Does Shared Capitalism Help the Best Firms Do Even Better?
Douglas Kruse, Rutgers University and NBER
Joseph Blasi, Rutgers University and NBER
Richard Freeman, Harvard University and NBER
May 2011
Prepared for British Academy Seminar, ―The Economics of Share Ownership and Gainsharing:
Findings and Policy Implications,‖ Centre for Economic Performance, London School of
Economics, May 26, 2011. This research was made possible with the gracious cooperation of
Amy Lyman and the Great Place to Work® Institute. It was supported by an Officer Grant from
the Sloan Foundation, and by the Foundation for Enterprise Development who funded the
Beyster Faculty Fellowships for Kruse and Blasi.
Abstract
We analyze the effects of employee ownership, profit and gain sharing, and broad-based
stock options (―shared capitalism‖) on employee attitudes, turnover, and performance among
applicants to the ―100 Best Companies to Work For in America‖ competition. The dataset
includes company surveys filled out by 780 firms for 1312 firm-year observations over 2005 to
2007, matched to representative employee surveys within each company totaling over 300,000
surveys in the three years. One finding is that company-reported shared capitalism is linked to
employee-reported empowerment measures (participation in decisions, information sharing,
high-trust supervision), and several measures of positive workplace culture. A second key
finding is that shared capitalism has favorable effects on voluntary turnover and employee intent
to stay, but only when combined with employee empowerment and positive workplace culture.
We find some similar patterns when examining return on equity and Tobin’s Q among the public
companies in the sample. The analysis provides a stringent test of the relationship between
shared capitalism and workplace performance, given that outcomes will be compressed in this
select sample of firms with the ―best‖ workplaces.
Introduction
About half of all private-sector workers have their pay or wealth tied to company or
workplace performance through employee ownership, profit sharing, gain sharing, or broadbased stock options (Kruse, Freeman, and Blasi, 2010). We refer to these collectively as ―shared
capitalism‖ plans. Over 100 studies indicate that these plans are linked to better average
workplace performance (see reviews in Kaarsemaker, 2007, Freeman, 2007, Freeman et al.,
2011). There is, however, considerable dispersion in the outcomes, with some plans linked to
very positive workplace outcomes and in other cases linked to null or negative outcomes. In the
past few years researchers have begun to uncover the individual- and firm-level factors that
moderate the effects of shared capitalism on employee attitudes, behavior, and performance.
These factors include complementary ―high-performance‖ workplace policies such as
participation in decisions, training, and job security that researchers have found to work well
together in improving workplace outcomes (e.g., Ichniowski et al., 1997; Appelbaum et al.,
2000; Kruse, Freeman, and Blasi, 2010).
While many studies have looked at outcomes under shared capitalism, there are very few
that have the data necessary to explore the mechanisms underlying its operation. Our recent
intensive studies had a large sample of over 41,000 employees but were based on only 14 firms.
We addressed the question of generalizability by replicating the findings in national employee
surveys (the 2002 and 2006 General Social Surveys), but the question remains of whether the
findings will extend to a larger firm-based sample.
In this paper we have a sample of 780 firms with up to 3 years of data on each firm,
including company reports on policies combined with representative employee surveys within
each company, totaling over 300,000 surveys. Unlike most firm-based studies that have data on
just one or two types of shared capitalism, this one has company-reported data on four key forms
of shared capitalism. The sample is unique in that all of the firms applied to the Great Place to
Work® Institute to be under consideration for inclusion on the list of the ―100 Best Companies
to Work for in America‖ published each year by Fortune magazine. The Institute compiles the
management surveys and administers the random employee surveys that go into this process of
selection. As such it is clearly not a representative sample of all firms, but is heavily skewed
toward firms that consider themselves candidates for the ―best companies‖ designation, so are
likely to have the best workplace policies and practices for attracting, motivating, and retaining
employees. As will be seen, about half of these firms have some form of shared capitalism plan
for employees. Our empirical task is to see whether even in this select group of firms: a) shared
capitalism is linked to better attitudes and outcomes, and b) the effects of shared capitalism are
moderated by high-performance policies.
If we created a distribution of all firms along a scale of workplace quality, these firms
would be concentrated in the ―upper tail,‖ and the variation in outcomes among them will be
more compressed than if we had a representative sample of all firms. As an analogy, the
variation in outcomes among professional basketball players will be more compressed than it
would be among a representative sample of the population, because these players have already
demonstrated their skill in one way or another. The greater compression makes it more difficult
to tease out differences among the subjects and the factors responsible for those differences.
Therefore our exploration of this sample can be seen as a stringent test of the effects of shared
capitalism, since the select nature of the sample creates a bias against finding significant
differences. While it is a select sample, it represents a noteworthy share of American workplaces
2
since these companies employ about 6 million workers or about 5% of the private-sector
workforce.
In the next section we summarize the theory and prior literature, followed by a
description of our model and the data. In the results section we first find that shared capitalism is
positively linked to measures of employee empowerment and workplace culture, and then find
that these measures moderate the effects of shared capitalism on voluntary turnover and
employee intent such that shared capitalism has favorable effects only when combined with
empowerment and positive workplace culture. We conclude with a discussion of limitations and
next steps for research.
Theory and Prior Literature
Establishing a closer link between compensation and performance is one solution to the
principal-agent problem of the employment relationship. The perception that compensation
plans based on workplace performance can help improve performance is widespread. In public
opinion polls, two-thirds to four-fifths of Americans believe that workers in employee-owned
firms ―work harder‖ and ―pay more attention to the quality of their work‖ than do other workers,
and four-fifths say that ―Employee stockholders are more likely than outside stockholders to vote
their shares in the best, long-term interest of the company‖ (summarized in Kruse and Blasi,
1999).
One of the key challenges facing all group incentives is the free rider or 1/N problem: in
a group incentive plan with N workers, individual incentives become diluted because each
individual receives on average only 1/N of the reward from his or her extra effort. It is possible,
however, for workers to establish a cooperative solution to the free rider problem by establishing
3
and enforcing work norms for high effort, with the results that all workers could be better off
(Axelrod, 1984; Fudenberg and Maskin, 1986). What it takes to establish and maintain such an
agreement, however, is not obvious. Laboratory experiments show that cooperation is more
common than would be predicted by simple models of self-interest, and is more likely when
participants can begin to form a group identity by talking with each other before making their
choices (Dawes and Thaler, 1988; Kahneman and Thaler, 1991). The empathy with co-workers
that may be built through communication can help develop and enforce worker norms that
support higher performance (Kandel and Lazear, 1992; Lazear, 1992).
These considerations make it clear that a shared capitalism plan cannot simply be
installed and expected to improve performance. ―To get the productivity-enhancing effects,
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‖ (Weitzman and Kruse, 1990: 100). As argued by Rousseau and
Shperling (2003), employee ownership may alter the psychological contract between the
employee and the organization, but only if combined with some of the perquisites of ownership:
Bundling equity and profit sharing with financial information and participation in
decision making can enhance worker contributions to the firm by creating employment
relationships based on congruent psychological contracts. Such a bundle can form the
basis of trust and aligned interests between workers and employer (2003: 564-565).
Most studies point toward a positive relationship between shared capitalism and
workplace performance. There have been over 100 studies on organizational performance
under shared capitalism policies, including not just comparisons among firms but also
pre/post studies following firms before and after the adoption of shared capitalism plans.1
For reviews of the employee ownership literature see Blasi 1988; Doucouliagos, 1995, Kruse and Blasi,
1997, Kruse, 2002, Kaarsemaker, 2006a,b, and Freeman, 2007. For a review of the broad-based stock option
literature see Blasi et al., 2003. For reviews of the profit sharing and gainsharing literatures see Weitzman
1
4
Two reviews of the employee ownership literature concluded that ―Two thirds of 129 studies on
employee ownership and its consequences found favourable effects relating to employee
ownership, while one tenth found negative effects‖ (Kaarsemaker, 2006), and ―research on
ESOPs [Employee Stock Ownership Plans] and employee ownership is overwhelmingly
positive and largely credible‖ (Freeman, 2007). Formal meta-analyses analyzing the combined
results of studies have found strong evidence of a positive association between shared capitalism
and performance (Doucialiagos, 1995; Kruse and Blasi, 1997; Weitzman and Kruse, 1990). A
number of the studies address firm-level selection effects by using panel data with fixed effects
to control for fixed unobservables, or by using other corrections to control for unobservables that
may not be fixed. Selection effects have also been addressed by random assignment: one field
study implemented random assignment of profit sharing at 3 of 21 establishments within a firm,
finding that the performance of those establishments improved relative to the control group
(Peterson and Luthans, 2006); also, laboratory evidence using a true experiment found higher
productivity among subjects organized into employee-owned ―firms‖ (Frohlich et al., 1998).
