![Sample our Politics & International Relations journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5947/TOC-Subject-Sample-2021-Politics-International-Relations.jpg"})
Abstract
Why do states comply with international human rights law? Scholars have looked to domestic politics to suggest a constituency-driven domestic compliance mechanism. However, it is unclear if strong public support for compliance leads to greater willingness to comply among elected officials. This study is the first to empirically demonstrate the complete causal chain of a constituency-driven domestic compliance mechanism. Using a survey experiment, this study first finds that a strong state commitment to UN Norms on the Responsibilities of Transnational Corporations (RTNC) increases public support for compliance. Using a parallel lab-in-the-field experiment of sitting legislators, the study then finds that a strong state commitment to RTNC is similarly associated with increased elite support for compliance. Finally, the study finds that public demand for compliance is associated with even greater elite support for compliance. Public demand for compliance seems to elicit elite willingness to comply by raising reputational concerns.
Notes
Acknowledgments
I would like to thank Beth Simmons, Josh Kertzer, Muhammet Bas, Seth Soderberg, David Romney, Boram Lee, participants of the Harvard International Relations Workshop, and two anonymous reviewers for offering helpful critique and advice. I am also grateful for the financial support of the Weatherhead Center for International Affairs and the Harvard Institute for Quantitative Social Science. Finally, this study would not have been possible without several Georgia and Massachusetts state legislators who took part in my lab-in-the-field experiment. All errors are my own. Harvard IRB Protocol #IRB16-0345.
Notes
1 It should be noted that although the averages differed between elites and the public in LeVeck’s (2014) study, a subsequent study demonstrated there is much more variation within groups than between groups (see Kertzer Citation2016).
2 The Norms on the Responsibilities of Transnational Corporations and Other Business Enterprises do not have a commonly used shorthand or acronym. The acronym RTNC is used for convenience in the context of this study.
3 Other than RTNC, which itself is not a formal legal instrument, there are no international legal instruments to bind states to regulate the human rights practices of business entities. After the study’s surveys were completed, the open-ended intergovernmental working group on transnational corporations and other business enterprises with respect to human rights (OEIGWG) published the Zero Draft of the UN Binding Treaty to regulate corporations in regard to international human rights law in July 2018 and discussed elements of the Draft in the Fourth Session of OEIGWG in October, 2018 (United Nations Citation2018). The outcome of the Draft remains to be seen.
4 For instance, Wallace (Citation2013, Citation2014) focused on torture. Chilton (Citation2014b) focused on indiscriminate bombings. Two rare exceptions are Chilton and Sarfaty (Citation2017), who focused on whether consumers understand supply chain disclosures designed to improve by corporate human rights behavior, and Putnam and Shapiro (Citation2017), who focused on economic sanctions for forced labor.
5 For the purposes of this article, I considered transnational corporations to be equivalent to big businesses as used in the Gallup poll. Gallup does not specify big business in its questionnaire, but Gallup does differentiate between big business and small business. As such, Gallup’s respondents would likely consider big business to mean transnational corporations, as used in this article. Also, Republicans are more likely to trust in big businesses (22 percent) compared to Democrats (15 percent), suggesting the possibility of partisan differences in trust of transnational corporations. In either case, however, public trust in big business is relatively low (Gallup Citation2017).
6 For comparison, the 2016 Gallup polls noted that 68 percent trust small businesses “quite a lot” or “some.” Seventy-three percent trust the military. Fifty-six percent trust the police. Thirty-six percent trust the president. Twenty-seven percent trust banks.
7 The particulars of the hypothetical vignette has some basis in reality. An Alabama-based US mining company (Drummond Company, Inc.) was accused of hiring paramilitary forces in its subsidiary in Colombia to forcibly displace residents from lucrative mining sites (see Happe and Petz Citation2012). However, to be clear, in accordance with IRB protocols against the use of deception, respondents were informed before reading the vignette that the questions they were about to read were hypothetical situations. The exact language of the disclosure was as follows: “The following questions are about hypothetical domestic and international laws. You will read about hypothetical scenarios our country has faced many times in the past and will probably face again. Different leaders have handled the situation in different ways. We will describe one approach US leaders are considering and ask whether you approve or disapprove.” The prevignette disclosures draw from other studies, including Kertzer (Citation2016: 188). It should be noted that an alternative option was to disclose at the end of the survey that the vignette was a hypothetical. However, this alternative would have required the temporary use of deception, which would have raised additional ethical concerns.
