concerns that "the entire planet will suffer from the overuse of natural resources, not just one community" or "the risk of spillover to workers in other communities as a result of abusive employers gaining a competitive advantage is tangible and serious." In many other cases, respondents described the problem using explicitly moral terminology (e.g., "immoral," "unconscionable") or broader injunctions, usually without specifying concrete consequences for those affected. An explicitly moral language of right versus wrong, moral versus immoral, good versus bad/evil was used in 11.5 percent of justifications (e.g., "child labor is immoral and it is disgusting"; "if a person does a hard job . . . and they do not get any sort of compensation, that is majorly wrong in my book"). 10 Among the justifications, 5.5 percent invoked broad categorical prohibitions (e.g., "child labor is unacceptable anytime and anywhere"; "modern slave labor should never even be talked about!"), often implying further justification was not needed. Others invoked normative principles (e.g., "I believe in a fair wage for your labor"), categorical distinctions (e.g., "overuse and illegal use are different"), or judgments of impropriety (e.g., "companies make enough profit without stealing from their workforce").
After starting inductively, we moved in two theoretical directions to better understand perceptions of the problems. First, we turned to March and Olsen's (2008) distinction between a "logic of consequences" and a "logic of "The entire planet will suffer from the overuse of natural resources, not just one community. I feel that the greater good of many outweighs the suffering of the few." Other problem consequences (14.2%) "Emission of toxic chemicals is far more dangerous comparing the two scenarios and I would certainly not like to be surrounded by it."

Logic of Appropriateness (23.5%)
Moral terms used to describe the problem (11.5%) "Child labor is immoral and it is disgusting. If corporations won't do anything about it then the government should impose their power on this horrible problem." Problem deserves categorical prohibition (5.5%) "Seafood companies should never over use natural resources so they can profit." Other statement about inappropriateness of the problem (7.7%) "I believe in a fair wage for your labor, I don't believe it is [our] right to tell other countries how they use their natural resources. On wages it must be freedom of choice to work for these companies, but if people are force[d] to work then our government must act." Other problem-centered rationales (20.7%) "The environment should be top priority right now." Other justification (7.2%) "My ex-husband's family lived in a location that involved a Superfund site that was created by a company that produced electronics." Agentic language (7.3% of problem-mentioning justifications) "The company is taking part in illegal activity which should be banned." appropriateness." A logic of consequences involves some estimation of benefits or harms, whereas a logic of appropriateness draws on "prescriptions of what is socially defined as normal, true, right or good, without, or in spite of, calculation of consequences and expected utility" (March and Olsen 2008:690). This distinction helps organize the vocabularies used by our respondents, while also allowing for variation within logics-such as the varying answers to the "consequences for whom?" question mentioned earlier.
As shown in Table 2, approximately onethird of codable justifications discussed the problem using a logic of consequences of some sort. This commonly meant discussing harms at the point of production, harms that spill over to "all of us," or specific harms without a precise specification of where they are felt. A logic of appropriateness was used in nearly one quarter of justifications. This often meant using explicit moral terminology without specifying concrete harms or invoking non-consequentialist normative principles (cf. Berman 2018).
We discovered that these logics were distributed unevenly across problems (see Figure 7). (For each problem in a prioritized scenario, we calculated the proportion of all codable justifications that utilized each logic. Figures S10 and S11 in the online supplement provide a disaggregated mapping.) A logic of appropriateness was far more likely to be used when justifying intervention into problems of child labor, modern slave labor, and unpaid wages. Although some respondents did use a logic of consequences to discuss these issues (e.g., "children need to attend school and get educated"; "their families are suffering because of this [unpaid wages]"), they more frequently drew on less consequentialist language of morality and fairness (e.g., "I don't think people should have to be unpaid while others are making tons of money off their unpaid labor").