Our study using an employee survey before and after the introduction of a profit sharing plan
also lends support to these findings (Freeman, Kruse, and Blasi 2010: 31). In addition, two
studies cast doubt on worker self-selection as a serious problem, finding that average worker
quality did not change as compensation was changed from individual to group incentives, while
average worker performance improved under the group incentives (Weiss, 1987; Hansen, 1997).2
While the studies overall have found better workplace performance on average for
firms with shared capitalism plans, there is dispersion in results both within and across
studies. This dispersion—with shared capitalism apparently improving performance in
and Kruse, 1990, Bullock and Tubbs, 1990, Kruse, 1993, OECD, 1995, Doucialiagos, 1995, Welbourne and
Mejia, 1995. For a listing of all studies see the appendix to Freeman, Blasi, and Kruse (2011).
2 For further discussion of selection bias in studying shared capitalism, see Kruse, Freeman, and Blasi (2010: 29-33).
5
some firms but sometimes not in others—indicates that there are likely to be important
pre-conditions or enabling factors for shared capitalism to have a positive effect, in line
with the idea that “something more” is needed to create a cooperative solution to the free
rider problem. In our 2010 book we analyzed two nationally representative surveys and over
41,000 employee surveys in shared capitalism companies, finding that these plans have the most
positive effects on performance-related attitudes and behaviors when they are combined with
high-trust supervision (trusting employees to work well without close supervision), base wages
at or above market, and high-performance policies (job security, training, and employee
involvement in decisions). Without these policies these plans can have no or negative effects,
indicating that the dispersion in results across the 100+ studies may be explained in part by the
types of policies that do or do not accompany shared capitalism. Other recent studies finding
evidence of complementarities between shared capitalism and workplace policies include Jones
et al. (2010) and Pendleton and Robinson (2010).3
Another key theoretical challenge to shared capitalism programs is employee risk
aversion. Employees may react negatively toward the pay variability in shared capitalism
programs (Rousseau and Shperling, 2003), and may require a compensating differential in
pay, benefits, or work conditions. The pay variability may, for example, be offset in part or
whole by greater employment security that can accompany shared capitalism (as
Weitzman, 1984, argued with respect to profit sharing). There is little evidence of a paybased compensating differential. While employee ownership has occasionally been
adopted as part of wage or benefit concessions in a few high profile cases, this appears to
Pendleton and Robinson find that employee ownership appears to interact with formal employee
involvement programs in affecting performance only when a minority of employees participate in employee
ownership, while it has a direct unmoderated effect on performance when a majority of employees
participate. The authors note that this latter result may reflect the importance of an informal ownership
culture for performance when most employees are owners.
3
6
happen in only 4% of adoptions ((U.S. General Accounting Office, 1986; Blasi and Kruse,
1991: 325-328). More broadly, there is no evidence of generally lower fixed pay or benefits in
exchange for employee ownership and profit sharing. Both cross-sectional and longitudinal
studies have found ESOPs to be associated with higher or similar levels of pay relative to nonESOP firms or pre-ESOP pay (Kim and Ouimet, 2009; Kardas, Scharf, and Keogh 1998; Scharf
and Mackin 2000), while other forms of employee ownership are associated with higher average
compensation levels (Blasi et al., 1996), pension assets (Kroumova, 2000), wage growth
(Renaud et al., 2004), and overall worker wealth (Buchele et al., 2010). Likewise, firms with
profit-sharing or broad-based stock options generally have higher average compensation than
otherwise-comparable firms (Carstensen et al., 1995; Estrin and Wilson 1989; Handel and
Gittleman 2004; Hart and Hubler, 1991; Hubler, 1993; Sesil et al., 2007), and profit-sharing
workers have faster wage growth than other workers (Azfar and Danninger, 2001).
Another challenge related to risk is whether the level of objective risk in worker
portfolios exceeds the prudent amount. Benartzi and Thaler (2001) found that when given a
choice in a company 401(k), workers over-invest in company stock. Among the 41,000
employee surveys in a recent sample of companies with a range of employee ownership formats,
the median amount of a worker’s portfolio in company stock was 5 percent and the mean was 5
percent, while 16 percent had more than twice the mean. Working with Harry Markowitz who
received the Nobel Prize in Economics in portfolio theory, it was determined that around 10 to
15 percent of a worker’s portfolio in company stock constituted a manageable level of risk. But,
in an analysis inspired by Kahneman’s work, we found that workers with high economic
insecurity or subjective risk did not respond well to shared capitalism (Blasi, Kruse, and
Markowitz 2010: 121, 125, 110-120.) To address this issue, the study found that companies can
7
use ESOPs where the workers do not purchase the shares but are granted them or stock options.
Different formats of employee ownership can be combined with cash profit sharing and gain
sharing in order to offset the shared capitalism risk (Kruse, Freeman, and Blasi 2010).
These results on pay levels and risk, combined with the performance literature
summarized above, indicate that shared capitalism generally appears to be part of a “gift
exchange” (Akerlof, 1983) in which the plans can be structured so that they do not
introduce excessive risk into the workers’ basic pay package but instead draw on feelings
of reciprocity that help establish norms for greater effort and cooperation (Kruse, Blasi,
and Freeman: 377-382).
The prior literature therefore points to average positive effects of shared capitalism plans
on performance, but with substantial dispersion in results and insufficient exploration of the
mechanisms and conditions under which they can have an effect. Here we contribute evidence
on this question from a new data source that combines management-reported and employeereported survey data drawn from companies that tend to have the best workplace policies.
Data and Method
Data source
The data come from surveys collected by the Great Place to Work® Institute, to which
we were given access under a special arrangement with the Institute. There are two types of
survey data: 1) a ―Culture Audit‖ with detailed information on company policies, filled out by a
company representative, and 2) an employee survey filled out by 200-300 representative
employees in each company, measuring employee attitudes and perceptions of the company.
The randomized surveys were administered by the Institute. The Culture Audit was developed
8
as a result of intensive field work with corporations over the last fifteen years by the Institute. It
contains information on the availability of shared capitalism plans, the company’s other work
practices, and performance measures. About 400 companies supply culture audit and employee
survey data each year as part of an application to be one of the ―100 Best Companies to Work
For in America.‖ The dataset contains all applicants including those who won this particular
competition and those who did not win the competition. We were provided access to the dataset
under a strict confidentiality agreement which allowed us to analyze the data on an Institute
server, but not download or copy the data.
We have data on 780 companies over a 3-year span, collected in 2005-2007, with a total
of 1312 company-year observations. The companies that submitted surveys for one year only
were 480, for two years were 168, and for three years were 182. We were able to match to public
company performance information for 375 or 48% of the 780 corporations (the remainder are
privately-held). While this collection of public companies was not randomly selected, the group
of companies studied do meaningfully represent the companies traded on the U.S. public market,
namely the NYSE and the NASDAQ. For example, based on Standard & Poors Compustat, our
public company sample constitutes 10% of the total sales, 10% of the total employment, and
20% of the market value of all publicly traded corporations in 2007. The total number of
employee surveys over the three years was 305,339 or an average of 232 surveys per companyyear. While firms can be tracked by year, individual employees cannot be tracked by year.
Model and measures
We examine the relationship between company-level and individual-level measures using
multilevel techniques, based on the model presented in Figure 1. Shared capitalism is measured
at the company level using management responses on the culture audit, and employee
9
empowerment and workplace culture are measured at the individual level using employee survey
responses. We examine the relationship of empowerment and culture to a) individual-level
intent to stay, and b) company-level performance outcomes, testing both the potential mediating
and moderating effects of empowerment and culture on the relationship of shared capitalism to
these outcomes.
One advantage of our data is that by using both employee and manager surveys, we are
able to avoid the problem of common method bias for many of our estimated relationships (in
which relationships may be inflated by spurious covariance due to the common method used in
collecting the data). Our shared capitalism measures come from the manager-reported data,
while the measures of empowerment, culture, and intent to stay come from the employee survey,
and the public company performance data come from Standard and Poor’s Compustat. We also
have manager-reported data on actual company voluntary turnover separate from the employee
survey measure of individual intent to stay, allowing us to compare both outcome variables on
this key measure.
Descriptive data are provided in Tables 1 to 3. Regarding overall prevalence of shared
capitalism, Table 1 shows that close to one-sixth (17.6%) of firms report ESOPs, just under half
(44.5%) report granting stock options in the previous year, and close to one-fifth report cash
profit/gainsharing plans (18.1%) and deferred profit sharing plans (22.3%). More detail on the
plans is provided in Table 2. In the average ESOP company, the ESOP owns one-sixth (17%) of
the company, with 9% of the ESOPs owning a majority of the company. Twelve companies were
100% owned by their ESOPs. In the companies with stock options, the average percent of
employees granted stock options was 20.6%, while 16.4% of the companies were broad-based in
that they granted stock options to more than half of their employees and 8.7% of the companies
10
granted stock options to between 25% and 50% of workers. The average profit/gain sharing plan
provided an employee 7.2% of annual pay as a profit/gain share, and the average deferred profit
sharing plan provided a contribution equaling 6.5% of employee pay.