8 For a similar experimental design, see Chilton (Citation2014a, Citation2014b).
9 The most prominent classification of international law employs three dimensions of legalization: obligation, precision, and delegation. Obligation is the degree to which states are bound by or committed to a particular law (see Abbott et al. Citation2000). Simply put, although some laws are “hard laws” that use the language of “must,” other “soft laws” use the language of “should” (see Simmons Citation2010). Laws of low obligation (i.e., soft laws) entail a weaker commitment, and laws of high obligations (i.e., hard laws) are usually codified in concrete terms and entail a strong commitment. The United States participated in the drafting of RTNC, and as such, this particular treatment condition is based on reality.
10 Unlike the weak commitment treatment, the United States has not made a strong commitment in reality. It is possible that more knowledgeable subjects were aware that such a commitment did not actually exist. However, none of the respondents receiving the strong commitment treatment discussed in open-ended answers an inability to suspend disbelief. Therefore, it is likely that none of the respondents was sufficiently knowledgeable about the actual state of international law; and if any were knowledgeable, there is no evidence they were unable to suspend their disbelief. The same applies for elite survey respondents.
11 It should be noted that the control treatment of no information left ambiguous the existence of international human rights law regarding unethical business practices. In that sense, the treatment did not represent a strict “no commitment” treatment. The existing literature nonetheless uses equivalent control conditions of no information to attain a baseline measure for comparisons with active treatment conditions (see Wallace Citation2013; Chilton Citation2014a). The alternative would have been to explicitly claim the nonexistence of any international law regulating unethical business practices. This alternative would have resulted in an unrealistic level of deception, given the status of RTNC. Consequently, the no information treatment instead offered the most reasonable research design.
12 The sanction was defined in the survey text as freezing the firm’s assets. There are admittedly other forms of sanctions the government can take against a disreputable company that could qualify as compliance. If respondents who chose not to support freezing the firm’s assets supported some other form of sanction, they would be willing to comply with RTNC but would have been incorrectly identified as unwilling to support compliance in this study. However, the survey text specified the type of sanction as freezing the firm’s assets because such an action was more readily understandable for the average survey respondent than other sanctions, and because leaving the sanction unspecified would have raised serious ambiguities for survey respondents regarding which sanctions they were supporting.
13 The vignette’s wording of the hypothetical firm “pressuring business owners to give up their assets below market value” was motivated by accusations that Drummond Company was hiring paramilitary forces in its subsidiary in Colombia to forcibly displace residents from lucrative mining sites. The wording of the text was not made more specific than that in order to capture a wider range of similar human rights abuses involving companies illegally pressuring other parties, and to obscure any direct identification of Drummond Company.
14 MTurk samples usually are not representative and tend to overrepresent younger, liberal, lower-income individuals. Fortunately, others have found that studies using samples from MTurk can be replicated using nationally representative samples (see Berinsky, Huber, and Lenz Citation2012). I have no theoretical reason to expect otherwise for the sample used in this study.
15 I used the same language as the public opinion survey informing respondents that the following questions were based on a hypothetical situation. “The following questions are about domestic and international law. You will read about one hypothetical scenario our country has faced many times in the past and will probably face again. Different leaders have handled the situation in different ways. We will describe one approach US leaders are considering and ask whether you approve or disapprove.”
16 The omission of the weak commitment treatment was due to statistical power concerns given the expected sample size of elite respondents.
17 Due to sample size constraints, I did not vary the level of public support. A promising future study could include an active treatment group that assigns elite respondents a high level of public opposition to compliance, as opposed to public support for compliance.
18 At the end of the survey, I reiterated to elite respondents who were assigned to either one of the two active treatments that the United States had not made a strong commitment to RTNC. The same was not done for public respondents. This second precautionary disclosure for elites was due to ethical concerns of misinforming legislators on the actual state of the law. This second disclosure was not included in the public opinion surveys, in which respondents were only told that the questions were hypothetical before the experiment. On the whole, elite respondents seemed generally unaware of the state of RTNC. If, however, elites were aware of the lack of a strong legal commitment to RTNC, the experiment would have raised validity concerns, as the hypothetical would have been less believable and elites may have had greater difficulty suspending their disbelief.