A logic of consequences was far more likely to be used when justifying intervention for all four environmental problems in our study-water pollution, overuse or illegal use of natural resources, and emission of toxic chemicals. This included claims about how the problem could harm "all of us," Americans (e.g., "pollution in one country can spread around the world, so it can affect the U.S.A. and its inhabitants"), or others (e.g., "pollution has many implications for drinking water and the health of many people"). In contrast, it was exceedingly rare for respondents to use a logic of appropriateness to justify state intervention into environmental problems. It is certainly possible to view it as immoral or unfair to externalize environmental damage to distant sites of production (Clapp 2015), but our study suggests Americans almost never describe environmental problems this way. Why might this be? One possibility is that concern about environmental problems is tightly linked to human health, which evokes a language of concrete harms more than a language of appropriateness or fairness. Note that dangerous working conditions also tended to be discussed using a logic of consequences (e.g., "a dangerous work place is going to result in unchecked deaths and dismemberment of employees"), lending support to this explanation. On the other hand, two of our environmental problems-overuse and illegal use of natural resources-do not have obvious and direct human health implications but nevertheless evoked a logic of consequences. It may be that a logic of consequences makes common sense when problems pertain to risks in the surroundings (whether social or ecological), whereas a logic of appropriateness is evoked when problems involve an experience of mistreatment (e.g., being forced to work, having wages stolen). As we discuss in the conclusion, the historical development of labor and environmental movements, and their relationship to discourses of justice, politics, and science (Latour et al. 2018), may help explain this pattern.
Second, we examined attributions of agency. To interpret a situation as morally problematic, some theorists argue, one must see an agent behind it. This enables "reactive attitudes" about the situation-such as blame, resentment, or condemnation-rather than allowing it to be viewed as a matter of fact or circumstance (Strawson 1974). Lukes (2010) argues that this distinction provides a transposable foundation for a sociology of moral fields.
Our experimental treatments simply named problems (e.g., modern slave labor, pollution of rivers and oceans) as they might be seen in a news article, but we noticed some respondents used agentic language when describing them (e.g., "the fast fashion industry needs to be stopped from exploiting people"; "we can't continue to kill a resource, the ocean"). 11 We coded for agentic language, defined as specifying an agent (either a particular type of actor or a personal pronoun) and an action (e.g., taking, using, forcing, exploiting) involved in the specified problem.
(We excluded justifications about "wage theft by employers" here, because this treatment already specified an agent.) Although agentic language was not common (appearing in 7.3 percent of all justifications that mentioned the problem), there is suggestive evidence of an association between agentic language and explicit moral terms (see Table 3). Examining this relationship while adjusting for the problem present in the scenario reveals an infrequent but statistically significant link between agentic and moral language (p = .037).
Despite the limits of our open-ended responses for probing nuanced assumptions about agency, culpability, and reactive attitudes, this intriguing pattern highlights the utility of bringing sociological theories of morality (Hitlin and Vaisey 2013) into research on global markets. Agentic language was not confined to either labor or environmental problems (see Table S20 in the online Note: Two-tailed tests, with standard errors clustered by respondent. Adjusted estimates use a regression model to control for the problem in the prioritized scenario and then generate a predicted probability of using explicit moral terminology if agentic language was/was not present. supplement) but instead seems to capture an additional axis of variation in styles of justification and assumptions about moralized markets. In the conclusion, we propose a strategy for further investigating and integrating these axes.

DISCuSSIon AnD ConCLuSIonS
Global supply chains distance consumers from environmental damage and labor exploitation experienced in the production process. Over the past three decades, transnational corporations have constructed vast global networks, and activists have sought to bridge these distances to evoke feelings of sympathy, solidarity, or interdependence. Partly as a result, markets have come to be understoodwithin the logic of a broader moral field-as sites of moral action and positioning. At the same time, voluntary initiatives are struggling to demonstrate their effects, and a rising wave of state intervention is complicating neoliberal visions of frictionless global markets. Through a novel examination of popular judgments about state intervention, our study reveals widespread assumptions about responsibility and regulation in moralized global markets. We find clear evidence that corporate assurances of private regulation make state intervention seem less important, even in a crowded discursive arena where advocacy groups urge the strengthening of state controls. Surprisingly, corporate assurances matter across the political-ideological spectrum and regardless of the credibility of private regulation. By contrast, the effects of advocacy frames are much more contingent. Advocacy frames emphasizing corporate greed, and (less consistently) scientific evidence, do increase the perceived importance of state intervention, but other advocacy frames have more mixed effects, and none are effective for political conservatives-who reacted negatively to being told that greater state involvement is the solution. Corporate voices resonate more clearly, but only when they assure the audience that relevant reforms are occurring. Prosocial language of corporate ethics, charity, and community, despite its theorized benefits for corporate impression management, does not decrease support for state intervention. Concrete assurances of reform may be riskier for firms (Carlos and Lewis 2018), but they are more potent in the public imagination.