Table 2 also introduces the shared capitalism index that we use as a convenient
thermometer of the intensity of shared capitalism within a company.
The shared capitalism
index is a summated rating with one point each for 1) an Employee Stock Ownership Plan
(ESOP), 2) an ESOP owning 50% or more of the company, 3) a stock option plan covering
25% or more of employees, 4) a stock option plan covering 50% or more employees, 5) a
cash profit/gainsharing plan, 6) a cash profit/ gainsharing paying more than the median
percent of pay, 7) a deferred profit sharing plan, and 8) a deferred profit sharing with a
contribution above the median percent of pay. As seen in Table 2, the average score is .835,
with 52% of firm-years having a score of zero (i.e., no shared capitalism) and 6% having scores
of 3 or more (with a maximum of 5). Along with using this index as a measure of the intensity
of shared capitalism, we also break out the individual shared capitalism variables in many of the
specifications.
Table 3 provides descriptive statistics on the other variables used in this study. We focus
on three workplace practices that both theory and prior research suggest are complementary to
shared capitalism: high-trust supervision, high participation in decisions, and high information
sharing. We examine these separately as dependent variables, but given the high correlation
among them we analyze their effects by combining them into a measure of employee
empowerment (alpha=.890). For workplace culture we use six measures: perceived ―family‖ or
―team‖ feeling, perceived employee cooperation, perceived employee willingness to ―give
extra‖, feeling that one receives a fair share of company profits, the Great Place to Work Trust
11
Index©, and the perception that ―this is a great place to work.‖ The Trust Index© is used by the
Institute as a summary measure of employee perceptions of the company; it includes all items
from the employee survey, grouped into five dimensions and then averaged across the
dimensions.4 While many of these items tap employee perceptions of company culture, in the
regressions we keep the presentation manageable by using the Trust Index© as our summary
measure of company culture.
Table 3 also presents descriptive statistics for our individual-level measure of intent to
stay, and our company-level performance measures of voluntary and involuntary turnover,
worker compensation claims, and return on equity (ROE) and Tobin’s Q for the public
companies. The ―adjusted‖ ROE and Tobin’s Q figures represent the residuals for the Great
Place to Work® applicants from a robust regression of ROE or Tobin’s Q on forty industry
dummies, separately by year, for the entire Standard and Poor’s Compustat dataset. The
resulting value therefore represents the company’s performance after subtracting the average
performance for all public companies in the company’s industry in that year. The mean value of
.081 for adjusted ROE indicates that the applicants had an average ROE that was 8.1 percentage
points higher than the industry-year average, consistent with the idea that these are betterperforming firms in general.
Regression specifications
Cross-sectional comparisons among companies may be tainted by unobservable firmspecific factors (e.g., management quality, market placement). The fact that many of the firms
applied in more than one year allows us to account for firm-specific factors using random-effects
The five dimensions are labeled camaraderie (10 items, alpha=.94), credibility (13 items, alpha=.96),
fairness (12 items, alpha=.93), pride (8 items, alpha=.93), and respect (13 items, alpha=.94). The Trust
Index© that averages these five dimensions has an alpha of .970. We exclude the “intent to stay” item since
we examine the relationship between the Trust Index© and intent to stay in Table 7.
4
12
specifications. While it would be ideal to use fixed effects, there was very little within-firm
variation in the shared capitalism measures over the three years of our data, so we rely on
random effects specifications that exploit both within-firm and between-firm variation (though
we hope to use fixed effects as further years of data become available).
There are three stages of analysis. First we examine the overall relationship of shared
capitalism to the other outcomes using the following equations:
(1)
Y1ijt= a + b1*SCjt + b2*X1ijt + b3*X2jt + b4*Rj + eijt
(2)
Y2ijt= a + b1*SCjt + b2*X1ijt + b3*X2jt + b4*Rj + eijt
(3)
Y3jt= a + b1*SCjt + b2*X2jt + b3*Rj + ejt
where
Y1ijt = perception of empowerment and culture for individual i, firm j, year t
Y2ijt = intent to stay for individual i, firm j, year t
Y3jt = firm performance measure for firm j, year t
SCjt = shared capitalism measure(s) for firm j, year t
X1ijt = employee-level controls for individual i, firm j, year t
X2jt = company-level controls for firm j, year t
Rj = firm-level random effect for firm j
eijt = error term for individual i, firm j, year t
ejt = error term for firm j, year t
The X1 controls include individual-level employee demographic information (gender dummy,
full-time dummy, 6 dummies for race/ethnicity, 5 dummies for age category, 7 dummies for
occupation, 6 dummies for tenure category) and the X2 controls include company-level
characteristics (defined benefit pension plan dummy, natural logarithm of total employment,
unionized percent of workforce, whether publicly-held, age of company, dummies for services
and manufacturing, and natural logarithms of average hourly pay for largest hourly-paid group
and average salary for largest salaried group).
13
In the second stage of analysis we examine how empowerment and culture may mediate
the effects of shared capitalism by inserting the Y1 term into equations 2 and 3 above, and
estimating with and without the shared capitalism term:
(4)
Y2ijt= a + b1*Y1ijt + b2*SCjt + b3*Xjt + b4*Rj + eijt
(5)
Y3jt= a + b1* Y 1 jt + b2*SCjt + b3*Xjt + b4*Rj + eijt
where
Y 1 = mean of Y1 variable across individuals in firm j, year t
Other variables as defined above
The mediation analysis is conducted using the method of Baron and Kenny (1986) by examining
a) whether the shared capitalism measures predict outcomes in equations 2 and 3, b) whether the
empowerment and culture measures predict outcomes in equations 4 and 5, both before and after
shared capitalism is included, and c) whether the magnitudes of the shared capitalism
coefficients are decreased from equations 2 and 3 to equations 4 and 5.
Finally in the third stage of analysis we examine the potential moderating effects of
empowerment and culture by using interactions with shared capitalism.
(6)
Y2ijt= a + b1*Y1ijt + b2*SCjt + b3*(Y1ijt*SCjt) + b4*Xjt + b5*Rj + eijt
(7)
Y3jt= a + b1* Y 1 jt + b2*SCjt + b3*( Y 1 jt*SCjt) + b4*Xjt + b5*Rj + ejt
The b3 coefficients will indicate whether empowerment and culture appear to moderate the
effects of shared capitalism.
As noted earlier, this is a select sample drawn from among companies that think they
have a chance to be on the list of the 100 best companies to work for. As such, they are likely to
have both formal and informal policies that help create a good work experience for employees.
Companies that lack recognized formal policies (e.g., shared capitalism plans, work-life
14
programs) may make up for this with informal policies that establish a distinct and positive
culture (e.g., a firm without shared capitalism plans may have a variety of nonfinancial
incentives that successfully engage employees). Both the formal and informal policies will lead
to a compression of outcomes relative to what we would observe across a representative sample
of U.S. companies. This compression is likely to lead to a downward bias in both the
magnitudes and statistical significance of within-sample differences, so that any significant
differences we do observe are more likely to represent true differences.
Results
Shared capitalism is positively related to the workplace practice and culture variables, as
shown in Table 4. The regressions show that employees in companies with greater amounts of
shared capitalism are more likely to report high-trust supervision, participation in decisions,
information sharing, and more favorable outcomes on all of the culture measures. When the
shared capitalism index is replaced first by the dummy variables decomposing the index, and
then by the continuous shared capitalism variables, the ESOP variables appear most consistently
as positive predictors of good outcomes. Furthermore the coefficients on ―ESOP 50%+ of co.‖
are always larger than the coefficients on ―ESOP 1-49% of co.,‖ indicating that outcomes are
better in majority-owned ESOP companies than in minority-owned ESOP companies.
Shared capitalism is also a positive predictor of an employee’s intent to stay with the
company, as shown in the last column of Table 4. Consistent with this, company-reported
voluntary turnover is also lower in firms with greater amounts of shared capitalism, as shown in
column 1 of Table 5. The other company-reported performance outcomes are not as consistently
related to shared capitalism. Involuntary turnover is not significantly tied to the shared
15
capitalism index, but is lower among firms with cash profit/gainsharing and deferred profit
sharing plans (column 2), as predicted by Weitzman (1984). Workers compensation claims are
significantly lower in ESOP companies, but the extent of claims is not related to the percent of
company owned by the ESOP (column 3). Within the publicly-held companies, the shared
capitalism index is a weak positive predictor of ROE, but there are no strong predictors of ROE
or Tobin’s Q among the shared capitalism variables without including measures of company
culture (columns 4 and 5).
Given the strong relationship of shared capitalism to voluntary turnover and intent to
stay, we focus on these two outcome variables in Tables 6 and 7 in order to disentangle the
routes through which shared capitalism may have an effect. These tables examine both the
mediating and moderating effects of empowerment and culture.
