Employee participation programs, group-based incentives, and company performance: a union-nonunion comparison.
Recent summaries of the literature suggest that both employee participation programs and group-based pay generally (but not always) have positive, albeit modest, independent effects on performance. Very few studies, however, have investigated the potential interaction effects of employee participation (EP) programs and group-based pay incentives on performance, and the findings of those few studies are mixed. Also, only two recent analyses investigate any differential effects by union status, and then only in regard to EP and again the findings are mixed. Finally, how the independent effects of group-based incentives and the interaction effects of EP and group-based incentives on performance may differ in unionized companies and nonunion companies has received very little theoretical attention and no empirical analysis. To examine these issues, I develop and test a model using data for 1989 on 841 manufacturing firms in Michigan.
Theory and Hypotheses
Central to a diverse literature is the proposal that employees have untapped knowledge, problem solving and leadership skills, creativity, or effort, which, if tapped appropriately, can lead to enhanced firm performance. Before identifying ways in which union representation positively or negatively influences the performance effects of EP and group-based incentives on firm performance, I briefly summarize the hypothesized performance effects absent any consideration of union status.
Employee Participation Programs and
EP comes in many different forms, with no two efforts completely alike. At one extreme are informal or short-lived ad hoc team efforts intended to address specific problems or inefficiencies. At the other extreme are self-directed or autonomous work teams in which employees make most production decisions. In-between these extremes are a variety of employee involvement or QWL efforts in which teams engage in regularly scheduled activities but limited efforts at identifying and resolving problems and inefficiencies. All these forms of EP, furthermore, are bound to differ in more detailed ways as they are structured to meet the specific circumstances of any given site. Because the data base used in the empirical analysis that follows does not provide information on the parameters, general type, or intensity of EP efforts, I do not attempt to develop hypotheses that might apply to the distinctions among EP programs. Instead, the propositions I draw from the literature or develop herein are applicable across all the various forms of EP activities.
At the heart of the basic proposition that EP programs enhance firm performance is the contention that employees generally have more complete knowledge and information about their work tasks and processes than do managers (Levine and Tyson 1990; Miller and Monge 1986) and are in a better position than managers to plan and schedule work, to organize work tasks and work flow, and to otherwise identify and resolve obstacles to achieving optimal performance (Hammer 1988). A second basic proposition is that EP provides employees with greater intrinsic rewards from work than do traditional forms of management. These greater rewards from work increase job satisfaction and, in turn, increase employees' motivation to achieve new production goals (Miller and Monge 1986; Hammer 1988). It has also been proposed that giving workers access to management information increases mutual trust and commitment to organizational goals (Hammer 1988). Hence, employee-supervisor relations improve (Cooke 1990a), employees are willing to be more flexible regarding changes in human resource policies (Delaney et al. 1993), and employees are more inclined to channel their power in positive ways than they otherwise would be (Strauss 1990).
The potential performance gains derived from EP, however, are offset, at least in part, by organizational costs associated with EP. These costs include reorientation and training costs (Cooke 1992) and added transaction costs associated with more decision makers being involved in making workplace decisions and greater required communication between these participants (Levine and Tyson 1990; Kelley and Harrison 1992). In addition, some employees place little or no value on participation and, hence, will not be motivated by such intrinsic rewards (Miller and Monge 1986), and the relaxation of supervisory monitoring of work allows some employees to "shirk" (Weitzman and Kruse 1990; Levine and Tyson 1990).
Group-based Pay and Company
There are three basic forms of group-based pay arrangements: profit sharing, gain sharing, and employee stock ownership. Because the data base utilized in this study does not provide information about the use of stock ownership plans, my focus is on profit and gain sharing effects.
The differences between profit sharing and gain sharing are found largely in the performance criteria used in making bonus calculations and, to a lesser extent, in the timing of bonus calculations and payments. Bonuses received under profit sharing are based on profits, typically calculated annually, semi-annually, or quarterly. Bonuses received under gain sharing, on the other hand, are based on measures of performance other than profits (for example, on reductions in cost-to-sales ratios that might include labor, materials, or overhead costs, or on improvements in quality, scrappage, safety, and so on). Gain sharing bonuses, moreover, are typically calculated on bonus periods of 4 to 5 weeks or quarterly. (See Kruse , Hammer , and Mitchell, Lewin, and Lawler  for descriptions and comparisons.)
The fundamental proposition underlying group-based incentives is that if employees' earnings are tied to performance, they will adjust their effort to optimize income. Because bonuses are tied to overall work force effort, moreover, employees have an incentive to work more cooperatively among themselves (Weitzman and Kruse 1990; Strauss 1990). Group-based incentives, consequently, also reduce the monitoring costs associated with supervisory control (Kruse 1993).
As proposed by Mitchell et al. (1990), group-based incentives indirectly increase employee effort and commitment by improving communication about company performance, by educating employees about the importance of profitability and organizational effectiveness, and by reducing the risk of layoffs (as Weitzman  argued would be the result of the "share economy"). As hypothesized by Strauss (1990), furthermore, group incentives make it easier for management to introduce new technologies and obtain changes in restrictive and unproductive work rules.