One question that remains is why so much credence is given to corporate assurances, even among groups that have faith in state regulation and that elsewhere register low levels of trust in business. One possibility is that individuals in the United States have far more implicit than explicit trust in corporations. Although a vast majority, especially on the left, report having "hardly any" or "only some" confidence in major companies in standard surveys (such as the General Social Survey), they may nevertheless uncritically believe companies when they hear promises of reform. A second possibility, which we see as more consistent with our findings, is that corporations have gained a kind of symbolic capital in the common sense of moralized global markets that elevates their assurances even in the absence of trust. (By symbolic capital, we mean a currency that derives from actors' position and monopolization of authority in a given domain [Bourdieu 1991].) Observers may pragmatically want someone to address distant problems-and see corporations as well-positioned to do this. They may not entirely trust companies to follow through-and may find advocacy frames about corporate greed appealing-but they see voluntary corporate action as an intuitive solution (regardless of its credibility) and state intervention as a second-best option. Advocacy campaigns may have contributed to this by highlighting the powers of well-known corporations more than the role of states in shaping labor and environmental conditions. Analytically, future research should assess whether the credence given to corporate assurances depends more on trust in business or pragmatic deference to position. Theoretically, scholars should consider what this configuration of the moral field means for the legitimacy of state action in a variety of global arenas.
Just as importantly, we found that distant labor and environmental problems differ substantially in both how much they are perceived as requiring state intervention and how intervention is justified. In addition to shocking abuses such as child labor and forced labor, reports of toxic emissions and water pollution generate high levels of interest in state intervention. Problems shape judgments more than corporate assurances do, and neither labor nor environment dominates. But the type of justification varies across problems. A logic of consequences predominates when justifying intervention into any of the environmental problems in our study, as individuals point out concrete harms to local people, to Americans, or to the world as a whole. A logic of appropriateness predominates for most labor issues, with individuals using explicit moral terms and categorical proscriptions in reacting to problems of mistreatment. Across problems, our evidence also suggests that explicitly moral terminology is linked to attributions of agency, which may indicate that individuals interpret the problem as condemnable rather than circumstantial.
This striking patterning of justifications raises questions about how labor and environmental issues have been framed historically-and in particular, whether different assumptions about agency and justice have come to pervade the public imagination of these issues. In discussing agency and environmentalism, Latour argues that social justice movements have worked for more than a century to "frame fights against inequalities and injustice" as amenable to political intervention, whereas environmental movements have been overly reliant on "objective science," such that "nature has indeed remained distinct and exterior to politics proper" even as environmental problems inspire fear and concern (Latour et al. 2018:354). Environmental justice movements have, of course, sought to counteract this (Bullard 1994), but our findings show a justice/fairness frame remains submerged when considering environmental problems in global supply chains. When confronted with problems that could be seen as imposing environmental burdens on distant populations for the benefit of consumers in affluent countries, almost none of our respondents used a language of fairness, injustice, or morality to justify intervention. What is needed, we believe, is a full analysis of how spatially dispersed labor and environmental concerns have been folded into moral fields over time, focusing on key oppositions between agency and circumstance, spillover and boundedness, and scientific and rightsbased claims. Such an analysis might also generate new insights into perceptions of spatially dispersed threats like climate change and into policies that attempt to meld labor and environmental concerns, such as green industrial policy (e.g., the Green New Deal).
Several limitations of our research should be kept in mind. First, we presented scenarios in textual form, but actual exposure to distant problems and corporate assurances may happen through visual imagery or audio narratives. To more fully understand responses to problems and claims, one could present them in non-textual ways. Second, our experiment was designed to register judgments across a range of scenarios but not to reveal the mechanisms behind them. To better understand the relevant mechanisms, studies could be designed to assess how responses to corporate assurances are mediated by trust or pragmatic deference (as described above), or how responses to problems are mediated by perceptions of spillover, injustice, vulnerability, or interdependence. Third, our conjoint experimental design is not well suited for examining interactions between treatments, so it is difficult to say whether advocacy frames about corporate hypocrisy can neutralize corporate assurances on specific issues, for instance.