The empowerment measure is a very strong predictor of voluntary turnover, both before
and after controlling for the shared capitalism index (Table 6, columns 2 and 3). When
controlling for empowerment, the coefficient on the shared capitalism index is reduced only
slightly (column 3), indicating that empowerment is only partially mediating the effects of shared
capitalism. The story is made richer, however, when considering the interactions in columns 4
and 5. The significant positive coefficient on the shared capitalism variable indicates that shared
capitalism has an unfavorable effect on voluntary turnover when empowerment is at a low level,
and the significant negative coefficient on the interaction term indicates that shared capitalism
has a favorable effect at higher levels of empowerment. The significant negative interaction is
found when specifying both as a continuous index (column 4) and as a dummy variable
indicating an index score above the upper quartile (column 5). We find the same pattern of
results when examining the interaction of shared capitalism with the Trust Index© (columns 6 to
16
9), although the interaction is statistically significant only with the continuous Trust Index©
(column 8).
These results are illustrated in Figure 2, based on regression 4 in Table 6.5 ―Low‖
empowerment is here defined as the empowerment score for the firm at the 10th percentile, while
―high‖ empowerment is the score at the 90th percentile. In a firm without any shared capitalism,
voluntary turnover is predicted to drop slightly from 15.3% to 14.4% as a firm moves from low
to high empowerment. In a firm with high shared capitalism (the maximum observed score of 5
on the index), voluntary turnover is predicted to drop dramatically from 17.6% to 6.3% as a firm
moves from low to high empowerment. It is noteworthy that the effects of shared capitalism
appear to be unfavorable with low employee empowerment, as predicted turnover is larger with
high shared capitalism (17.6%) than without shared capitalism (15.3%). This is consistent with
the results in Kruse, Freeman, and Blasi (2010), where we found positive effects of shared
capitalism on employee behavior and attitudes only when shared capitalism was combined with
high-trust supervision and high-performance practices. Our interpretation is that employees may
react badly to shared capitalism when they are over-supervised and not given the tools to
improve performance (―we want you to be inspired by the shared rewards, but we’re still going
to keep a close eye on you‖); in this case the shared capitalism may be seen primarily as shifting
financial risk onto employees.
An important finding is that these results from company-reported data are consistent with
employee reports on intent to stay, as seen in Table 7. The positive interaction coefficients in
columns 4 and 8 indicate that intent to stay increases when shared capitalism is combined with
high levels of the empowerment and trust. The coefficient on the base shared capitalism variable
5
A very similar figure is created when the results of regression 8 are used, replacing empowerment with the Trust
Index© (not shown but available).
17
is negative when the interactions are included, indicating lower intent to stay when shared
capitalism is implemented in companies where empowerment and the Trust Index© are low.
One difference from the pattern in Table 6 is that the shared capitalism coefficient is no longer
significant when controlling for empowerment or the Trust Index© (columns 3 and 7), which
would indicate either that these variables fully mediate the effects of shared capitalism or that
shared capitalism is a proxy for them in column 1. The interaction results from both Tables 6
and 7, however, point toward a richer story of moderation rather than mediation in analyzing the
effects of shared capitalism.
Do we find such patterns in other performance-related outcomes? Involuntary turnover
reflects decisions by the company and is not under control of the employee, so we do not
investigate how employee attitudes may mediate of moderate the effects of shared capitalism on
involuntary turnover. As found in Table 5, shared capitalism is not a predictor of workers
compensation claims, and in further tests we do not find that it interacts with empowerment or
culture in predicting workers compensation claims. We do find some positive interactions in
predicting ROE and Tobin’s Q in the public companies, as shown in Tables 8 and 9. In Table 8,
the shared capitalism index has a positive effect on Tobin’s Q only for firms in the top quarter of
employee empowerment (column 5), although the interactions with the continuous
empowerment measure and the Trust Index© are not statistically significant (columns 4, 8, 9).
In Table 9, the shared capitalism index has a positive effect on ROE only for firms in the top
quarter of the Trust Index© (column 9), although the interactions with the continuous Trust
Index© and the empowerment measures are not statistically significant (columns 4, 5, 8). Apart
from the question of shared capitalism’s effects, it is noteworthy that the Trust Index© is a
18
positive predictor of both Tobin’s Q and ROE, and empowerment is a significant predictor of
ROE, both with and without controlling for shared capitalism (columns 2, 3, 6, and 7).
We also explored a productivity measure for the public companies based on sales per
employee, using a standard Cobb-Douglas production function.6 While it would be better to use
value-added per employee, the Compustat dataset has necessary information for value-added for
only 20% of the companies, or 75 of the applicants, which is not a sufficient sample for
meaningful analysis. The sales per employee measure was not related to any of the shared
capitalism, empowerment, or trust measures (not reported here but available). We have concerns
about the quality of the productivity measure, since we would expect it to be related to voluntary
turnover and other performance measures. The lack of significant relationships may also
indicate that the effects of shared capitalism, empowerment, and the culture measures are
swamped by the influence of other factors inside and outside the company (e.g., type of
technology, industry shocks). Nonetheless the findings provide an important caution in
interpreting the other favorable results on the effects of shared capitalism.
Conclusion
Our key findings are: 1) shared capitalism predicts greater employee empowerment
(high-trust supervision, participation in decisions, and information sharing) and a variety of
positive perceptions of company culture, and 2) shared capitalism has a favorable effect on
voluntary turnover and employee intent to stay only when combined with employee
empowerment and a positive culture. These results are consistent with the findings from our
To obtain industry-specific parameters, a Cobb-Douglas specification was run separately by industry for
each of 40 industries over the 2000-2009 period for all Compustat companies, then the residuals for each
year for the applicants were used to form the dependent variable in specifications similar to those in Tables 8
and 9.
6
19
intensive analysis of two national random datasets from the General Social Survey and 41,000
employees in 14 companies with shared capitalism plans.
The results are strengthened by two features of our data. First, unlike most studies in this
area, we can match management and employee responses, helping avoid common method bias.
The fact that the voluntary turnover results are highly consistent with the results for employeereported intent to stay, and that one is based on management-reported data while the other is
based on employee-reported data, provides strong cross-validation that increases our confidence
in the findings. A second strength derives from the fact that this is a select sample of companies
who think they have a chance of being on the list of the 100 best companies to work for. As
described earlier, the outcomes within this sample will be more compressed than among a
representative sample of U.S. companies, making it more difficult to find significant differences.
Indeed, we would not have been surprised to find few significant differences among these
companies with the ―best‖ workplaces, just as the top basketball players may be roughly
equivalent. This makes it all the more noteworthy that we found positive effects of shared
capitalism when combined with empowerment and a positive culture.
There are of course a number of limitations. First, while we found consistent evidence in
how shared capitalism appears to affect voluntary turnover and employee intent to stay, with
some supportive findings in analyzing ROE and Tobin’s Q, we did not find these patterns in
analyzing workers compensation claims and the productivity measure. This may reflect
difficulty in sorting out the variety of influences on workers compensation claims and
productivity, whereas voluntary turnover is fully under control of employees. We think it also
reflects the poor data on productivity available in Compustat. A second limitation is the select
sample: while this provides a stringent test, it would clearly be valuable to have such high-
20
quality company-reported and employee-reported data from a fully representative sample of U.S.
companies. The select sample raises questions of generalizability, though the positive results are
consistent with the general findings of favorable outcomes for firms with shared capitalism. The
question of selection would be more acute if we had only those companies that made it onto the
100 best list; the fact that we have both those that did and did not make it onto the list gives us
greater variation to exploit. A third limitation is that there may be unobservable companyspecific factors that help account for our results. We have two or three years of observations on
many of the companies, allowing us to do random-effects specifications that help account for
unobservable factors in analyzing both within-firm and between-firm variation. We do not,
however, have sufficient within-company variation in shared capitalism over time to reliably use
company fixed effects, which would fully remove the effects of any fixed unobservable factors at
the company level. The addition of further years of data may provide us such variation.
The ―100 Best Companies to Work for In America‖ annual listing has itself become a
part of American business culture. Companies vie for the designation and talk about being on
the list. This study indicates that shared capitalism plays an important role among those who
apply for the designation. In addition, shared capitalism has a meaningful relationship to the
Trust Index©—an important measure of ―a good place to work‖—and to voluntary turnover and
some financial measures of performance. These results, using a large sample of banner U.S.
corporations and thousands of randomly administered worker surveys, both confirm and help
explain the overall findings from prior studies. The results strongly point to the value of further
research on how shared capitalism interacts with other workplace policies, particularly policies
that give workers the means to make a difference in the workplace and help engender a climate
of cooperation.
21
22
Bibliography
Akerlof, George. 1982. "Labor Contracts as Partial Gift Exchange," Quarterly Journal of
Economics, Vol. 97, pp. 543-569.
Appelbaum, Eileen, Thomas Bailey, Peter Berg, and Arne Kalleberg. 2000. Manufacturing
Advantage. Ithaca: Cornell University Press.
Axelrod, Robert M. 1984. The Evolution of Cooperation. New York: Basic Books.
