دستمزد سرریزها در مذاکرات قرارداد بخش دولتی : اهمیت مقایسه های اجتماعی

We explore the existence of wage spillovers in public sector teacher contract negotiations. We focus on the role that informal social comparisons have in determining wages. Using a combination of survey and administrative data, we estimate the relationships among a district's negotiated salary and the wages negotiated in that district's reference districts. Using panel data and spatial econometrics, we control for observed and unobserved factors that jointly determine salaries in a local labor market to isolate the causal influence of wage spillovers. We find that there are indeed causal relationships among salaries and that union “strength” influences these relationships.

Estimating the magnitude of wage spillovers in collective bargaining has been a popular research area for the past 50 years. However, most previous research has been quite narrow in its focus, concentrating on the role that formalized union pattern bargaining has had on negotiated outcomes Budd, 1992, Budd, 1995, Budd, 1997, Erickson, 1992, Erickson, 1996, Ready, 1990, Levinson, 1960 and Alexander, 1961. In pattern bargaining, the union selects a target employer to negotiate with first and then attempts to impose this “pattern” on other employers in the industry. While pattern bargaining may be an important determinant of wages in some sectors, it ignores the more pervasive, less formal, convention of referring to sets of wage “comparables” during collective bargaining (Horowitz, 1994). Both union and management engage in this practice of “social comparisons,” and it can have significant implications for the outcomes that are negotiated. There are several reasons why both sides refer to wages negotiated in other organizations. First, management and union negotiators are able to reduce uncertainty and minimize transaction costs by referring to a set of mutually agreed upon referents. This allows them to more quickly narrow in on a smaller set of potential outcomes. However, each side will also introduce referents that deviate from the common set in ways that serve their own constituencies. They may do this to gain a strategic advantage or because they have come to believe this biased set is the appropriate comparison group (Babcock et al., 1996). Second, union negotiators may use reference groups because workers are concerned explicitly with how their wages compare to those of workers in similar organizations. Research has indicated that workers care about the fairness and equity of relative wages Easterlin, 1995, Dusenberry, 1949, Frank, 1985, Kahneman et al., 1986, Felstinger, 1954 and Pollack, 1976. This will put pressure on elected union negotiators to bring the workers' wages in line with those of their referents. Third, success at the negotiating table relative to comparable organizations may provide a useful signal to the negotiators' constituencies. This is especially important in public organizations where direct measures of managerial competence may be less available than, say, the stock price of a private company. In agency models, such as the one laid out by Besley and Case (1995), voters compare local outcomes to the “yardstick” of outcomes in similar jurisdictions and base votes on this evaluation of their elected officials' performance. If constituents base their votes on relative wages, then public management negotiators will explicitly care about these yardsticks. Indeed, union members may also use relative wages as a signal of their leadership's skills in dimensions for which they lack direct measures. This paper explores the extent to which the references that both sides make to wages in other organizations affect negotiated outcomes of teacher salary contracts in public school districts. In doing so, we extend the literature on wage spillovers in several ways. First, rather than focusing only on the union's use of pattern bargaining, we examine how the choice of referents that both sides make may affect the negotiated outcome. Our effort distinguishes the impact of the set of referents that both sides agree upon from the impact of those chosen by only the union and those chosen by only the employer. Second, we explore whether the impact that each sides' referents have on the negotiated outcome depends on each side's “bargaining power” in the negotiations. For example, it may be that in environments in which the union is very powerful, their list of referents greatly affects the negotiated outcome. In environments in which the union is not powerful, their referents may hardly matter. By uncovering the circumstances under which unions can use their strength to achieve their goals, we will gain insight into the mechanisms by which union strength is exerted. There has been a lively debate about whether it is possible to obtain legitimate estimates of wage spillovers. Contributors to the pattern bargaining literature have recognized that it is difficult to isolate true wage spillovers that arise from institutional factors in the negotiations process from spurious correlation due to market forces Mitchell, 1982, Budd, 1997 and Addison and Burton, 1979. Because factors affecting labor demand and labor supply will affect wages for many firms within a single industry and even for firms from different industries whose markets are related, empirical estimates of spillovers can be biased by the inability to control for these labor market factors. Indeed, in a more general model of endogenous social effects, Manski (1993a) has shown that such effects are not identified in cross-sectional models.1 Public sector unionized labor markets are an especially promising setting in which to study wage spillovers. It is commonplace for the two sides to refer to outcomes negotiated in “comparable” municipalities during public sector contract negotiations (Horowitz, 1994). Because the competitive pressures are less strong in the public sector than in the private sector, issues such as concern for equity may have a larger impact and increase the potential for wage spillovers. This concern for equity has been institutionalized into the collective bargaining statutes of many states. These statutes often require that wages in comparable communities be explicitly considered during the contract negotiations process Brown and Medoff, 1988 and Horowitz, 1994. Additionally, public sector managers, who are accountable to the voting (and potentially migrating) public, will care very much about how the wages they negotiate compare to wages offered by comparable communities. Our effort to distinguish spillover from other sources of wage correlation is made possible by two important aspects of the negotiation process in the industry we examine. First, the union and employer negotiators in our industry use distinct criteria for choosing their referent organizations. These criteria include, but are not limited to, geographic proximity. We have unique survey information on referent choices that allows us to distinguish the sets of referents from the set of organizations that merely share a geographic labor market. Second, the staggered longitudinal structure of negotiations in our industry provides a natural division in any given year between the organizations that are currently negotiating and the organizations that are covered by contracts negotiated in previous years. Spillover arises from the reference by negotiators to existing contracts and is empirically distinguishable from correlation among concurrently negotiated contracts within the same geographic area that arises due to common labor market factors. Our estimation procedure exploits the overidentifying restrictions that arise from these institutional aspects of the negotiation process. We implement some specification checks that demonstrate our estimates of spillover are not merely a reflection of unmodeled or unobserved exogenous sources of correlation. As an extension of our basic model, we explore the circumstances under which the union and the management sides can successfully negotiate a contract that resembles that of their preferred set of referent contracts. We create an exogenous measure of relative union strength based on time-varying geographic-specific objective measures of support for union activity. By interacting this measure of union strength with the average wages of each side's referent contracts, we provide the first evidence regarding this avenue for the exercise of union power. To investigate public sector wage spillovers, we utilize data from Pennsylvania school districts during the mid-1980s. The spatial model presented in the next section uses the time-series and cross-section structure of panel data to account for observed and unobserved factors that jointly determine salaries in local labor markets. This enables us to isolate the causal influence of wage spillovers. Section 3 describes the survey and administrative data, Section 4 discusses our estimation strategy, and Section 5 presents the empirical results. We find that the salaries of districts that unions use as referents have a large impact on negotiated outcomes. Section 6 offers some concluding comments.

