UC Irvine

This dissertation contains three chapters on the political economy and economic history of labor unions. It uses microeconomic theory to explore the strategic environment in which two players (such as a union and firm) bargain in the shadow of conflict. It also uses newly compiled data on union formation in the United States to estimate how import competition and automation contributed to union decline since the 1950s.

The first chapter specifies a game theoretic model for settings such as litigation, labor relations, or arming and war in which players first make non-contractible up-front investments to improve their bargaining position and gain advantage for possible future conflict. Bargaining is efficient ex post, but we show that a player may prefer conflict ex ante if there are sufficient asymmetries in strength. There are two sources of this finding. First, up-front investments are more dissimilar between players under conflict, and they are lower than under bargaining when one player is much stronger than the other. Second, the probability of the stronger player winning in conflict is higher than the share received under Nash bargaining. We thus provide a rationale for conflict to occur under complete information that does not depend on long-term commitment problems. Greater balance in institutional support for different sides is more likely to maintain peace and settlements.

The second chapter investigates why private sector union formation fell away so much in the United States since the late 1950s. Featuring an improved dataset on National Labor Relations Board (NLRB) representation elections, I present evidence that import penetration accounts for 42-55% of the decline of U.S. manufacturing unions. Furthermore, using a shift share analysis, I show that employment migration away from manufacturing, a traditionally unionized sector, and into services, a scarcely unionized sector, accounts for at most 2/5 of the overall decline.

The third chapter evaluates how automating technologies may have contributed to union decline. After discussing the history of post WWII automation and its interaction with routineness in labor markets, I outline three hypotheses for how it affected union formation in the United States. I then present the available measures of automation such as Routine Task Intensity to set up an empirical analysis using the same panel data as in chapter 2. Results suggest that automation did not have a significant effect on formation rates between 1975 and 2010. It is possible, however, that these regressions capture competing forces that cancel each other out and mask more nuanced effects.