A number of factors, however, could reduce gains. First, there is the "free rider" or employee shirking problem, whereby some employees will merely take unfair advantage of group-generated performance gains and consequent bonuses. Levine and Tyson (1990) argued, however, that group-based incentives induce employees to monitor the behavior of co-workers and impose social sanctions on those employees who shirk from cooperative work group norms.
Second, employees may be reluctant to exchange some fixed earnings for variable earnings, because of the higher risks attached to the latter. The optimal ratio of variable earnings to fixed earnings, therefore, must provide sufficient income stability on one hand, and sufficient incentive to induce greater effort, on the other hand (Weitzman and Kruse 1990).
The strength or magnitude of the potential effects of group-based incentives is likely to be moderated, furthermore, by the specific incentive formulas and the timing of bonus payments. First, what employees accomplish in their respective work areas is apt to be only imperfectly correlated with overall company profits. Although the bonuses provided by gain sharing formulas are more directly tied to actual employee effort than are bonuses based on profits, employees still may not reap the full benefit of improved production performance. Take, for example, the modified Scanlon plan used by Eggers Custom Plywood. The ratio by which gains are calculated includes labor, material, and overhead costs in the numerator and the sales value of the product in the denominator. Any drop in the sales value of the product or increase in materials costs caused by market forces outside the control of the work force would offset some of the gain achieved by reducing labor costs.
Second, most profit sharing bonus schedules defer payments, often as promised pension endowments. It appears that only about 19% of profit sharing plans in 1988 provided bonuses on an annual or more frequent basis (Hewitt Associates 1988). Gain sharing bonuses, on the other hand, are typically paid every several weeks. Given the greater frequency and timeliness of gain sharing bonuses in comparison to profit sharing bonuses, it seems likely that they provide workers with a stronger incentive to be more productive (Strauss 1990; Eaton and Voos 1992). Some have argued that bonuses paid at retirement or upon otherwise leaving the firm are likely to have little or no incentive effects (Hammer 1988). Assuming that employees are not so myopic as to completely discount the value of retirement income (received at more favorable tax rates), however, even deferred profit sharing can be hypothesized to influence work-related behavior.
Interaction Effects of EP and Group-Based
The potential performance gains from combining EP and group-based incentives may exceed the potential gains of either by itself. First, as argued by Levine and Tyson (1990), management must tie financial incentives to EP, or else employees have little reason to share their performance-enhancing knowledge with management or to monitor and sanction shirking behavior of co-workers. Second, the less freely employees are able to participate in workplace decisions affecting performance, the less able they will be to respond effectively to group-based pay incentives. The introduction of group-based pay incentives, therefore, is likely to be followed by employee demands for greater participation (Weitzman and Kruse 1990). Conversely, employees who have been allowed to participate in decision making are likely to eventually demand from management that the gains they have helped generate through participation be equitably shared (Cooke 1990b, Chap. 6).
Union Effects on Employee
Participation, Group-Based Incentives,
Plausible arguments can be made to support widely varying predictions of the effect of unions on company performance. In this section, I summarize the potential positive and negative effects of union representation on EP and on group-based incentives.
Union effects on EP. The collective voice thesis (Freeman and Medoff 1984) suggests that unions provide an avenue for employee input about the terms and conditions of employment, which, in turn, leads to better management practices and more productive employees. It can be argued that by similarly establishing more direct and open channels for collective voice through EP, unions may also enhance the value of EP.
Several considerations support that view. First, compared to nonunion settings, unionized workplaces provide greater insurance that a serious hearing will be given to employees' ideas and concerns about the design and processes governing participation - training and acquisition of skills, job assignments, rotation, team leader selection, the supporting roles of supervision and staff, and so on. To the extent that unions allow team proposals to reorganize work and eliminate inefficient work rules (either directly through participation or indirectly through contractual negotiations), EP becomes a means for eliminating long-standing restrictive practices.
Second, as proposed by Costanza (1988), unionized employees are more secure than nonunionized employees in requesting information from supervisors, and in challenging management's interpretation of data in proposed solutions. Union leaders, furthermore, have access to higher levels of management than do nonunion employees when team proposals are ignored or blocked by supervision or when ground rules and accepted policies and philosophies governing EP are not honored. Unions, therefore, help ensure that more thorough problem solving occurs and that employees remain committed to participatory programs. Third, as discussed by Eaton and Voos (1992), unions help create a better participatory environment by placing emphasis on a more acceptable balance between improving performance and the quality of work life and by providing contractual protections against arbitrary or inequitable treatment and managerial reprisal.
Fourth, as proposed by Levine and Tyson (1990), the longer-term employment relations and narrower pay and status differentials in unionized settings than in nonunion settings strengthen sustained commitment to EP. In particular, unions have long negotiated various contractual rules governing employment security and longterm employment relations; unionized employees are less likely to quit than their nonunion counterparts, other things equal (Freeman 1980); and wage differentials are significantly narrower in unionized firms than in comparable nonunion firms (Freeman and Medoff 1984).
On the other hand, economic notions of monopoly-like price theoretic behavior, suboptimal production functions, and hostile labor-management relations suggest reasons to believe that unions have negative effects on EP activities (Addison and Hirsch 1987; Freeman and Medoff 1984; Allen 1986; Boal 1990). First, except for cases in which unions are willing to negotiate away restrictive work rules, they are likely to insist that members of employee participation programs not make proposals that infringe on existing contractual language, however restrictive that language may be (Eaton 1990). Second, unions are inclined to make EP voluntary, defending the choice of employees not to participate in EP activities. The refusal of employees to participate is likely to frustrate the planning, scheduling, or assignment of work decided by teams and to demoralize team members.
