The organization of state and local government employees: Reply

Abstract

Joseph D. Reid, Jr. and Michael M. Kurth (hereinafter R-K) raise interesting questions concerning the assumptions made and the explanatory variables used in my model of the organization of state and local government employees. I assert that models of private sector unionization are not directly transferable to the public sector because public employees participate in the hiring of their own employer. R-K believe that my regressions do not address this issue. The regressions were designed to explain the growth of public labor organizations, not the difference in sectors of the economy. In developing a model, economists typically make certain assumptions about the conditions surrounding the area under consideration and proceed to test within these parameters. We do not test each assumption, but make what seems to be reasonable assumptions. R-K employ the same logic when they exclude state employees from their regressions because "state employees are less likely to strike and are more supportive of civil service commissions than are local government employees." Their regression does not test for this condition. R-K are concerned with the dates of the variables used. "... if analyzing levels, make sure that the times for generation of these levels are normalized appropriately." The dependent variable and three independent variables are for 1977, while the law variables and percentage of nonagricultural employees organized are for 1976. These are the dates for which the data are published. Unless substantial changes occur within one year, which seems doubtful, the results would not seem to be biased by using the available data. R-K would substitute percentage change in average salaries from 1957 to 1972 for the current level of earnings. My use of the level of current income to measure how much of union services employees will buy is meant to be in accordance with standard economic theory. Higher income results in a larger quantity purchased of normal goods. In some sections of the country union membership may be considered an inferior good. To avoid using Two Stage Least Squares, they used as a wage variable the "change in wages from 1957 to 1972, which is unaffected by union growth in the following decade." This implies that union growth from 1972 to 1980 is unaffected by changes in wages from 1972 to 1980. I question this implication. R-K use their earnings variable as a measure of satisfaction with compensation.