Electronic copy available at: http://ssrn.com/abstract=1556746
SBR 61 October 2009 393412 393
Oliver Ludewig/Dieter Sadowski*
Measuring OrganizatiOnal Capital**
abs tr aCt
Firms develop their organizational practices to realize returns from given and market-
able resources. Implementing effective practices requires substantial up-front invest-
ment. We approximate the economic relevance of establishment-specific organizational
capital by using a two-step procedure. First, we extract an establishment-specific per-
formance differential from a within-panel estimator. Second, we explain the variation in
this differential by using organizational and control variables. Our results make it possi-
ble for us to predict the contribution of organizational practices to the performance dif-
ferential. We label this part of the firm-specific performance differential “organizational
capital”. Our results indicate that organizational capital has a substantial impact on per-
JEL-Classification: D24, L23, M29.
Keywords: Managerial Economics; Organizational Capital; Organizational Design;
1 Org ani zati Ona l Cap ital
Organizations develop their organizational practices to realize returns from given and
marketable resources like real capital and labor. Organizational practices that facilitate the
creation of sustainable above-average returns must be durable and idiosyncratic, i.e., hard
to imitate. In cases in which their implementation requires substantial investments, these
organizational practices, routines, and processes represent “organizational capital”.
e economics and management literature attaches diﬀerent meanings to the term “orga-
nizational capital”. ere are at least two schools of thought. One views organizational
* Oliver Ludewig, Institute for Employment Research (IAB), Regensburger Straße 104, D-90478 Nürnberg.
Dieter Sadowski, Institute for Labour Law and Industrial Relationships in the European Community (IAAEG),
University of Trier, D-54286 Trier.
** We thank two anonymous referees, Douglas C. Bice, Ruslan Gurtoviy, Susanne Warning, and Dodo zu Knyphau-
sen-Aufseß as well as the participants of seminars, workshops and conferences at the universities of Zurich, Ha-
nover, Lima, and Munich. We greatly appreciate Sandra Sizer’s help to improve our English.
Electronic copy available at: http://ssrn.com/abstract=1556746
O. LUDEWIG/D. SADOWSKI
SBR 61 October 2009 393412
capital as residing in the organization’s members and their social networks (Prescott and
Visscher (1980)). e other ascribes organizational capital to the organization itself, not
to its members, considering it to be embodied in the company’s routines and practices
(Tomer (1986; 1987); Lev and Radhakrishnan (2005)). If we were to take this second
view to the extreme, then the stock of organizational capital would remain unchanged,
even if all employees were replaced.
To support the second viewpoint, which is our approach, we provide two examples:
If a whole football team is replaced, including the coach, but the playing strategy
and the performance stay unchanged, we can speculate that the organizational prac-
tices, such as their incentive or playing systems, are responsible for resulting in the
same outcome as before the replacement. ere is a famous example: Until the end
of the 1990s, Ajax Amsterdam applied the same playing system to all teams, from
the youngest junior team to the professionals. ere was always a pool of young,
motivated players who knew the system and could replace any injured or departing
player from the top team quickly and without any substantial loss in quality. It was
the system that made the diﬀerence, not the individual players. It was the base of
Ajax’s golden years.
During the 1990s, Wal-Mart was hailed as having successfully deployed information
and communication technology. However, this success was not because Wal-Mart had
exclusive access to special equipment linking the checkout registers directly with the
vendors. Rather, Wal-Mart’s use of this technology facilitated innovations in manage-
ment and organizational structure1. Wal-Mart’s competitors were slow to imitate this
use of freely available information and communication technology, giving Wal-Mart
a competitive advantage for quite some time.
In their 2005 study, Lev and Radhakrishnan state that “Organization capital is … an
agglomeration of technologies – business practices, processes and designs, ... – that
together enable some ﬁrms to consistently and eﬃciently extract from a given level of
physical and human resources a higher level of product than other ﬁrms ﬁnd possible to
obtain.” Sadowski (2002) provides a similar, but more speciﬁc, deﬁnition: “If an enter-
prise succeeds in giving itself an order, including an amount of rules to share informa-
tion, settle conﬂicts, secure the willingness to cooperate, then we can call this order with
good reason ‘organizational capital’.” is idea holds for labor relations and for the rela-
tionships with other shareholders and stakeholders.
e metaphorical use of the term “capital” has a long and successful history. For example,
the concepts of establishment-speciﬁc and general human capital are basic in modern
economics, although they do not have a simple empirical correlate. From a ﬁrm’s perspec-
tive, “human capital” is the economic value of the knowledge and competencies of the
employees, deployed in favor of the ﬁrm.