Azfar, Omar, and Stephan Danninger. 2001. ―Profit Sharing, Employment Stability, and Wage
Growth,‖ Industrial and Labor Relations Review, 54(3), pp. 619-630.
Baron, Reuben M., and David A. Kenny. 1986. ―The Moderator-Mediator Variable Distinction
in Social Psychological Research.‖ Journal of Personality and Social Psychology, Vol.
51, No. 6, pp. 1173-82.
Benartzi, S. and R. Thaler. 2001. Naïve Diversification Strategies in Defined Contribution
Plans,‖ American Economic Review Vol. 91, No. 1, pp. 79-98.
Berger, Johannes, Claus Herbertz, and Dirk Sliwka, 2011. ―Incentives and Cooperation in
Firms: Field Evidence,‖ IZA DP No. 5618, Institute for the Study of Labor, April 2011.
Blasi, Joseph R. 1988. Employee Ownership: Revolution or Ripoff? New York: HarperCollins,
Ballinger Books.
Blasi, Joseph, and Douglas Kruse. 1991. The New Owners The Mass Emergence of Employee
Ownership in Public Companies and What it Means to American Business. New York:
HarperCollins.
Blasi Joseph, Michael Conte, and Douglas Kruse. 1996. ―Employee Stock Ownership and
Performance Among Public Companies,‖ Industrial and Labor Relations Review,
Volume 50, Number 1, 60-79.
23
Blasi, Joseph, Douglas Kruse, and Aaron Bernstein. 2003. In the Company of Owners: The Truth
About Stock Options (And Why Every Employee Should Have Them). New York: Basic
Books.
Blasi, Joseph, Douglas Kruse, and Harry Markowitz. 2010. ―Risk and Lack of Diversification
under Employee Ownership,‖ in Douglas Kruse, Richard Freeman, and Joseph Blasi, eds.
Shared Capitalism at Work: Employee Ownership, Profit Sharing, Gainsharing, and
Broad-based Stock Options. Chicago: University of Chicago Press.
Buchele, Robert, Douglas Kruse, Loren Rodgers, and Adria Scharf. 2010. ―Show Me the
Money: Does Shared Capitalism Share the Wealth?‖ in Douglas Kruse, Richard Freeman,
and Joseph Blasi, eds. Shared Capitalism at Work: Employee Ownership, Profit Sharing,
Gainsharing, and Broad-based Stock Options. Chicago: University of Chicago Press.
Bullock, R.J. and Mark E. Tubbs. 1990. "A Case Meta-Analysis of Gainsharing Plans as
Organization Development Interventions," Journal of Applied Behavioral Science, 26, 3,
pp. 383-404.
Cappelli, Peter, and David Neumark. 2001. ―Do 'High-Performance' Work Practices Improve
Establishment-Level Outcomes?‖ Industrial and Labor Relations Review, July 2001, v.
54, iss. 4, pp. 737-75.
Carstensen, Vivian, Knut Gerlach, and Olaf Hübler. 1995. ―Profit sharing in German firms,‖ in
Friedrich Buttler, Wolfgang Franz, Ronald Schettkat, and David Soskice (eds.),
Institutional frameworks and labor market performance: Comparative views on the U.S.
and German economies. London and New York: Routledge, 168-207.
Dawes, Robyn M., and Richard H. Thaler. 1988. ―Anomalies: Cooperation,‖ Journal of
Economic Perspectives, Summer 1988, v. 2, iss. 3, pp. 187-97
24
Doucouliagos, C. 1995. ―Worker participation and productivity in labor-managed and
participatory capitalist firms: a meta-analysis,‖ Industrial and Labor Relations Review,
49(1), 58-77.
Estrin, Saul, and Nicholas Wilson. 1989. "Profit Sharing, the Marginal Cost of Labour and
Employment Variability," Draft, Dept. of Economics, London School of Economics
(May).
Freeman, Richard B., Joseph R. Blasi, and Douglas L. Kruse. 2011. ―Inclusive Capitalism for the
American Workforce: Reaping the Rewards of Economic Growth through Broad-based
Employee Ownership and Profit Sharing,‖ Center for American Progress, Washington,
D.C., March.
Freeman, Richard B., Douglas L. Kruse, and Joseph Blasi, 2011. ―Worker Responses to
Shirking under Shared Capitalism,‖ ,‖ in Douglas Kruse, Richard Freeman, and Joseph
Blasi, eds. Shared Capitalism at Work: Employee Ownership, Profit Sharing,
Gainsharing, and Broad-based Stock Options. Chicago: University of Chicago Press.
Freeman, Steven F. 2007. ―Effects of ESOP Adoption and Employee Ownership: Thirty years
of Research and Experience,‖ Working Paper #07-01, Organizational Dynamics
Programs, University of Pennsylvania.
Frohlich, N., J. Godard, J. A. Oppenheimer, and F. A. Starke. 1998. Employee versus
conventionally- owned and controlled firms: An experimental analysis. Managerial and
Decision Economics 19 (4/ 5): 311– 26.
Fudenberg, Drew, and Eric Maskin. 1986. "The Folk Theorem in Repeated Games with
Discounting or with Incomplete Information." Econometrica, Vol. 54, No. 3 (May), pp.
533-54.
25
Handel, Michael, and Maury Gittleman. 2004. "Is There A Wage Payoff to Innovative
Practices?" Industrial Relations, Vol. 43 (1), pp. 67-97.
Hansen, D. G. 1997. ―Worker performance and group incentives: A case study,‖ Industrial and
Labor Relations Review 51 (1): 37– 49.
Hart, R.A., and Olaf Hubler. 1991. ―Are Profit Shares and Wages Substitute or Complementary
Forms of Compensation?‖ Kyklos 44, pp. 221-231.
Hubler, Olaf. 1993. ―Productivity, Earnings, and Profit Sharing: An Econometric Analysis of
Alternative Models,‖ Empirical Economics, 18, pp. 357-380.
Ichniowski ,Casey, Kathryn Shaw, and Giovanna Prennushi. 1997. ―The Effects of Human
Resource Management Practices on Productivity: a Study of Steel Finishing Lines,‖
American Economic Review, Volume 87. Number 23, 291-313.
Jones, Derek C.; Kalmi, Panu; Kauhanen, Antti. 2010. ―Teams, Incentive Pay, and Productive
Efficiency: Evidence from a Food-Processing Plant,‖ Industrial and Labor Relations
Review, July 2010, v. 63, iss. 4, pp. 606-626.
Jones, D., T. Kato, and J. Pliskin. 1997. Profit sharing and gainsharing: A review of theory,
incidence, and effects. In Handbook of human resources, ed. D. Lewin, D. Mitchell, and
M. Zaidi. Greenwich, CT: JAI Press.
Kaarsemaker, Eric C.A. 2006. ―Employee ownership and its consequences: Synthesis-generated
evidence for the effects of employee ownership and gaps in the research literature.‖
York, UK: University of York.
Kahneman, Daniel, Jack L. Knetsch, and Richard Thaler. 1991. ―Anomalies: The Endowment
Effect, Loss Aversion, and Status Quo Bias,‖ Journal of Economic Perspectives, Volume
5, Issue 1, Winter 1991, 193-206
26
Kandel, Eugene, and Edward P. Lazear. 1992. "Peer Pressure and Partnerships," Journal of
Political Economy, Vol. 100, No. 4 (August), pp. 801-17.
Kardas, Peter; Adria L. Scharf; and Jim Keogh. 1998. ―Wealth and Income Consequences of
ESOPs and Employee Ownership: A Comparative Study from Washington State,"
Journal of Employee Ownership Law and Finance. Vol. 10, No. 4, Fall.
Kaufman, Roger T. 1998. ―The Effects of Gain Sharing on the Basic Wage: The Case of
IMPROSHARE ,‖ Labour, Autumn 1998, v. 12, iss. 3, pp. 595-610.
Kim, E. Han and Paige Ouimet. 2011 ―Employee Stock Ownership Plans: Employee
Compensation and Firm Value,‖ working paper, Ross School of Business, University of
Michigan.
Kroumova, M. 2000. ―Investment in employer stock through 401(k) plans: Is there reason for
concern?‖ PhD dissertation. New Brunswick, NJ: Rutgers University.
Kruse, Douglas. 1993. Profit Sharing: Does It Make A Difference? Kalamazoo, MI: W.E.
Upjohn Institute for Employment Research.
Kruse, Douglas, and Joseph Blasi. 1997. "Employee Ownership, Employee Attitudes, and Firm
Performance: A Review of the Evidence." In David Lewin, Daniel J.B. Mitchell, and
Mahmood A. Zaidi, eds., Human Resources Management Handbook, Part 1. Greenwich,
CT.: JAI Press.
Kruse, Douglas, and Joseph Blasi. 1999. ―Public Opinion Polls on Employee Ownership and
Profit Sharing,‖ Journal of Employee Ownership Law and Finance, Vol. 11, No. 3,
Summer 1999, pp. 3-25.