Our findings carry practical and theoretical implications for a number of audiences. For practitioners and policymakers, our results reveal a unique terrain of public opinion that can serve as a resource for or constraint on policy change. We would not expect the perceptions we measured to translate directly into political behaviors or policy outcomes. Instead, we see them as pools of sentiment that can be mobilized by policy entrepreneurs at a moment of heightened attention to global supply chains. On one hand, our research indicates there is a substantial pool of public support for state intervention to address a subset of distant labor and environmental problems. This is especially apparent among liberals and women but can also be found in a substantial number of pro-regulation centrists (roughly 11 percent of our sample), who were highly supportive of bans in our scenarios (see Table S4c in the online supplement).
On the other hand, corporate assurances undermined support for state interventionacross the political spectrum and regardless of exactly how companies promised to regulate their suppliers. If policy entrepreneurs want to expand the role of the state in this context, they may need to consider two options. One is to leverage but transform private action-for instance, by strictly penalizing private performance failures or by providing incentives for effective remediation of problems rather than just risk management and reporting. A second option is to neutralize corporate assurances of private regulation-that is, to render them the equivalent of symbolic (and ineffectual) prosocial statements. In addition, some problems that are endemic in global supply chains-such as overuse of natural resources and non-payment of wages-may fail to stimulate interest in state intervention, at least without further elaboration. These barriers and pathways to reform should also be incorporated on a theoretical level in research on shifts from voluntary to mandatory approaches to corporate accountability (Seidman 2007;Tzankova 2021).
For the sociology of markets and morality, a key implication of our findings is that assurances of reform, even from market actors themselves, can grease the wheels of potentially objectionable markets. To date, this literature has advanced from highlighting contested commodifications to theorizing how objections are diminished by particular cultural frames (Almeling 2007) and structural arrangements (Schilke and Rossman 2018). Given the expansion of formal assurances in both global and national markets, it is worth moving into questions about how assurances alter perceptions of impropriety in different contexts, when their credibility is questioned, and whether their influence extends to highly contested commodifications (e.g., of sex, human tissues, or surveillance data).
Our findings suggest a collective sense of ambivalence about state intervention in global arenas, with implications for political sociology. On one hand, relatively few Americans are hostile to the idea of restricting markets to address (a subset of) distant labor and environmental problems, and individuals use a variety of rationales for doing so. On the other hand, corporate assurances alter perceptions to some degree-and it is clear that calls to restrict markets can provoke resentment, as in debates about consumption and climate change. A logical next step is to investigate how this ambivalence plays out in concrete policy reform projects, where perceptions of whether and why markets need restriction intersect with dynamics of voting, protest, and partisanship.
For research on perceptions of globalization, our study illustrates how fresh insights can be generated by attending to moral rationales and situational preferences. Our research reveals more morally-inflected perceptionsand to some extent, more similarity across the political-ideological spectrum-than one would expect based on analyses that root perceptions of globalization in material interests and ethnocentrism (Mansfield et al. 2019). But these are not mutually exclusive accounts. We suspect globalization is interpreted through a mix of concerns about distant problems, experiences of occupational dislocation or advancement, assurances that problems are being addressed, and in-group/ out-group boundaries.
Current concerns about forced labor in the Xinjiang region of China provide a case in point. Geopolitical rivalry and nationalistpopulist economic agendas may help explain the bipartisan wave of concern in the United States, but this certainly should not lead one to see objections to forced labor as epiphenomenal. Nor should it lead one to discount concerns that companies are lobbying against tighter restrictions by "claim[ing] they have deployed third-party auditors to verify that their supply chains are free of forced labor" (Turkel and Van Schaack 2021). The relative weight of different rationales may vary across individuals and policymakers, but a potent combination of rationales is likely to include moral objections to exploitation. What remains to be seen is whether the current focus on China is likely to overshadow concerns elsewhere or whether it could bring the U.S. state back in broader ways-as general principles get written into law, enforcement bodies expand their mandates, and a wider range of problems get defined as too damaging or inappropriate to leave to the market.

Tim Bartley is Professor of Sociology at Washington
University in St. Louis and Senior Lecturer in Sociology at Stockholm University. He is an organizational, political, and economic sociologist who uses a range of methods to examine environmental and labor standards, corporate power, and transnational governance. His 2018 book, Rules without Rights: Land, Labor, and Private Authority in the Global Economy, compares the implementation of fair labor and sustainable forestry standards in Indonesia and China. He is currently conducting research on parallel perceptions of globalization and immigration and on emerging regimes to regulate corporate accountability and artificial intelligence.