The comment by John Barron and Dan Black and the note by Duane Leigh (published December 1981, AER) are on the theoretical model and empirical evidence, respectively. We will respond to their comments in this order and offer new empirical evidence based on a panel component of the national Quality of Employment Survey of the Survey Research Center of the University of Michigan. The central point of the theoretical model is that given a world where there are unforeseen changes in the parameters facing a firm,' such as changes in prices or production technology, two conditions will necessitate some intragroup mechanisms, such as union procedures for grievances and reaching consensus, in order to realize efficiency. These two conditions are firm-specific attachment costs, such as search and training costs, and workplace public goods, such as the speed of the production process in assembly lines or the scheduling of workshifts.2 Firms differ in both attachment costs and extent of workplace public goods. At one extreme, suppose firm-specific attachment costs are so large that no workers can leave or come into the firm. Then in the presence of workplace public goods, the workers are faced with a traditional public finance problem of selecting the optimal level of the public good. At the other extreme, suppose there are no firm-specific attachment costs. Then we have the common assumption of a frictionless equilibrium in which workers move across firms and the public goods problem may be solved in the manner suggested by the Tiebout (1956) model: each worker selects his desired bundle of labor income and the shared working condition by choosing among firms. This process is consistent with the type of hedonic (labor) market equilibrium characterized by Sherwin Rosen (1974). From the point of view of an individual firm, efficiency in the intermediate case of some firm-specific attachment costs combined with shared working conditions or workplace public goods leads to a blend of market adjustments and the solution to a public goods problem for a less heterogeneous group of workers than in the pure public goods equilibrium. The comings and goings of workers can be thought of as one part of the process, and solving the public goods problem for those who are there after the comings and goings is another. In our formal model the role of attachment costs (in the form of a search cost, c) is highlighted. Our approach to modelling working conditions and unions is consistent with recent empirical work, including the observation that unionized workers have lower quit rates and thereby can have greater firm-specific human capital as indicated by higher productivity (Charles Brown and James Medoff, 1978). However, our approach adds the additional prediction that unions will be more likely to be present in certain work settings; specifically those with shared working conditions. Our approach therefore differs somewhat from that proposed by Richard Freeman (1980) since he does not emphasize the role of shared working conditions in encouraging unionism. *University of Michigan. 'More precisely, a workgroup. Some working conditions are firm-wide but many apply at the level of the workgroup or production site. 2An interesting historical example of workshift scheduling is the transition from two twelve-hour shifts to three eight-hour shifts in the steel industry. For some of the inherited workers an eight-hour day was not long enough (at the old wage rate), and even if none of the inherited workers quit, the firms needed 50 percent more workers. The transition was characterized by acrimony and bloodshed (William Hogan, 1971).
Stored only when both papers are in the corpus — 156 citations exist upstream.