27
Kruse, Douglas, Richard Freeman, and Joseph Blasi (Eds). 2010. Shared Capitalism at Work:
Employee Stock Ownership, Profit and Gain Sharing, and Broad-based Stock Options.
Chicago: University of Chicago Press.
Lazear, Edward P. 1992. ―Compensation, Productivity and the New Economics of Personnel,‖
in Lewin, David, Olivia Mitchell, and Peter Sherer, eds., Research frontiers in industrial
relations and human resources. Madison, WI: Indistrial Relations Research Association,
pp. 341-80
Organization for Economic Cooperation and Development (OECD). 1995. ―Profit sharing in
OECD countries,‖ OECD Employment Outlook, pp. 139– 69.
Pendleton, Andrew; Robinson, Andrew. 2010. ―Employee Stock Ownership, Involvement, and
Productivity: An Interaction-Based Approach,‖ Industrial and Labor Relations Review,
October 2010, v. 64, iss. 1, pp. 3-29.
Poutsma, Erik, Panu Kalmi, and Andrew Pendleton. 2006. ―The Relationship between Financial
Participation and Other Forms of Employee Participation: New Survey Evidence from
Europe,‖ Economic and Industrial Democracy, November, vol. 27, no. 4, pp. 637-67.
Peterson, Suzanne J., and Fred Luthans. 2006. ―The Impact of Financial and Nonfinancial
Incentives on Business-Unit Outcomes Over Time,‖ Journal of Applied Psychology, Vol.
91, No. 1, 156–165
Renaud, S., S. St- Onge, and M. Magnan. 2004. ―The impact of stock purchase plan participation
on workers’ individual cash compensation,‖ Industrial Relations 43(1): 120– 47.
Scharf, Adria, and Christopher Mackin. 2000. "Census of Massachusetts Companies with
Employee Stock Ownership Plans (ESOPs)." Boston: Commonwealth Corporation.
28
Sesil, James, Maya Kroumova, Douglas Kruse, and Joseph Blasi. 2007. ―Broad-based Employee
Stock Options in the United States: Company Performance and Characteristics,‖
Management Revue, Vol. 18, No. 2, 2007, pp. 5-22.
U.S. Bureau of Labor Statistics. 2007. National Compensation Survey: Employee Benefits in
Private Industry in the United States, 2005, Bulletin 2589. Washington, D.C.: U.S.
Bureau of Labor Statistics, May.
U.S. Bureau of Labor Statistics. 2010. National Compensation Survey: Employee Benefits in
the United States, March 2010, Bulletin 2752. Washington, D.C.: U.S. Bureau of Labor
Statistics, September.
United States General Accounting Office. 1987. Employee Stock Ownership Plans.
Washington, D.C.: U.S. General Accounting Office, Report GAO/PEMD-88-1. END
Weiss, Andrew. 1987. "Incentives and Worker Behavior," in Haig Nalbantian, ed., Incentives,
Cooperation, and Risk Sharing (Totowa, N.J.: Rowman and Littlefield, 1987), pp. 137-150.
Weitzman Martin. 1984. The Share Economy. Cambridge, MA: Harvard University Press.
Weitzman, Martin L, and Douglas Kruse. 1990. ―Profit Sharing and Productivity,‖ in Alan
Blinder, ed., Paying For Productivity: A Look at the Evidence. Washington, D.C.:
Brookings Institution.
Welbourne, T. M., and L. R. G. Mejia. 1995. Gainsharing: A critical review and a future research
agenda. Journal of Management 21 (3): 559– 610.
29
Individual level
Company level
Figure 1: Overall model
Shared capitalism
Employee ownership
Cash profit/gain sharing
Deferred profit sharing
Stock options
Employee
Empowerment and
Culture
1. Supervision
2. Participation in
decisions
3. Information
4. Perceptions of
culture
Company outcomes
1.
2.
3.
4.
5.
Voluntary turnover
Involuntary turnover
Workers comp claims
Tobin’s Q
ROE
Intent to stay
Figure 2: Shared Capitalism,
Empowerment, and Voluntary
Turnover
20.0%
18.0%
16.0%
14.0%
12.0%
10.0%
8.0%
6.0%
4.0%
2.0%
0.0%
17.6%
15.3%
14.4%
No shared capitalism
6.3%
High shared
capitalism
Low employee
empowerment
High employee
empowerment
Table 1: Prevalence of shared capitalism plans
All years
17.6%
2006
17.3%
2007
18.2%
2008
17.3%
Stock options granted in past year
44.5%
52.2%
40.2%
40.5%
Cash profit/gainsharing plan
18.1%
20.1%
19.3%
14.4%
Deferred profit sharing plan
22.3%
22.5%
23.0%
21.5%
1312
463
444
405
ESOP
n
32
Table 2: Characteristics of shared capitalism plans
Combining all company-years
If have ESOP, % of co. owned by ESOP:
Mean
(Std. dev.)
Median
<50% of co.
50%+ of co.
n
If granted stock options, % of employees granted:
Mean
(Std. dev.)
Median
<25% of employees
25-49% of employees
50%+ of employees
n
If have cash profit/gainsharing plan, bonus as avg. % of pay:
Mean
(Std. dev.)
Median
n
If have deferred profit-sharing plan, contribution as avg. % of pay:
Mean
(Std. dev.)
Median
n
Shared capitalism index
Mean
(Std. dev.)
Distribution
0
1
2
3
4
5
n
33
17.4%
(26.9)
5.9%
90.9%
9.1%
154
20.6%
(29.1)
6.5%
74.9%
8.7%
16.4%
538
7.2%
(10.5)
4.7%
201
6.5%
(8.7)
3.9%
212
0.835
(1.05)
52.3%
20.9%
20.4%
4.0%
2.0%
0.4%
1089
Table 3: Descriptive Statistics
Var. name
Shared capitalism index
Employee empowerment
High-trust
supervision
Participation in
decisions
Info sharing
Culture
Family
Cooperation
Give extra
Fair share
Trust Index©
Great place
Intent to stay
Company performance
Voluntary turnover
Variable definition/survey statement^
One point each for: any esop, esop>50% of co., stock options to >25% of
employees, stock options to >50% of employees, any cash
profit/gainsharing, cash profit/gainsharing>5% of pay, any deferred profit
sharing, deferred profit sharing>4% of pay
Management trusts people to do a good job without watching over their
shoulders.
Management involves people in decisions that affect their jobs or work
environment.
Management keeps me informed about important issues and changes.
There is a "family" or "team" feeling here.
You can count on people to cooperate.
People here are willing to give extra to get the job done.
I feel I receive a fair share of the profits made by this organization.
Great Places to Work Trust Index©
Taking everything into account, I would say this is a great place to work.
I want to work here for a long time.
Voluntary separations (excluding retirements) of full-time employees as
proportion of full-time employment (company-reported)
Workers comp claims Workers comp claims per 100 FTE (company-reported)
ROE
Return on equity (from S&P Compustat)
ROE adjusted
Return on equity relative to year-industry mean (upper and lower 1%
trimmed)
Tobin's Q
Tobin's Q (from S&P Compustat)
Tobin's Q adjusted
Tobin's Q relative to year-industry mean (upper and lower 1% trimmed)
^ All employee survey items scored on 1-5 scale (1="almost always untrue", 5="almost always true")
Survey statements are protected by copyright by Great Place to Work® Institute
34
Sample
Mean (S.d.)
size
0.835 (1.05)
1089
Mean (s.d.)
n
4.246
(0.96) 303928
3.821
(1.09)
302608
4.026
(1.02)
303855
4.215
4.116
4.200
3.671
(0.99)
(0.88)
(0.88)
(1.21)
303517
303232
303503
299067
4.180
4.351
(0.72)
(0.95)
305339
304160
4.266
(1.04)
302863
0.139
(0.12)
1258
4.774
(5.45)
1035
0.156
0.081
(0.18)
(0.15)
415
406
1.965
0.737
(1.64)
(1.31)
360
351
Table 4: Predicting Workplace Practices, Culture, and Intent to Stay under Shared Capitalism
Workplace practices
Dep. var.: High-trust supervision
Part. in decisions
Info sharing
(1)
(2)
(3)
Shared capitalism index
0.019 (3.07) ***
0.012 (2.12) **
0.015 (2.09) **
n
Wald chi-sq. (38)
Elements of index
Stock options 1-25%
Stock options 25-49%
Stock options 50%+
ESOP 1-49% of co.
ESOP 50%+ of co.