Third, because of the often confrontational nature of contract negotiations and administration and sometimes hotly contested union officer elections, a spirit of trust and commitment in unionized settings may be harder to accomplish or maintain than in nonunion firms. Union leaders may signal employees to relax team efforts or simply hold EP activities "hostage" until certain union demands are met, internal union leader disputes over cooperation or participation are resolved, or other labor-management conflicts are settled (for example, see Hammer and Stern 1986). The likelihood that union leaders will engage in disruptive behavior appears to be relatively high when union leaders either are excluded from EP activities or choose not to jointly administer them (Cooke 1992). Indeed, some union leaders will attempt to undermine or destroy EP activities (Eaton and Voos 1986). Finally, unions may increase transaction costs by involving union leaders and appointed representatives in the design, implementation, and oversight of EP activities.
Union effects on group-based incentives. The formulas used in profit sharing plans differ from those used in gain sharing plans in ways that may affect the productivity-enhancing behavior of employees. In this section, I briefly remark on the possible differences in union effects on profit sharing vis-i-vis gain sharing. In the empirical analysis that follows, however, I am unable to distinguish profit sharing from gain sharing, and the propositions I develop apply to both forms of group-based incentives.
On the positive side, there is reason to believe that profit sharing will have a greater incentive effect in unionized firms than in nonunion firms. First, assuming that unions will resist substituting profit sharing bonuses paid out at retirement for existing pension plans, profit sharing bonuses are more likely to be paid on an annual or more frequent basis in unionized firms than in nonunion firms. Profit sharing plans in unionized firms, therefore, can be expected to have greater incentive effects, on average, than profit sharing plans in nonunion firms.
Second, given that the union leaders have negotiated profit or gain sharing, they have a political stake in seeing that bonus plans pay off. Hence, they have an incentive to encourage employees to vigorously pursue improvement in company performance. In addition, because of the union's involvement in the negotiation and administration of the plans, any communication, education, and symbolic value derived from group-based incentives (Mitchell et al. 1990) may be greater in unionized firms than in nonunion firms.
On the negative side, there are several reasons to conclude that union representation diminishes the full incentive effects of group-based pay. First, John Zalusky of the AFL-CIO concludes that to union leaders, "arguments supporting profit sharing as an incentive ... seem weak, unrealistic, or unnecessary" (Zalusky 1990:65). He argues that when unions negotiate group-based bonus plans, these plans are treated as an added benefit, or as a mechanism to maintain fixed wage patterns in an industry in which the ability to pay varies across employers, or as a necessary substitute for wage concessions. In economic terms, unions are applying their monopoly-like power to raise or maintain the price of labor. Group-based pay may therefore be unlikely to provide employees with a direct incentive to improve performance in unionized firms.
Second, given the generally higher compensation rates in unionized firms than in comparable nonunion firms, the ratio of variable earnings (based on profit or gain sharing) to fixed earnings (based on hourly compensation rates or salaries) can be expected to be smaller in unionized firms. Although lower risk-taking associated with smaller variable-to-fixed earnings ratios may help gain employee acceptance of group-based bonus plans, employees' incentive to enhance performance, nonetheless, may be reduced.
Third, although employees may attempt to impose subtle social sanctions on shirking co-workers in unionized settings, union leaders can be expected to strongly discourage members from reporting shirking members to supervisors and strictly prohibit team members from disciplining other team members. Group-based incentives are therefore likely to induce much less co-worker monitoring in unionized firms than in nonunion firms. Alternatively, to the extent that co-worker monitoring and sanctioning remains informal or cannot be constrained by the union leadership, disruption and conflict among co-workers and within the union leadership may negatively affect the ability of team members to work cooperatively.
Summary of Potential Effects
Predictions about the effects of EP and group-based incentives on company performance in union and nonunion firms cannot be readily deduced from the diverse and competing theoretical propositions summarized above. Any net effects depend on whether union representation, on balance, yields positive collective voice effects or negative union restriction effects. The nature of the net effects is clearly an empirical question.
Only two previous studies of this subject have examined the effect of union status, and they reached conflicting conclusions. Recent reviews of the literature conclude that both EP (Levine and Tyson 1990; Miller and Monge 1986) and group-based incentives (Weitzman and Kruse 1990) generally have modest positive independent effects on performance. Only three recent analyses, however, addressed potential interaction effects between EP and group-based incentives on performance. Examining effects on value added, Conte and Svejnar (1988) obtained evidence that EP and profit sharing had both significant independent effects and significant interaction effects on performance. Examining effects on sales per employee and returns on investment and assets, Mitchell et al. (I 990) found that EP and group-based incentives had independent effects but no interaction effects on all three performance measures. Finally, Kruse (I 993), examining effects on sales and value added per employee, found positive independent effects of profit sharing, but no independent or interaction effects of EP. None of those several studies examined the effects of union status.
