Abstract
This article explores the association between various workplace-level HRM practices and a decline in union strength using a unique longitudinal survey data-set gathered in South Korea from 2005 to 2013. It addresses the underlying theoretical mechanisms by which HRM programs substitute for the roles labor unions have traditionally played. Suggesting more nuanced theoretical implications about HRM practices and union decline, statistical analyses reveal that workplaces that have implemented HRM practices have unions with a weaker organizational base than those without such practices, but that certain HRM programs correlate with unions with a strong collective voice in management decision-making. This article identifies the new roles for unions in South Korea in the era of HRM.
Acknowledgements
The research was funded by the RGSO Dissertation Competition Funding at the Pennsylvania State University (Fall 2015). I appreciate the helpful comments by Gary Adler, Laurie Ajan, Mark Anner, Forrest Briscoe, Kate Epstein, Roger Finke, John D. McCarthy, and Wayne Osgood on an earlier draft of this paper.
Notes
1. The use of the 1.5 score for neutrality in the union voice indicator may introduce some bias in the data. For a sensitivity test, I generated a random value between 0 and 3, and imputed the value for neutrality. This imputation may cause the estimates to be less efficient (due to a larger standard error of the predictor), but not biased (because the dependent variable does not have a specific value that may systematically affect the size of coefficients). The result is almost identical. The comparative table is available upon the request.
2. To address a different point of view, an unobserved factor might affect performance appraisal and union density and there might be no direct causal relationship between two variables. First, the establishment size is an important predictor of the adoption of HRM practices, and it is also directly related to the union density in South Korea (Jeong, Citation1995). The establishment size is, however, included as a control variable in the analysis. Second, specific types of unionized workforce could be in favor of (or at least not against) performance appraisal. The industrial type to which the workplace belongs could be considered a control variable in this context. However, the fixed-effect model employed in this manuscript excludes any possible effects of time-invariant predictors (such as industrial types) from the analysis.