Abstract
This study argues that institutionalized authoritarian states, those with established political parties and legislatures, tend to pursue a two-pronged strategy concerning labor rights: they severely restrict collective labor rights but moderately protect substantive labor rights at the same time. This argument leads to an expectation of a decoupling between collective labor rights protection and substantive labor rights protection in institutionalized autocracies, or a strategic bifurcation. Using a novel global panel data on labor rights, it finds that (1) institutionalized authoritarianism does not protect collective labor rights better than noninstitutionalized authoritarianisms, (2) institutionalized authoritarianism protects substantive labor rights more than noninstitutionalized authoritarianisms, and (3) better protection of collective labor rights has no impact on the protection of substantive labor rights under institutionalized authoritarianism. The findings remain largely robust to alternative estimators, measures, reduced samples, and model specifications.
Notes
Notes
1 It would be preferable to have a unified measurement to test the hypothesized expectations. However, a valid measurement is difficult to construct. Instead, in robustness checks, I used seemingly unrelated regressions to redo the analysis, findings from which comport to the main results. Thanks to Reviewer 1 for suggesting this point.
2 In the original data, which cover all states, the maximum value of this variable is 10.
3 Cheibub et al. (Citation2010) also compiled data on authoritarianism, which code parties and legislatures separately until 2008, whereas Geddes et al.’s (Citation2014) data do not make such distinction but are more recent, with the exit year being 2010. Comparing the two, less than 1 percent of the total observations in Geddes et al.’s data for the overlapping time period are identified as party autocracies without legislatures as coded in Cheibub et al.’s dataset. In the section on robustness checks, I rerun all analyses using Cheibub et al.’s data.
4 There is a trade-off between adding LDV to control for a dynamic process (Keele and Kelly Citation2006) and minimizing the Nickell bias, due to including a LDV in a fixed effects model (Nickell Citation1981). I therefore did both. I also tried the weight of the past approach (Sullivan, Loyle, and Davenport Citation2012) employing the five-year moving averages of the two dependent variables in the robustness checks. The inferences remained unchanged. Thanks to Reviewer 2 for suggesting this point.
5 Due to space limitations, the tables are available in an Online Supplemental Appendix.
Additional information
Notes on contributors
Zhiyuan Wang
Zhiyuan Wang is an assistant professor in the Department of Political Science at the University of Florida. He earned his PhD in political science from Binghamton University, SUNY. His research interests are international and comparative political economy, human rights (including labor rights), international law and organizations, and law and politics in China. His work is forthcoming or has appeared in the Political Research Quarterly, New Political Economy, Political Studies, Social Science Quarterly, The Journal of Human Rights, China: An International Journal, Review of Law and Economics, and Human Rights Review.