Only the recent analyses by Kelley and Harrison (1992) and Cooke (1992) addressed any differential interaction effects by union status, and then only in regard to EP. Whereas Kelley and Harrison did not find any significant effects of EP on efficiency in either union or nonunion firms, in my 1992 study I found that EP had significant positive effects on quality improvements of about the same magnitude in both nonunion firms and those unionized firms in which unions had a share in the administration of EP activities. Finally, there are no published empirical studies examining the independent and interaction effects of EP and group-based incentives on performance as moderated by union status.
In the next section, I specify and estimate a model of the average net effects of participation, group-based pay, and union representation on firm-level performance. The data available at the firm level, unfortunately, do not allow one to test the many specific hypotheses reviewed and developed herein. But the data do allow for tests of the average aggregated or summary net effects on performance of the various interactions between EP, group-based pay, and union representation.
Data Base and Model Specification
The Industrial Technology Institute (ITI), a nonprofit research and service organization located in Ann Arbor, Michigan, conducted an extensive telephone and mail survey of Michigan manufacturers during 1989-90. The primary purpose of the survey was to study both the extent of recent and planned capital investments in computer-aided technologies and the utilization of these investments. Questions were also asked about sales; costs of materials, parts, and services; labor cost/total cost ratios; wages; union status; and other firm characteristics central to this inquiry.
In conjunction with local economic development agencies, ITI surveyed firms in five targeted Michigan counties. All capital goods-producing manufacturers with 10-19 employees and all manufacturers with 20 or more employees were surveyed (ITI 1991). Although data about nonrespondent are not available in the ITI data base, responses were received from 70% or more of the sampled firms in each of the five Michigan counties targeted for the data collection. The data base provides information on 2,431 of these establishments. With the exclusion of cases with missing data on any of the variables included in the regression equations that follow, however, the final sample includes 841 firms.(1)
Given the data available, company performance is measured as value added net of labor cost per employee for 1989. To calculate the differential effects of the various combinations of EP, group-based pay, and union status on firm-level performance (as defined), I first estimate three separate semi-logarithmic regression equations against value added per employee, wage rates, and labor cost/total cost (LC/TC) ratios. Using the estimated differentials associated with each combination of EP, group-based pay, and union status, I then calculate performance by subtracting the estimated wage differential (adjusted by the LC/TC ratio differential) from the estimated value added per employee differential (that is, value added differential - (wage differential x [1 + LC/TC differential])).(2) The result is an estimated average proportional gain or loss in performance associated with each combination of EP, groupbased pay, and union status.
The right-hand-side variables are treated as exogenous variables in reduced form single equations. Several authors have argued, however, that a firm's choice to offer profit or gain sharing or to establish work teams is endogenous. Any endogeneity may bias coefficients either upward or downward, since more profitable firms can better afford to engage in or experiment with such innovations (Ehrenberg 1990; Kruse 1993) and less profitable firms may adapt these kinds of innovations in their efforts to become more competitive (Cooke and Meyer 1990; Meyer and Cooke 1993). The data base, however, provides no information on when teams or group-based incentives were adopted or any longitudinal information on firm performance. Any efforts to specify a nonrecursive model so that teams, group-based incentives, and union status are treated strictly as exogenous variables would lead, in any case, to serious misspecification of the theoretical assumptions and propositions about the independent and interaction effects of these variables on performance.
EP, group-based pay, and union status. The ITI survey only asked the following questions about the incidence of teams, profit sharing, and gain sharing: "Do you have any of the following labor-management programs in place? ... Work Teams? ... Scanlon or Other Gain or Profit-Sharing Plans?" The survey results therefore provide no information about the structure, focus, extent, degree of autonomy, or other parameters of work teams. Consequently, all forms of labor-management programs that include work team activities (whether or not they are jointly administered with unions in unionized firms) are presumed to be included in the current measure of work teams.
The survey questions do not distinguish between profit sharing and gain sharing plans. No information about the provisions of these plans is provided, nor can the incidence of profit sharing be distinguished from gain sharing. Consequently, all forms of profit sharing and gain sharing are combined as one measure of group-based incentive plans (hereafter referred to as profit/gain sharing).
As reported in Table 2, 46% of unionized firms and 44% of nonunion firms reported having work teams. A substantially larger proportion of nonunion firms (52%) than unionized firms (36%), however, reported having some form of profit or gain sharing. These proportions mirror closely the extent of EP, profit sharing, and gain sharing reported in other surveys.(3) Given the evidence that gain sharing has been adopted by very few firms (as reported by surveys described in footnote 3), it is reasonable to assume that the average effects of profit sharing greatly outweigh the average effects of gain sharing in this measure of group-based pay incentives.
All the various combinations of work teams, profit/gain sharing, and union status are constructed as 0-1 dummy variables. Employing these categorical variables in the regression equations (nonunion firms with neither teams nor profit/gain sharing comprise the omitted benchmark category) makes it possible to estimate the independent as well as interaction effects of teams, profit/gain sharing, and union status.