Def. PS <4% of pay
Def. PS 4%+ of pay
Cash PS/GS <5% of pay
Cash PS/GS 5%+ of pay
241564
3863.5
n
Wald chi-sq. (45)
Continuous variables
Any stock options
% of ees. given options
ESOP
% of co. owned by ESOP
Deferred PS plan
Def. PS as % of pay
Cash PS/GS plan
Cash PS/GS as % of pay
241564
3888.7
-0.005
0.029
0.015
0.076
0.149
0.050
0.082
-0.013
-0.011
-0.005
0.000
0.066
0.001
0.048
0.266
-0.007
-0.001
240507
7299.8
(0.27)
(1.03)
(0.55)
(3.36)
(2.74)
(2.08)
(3.49)
(0.71)
(0.53)
***
***
**
***
-0.015
0.038
0.011
0.066
0.128
0.035
0.081
-0.006
-0.011
241499
6390.4
(0.67)
(1.04)
(0.30)
(2.31) **
(1.82) *
(1.15)
(2.72) ***
(0.27)
(0.41)
240507
7311.9
(0.28)
(0.87)
(2.71) ***
(1.70) *
(2.14) **
(1.41)
(0.41)
(0.95)
-0.019
0.001
0.060
0.001
0.038
0.322
-0.009
0.000
-0.022
0.016
-0.021
0.071
0.109
0.042
0.061
-0.005
0.002
(1.04)
(0.47)
(0.61)
(2.69) ***
(1.69) *
(1.51)
(2.24) **
(0.23)
(0.08)
241499
6402.2
(0.80)
(1.39)
(1.93) *
(1.10)
(1.33)
(1.32)
(0.44)
(0.10)
-0.024
0.000
0.066
0.001
0.033
0.294
-0.006
0.000
Family
(4)
0.015
Culture
Cooperation
(5)
(2.07) **
0.012 (2.12) **
241227
6675.6
240999
5136.1
-0.026
0.012
-0.020
0.066
0.139
0.034
0.053
-0.007
0.006
(1.28)
(0.38)
(0.61)
(2.56) **
(2.22) **
(1.23)
(1.97) **
(0.34)
(0.26)
241227
6686.5
(1.11)
(0.69)
(2.31) **
(0.83)
(1.27)
(1.32)
(0.31)
(0.38)
-0.029
0.000
0.058
0.001
0.026
0.268
-0.001
0.000
0.000
0.030
0.017
0.038
0.114
0.041
0.064
-0.016
-0.018
(0.01)
(1.14)
(0.63)
(1.83)
(2.23)
(1.82)
(2.95)
(0.92)
(0.91)
*
**
*
***
240999
5153.8
(1.40)
(1.06)
(2.09) **
(1.35)
(1.02)
(1.23)
(0.07)
(0.17)
-0.003
0.000
0.032
0.001
0.029
0.371
-0.017
0.000
(0.18)
(1.36)
(1.42)
(1.62)
(1.39)
(2.11) **
(1.11)
(0.07)
n
241564
240507
241499
241227
240999
Wald chi-sq. (45)
3890.8
7312.2
6404.4
6690.1
5158.8
* p<.10 ** p<.05 *** p<.01
Each column has three regressions: 1)with the shared capitalism index, 2) breaking out the elements of the shared capitalism
index, and 3) using the continuous variables underlying the shared capitalism index. All regressions are run using a random-effects
maximum likelihood multilevel model, with random effects both at the company-year level and the company level. Companylevel control variables include defined benefit pension, ln(total employment), ln(avg. hourly pay), ln(avg. salaried pay), company
age, whether publicly-held, manufacturing, services, and percent of employees who are unionized. Individual-level control
variables include gender, race/ethnicity (7 dummies), age (5 dummies), tenure (7 dummies), and occupation (7 dummies).
Regressions were run using Stata's xtmixed command.
35
Table 4 (cont.)
Culture
Dep. var.: Give extra
Fair share
Trust Index©
(6)
(7)
(8)
Shared capitalism index
0.020 (3.33) ***
0.073 (6.29) ***
0.069 (2.34) **
n
Wald chi-sq. (38)
Elements of index
Stock options 1-25%
Stock options 25-49%
Stock options 50%+
ESOP 1-49% of co.
ESOP 50%+ of co.
Def. PS <4% of pay
Def. PS 4%+ of pay
Cash PS/GS <5% of pay
Cash PS/GS 5%+ of pay
241241
7054.4
n
Wald chi-sq. (45)
Continuous variables
Any stock options
% of ees. given options
ESOP
% of co. owned by ESOP
Deferred PS plan
Def. PS as % of pay
Cash PS/GS plan
Cash PS/GS as % of pay
241241
7077.7
-0.003
0.048
0.069
0.041
0.097
0.045
0.075
-0.024
-0.013
-0.008
0.001
0.036
0.001
0.031
0.447
-0.024
0.000
237488
8825.5
(0.17)
(1.83)
(2.56)
(1.90)
(1.88)
(1.93)
(3.30)
(1.35)
(0.62)
*
**
*
*
*
***
0.022
0.074
0.083
0.160
0.305
0.078
0.237
0.007
0.100
242587
9910.2
(0.66)
(1.48)
(1.59)
(3.83)
(3.02)
(1.72)
(5.42)
(0.22)
(2.56)
***
***
*
***
**
237488
8843.6
(0.49)
(3.58) ***
(1.57)
(1.31)
(1.45)
(2.49) **
(1.54)
(0.53)
0.019
0.001
0.134
0.002
0.110
0.773
0.029
0.002
-0.024
0.158
0.037
0.264
0.584
0.166
0.317
-0.101
-0.029
241751
4860
(0.29)
(1.23)
(0.28)
(2.50) **
(2.29) **
(1.46)
(2.87) ***
(1.17)
(0.29)
242587
9928
(0.57)
(1.70)
(2.99)
(2.04)
(2.59)
(2.18)
(0.96)
(1.26)
*
***
**
***
**
-0.032
0.002
0.230
0.004
0.166
1.186
-0.073
-0.001
Intent to stay
Great place
(9)
0.022 (3.03) ***
-0.007
0.050
0.001
0.067
0.159
0.025
0.085
-0.022
0.025
-0.007
0.000
0.056
0.001
0.023
0.495
-0.004
0.000
(1.73) *
240715
7958.6
(0.32)
(1.58)
(0.04)
(2.56) **
(2.53) **
(0.88)
(3.10) ***
(1.04)
(1.01)
241751
4877.5
(0.39)
(1.16)
(2.03) **
(1.42)
(1.56)
(1.33)
(0.95)
(0.15)
(10)
0.012
0.006
0.043
-0.017
0.056
0.175
0.009
0.063
-0.026
0.005
(0.28)
(1.37)
(0.52)
(2.22) **
(2.86) ***
(0.33)
(2.35) **
(1.21)
(0.19)
240715
7978.4
(0.34)
(0.99)
(1.99) **
(1.67) *
(0.88)
(2.25) **
(0.22)
(0.07)
0.006
0.000
0.043
0.001
0.021
0.233
-0.012
0.000
(0.30)
(0.17)
(1.56)
(2.11) **
(0.81)
(1.09)
(0.64)
(0.27)
n
241241
237488
242587
241751
240715
Wald chi-sq. (45)
7087.3
8840.4
9927.7
4874.2
7973.4
* p<.10 ** p<.05 *** p<.01
Each column has three regressions: 1)with the shared capitalism index, 2) breaking out the elements of the shared capitalism
index, and 3) using the continuous variables underlying the shared capitalism index. All regressions are run using a random-effects
maximum likelihood multilevel model, with random effects both at the company-year level and the company level. Companylevel control variables include defined benefit pension, ln(total employment), ln(avg. hourly pay), ln(avg. salaried pay), company
age, whether publicly-held, manufacturing, services, and percent of employees who are unionized. Individual-level control
variables include gender, race/ethnicity (7 dummies), age (5 dummies), tenure (7 dummies), and occupation (7 dummies).
Regressions were run using Stata's xtmixed command.
36
Table 5: Predicting Firm-level Outcomes with Shared Capitalism
Dep. var.: Voluntary turnover, Involuntary
full-time workers
turnover
(1)
(2)
Shared capitalism index
-0.008 (0.004) **
-0.004 (0.002)
R-squared
n
Elements of index
Stock options 1-25%
Stock options 25-49%
Stock options 50%+
ESOP 1-49% of co.
ESOP 50%+ of co.
