Density delay in the evolution of organizational populations: a model and five empirical tests

Administrative Science Quarterly,  Sept, 1989  by Glenn R. Carroll,  Michael T. Hannan

• E-mail
• Print
• Link

National Labor Unions

Unlike the other kinds of organizations under study, labor unions do not rely on commercial viability in the marketplace in order to survive, yet unions are historically very important organizations. The existence of unions and their actions have shaped American industrial structure and politics. Consequently, unions have attracted the attention of historians and other scholars, and their histories, too, can be reconstructed from archival sources. The data on national unions used here come from an effort to collect information about every national labor union that has existed, however briefly, in the United States during the period from 1836 (the year of the first national unions) to 1985. This coding began with lists of names and starting dates from reports published in various years that claimed to cover the population of unions. These publications yielded an initial master list. This set was extended by consulting standard histories of the labor movement, which were especially useful for the early period, 1830-1870. This procedure identified 621 national unions with usable histories. More detailed discussion of the design used in this study can be found in Hannan (1988b) and Hannan and Freeman (1989: ch. 7). Figure 4 shows the trajectory of density over the 150-year period, beginning with the founding of the first national union. The number of national unions fluctuated near zero until the Civil War era, rose modestly until about 1881, and then grew explosively until 1905. From that point on, growth in the number of unions was slower and more erratic until the series reached its peak level of 211 in 1954. The last portion of the series shows consistent but modest contraction in the number of unions. This trajectory, like those for newspaper populations, fits the pattern discussed at the outset. In analyzing mortality, we considered events of "disbanding," by which we mean that a union ceases to operate as a national union without merging with another union. We concentrated on disbanding here rather than mortality due to all causes (including mergers) because earlier research reveals that the pattern of density dependence is quite different for the various forms of mortality in the population of national labor unions (Hannan, 1988a). Roughly one-third of the unions studied were disbanded (191). Because previous research indicates that rates of mortality have varied over time in response to changes in the legal standing of unions and major organizational changes in the national union movement (Hannan and Freeman, 1987, 1988a), we used a set of five period effects in the analysis reported below. The first period begins in 1836, the start of national unionization. The second begins in 1887, following the formation of the American Federation of Labor (AFL). The third begins in 1932 with the New Deal, when unions acquired substantial legal protection. The fourth begins in 1948, following passage of the Taft-Hartley Act, which eliminated some New Deal provisions. The fifth period begins in 1955, the year of the merger of the AFL and the CIO. We also included a dummy variable that distinguishes years of economic depression and a pair of dummy variables that indicate whether a union began with a founding or a secession from an existing union (the omitted category is "started by merger"). We did not include counts of prior foundings and disbandings in these analyses because earlier research revealed that such effects are negligible for the population of labor unions (Hannan and Freeman, 1989: ch. 11).