Control variables. To indirectly and partially account for differences in capital intensity, assets per employee at the 2-digit industry level (ASSET/EE), the percentage of equipment that is computer-aided at the firm level (% COMPUTER-AID), and 2-digit SIC codes are controlled for in the regressions. The quality of the work force across firms is partially controlled for with measures of the percentage of hourly employees who are skilled (% HRLY SKILLED), the percentage of workers who have "journeyman" status (% JOURNEY), the percentage of the operating budget allocated to training (% TRAIN BUDG), and whether the firm has pay-for-knowledge compensation (PAY-FOR-KNOW). Controlling for any economies or diseconomies of scale, size of firm (FIRM SIZE) is included. To partially control for any advantages associated with newer establishments, the age of the plant (FIRM AGE) is held constant. To control for industry market factors, controls for recent changes in industry shipments (CIND SHIP) and employment (CIND EMPLOY) are included, as well as four-firm concentration ratios (CONCENTRATE).
Variables are defined in Table 1. Mean statistics by the various categories of teams, profit/gain sharing, and union status are provided in Table 2. Table 3 reports the regression estimates corrected for heteroskedasticity, and Table 4 provides net effect calculations.
[TABULAR DATA OMITTED]
As reported in Table 2, about 27% of the sample of Michigan manufacturers in 1989 were unionized. Unionized firms, on average, had higher value added per employee, paid much higher hourly wages, had lower LC/TC ratios, were less likely to have profit/gain sharing plans, were older and larger, and employed a substantially larger proportion of skilled trades journeymen than did nonunion firms.
Although the equations are estimated in semi-logarithmic form, the use of categorical variables and the distribution of the value added and wage rates suggest the potential for violating the OLS assumption of homogeneity. A general test of the homogeneity assumption (White 1980) indicates that the estimated standard errors are heteroskedastic in both the value added and wage equations. To correct for this inefficiency in estimation, the equations are estimated using White's efficient estimator as employed by Greene (1991,Chap. 26). Although the standard errors have been corrected for heteroskedasticity and differ from the OLS standard errors, the levels of significance obtained for the salient coefficients remain unchanged between the uncorrected and corrected estimates. The results of the regressions as corrected for heteroskedasticity are, nonetheless, reported in Table 3.
Except for the insignificant coefficient of NONUNION*TEAMS*NO PS, all other coefficients estimated for the various combinations of teams, profit/gain sharing, and union status obtain significance at the .05 level of confidence or better (using two-tailed tests) in both the value added and wage equations. In comparison to nonunion firms without teams or profit/gain sharing, the estimated LC/TC differentials are negative for all combinations, but significance is not obtained for the NONUNION*TEAMS*NO PS and NONUNION*NO TEAMS*PS coefficients. In addition, the coefficient for UNION*NO TEAMS*PS reaches significance only at the .10 level.
Among the control variables, only FIRM SIZE, % JOURNEY, and several SIC variables obtain significance at the.05 level of confidence in more than one equation. Joint significance tests for the set of SIC dummy variables obtain significance at the .05 level in all equations. joint significance tests for the several industry market variables (ASSET/EE, CONCENTRATE, CIND SHIP, and CIND EMPLOY), however, fail to obtain significance in all three equations. Sensitivity tests on the functional forms of FIRM SIZE, ASSET/EE, and CONCENTRATE (estimated in logarithmic form) and FIRM AGE (estimated in both dummy and logarithmic forms) yield inferior goodness-of-fit statistics.
To obtain the percentage differential effects of the various combinations of work teams, profit/gain sharing, and union status on the dependent variables, the exponential functions of the coefficients (100[exp([b.sub.1])-1]) are reported in Table 4 (columns 1-3). The estimated differentials indicate that value added per employee is substantially higher in unionized firms with work teams or profit/gain sharing (or both) than in similar nonunion firms. The average value added per employee is even somewhat higher in unionized firms with teams or profit/gain sharing than in nonunion firms with profit/gain sharing and work teams.(4)
The wage differentials, on the other hand, are substantially greater in unionized firms than in comparable nonunion firms. In comparison to their counterparts in nonunion firms without teams or profit/gain sharing, unionized employees earn, on average, 18-23% higher wages. Among nonunion firms, firms having teams or profit/gain sharing (or both) pay roughly 6-7% higher wages than firms without teams or profit/gain sharing. Overall labor costs associated with higher union wages, however, are partially offset by lower average LC/TC ratios in unionized firms than in nonunion firms.
To calculate the net differential gain in value added per employee, the estimated differential cost of labor (wage differential x [1 + LC/TC differential]) is subtracted from the estimated differentials in value added per employee (see column 4, Table 4). Inferences about the independent and interaction effects of work teams, profit/gain sharing, and union status on firm performance are based on these estimates.
First, performance is roughly 13% higher in unionized firms with neither teams nor profit/gain sharing (UNION*NO TEAMS*NO PS) than in comparable nonunion firms. The addition of teams in unionized firms (UNION*TEAMS*NO PS) appears to increase performance by another 22 percentage points (that is, 34.8% - 12.7%). In contrast, the addition of work teams in nonunion firms (NONUNION*TEAMS*NO PS) is associated with no net gain in performance. Indeed, the estimates suggest that performance is slightly reduced (-1.7%).
