the union deals with only half of the issue. We argued earlier that the union effect arises through a complex process of organizational and behavioral change. A complete analysis requires not only measurement, but identification of the channels of union influence.

We attempted to identify worker and management response to changes in the labor contract brought on by unionization through interviews with union and management officials. The interviews were structured around the distinction between changes in contract provisions, on the one hand, and the response or adjustment to the changes, on the other. In the first phase of the interview we examined existing prac tices under the union contract in terms of the following categories: compensation, internal mobility (promotions and transfers), exit and entry (hiring and layoffs), dispute settlement, technological change, and work practices. The interviews were supplemented with evidence from written contracts. After the existing practices were described, the interviewees were asked to contrast experience under collective bargaining with practice before unionization, using the same categories of subjects.

The second phase of the interview dealt with responses and adjustments to changes in the rules. Information was sought from both union and management representatives on the behavior and adjustment of both groups. With management representatives, however, most of the second phase was devoted to a discussion of changes in management procedure and personnel, while the union interviews tended to focus on the grievance problem and the implicit contract in the nonunion era. It was apparent from these discussions that retrospective interviews provide only weak evidence on the more subtle effects of unionization. Changes in morale and motivation, differences in work group efficiency, and the amount and quality of on-the-job training appear to be difficult to assess without carefully designed questionnaires administered before and after unionization and without measurement and observation of the process in question.

Changes in the labor contract. Differ

ences in the labor contract serve as necessary or enabling conditions for a union productivity effect. The extent of change in rules governing the employment relation is an important determinant of the scope and magnitude of union influence. Analysis of the labor contract thus serves as a useful check on the plausibility of the statistical analysis and the examination of adjustments in operations. Table 4 summarizes contract changes in the before-after plants following unionization.

The evidence on the labor contract under the union and nonvnion regimes reveals a fundamental change in rules covering exit and entry, internal mobility, and dispute settlement with a more moderate impact on compensation. The moderate union im. pact on compensation evident in line 1 is consistent with evidence on the union wage effect found in cross-section data and with information obtained through interviews and discussions with union and manage. ment officials not connected with the before-after plants. The apparent increase in compensation occurs through a variety of forms, with direct adjustments in straighttime rates often assuming a minor role. Four of six plants studied followed a policy of paying union-scale wages before they were organized, and in a fifth plant substantial bonuses were used to bring earnings on par with union plants. Even where union scale had been paid before a plant was unionized, however, some change in the average wage occurred after unionization as a result of changes in the number of job classifications and in re-assignment of workers to different (higher) classifications. Fringe payments were less closely linked to the union contract, but, even here, changes more often took the form of liberalizing the existing package rather than adding totally new benefits.

In contrast to compensation, somewhat greater changes occurred in provisions governing entry and exit and internal mobility. In both union and nonunion settings a distinction is made between internal and external sources of labor. Under both regimes, promotion from within was the most common form of filling job openings, but the internal-external distinction was much

looser in the nonunion situation. We found a policy of hiring "outside" workers into maintenance jobs in every plant, and in three of the six plants hiring from cutside was practiced at all levels of the job struc ture. Furthermore, in situations involving internal mobility the scope of managemenɩ discretion was broader and the rules governing internal movement were less explicit and specific than was the case in the union setting. Without the union, job mobility within the plant depended solely on the decisions of the department foreman or plant manager, with no restrictions on the process of matching workers and jobs. The criteria for making judgments about job requirements and qualifications were not explicit; we found no evidence of formal evaluation systems or attempts to communicate management's views of ideal qualifications.

The contrast with the union setting is quite sharp. Internal mobility under col

lective bargaining was governed by formal plantwide job posting and bidding procedures, with explicit criteria for selection. Contracts in some of the plants contained provisions allowing management to take ability into account, but in practice plantwide seniority was the principal factor. The existence of a formal procedure with explicit criteria for choosing among workers had several organizational consequences. Perhaps the clearest change was the reduced role of management judgment. In the nonunion setting, the foreman (or plant manager) decided both the criteria for filling jobs and those who met the criteria in specific situations. Ability was given greater weight in the nonunion setting, but there is some evidence that personal factors were also taken into account. Union officials and some representatives of management generally agreed that personalities were important in determining who was promoted and that foremen sometimes "played favor

ites." Whether such preferences came at the expense of ability or whether personalities were important only in choosing among people of equal ability is not clear. It is fair to say, however, that the introduction of the union essentially eliminated personal considerations in decisions regarding post. hiring job mobility.

The changes in rules governing internal mobility underscore the fundamental shift in authority and power that occurs with unionization. The limits on management discretion are clearly revealed in line 4 of Table 4, which summarizes the different methods of resolving disputes in the union and nonunion regimes. In each plant unConization was followed by the introduction of a formal grievance procedure with out side arbitration. The grievance procedure replaced a "system" in which workers' problems were dealt with in an ad hoc fashion. Prior to unionization, none of the plants i the before-after sample had regular channcls of communication through which grievances or complaints could be expressed. Individuals with grievances had to raise them with supervisors, who heard the dispute and rendered judgment. In contrast to the union regime, under which too-írequent use of the system is often a problem, information from both union and management officials suggests that in the nonunion setting few problems were ever raised. The evidence implies that the absence of guarantees against recrimination was a strong deterrent.