Def. PS <4% of pay
Def. PS 4%+ of pay
Cash PS/GS <5% of pay
Cash PS/GS 5%+ of pay
R-squared
n
Continuous variables
Any stock options
% of ees. given options
ESOP
% of co. owned by ESOP
Deferred PS plan
Def. PS as % of pay
Cash PS/GS plan
Cash PS/GS as % of pay
0.34
1011
0.004
-0.025
-0.012
-0.015
0.036
0.029
-0.018
-0.013
-0.023
0.317
968
(0.011)
(0.018)
(0.018)
(0.012)
(0.034)
(0.013) **
(0.013)
(0.011)
(0.013) *
0.353
1011
0.003
0.000
-0.020
0.001
0.019
-0.225
-0.015
0.000
Workers' comp.
claims
(3)
0.036 (0.17)
0.014
0.008
0.026
0.000
0.018
0.008
-0.018
-0.012
-0.017
0.105
848
(0.007)
(0.011)
(0.011)
(0.008)
(0.021)
(0.008)
(0.008)
(0.007)
(0.008)
**
**
**
*
**
0.335
968
(0.011)
(0.000)
(0.014)
(0.000)
(0.012)
(0.109) **
(0.010)
(0.001)
0.013
0.000
-0.001
0.000
-0.002
-0.065
-0.013
0.000
-0.774
-0.686
-0.270
-1.937
-1.543
-0.341
0.793
0.871
0.057
-0.761
0.001
-2.038
0.007
0.543
-4.300
0.724
-0.032
(4)
0.0149 (0.008) *
0.107
406
(1.15)
(0.60)
(0.24)
(2.60) *
(0.78)
(0.42)
(1.07)
(1.24)
(0.08)
0.123
848
(0.007) *
(0.000)
(0.009)
(0.000)
(0.008)
(0.066)
(0.006) **
(0.000)
Return on equity
-0.013
-0.007
-0.017
0.0253
na
-0.001
0.0553
0.055
0.0379
-0.016
0.0003
0.0293
-5E-04
0.0035
0.4801
0.0496
-7E-04
(5)
0.0114 (0.064)
0.202
351
(0.023)
(0.034)
(0.034)
(0.024)
(0.030)
(0.033)
(0.035)
(0.029)
0.109
406
(1.12)
(0.07)
(2.47) *
(0.33)
(0.76)
(0.70)
(1.21)
(0.96)
Tobin's Q
0.055
0.054
0.113
-0.04
na
-0.29
-0.04
-0.05
0.112
(0.209)
(0.297) *
(0.289)
(0.197)
(0.254)
(0.267)
(0.277)
(0.253)
0.232
351
(0.024)
(0.000)
(0.028)
(0.003)
(0.037)
(0.605)
(0.034)
(0.003)
0.0774
0.0005
-0.056
0.002
-0.547
7.747
0.0447
-0.001
(0.213)
(0.003)
(0.232)
(0.020)
(0.304) *
(4.878)
(0.271)
(0.024)
R-squared
0.343
0.329
0.123
0.11
0.217
n
1011
968
848
406
351
* p<.10 ** p<.05 *** p<.01 (s.e. in parentheses)
All regressions use random effects with correction for autogression. Control variables include defined benefit pension, ln(total
employment), ln(avg. hourly pay), ln(avg. salaried pay), company age, whether publicly-held, manufacturing, services, and percent
of employees who are female, black, Hispanic, asian, other race, age 26-34, age 35-44, age 45-54, age 55+, and unionized.
37
Table 6: Interacting shared capitalism with practices and culture in predicting voluntary turnover
Dep. var.: Voluntary turnover rate among full-time workers (company level)
(1)
(2)
(3)
(4)
(5)
Shared capitalism index
-0.008 **
-0.007 * 0.1374 ** -0.0033
(0.004)
(0.004)
(0.056)
(0.004)
Employee empowerment
-0.0454 *** -0.044 *** -0.016
(0.048) **
(0.015)
(0.016)
(0.020)
(0.020)
* shared capitalism index
-0.036 ***
(0.014)
High level of employee
0.014
empowerment (dummy)
(0.010)
* shared capitalism index
-0.013 **
(0.006)
Trust Index©
* shared capitalism index
High trust dummy
* shared capitalism index
(6)
(7)
(8)
(9)
-0.007 * 0.154 ** -0.005
(0.004)
(0.065)
(0.004)
-0.056 *** -0.054 *** -0.023
-0.042
(0.017)
(0.019)
(0.023)
(0.024)
-0.039 **
(0.016)
-0.005
(0.004)
-0.008
(0.006)
n
1011
1168
1011
1011
1011
1168
1011
1011
1011
R-squared
0.348
0.346
0.348
0.353
0.35
0.349
0.351
0.355
0.353
* p<.10 ** p<.05 *** p<.01 (s.e. in parentheses)
All regressions use random effects with correction for autogression. Control variables include defined benefit pension, ln(total
employment), ln(avg. hourly pay), ln(avg. salaried pay), company age, whether publicly-held, manufacturing, services, and percent of
employees who are female, black, Hispanic, asian, other race, age 26-34, age 35-44, age 45-54, age 55+, and unionized.
38
Table 7: Interacting shared capitalism with practices and culture in predicting intent to stay
Dep. var.: Intent to stay at company (individual level)
(1)
(2)
Shared capitalism index
0.012 *
(0.007)
Employee empowerment
0.7347 ***
(0.002)
* shared capitalism index
High level of employee
empowerment (dummy)
* shared capitalism index
(3)
(4)
(5)
0.0006
-0.03 *** -0.002
(0.004)
(0.008)
-0.004
0.7365 ***
0.73 *** 0.8138 ***
(0.002)
(0.002)
-0.003
0.0076 ***
(0.002)
-0.206 ***
-0.006
0.0064 **
-0.003
Trust Index©
(6)
(7)
-0.003
(0.003)
(8)
(9)
-0.047 *** -0.006 *
(0.008)
(0.003)
1.114 *** 1.115 *** 1.106 ***
1.18 ***
(0.002)
(0.002)
(0.003)
-0.002
0.011 ***
(0.002)
-0.202 ***
(0.005)
0.0108 ***
(0.003)
* shared capitalism index
High trust dummy
* shared capitalism index
n
240715
278667
240550
240550
240550
278852
240714
240714
240714
Wald chi-sq.
7958.6
191767
166966
166998
169943
391595
340643
340722
346887
* p<.10 ** p<.05 *** p<.01 (s.e. in parentheses)
All regressions are run using a random-effects maximum likelihood multilevel model, with random effects both at the company-year level
and the company level. Company-level control variables include defined benefit pension, ln(total employment), ln(avg. hourly pay), ln(avg.
salaried pay), company age, whether publicly-held, manufacturing, services, and percent of employees who are unionized. Individual-level
control variables include gender, race/ethnicity (7 dummies), age (5 dummies), tenure (7 dummies), and occupation (7 dummies).
39
Table 8: Interacting shared capitalism with practices and culture in predicting Tobin's Q
Shared capitalism index
Employee empowerment
* shared capitalism index
(1)
0.011
(0.064)
(2)
0.462 *
(0.282)
(3)
0.003
(0.064)
0.546
(0.335)
(4)
-0.638
(1.007)
0.369
(0.435)
0.162
(0.253)
High level of employee
empowerment (dummy)
* shared capitalism index
(5)
-0.0573
(0.067)
0.827 **
(0.385)
(6)
(7)
-0.002
(0.064)
(8)
-0.582
(1.143)
(9)
-0.024
(0.069)
0.7292
(0.490)
0.1406
(0.277)
0.648
(0.460)
(0.647)
(0.220) ***
0.350 ***
(0.117)
Trust Index©
0.834 *** 0.8819 **
(0.327)
(0.395)
* shared capitalism index
High trust dummy
0.0591
(0.227)
0.0834
(0.124)
* shared capitalism index
n
351
452
351
351
351
452
351
351
351
Wald chi-sq.
0.202
0.194
0.212
0.216
0.236
0.202
0.223
0.225
0.226
* p<.10 ** p<.05 *** p<.01 (s.e. in parentheses)
All regressions use random effects with correction for autogression. Control variables include defined benefit pension, ln(total
employment), ln(avg. hourly pay), ln(avg. salaried pay), company age, whether publicly-held, manufacturing, services, and percent of
employees who are female, black, Hispanic, asian, other race, age 26-34, age 35-44, age 45-54, age 55+, and unionized.
40
Table 9: Interacting shared capitalism and culture in predicting return on equity
Shared capitalism index
Employee empowerment
* shared capitalism index
(1)
0.015 *
(0.008)
(2)
0.088 **
(0.037)
(3)
(4)
0.014 * -0.151
(0.008)
(0.126)
0.116 *** 0.071
(0.041)
(0.053)
0.042
(0.032)
High level of employee
empowerment (dummy)
* shared capitalism index
(5)
0.0104
(0.008)
0.101 **
(0.048)
(6)
(7)
0.013 *
(0.008)
(8)
-0.217
(0.143)
(9)
0.0066
(0.008)
-0.012
(0.027)
0.022
(0.015)
Trust Index©
0.1426 *** 0.166 *** 0.1067 *
(0.043)
(0.047)
(0.060)
0.0557
(0.035)
* shared capitalism index
High trust dummy
0.1573 ***
(0.056)
-0.03
(0.026)
0.0339 **
(0.015)
* shared capitalism index
n
406
512
406
406
406
512
406
406
406
Wald chi-sq.
0.107
0.088
0.137
0.135
0.14
0.101
0.152
0.150
0.16
* p<.10 ** p<.05 *** p<.01 (s.e. in parentheses)
All regressions use random effects with correction for autogression. Control variables include defined benefit pension, ln(total
employment), ln(avg. hourly pay), ln(avg. salaried pay), company age, whether publicly-held, manufacturing, services, and percent of
employees who are female, black, Hispanic, asian, other race, age 26-34, age 35-44, age 45-54, age 55+, and unionized.
41