Second, the addition of profit/gain sharing, in contrast, appears to have substantial positive effects on performance in nonunion firms (NONUNION*NO TEAMS*PS) but modest effects in unionized firms (UNION*NO TEAMS*PS). In nonunion firms, the addition of profit/gain sharing alone is associated with an 18.3% increase in performance over comparable nonunion firms without profit/gain sharing incentives or work teams. In unionized firms, profit/gain sharing incentives are associated with a 6.5% increase in performance over unionized firms without profit/gain sharing or work teams (that is, 19.2% - 12.7%).
Third, the interaction effects of work teams and profit/gain sharing are fairly modest in nonunion firms and appear to be negative in unionized firms. With respect to nonunion firms, the estimated performance differential associated with the combination of work teams and profit/gain sharing (NONUNION*TEAMS*PS) is about 21%. Bearing in mind that the estimated independent effect of work teams in nonunion firms is -1.7% but that the estimated independent effect of profit/gain sharing is +18.3%, it appears that the interaction effect raises performance about 4 percentage points.
With respect to unionized firms, on the other hand, the interaction of work teams and profit/gain sharing (UNION*TEAMS*PS) appears to have negative consequences. The interaction effect yields only a 5.8 percentage point improvement in performance over traditional unionized firms (18.5% - 12.7%), which is only slightly lower than the independent effect of profit/gain sharing (6.5%) but substantially lower than the independent effect of work teams (22%).
Although the empirical analyses do not disentangle the many hypothesized ways in which work teams, profit/gain sharing, and union representation may influence performance, the results allow several general inferences. In unionized firms, work teams appear to have fairly substantial positive effects on performance. As discussed earlier, this positive effect may occur because unions ensure that employees have substantial input in the design and processes governing EP, that more thorough problem solving occurs, and that employees remain committed to work team efforts. In addition, given that work teams have limited positive effects on performance in nonunion firms (and then only when teams are combined with group-based incentives), it appears that unions have collective voice effects that help to create environments more conducive to effective participation.
The evidence also indicates that profit/gain sharing has modest effects on performance in unionized firms, but rather substantial effects in nonunion firms. This finding suggests that unions in general treat profit and gain sharing arrangements less as group-based pay incentives and more as added benefits, substitutes for wage concessions, or mechanisms for maintaining fixed industry wage patterns (as argued by Zalusky 1990). In nonunion settings, on the other hand, the evidence is consistent with the notion that group-based pay provides a valuable incentive to employees to improve performance. Although there is no direct test of the hypothesis that employees also monitor and sanction shirking employees, to the extent that such behavior is induced by profit or gain sharing, the evidence would suggest that it is more prevalent or effective in nonunion than in unionized settings. Another possible explanation for why profit and gain sharing appear to have substantially different effects between unionized and nonunion firms is that the ratio of variable earnings to fixed wages differs substantially between sectors. Because nonunion employees have a lower average base wage than union employees, the ratio of variable earnings to fixed wages will be higher for them, giving them a significantly greater incentive.
Only in nonunion firms does the combination of work teams and profit/gain sharing yield positive, albeit relatively modest, interaction effects on performance. In nonunion firms, it appears that group-based pay arrangements either generate greater effort from work teams or lead to performance-enhancing modifications in the structure or processes of work team activities. In sharp contrast, the evidence indicates a large negative interaction effect in unionized firms. In unionized firms in which employees are engaged in work team activities, employees appear to be responding negatively to group-based incentives, even though they do not appear to respond negatively to either work teams or profit/gain sharing alone. The theoretical propositions developed above suggest that this negative interaction effect may be attributable to the inherent potential for co-worker monitoring and sanctioning of team members induced by group-based incentives. Monitoring and sanctioning of shirking team members or other employees choosing not to participate, that is, may lead to conflict within the union and, hence, prove counter-productive to creating or maintaining a spirit of cooperation necessary for achieving the full potential of work team activity.
This analysis of survey data on a large sample of Michigan firms in 1989 has produced fairly strong evidence that employee participation programs and group-based incentives yielded substantial gains in firm-level performance - measured as value added net of labor cost per employee - but that these gains differed by union status. Among the various combinations of work teams, group-based pay, and union status, unionized companies with work teams but no group-based pay appeared in 1989 to have achieved the highest level of performance - estimated at 35% higher than comparable nonunion firms without teams or group-based pay.
Four categories of firms appear to have achieved performance levels in the range of 18-21% above nonunion firms without teams or group-based incentives: both unionized and nonunion firms having both teams and group-based pay incentives, and both unionized and nonunion firms with group-based pay incentives but no work teams. Among the remaining firm categories, unionized firms without work teams or group-based incentives appear to have achieved performance levels about 13% higher than both nonunion firms without teams or group-based incentives and nonunion firms with work teams but without group-based incentives.
More generally, the evidence implies that unionized firms, on average, provide a much better environment for tapping the benefits of employee participation programs than do nonunion firms; and, by the same token, nonunion firms generally provide a better environment for tapping the performance incentive effects of group-based pay than do unionized firms.
Given that no other published research has examined the interaction effects on company performance of union representation, employee participation programs, and profit or gain sharing, and given the unique sample and limitations of the data utilized in this empirical study, the findings reported here are tentative. The results do raise questions, however, about previous findings based on analyses that have not accounted for the effects on firm performance of employee participation, group-based incentives, and union status.
