Between-Company Comparisons on Performance and Human Resource Policies
Over one-fifth of U.S. private-sector employees - 24 million workers -- own stock in their own companies. Over the past 25 years employee ownership and in other compensation arrangements linking worker pay to company performance, including profit sharing, gain-sharing, and broad-based stock options have increased substantially (Freeman and Dube 2000; Blasi, Kruse, and Bernstein 2003). While on average employee ownership firms tend to match or exceed the performance of other similar firms (Kruse and Blasi 1997), there is considerable dispersion of outcomes among employee owned firms, as evidenced by the bankruptcy of United Airlines (Mackin 2002) contrasted with the continued success of firms like SAIC.
By tying pay to firm performance, employee ownership can help improve performance by reducing workplace principal-agent problems. But group incentive systems face the free rider problem, highlighting the weak link between individual effort and rewards. As standard economic analysis provides no way to resolve the free rider problem, many researchers agree with Weitzman and Kruse that "something more may be needed-something akin to developing a corporate culture that emphasizes company spirit, promotes group cooperation, encourages social enforcement mechanisms, and so forth" (1990: 100). A three-pronged combination of i) incentives, which must be sufficiently meaningful to workers to motivate them; ii) participation, which must be sufficiently meaningful for workers to make critical decisions; and iii) a workplace environment or company ethos that overcomes, or at least reduces, the free rider problem appears to be the key to improving performance through employee ownership.
Over 70 large-sample studies have been done on employee ownership in the past two decades (Kruse 2002). Studies of both firm performance and employee attitudes and behavior are split between neutral and favorable findings for employee ownership, with very few negative findings. Meta-analysis indicates that the average increase in productivity associated with ESOP adoption is 4.5%. But the wide band of outcomes around the average makes it clear that giving employees an ownership stake does not, in itself, ensure superior employee or firm performance.
This study links employee reports on how ownership plans actually affect their attitudes and behavior to objective company performance measures. Such an approach is rare in productivity studies in part because employee surveys lack the quantitative output data necessary for a productivity analysis. Employees in worker-owned and participative firms report that their firms perform better than do employees in other firms (Freeman and Dube 2000), but the workers may not have an accurate assessment of their firm's actual performance. It is only by combining evidence from workers and firms - using matched employee-employer data files - that we are likely to make progress in understanding why some ownership plans work while other fail, and thus to explain the diversity of outcomes from companies with at least nominally similar ownership structures (though absent a true experiment, there will always remain questions of selection bias and generalizability).
This paper uses survey data from 13 ESOP companies to examine the factors that affect the differential impact of employee ownership on productivity and work behavior. The surveys were conducted at different periods of time by Ownership Associates, a consulting firm , and by our research team. The Ownership Associates (OA) survey covers employees and managers in eleven ESOP companies over the period 1996-2002, asking employees about their views and attitudes toward various aspects of their workplace. Company managers filled out a survey on human resource policies, firm performance, and ESOP characteristics. The firms in this survey are relatively small, with an average of 396 employees. There are a total of 2139 survey respondents from the 11 companies, giving a response rate for workers of 71% across all companies. The second dataset contains information on employees in two firms that the NBER's Shared Capitalism Research Project surveyed in 2002. Here the focus is on individual variation in how workers respond to ownership and participation and to free rider behavior on the part of coworkers. While selection bias in the types of firms that adopt ESOPs and take part in our surveys is a legitimate concern, prior ESOP literature indicates that results are not much affected by selection corrections (Kruse and Blasi 1997). Moreover, by basing our analysis on comparisons within ESOP firms, we potentially avoid errors in interpretation due to selectivity. Also, a set of fairly similar firms with comparable ownership structure provides just the right sample to assess variation within the employee ownership structure.
Our samples are small and thus give results that should be viewed as suggestive. The National Opinion Research Center (NORC) has completed a national survey using questions that we devised analogous to those in the current study. This survey has both a representative sample of workers and data that match workers with firms. Thus, this paper is a foray with limited data into an area that will offer new and nationally representative information in the near future.
Between-company Comparisons on Performance and Human Resource Policies
A growing literature has documented that "innovative human resource practices can improve business productivity," particularly when the firm combines complementary practices (Ichniowski et al 1996: 322; also see Appelbaum et al. 2000 and Becker et al. 2001). The OA data allow us to analyze the relationship between HR policies and performance, to see if HR policies can help create a climate that overcomes the free rider problem in employee ownership firms. As in Freeman and Dube (2000), we use employee-reported performance, but one advantage we have is data on objective company outcomes that can validate the employee reports. We construct three measures based on six items reflecting employee assessments of co-worker performance (listed at the bottom of Table 1). These measures vary significantly between firms and (as reported in our longer paper) are positively correlated with industry-adjusted firm performance, particularly with profit margin (.582 to .630), 3-year employment growth (.481 to .621), and 3-year productivity growth (.328 to .373).
There is also substantial dispersion in company-reported HR policies among the eleven OA companies. For example, seven firms have employee task forces, five have employee involvement in new hires, and three have employee representation on the board of directors. Firms also reported on nine methods of sharing information with employees (e.g., newsletters, regular employee meetings, new employee orientations), and other policies such as non-ESOP pension plans, grievance procedures, labor-management training, employee surveys, and bonuses. Since the HR policies are highly correlated and there are more policies than companies in the OA survey, we added together seven of the policies to form an HR index (described at the bottom of Table 1), which has a mean of 3.55 and standard deviation of 1.97.
Are the HR variables linked to performance? Table 1 reports regressions of the three employee-reported performance measures on the HR index and on the use of two practices that did not fit in the index, individual bonuses and a suggestion system. The results show that the HR index is positively related to worker-reported work effort, and significantly different from zero in five of the six regressions. Individual bonuses are positively related to the outcome variables, while suggestion systems are negatively related. We estimate that an increase of one standard deviation in the HR index increases the score on "People at [OurCo] work hard" by about .2, and increase the scores on performance indices 1 and 2 by about .8 and 1.2 (respectively). These represent increases of about 15-20% of a standard deviation in the performance measures.
In the even-numbered columns we include a variable for whether the firm introduced the ESOP because of economic performance concerns, which helps address questions of selection bias. Inclusion of this variable makes little difference in the coefficients on the HR variables, though it does slightly weaken the link between performance index 2 and the HR index.
Additional analysis of employee survey data shows that the HR index is positively and significantly linked to perceptions of fairness, co-worker relations, good supervision, and worker input and influence. It is not, however, linked to a sense of ownership. While this could indicate `that ownership is irrelevant to actual work performance, we find a strong positive correlation between the sense of ownership and our three outcome measures, both by itself and with the inclusion of the nearly independent HR index. Thus, it appears that feeling of ownership is judged by employees' actual financial stake in their company, as indicated by positive correlations with percent of company shares owned by the ESOP and ESOP value per employee. A sense of ownership must be backed up by practical implications for the individual employee.