In this paper, we estimate the magnitude of wage spillovers in teacher contract negotiations. Because there has been a large shift in union presence from the private to the public sector, it is appropriate that the analysis of wage spillover among bargaining units should also shift to the public sector. We have formulated spillover not as a single leader with many followers, as usually is the case in the private sector, but as an ongoing process in which local public bargaining units look to recently negotiated contracts in similar jurisdictions. During the contract negotiation process, each side enters into bargaining with a list of referents to be used as comparables. We estimate the separate influence on the negotiated outcome of the referents that are common to both sides, those used only by the union and those used only by the board. We found that the salaries of districts referred to by the union have the largest impact on the negotiated outcome. Using our preferred specification, we found spillovers of approximately 87%; therefore, when salaries of the union-only referents increase by 10%, the negotiated outcome increased by approximately 8.7%. This specification, reported in the first column of Table 2, controls for potential endogeneity of referents' salaries by using an instrumental variable estimator and controls for unobserved spatially autocorrelated factors that affects wages. Our results are also insensitive to the inclusion of the salaries of neighboring districts as a regressor. Our estimates are quite similar to those found in the recent wage spillover literature. Budd (1992, 1997) finds estimates in the range of 65% to 90% for spillover in UAW contracts. This is somewhat surprising given the very different mechanisms by which spillovers are modeled. Budd focuses on formal pattern bargaining, and our interest in this paper is on informal social comparisons made by the negotiators. We also found that the impact of each side's referents on the negotiated outcome depends on each side's “strength” in the negotiations. Using an exogenous measure of union strength, our results suggest that in environments supportive of the goals of organized labor, the influence of union-only referents is much stronger than in environments less supportive of organized labor. In these less-supportive environments, the board-only referents and the referents the union and board have in common become more influential. For districts with high union strength, only the union's list influences the negotiated outcome. For districts with union strength below the median, the union's list has approximately the same impact as the combination of the board-only and common lists. Why does the union-only list always have an influence, while the other lists only have an influence when the union is weak? Perhaps, the explanation lies in the range of union strength values contained in the sample. Because all districts are unionized and have the right to strike, presumably all have some strength. It is possible that our sample only contains cases that range from negotiations with balanced strength to negotiations with complete union dominance. An alternative explanation involves differences in the motivation for the two sides' reliance on social comparisons. Board negotiators will care about relative wages if voters base their subsequent votes on this signal of their elected officials' competency (Besley and Case, 1995). For this explanation to be credible, it must be difficult for voters to obtain direct evidence of their official's competency and relatively easy for voters to use the voting mechanism to reward or punish the officials. Our finding that the board-only list is only important when the union is very weak suggests that these conditions may not be fully met. Voters have other mechanisms besides salaries in referent districts, such as quality of education and the level of taxes, upon which to base their evaluations of the board leadership. Union members are more likely to base their evaluations of the leadership on how their salaries stack up against those of their referents. Furthermore, union members may care a great deal about relative wages if such relative measures directly enter their utility function due to concern about equity and fairness (Frank, 1985). Therefore, the overall importance of referent outcomes is likely to be greater for unions than for boards.