To examine the hypothesized effect of ILO conventions on worker rights, we gathered time-series, cross-sectional data. Our unit of analysis is country-year and the time coverage of the analysis is 1981 to 2011, which is determined by the availability of data for worker rights. We first estimate models for a global sample of 152 countries for which the data are fully available. The inclusion of developed economies in the sample might bias the results since they tend to have relatively few incidents of major worker rights violations (Mosley and Uno 2007). Hence, to check the robustness of the findings, we also run models restricted to a sample of 128 emerging markets and developing economies as defined by the IMF’s World Economic Outlook Database (2013) during the time period of our analysis. A summary statistics table for all the variables used in the analysis appears in Table 3 in Appendix.

3.1 Outcome variables Our first dependent variable is the Worker Rights index, drawn from the Cingranelli-Richards (CIRI) dataset (Cingranelli and Richards 2012). The worker rights measure captures the extent to which workers are able to exercise and enjoy the outcomes of the following internationally recognized rights: “(A) the right of association; (B) the right to organize and bargain collectively; (C) a prohibition on the use of any form of forced or compulsory labor; (D) a minimum age for the employment of children; and (E) acceptable conditions of work with respect to minimum wages, hours of work, and occupational safety and health” (CIRI 2012, 65). It is an ordinal measure ranging from 0 to 2. A score of 0 suggests that worker rights are severely restricted whereas a score of 2 indicates that worker rights are vigorously enforced. The data for the variable are based on the U.S. State Department’s annual country reports on various political and socioeconomic rights practices.3 Our second outcome variable, Collective Labor Rights Practices, comes from Mosley and Uno (2007). This variable captures the extent to which the following collective worker rights are respected in practice: freedom of association and collective bargaining-related liberties; the right to establish and join worker and union organizations; other union activities; the right to bargain collectively; the right to strike; and rights in export processing zones. The data were originally gathered based on detailed country surveys and reports from three different sources: U.S. State Department of Annual Reports on Human Rights Practices, International Labor Organization Committee of Experts on the Applications of Conventions and Recommendations, and the International Confederation of Free Trade Unions Annual Survey of Violations of Trade Union Rights. The coding scheme used by Mosley and Uno is derived from Kucera (2002) and assigns a weight to each labor rights abuse for each unit of analysis (country-year). In our sample, the labor rights practice variable ranges from 1.5 to 27.5 with higher scores indicating better labor rights practices.4

3.2 Explanatory variables While the ILO has passed over 180 conventions, we focus our analysis on those that are recognized as its “fundamental” conventions to promote and protect worker rights. Specifically, we include the following: C29 (Forced Labour Convention) created in 1930; C87 (Freedom of Association and Protection of the Right to Organise Convention) created in 1948; C98 (Right to Organise and Collective Bargaining Convention) created in 1949; C100 (Equal Remuneration Convention) created in 1951; C105 (Abolition of Forced Labour Convention) created in 1957; C111 (Discrimination (Employment and Occupation) Convention) created in 1958; C138 (Minimum Age Convention) created in 1973. The data come from the ILO’s NORMLEX database.5 To test our hypotheses, we created the Number of Ratified ILO Convention variable. It accounts for the number of the core ILO conventions adopted by a country in a given year. Given that we focus on seven different ILO conventions, this variable enables us to capture the presence of any ratified conventions as well as the total core conventions ratified by a country. We also control for the impact of related treaties and international law that may also influence labor rights. As labor rights are ostensibly part of the broader array of human rights (Kang 2012), we include Total Human Rights Treaties to assess the possible impact of core human rights treaties on labor rights. This variable is the total number of the five core human rights conventions ratified by each country, specifically the International Covenant on Civil and Political Rights; the International Covenant on Economic, Social and Cultural Rights; the International Covenant on the Elimination of Racial Discrimination; the Convention on the Elimination of All Forms of Discrimination against Women; and the Convention against Torture and Other Cruel, Inhuman or Degrading Treatment or Punishment. The data are from the UCL’s National Commitment to Human Rights database6 and from various UN websites for the post-2009 period. The impact of global economic integration upon labor rights is contentious, and several studies have assessed the impact of economic globalization on these rights (i.e., Blanton and Blanton 2012; Mosley 2011; Neumayer and De Soysa 2006). To account for these impacts, we include measures of Trade and Foreign Direct Investment (FDI) Inflows. The trade measure is the natural log of total trade flows (imports and exports) as a percentage of GDP and the FDI variable is total FDI inflows as a percentage of GDP. Macroeconomic factors within a country may also influence its level of worker rights, so we include measures of GDP per Capita (logged) and annual Economic Growth. Earlier research indicates that countries that are less developed or face stagnant growth tend to be more prone to labor rights abuses than wealthier countries (Mosley and Uno 2007). The data for all the economic indicators come from the World Bank (2013). Political characteristics may also influence labor rights, in that they provide insights into the levels of freedom enjoyed by societal actors as well as the balance of power between government, labor, and business. First, studies have found a positive correlation between democratic regimes and the respect for socio-economic and political freedoms (Cole and Ramirez 2013; Mosley and Uno 2007; Poe et al. 1999). To account for the effect of political regime type, we include a Democracy variable. The index ranges from −10 to 10, where a score of 10 represents the highest level of democracy. The data for regime type come from the Polity IV dataset (Marshall et al. 2007). We also control for the possible positive influence of governments controlled by left-leaning political parties on worker rights using a binary variable, Left Government. We also include the Communist variable to control for communist regimes. Though they had a strong official structure for workers, they were nonetheless repressive of their workers in practice. We therefore expect to find that communist regimes might be more prone to violating labor rights. Another control variable included in the models is Past Worker Rights. In the models using the CIRI worker rights data, this variable is the initial (1981) value of the worker rights variable as a proxy for the history of each country’s treatment of its workers. In the models using the collective labor rights data, we include the initial (1985) value of the collective labor rights index variable (Mosley and Uno 2007) that allows us to thoroughly account for both the legal and practical environment in the protection of collective labor rights. Theoretically, countries that have a long history of respect for worker rights are more likely to have the institutions, procedures, and safeguards in place to protect and enforce worker rights. As the CIRI dataset relies on the U.S. State Department’s annual country reports, it is likely susceptible to the same problem as the CIRI physical integrity rights index, namely that the standards of accountability have changed as the State Department reports have become more rigorous over time (Fariss 2014). To control for this, we follow the strategy recommended by Fariss (2014), and include two additional variables in the models that use the CIRI worker rights variable as the outcome variable. Namely, we add a time trend variable that counts the number of years since the first year for which the worker rights data are available. We also include a Latent Empowerment variable. This variable was originally generated by Schnakenberg and Farriss (2014) using a dynamic ordinal item-response theory model and based on seven different empowerment rights in the CIRI dataset that include worker rights, freedom of assembly and association, freedom of speech, freedom of foreign movement, freedom of domestic movement, freedom of religion, and free and fair political participation. We lag all of the time-variant explanatory variables one year in order to ensure that they temporally precede our outcome variables. While the suggested bias may be a lesser issue for the collective labor rights data since they were collected using multiple sources, we include a trend variable in the models using the collective labor rights variable as the dependent variable to minimize any similar bias that may exist.