If we are to truly understand the conditions under which employee participation and group-based incentives lead to enhanced performance, much more research is required. In particular, future research needs to examine the many detailed hypotheses summarized but not tested above, and investigate how the full range of human resource and labor-management practices bolster or limit the intended effects of employee participation and group-based incentives on firm performance.
(1) Approximately 30% of the observations have missing data on firm characteristics - in particular, with respect to controls for the percentage of workers who are skilled or have journeyman status, the percentage of the operating budget allocated to training, and LC/TC ratios. Nearly half of the sample does not report the cost of materials, parts, and services, which is used in calculating value added. Nearly all firms, however, reported on the existence of work teams, profit or gain sharing, and union representation. The distribution of those three key variables in the full sample is nearly identical to their distribution in the reduced sample used for estimation. (2) Since the cost if labor includes direct compensation and labor's productive utilization of parts, materials, services, and capital (taking into account productivity, efficiency, quality of work, scrappage, and so on), I attempt to adjust the wage differential costs by difference in the productive utilization of labor. Since I was unable to directly isolate or account for differences in labor costs or in the productive utilization of labor in this sample of data, I used estimates of the LC/TC differentials to adjust the effects of wage differentials. Although the estimated LC/TC differentials (associated with the various combinations of FP, group-based pay, and union status) are imperfect measures of the productive utilization of labor in the production process, they do provide a plausible means of adjusting the cost of estimated wage differentials. Once the estimated wage differentials are adjusted by the estimated LC/TC differentials, LC/TC differentials become positively correlated with the effect on costs of wage differentials - and, therefore, with the estimated value added differentials minus the wage differentials. (3) The 1987 GAO survey of 326 Fortune 1000 firms found that 44% of nonunion firms and 56% of unionized firms had established participation programs by 1987 (U.S. General Accounting Office 1987). A 1987 survey conducted by researchers at Columbia University of 448 manufacturing business units found that 43% of nonunion units and 49% of unionized units had established participation programs as of 1987 (Delaney et al. 1989). The GAO sample also reported that in 1987, 54% of nonunion firms and 39% of unionized firms had profit sharing for hourly employees. The Columbia University sample reported that in 1981 53% of nonunion and 29% of unionized business units had profit sharing for hourly employees. Gain sharing had been established in 8% of the nonunion firms and 6% of unionized firms in the GAO sample. In the Columbia University sample, gain sharing had been established in 2% of the nonunion and 4% of the unionized business units. (4) As additional checks on the robustness of these estimates, the sample was partitioned into six subsamples and three pairs of regression were estimated against log (VALUE ADDED/EE); teams vs. no teams, profit/gain sharing vs. no profit/gain sharing, and union vs. nonunion. In five of the six equations, the estimates closely mirror the estimates obtained using the full sample. In the sample including only unionized firms, however, although the estimated differentials for teams and profit-gain sharing are consistent with full sample estimates, they do not obtain significance. Colinearity with other variables does not appear to explain the lack of significance, as the coefficients remain insignificant with and without the exclusion of industry SIC dummy variables and FIRM SIZE. Given that the sample is reduced substantially (to 231 observations), the lack of significance is likely attributable to the loss in degrees of freedom.
Addison, John T., and Barry T. Hirsch. 1989. "Union Effects on Productivity, Profits, and Growth: Has the Long Run Arrived?" Journal of Labor Economics, Vol. 7, No. 1 (January), pp. 72-105. Allen, Steven G. 1986. "Unionization and Productivity in Office Building and School Construction." Industrial and Labor Relations Review, Vol. 39, No. 2 (January), pp. 187-201. Boal, William M. 1990. "Unionism and Productivity in West Virginia Coal Mining." Industrial and Labor Relations Review, Vol. 43, No. 4, pp. 390-405. Conte, Michael A., and Jan Svejnar. 1988. "Productivity Effects of Worker Participation in Management, Profit-Sharing, Worker Ownership of Assets, and Unionization in U.S. Firms." International Journal of Industrial Organization, Vol. 6, No. 1 (March), pp. 139-51. Cooke, William N. 1990a. "Factors Influencing the Effect of Joint Union-Management Programs on Employee-Supervisor Relations." Industrial and Labor Relations Review, Vol. 43, No. 5 (July), pp. 597-603. _____. 1990b. Labor-Management Cooperation: New Partnerships or Going in Circles? Kalamazoo, Mich.: W.E. Upjohn Institute for Employment Research. _____. 