19 Similar to the public opinion surveys, the vignette did not specify precisely where the jobs would be lost—or, specifically, whether the jobs would be lost in the legislator’s district. The alternative of specifying that the legislator’s district would lose the jobs would have required most elite respondents—whose districts do not have a major company that employs thousands of workers—to suspend their disbelief and to think as if there were such a firm that employed thousands of constituents. Such a level of abstraction would have been unrealistic and unscientific. Thinking in abstract terms is already a concern with hypothetical vignettes in survey experiments. Admittedly, having the domestic distribution of costs be such that the legislators’ constituents are not affected does reduce the costs of compliance somewhat, as the economic interests of the legislator’s district will not be directly affected by compliance. One might expect reduced levels of compliance from elites whose constituents would be losing jobs as a result of compliance, and as such, the results here may not be generalizable to such elites.
20 A total of 102 elites were randomly selected from 200 sitting Massachusetts legislators and 236 sitting Georgia legislators. Each of the legislators was invited to participate in an in-person interview. More specifically, 31 Massachusetts legislators (six conservatives and 25 liberals) and 42 Georgia legislators (27 conservatives and 15 liberals) agreed to participate. This resulted in a relatively high response rate of 72 percent. Maintaining a high response rate was important because elites who chose not to participate could theoretically be elites who were less responsive to public opinion and thus less affected by the final active treatment. I made repeated visits to the offices of the legislators rather than making phone calls or writing emails, to which legislators are understandably less responsive.
21 Under most circumstances, researchers interested in elite opinion are forced to rely on quasi-elite subjects and extrapolate from their findings. My sample meets, if not exceeds, such standards. For instance, others have used graduate students drawn from the Harvard Kennedy School as substitutes for elites who regularly make foreign policy decisions (see Renshon Citation2014). Others have drawn from business executives, former elected officials, and trade negotiators (see Hafner-Burton et al. Citation2014). Rare instances in which elite respondents are drawn directly from the population of interest include Doshi, Kelley, and Simmons (Citation2016); Bayram (Citation2017a); and Renshon, Yarhi-Milo, and Kertzer (Citation2018).
22 In Crosby v. National Foreign Trade Council, 530 U.S. 363 (2000), the US Supreme Court unanimously struck down the 1996 Massachusetts Burma Law that restricted state entities from doing business in Burma due to Burmese human rights concerns. The court reasoned that the state law violated the Supremacy Clause of the US Constitution.
23 In 2011, former Massachusetts House Speaker Salvatore DiMasi was sentenced to eight years in federal prison for political corruption (see Valencia Citation2011). DiMasi was the third consecutive Massachusetts House Speaker found guilty of federal charges, and the Massachusetts state legislature has since proposed measures to curb unethical business practices involving state contracts (see Eldridge Citation2009).
24 Some interviews took place in conference rooms near their offices, and some meetings took place after business hours to accommodate legislators’ schedules.
25 I did not time each interview, but my best estimation of the length of the interviews range from twenty minutes to over an hour, which suggests that elites seriously contemplated the policy conundrum I presented them.
26 See Grossman (Citation2011) for an overview of the advantages of lab-in-the-field experiments.
27 The response variable, willingness to comply, was converted from the original Likert scale to percentage points, in which “strongly agree” on the Likert scale was converted to 1.0, “agree” to .833, “somewhat agree” to .667, “neither agree nor disagree” to .500, “somewhat disagree” to 0.333, “disagree” to .167, and “strongly disagree” to 0.
28 Responses from three elites could not be classified into any of the four categories. Those three were removed for the following analysis only. Putting aside the three responses, all of the remaining responses were grouped into one of the four mechanisms. The groups were mutually exclusive.
29 It is possible that some elites were concerned that their responses would not remain anonymous, despite my assurances that their responses would be deidentified. Such elites may have chosen to support costly compliance while waxing about general rule of law or reputational concerns. In order to assure elites that their responses would remain anonymous, I made clear that their responses were not being audio-recorded and that the hand-scribed responses would remain within my sole possession before deidentification. As stated above, I also assured elites that their deidentified responses would not be quoted.
30 It should be noted that respondents in the market mechanism category who supported other market measures, such as public shaming, rather the freezing of company assets generally opposed compliance as defined in this article. There are admittedly other nongovernmental forms of sanctioning firms that could arguably be defined as measures that amount to compliance with RTNC. However, those nongovernmental sanctions are not the form of compliance theorized in the constituency-driven domestic compliance mechanism that requires constituents to pressure political elites to use the power of the government to sanction human rights violations.
Additional information
Notes on contributors
Matthew Kim
Matthew Kim is a student at Harvard Law School and a graduate affiliate at the Institute for Quantitative Social Science at Harvard. He earned his PhD in political science from Harvard University in 2017.