[Author Affiliation]Charlene M Kalenkoski, Department of Economics, Ohio University, Bentley Annex 351, Athens, OH 45701, USA; kalenkos@ohio.eduDonald J Lacombe, Department of Economics, Ohio University, Bentley Annex 345, Athens, OH 45701, USA; lacombe@ohio.edu; corresponding author[Acknowledgment]The authors gratefully acknowledge invaluable comments from Barry Hirsch, Jim LeSage, and an anonymous reviewer, and would also like to thank Tatevik Sekhposyan and Nicholas Prala for their excellent research assistance.1. IntroductionRight-to-work (RTW) laws prohibit the requirement that a person become a union member as a condition of employment. Such a prohibition, if effective, raises the cost of union organizing activity, leading to a decline in union membership and thus in union bargaining power. To the extent that this reduction in the bargaining power of unions occurs, firms considering locating in an RTW state may expect lower wages and a more favorable business climate than would be the case in a non-RTW state, leading to greater employment in RTW states, all else equal. Therefore, it is important to determine whether these laws are effective.Many studies have documented a negative correlation between RTW laws and unionization rates, suggesting that RTW laws do indeed decrease unionization rates, although some of this correlation is thought to be due to negative public perceptions of unions in right-to-work states rather than the effect of the laws themselves. There have also been a few studies that have investigated the effects of RTW legislation on the level of and growth in manufacturing employment, but the results of these studies are mixed, with the results hinging on the particular econometric specification used. A regression of the manufacturing employment share on an RTW dummy variable often suffers from omitted variable bias, and each study attempts to deal with this bias differently. Potential omitted variables are climate, soil quality, the availability of natural and labor resources, infrastructure, and public attitudes toward unions and business, all of which are thought to be determinants of employment but are not easily measured. If public attitudes are probusiness, then omitting measures of public attitudes may positively bias the RTW coefficient. This is because probusiness attitudes may lead both to passage of RTW legislation and to other probusiness policies that increase employment. Another example may be the percentage of the population that consists of recent immigrants. Businesses that employ recent immigrants may be more likely to vote for RTW legislation and may employ more people. In addition, there may also be spatial correlation in the errors, because these omitted variables are likely correlated across counties. Weather, resources, infrastructure, and public attitudes usually do not change abruptly at political borders, and therefore these omitted factors would necessarily be geographically correlated. Another possibility is that manufacturing employment itself may be correlated across counties as a result of agglomeration economies, which are cost savings that result when firms locate in close proximity to one another. For example, firms may locate close to bodies of water such as rivers to take advantage of natural shipping routes, and any variables associated with firm activities would be geographically correlated. Finally, measurement error is also a possibility if the relevant unit of measurement is the city but we are measuring our variable at the county level.1 The potential presence of omitted variables bias and spatial dependence and the techniques used to address them are the primary concern of this paper.As will be seen in the literature review below, manufacturing has been the primary industry of interest with respect to analyzing the effects of RTW legislation because of the high percentage of manufacturing workers that are unionized. …
Read full abstract