1992. Product Quality Improvement through Employee Participation: The Effects of Unionization and Joint Union-Management Administration." Industrial and Labor Relations Review, Vol. 46, No. 1 (October), pp. 119-34. Cooke, William N., and David G. Meyer. 1990. "Structural and Market Predictors of Corporate Labor Relations Strategies." Industrial and Labor Relations Review, Vol. 43, No. 2 (January), pp. 280-93. Costanza, Anthony J. 1989. "Participatory Action Research: A View from the ACTWU." American Behavioral Scientist, Vol. 32, No. 5 (May-June), pp. 566-73. Delaney, John, Jack Fiorito, Paul Jarley, and Susan Schwochau. 1993. "Employee Cooperation Program Structures and Organizational Flexibility." Unpublished paper, University of Iowa. Delaney, John T., David Lewin, and Casey Ichniowski. 1989. Human Resource Policies and Practices in American Firms. Washington, D.C.: U.S. Department of Labor, BLMR 137. Eaton, Andrienne E. 1990. "The Role of the Local Union in a Participative Program." Labor Studies Journal, Vol. 15, No. 1 (Spring), pp. 33-53. Eaton, Andrienne E., and Paula B. Voos, 1992. "Productivity-Enhancing Innovations in Work Organization, Compensation, and Employee Participation in the Union Versus Nonunion Sectors." Unpublished paper, Rutgers University. Eaton, Andrienne E., and Paula B. Voos. 1989. "The Ability of Unions to Adapt to Innovative Workplace Arrangements. " American Economic Review, Vol. 79, No. 2 (May), pp. 172-76. Ehrenberg, Ronald E. 1990. Introduction to the symposium, "Do Compensation Policies Matter?" Industrial and Labor Relations Review, Vol. 43, No. 3 (February), pp. 3s-10s. Freeman, Richard B. 1980. "The Exit-Voice Tradeoff in the Labor Market: Unionism, Job Tenure, Quits, and Separations." Quarterly Journal of Economics, Vol. 94, No. 4 (June), pp. 643-73. Freeman, Richard B., and James L. Medoff. 1984. What Do Unions Do? New York: Basic Books. Greene, William H. 1991. LINWEP. Beliport, N.Y.: Econometric Software. Hammer, Tove H. 1988. "New Developments in Profit Sharing, Gainsharing, and Employee Ownership." In John P. Campbell and Richard J. Campbell, eds., Productivity in Organizations. San Francisco: Jossey-Bass, pp. 328-66. Hammer, Tove H., and Robert N. Stern. 1986. "A Yo-Yo Model of Cooperation: Union Participation in Management at Rath Packing Company." Industrial and Labor Relations Review, Vol. 39, No. 3 (April), pp. 337-49. Hewitt Associates. 1988. 1988 Profit Sharing Survey Chicago: Profit Sharing Council of America. ITI. 1991. The Michigan Foundation: A Study of the Modernization Process. Ann Arbor, Mich.: Industrial Technology Institute. Kelley, Maryellen R., and Bennett Harrison. 1992. "Unions, Technology, and Labor-Management Cooperation." In Lawrence Mishel and Paula B. Voos, eds., Unions and Economic Competitiveness. Armonk, N.Y.: M.E. Sharpe, pp. 247-86. Kruse, Douglas L. 1993. "Does Profit Sharing Affect Productivity?" Unpublished paper, Rutgers University. Levine, David I., and Laura D. Tyson. 1990. "Participation, Productivity, and the Firm's Environment." In Alan S. Blinder, ed., Paying for Productivity. Washington, D.C.: Brookings Institution, pp.183-243. Meyer, David G., and William N. Cooke. 1993. "U.S. Labour Relations in Transition: Emerging Strategies and Company Performance." British Journal of Industrial Relations, Vol. 31, No. 4 (December), pp. 531-52. Miller, Katherine I., and Peter R. Monge. 1986. Participation, Satisfaction, and Productivity: A Meta-Analytic Review." Academy of Management Journal, Vol. 29, No. 4 (December), pp. 727-53. Mitchell, Daniel J.B., David Lewin, and Edward E. Lawler, Ill. 1990. "Alternative Pay Systems. Firm Performance, and Productivity." In Alan S. Blinder, ed., Paying for Productivity. Washington, D.C.: Brookings Institution, pp. 15-88. Strauss, George. 1990. "Participatory and Gain Sharing Systems: History and Hope." In Myron J. Roomkin, ed., Profit Sharing and Gain Sharing. Metuchen, N.J.: Scarecrow Press, pp. 1-45. U.S. Department of Commerce. 1991a. 1987 Census of Manufactures. General Summary, MC87-S-1 1. Washington, D.C.: Bureau of the Census. _____. 1991b. 1989 Annual Survey of Manufactures. Statistics for Industry Groups and Industries. Washington, D.C.: Bureau of the Census. _____. 1991c. 1987 Census of Manufactures. Concentration Ratios in Manufacturing, MC87-S-6. Washington, D.C.: Bureau of the Census. U.S. General Accounting Office. 1987. Survey of Corporate Employee Involvement Efforts (data set provided on disk). Washington, D.C. Weitzman, Martin L. 1984. The Share Economy: Conquering Stagflation. Boston: Harvard University Press. Weitzman, Martin L., and Douglas L. Kruse. 1990. "Profit Sharing and Productivity." In Alan S. Blinder, ed., Paying for Productivity. Washington, D.C.: Brookings Institution, pp. 95-140. White, Halbert. 1980. "A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity." Econometrica, Vol. 48, No. 4 (May), pp. 817-38. Zalusky, John L. 1990. "Labor-Management Relations: Unions View Profit Sharing." In Myron J. Roomkin, ed., Profit Sharing and Gain Sharing. Metuchen, NJ.: Scarecrow Press. pp. 65-78.
|Printer friendly Cite/link Email Feedback|
|Author:||Cooke, William N.|
|Date:||Jul 1, 1994|
|Previous Article:||Long-run convergence of ethnic skill differentials: the children and grandchildren of the Great Migration.|
|Next Article:||The credibility of drug tests: a multi-stage Bayesian analysis.|