Changes in workers' behavior. For work. ers, the introduction of a grievance procedure and an increase in compensation augur for adjustments of several kinds. In the interviews we sought to uncover evidence of changes in worker morale and exit behavior. including permanent separations initiated by workers, absenteeism, and subpar or disruptive work effort evidenced in disci, pline problems. The evidence available through the interviews is summarized in line I of Table 5. The table presents a statement of prior expectations about workermanagement adjustments in the first column and evidence of observed changes in each of the six plants in the other columns. The evidence on worker behavior is rela

tively weak; cost considerations prevented a survey of workers, and measures of turnover or absenteeism before and after unionization either did not exist or were not made available. As the summaries in line la suggest, the interviews revealed a decline (3) or no change (2) in the quit rate in five of the six plants studied; in one plant quits were perceived to have risen. We found some evidence of an increase in absenteeism, and a decline in major discipline problems. In the main, these results are indicative of very nioderate changes in exit behavior. It seems likely that had quits or absenteeism changed substantially, we would have uncovered .nuch greater awareness of that adjustment. The interviews thus lead to the conclusion that reductions in turnover and other forms of exit behavior were not a principal connec tion in the union-productivity nexus.

The question of worker morale, examined in line Ib, was one of the few issues over which substantial disagreement between union and management representatives emerged. In two of the plants examined, management officials felt that morale had declined or lad not changed while union representatives perceived definite improvements. In the four other situations, morak apparently improved in two and was not much affected in two. These conclusions are based on impressionistic and perceptual information and underscore the point made earlier, that the interview format is not well suited to the analysis of morale and motiva tion. Although the changes in the labor contract provide strong prior grounds for expecting substantial changes in morale, there appears to be insufficient evidence to confirm or disprove those expectations.

Management adjustments. The introduction of collective bargaining fundamen:ally changes the task of management. Managers are faced with constraints on old procedures and practices, and the processes of negotiation and contract administration constitute a reduction of management power. Decisions traditionally within the purview of management are often challenged as a matter of routine. The magnitude of the change that unionization entails deper.ds on the style of management in the nonunion era. As the summaries in line 2a

suggest, in five of the six plants studied an essentially authoritarian management was confronted by a significant shift in power and authority. Previous methods, particularly the manner of handling and dealing with workers, were no longer effective (or, at least, were much more expensive). The evidence suggests that successful management in the union context required new management procedures and practices. Perhaps the most cogent description of the differences in the management process before and after unionization was given by a plant manager who remarked:" ... before the union this place was run like a family; now we run it like a business."

The major change in plant management uncovered in the interviews was the introduction of a new plant manager and, in some instances, new supervisors. Given the substantial change in the nature of industrial relations, the identification of the old manager with the nonunion regime, and the likelihood that previous management was involved in attempts to block unionization, the change in plant management is not surprising. While retraining permitted many front-line supervisors to make the adjustment to a union regime, training was not a viable option in the case of plant managers. The interviews suggest that a new manager was in some sense a prerequisite for innovation in management methods.

Before the interviews were conducted, changes in management procedure were expected on theoretical grounds to have occurred after unionization. Apart from capital-labor substitution and labor-quality adjustments, the unton wage effect creates incentives for management to extract more work effort from a given level of employees. These expectations were clearly realized. As line 2b reveals, the interviews uncovered changes in management methods in all plants. The magnitude of the change varied from situation to situation, with a more professional, businesslike approach to labor relations by front-line supervisors the most common adjustment. In four of the six plants we found attempts to increase work effort and work group efficiency primarily through introduction of formal methods of organizational control. The adjustments in

formal control procedures took several forms. In essence, however, they amounted to a system of production goals or targets accompanied by procedures for the review and monitoring of performance. The evaluation often occurred in newly introduced staff meetings, which were used for com munication, training, and assessment of conditions and progress. Substantial changes in fwmal procedures were not introduced in all plants. Yet, even where formal procedures were changed only mod. erately, the interviews suggest that management monitored work performance and manning requirements more closely.

The evidence summarized in Table 5 suggests that unionization led to substantial changes in management in each of the before-after plants. Not all adjustments noted were observed in all plants, but each plant experienced change in a number of dimensions. The existence of a pattern of management adjustment across plants organized at different points in time suggests that the observed changes are not due solely to general technical change. While technical change may be at work in the proc· esses we observed, it seems clear from the interviews that unionization had a significant independent effect. Our tentative con clusion, therefore, is that an improvement in plant management is one of the key adjustments to unionization. These results inay be interpreted as evidence of a modern union "shock effect." The institutional analysis is consistent with a broad range of earlier studies on the effects of unionization and provides a partial explanation of the union effect estimated earlier.

Conclusions and Implications

The examination of collecti ve bargaining and productivity in this study has yielded empirical results on the magnitude of the union productivity effect, and the case studies have provided some insight into the channels through which unions influence productivity. The empirical eviden e suggests that in the cement industry unionization has led to gains in productivity of 6 – 8 percent.” This finding appears to be rela

"This gain in productivity may be compared to the