1 Brynjolfsson, Hitt, and Yang (2002, 146); Lev and Radhakrishnan (2005); and Ramirez and Hachiya (2006b).
SBR 61 October 2009 393412 395
As far as social capital is concerned, Coleman (1988) states that it “[…] is created [...]
when the relations among persons change in ways that facilitate action.” Moreover, social
capital secures the ﬂow of information, helps to coordinate actions, and facilitates coop-
eration (Matiaske (1999))2. us, human capital rests on the skills, knowledge, and abil-
ities of people, and social capital is based on individuals’ relationships with other persons
both inside and outside the ﬁrm.
If ﬁrms can beneﬁt from the human and social capital of their personnel via purposeful
organizational practices, then such practices have a positive economic value and they
contribute also to “organizational capital” (Hardin’s 1999 concept of institutional capital is
analogous). Social capital, human capital, and organizational capital are linked, but theo-
retically separate. Bernd Schauenberg (1983) provided an early and thoughtful analysis of
these links; Bounfour (2009) shows how hard it still is to develop a metrics and statistical
reports to identify them empirically.
e research on organizational capital is closely connected to, and inspired by, the
resource-based view (Sadowski and Ludewig (2004); Schneider (2008)). In short, this
view states that strategic resources that generate a lasting competitive advantage have
to be scarce, hard to imitate, and hard to replace (Barney, Wright, and Ketchen (2001);
Knyphausen (1993)). ese conditions are fulﬁlled by many resources, especially those
that are intangible, such as human capital and social capital. Our interpretation of orga-
nizing as a company as a resource is related to the Dynamic Capabilities View, in which
organizational routines are the decisive resource that enables ﬁrms to survive in changed
environments better than their competitors (Teece, Pisano, and Shuen (1997)).
Organizational capital is as intangible as human and social capital. Organizational routines
are usually stable for intermediate time periods. Despite this stability, organizational capital
is exposed to the risk of becoming obsolete due to imitation or innovation of competi-
tors (Lev and Radhakrishnan (2005)). Only the part of organizational capital that cannot
be imitated can generate a sustainable competitive advantage. Organizational capital is
idiosyncratic and cannot be traded, unless the whole organization itself is sold (Black and
Lynch (2005)). erefore, there is no market price on organizational capital (Ramirez and
Hachiya (2006b)). Attempts to identify organizational capital must be closely linked to
its eﬀects on proﬁt, added value, or other performance measures.
We ask empirically what the importance is of the organizational capital of business units,
and which organizational practices generate a relatively lasting competitive advantage.
2 Meas ur e Me nt pr Obl eMs
We believe that organizational capital can explain why ﬁrms with identical resource
endowments diﬀer in performance. Measuring the extent to which the performance of
2 Examples are obligations and favors that can be called on or information that can be obtained through these re-
O. LUDEWIG/D. SADOWSKI
SBR 61 October 2009 393412
establishments depends on their ability to organize their activities is obviously of both
practical and theoretical interest. We want to know in which way the diﬀerent practices
contribute to the formation of organizational capital.
Accountants measure physical capital by its purchase costs minus depreciation. is
approach is not feasible for the measurement of organizational capital, because the acqui-
sition costs are not known. As noted above, it cannot be traded; therefore, it has no
market price (Black and Lynch (2005); Ramirez and Hachiya (2006b)). To a great extent,
the costs of organizational capital formation are opportunity costs, thus they are indirect
and hidden, and it is not possible to directly assess organizational capital by using acqui-
sition costs. But indirect approaches are also problematic. For example, subtracting the
book value of all its other assets from a ﬁrm’s market value isolates the value of organiza-
tional capital. However, this “goodwill” comprises not only the organizational capital, but
also all assets not accounted for in the balance sheet, including human and social capital
ere is an additional obstacle to measuring the stock of organizational capital held
by any particular ﬁrm. It is diﬃcult to decompose the organization’s performance at a
particular point in time into the diﬀerent contributions of the corresponding assets, such
as real capital, labor, human capital, social capital, and organizational capital (Lev and
Instead, we measure the value of organizational capital by comparing the performance of
establishments with diﬀerent organizational practices. We can either observe the same ﬁrm
across diﬀerent points in time, or compare diﬀerent organizations at the same point in time.
If they diﬀer only in their organizational practices, then we can attribute the diﬀerences in
ﬁrm performance to their organizational diﬀerences. Such an indirect approach has impor-
tant consequences. First, instead of obtaining information about the total value of orga-
nizational capital, we will only know the diﬀerences in value caused by alternative sets of
practices. Second, because it is very unlikely that there are two identical ﬁrms that vary in
only a few organizational practices, econometric methods must control for other factors.
ere are many studies that apply such econometric approaches to determine the eﬀects
of diverse organizational practices on performance. Only a few of these studies use the
concept of organizational capital. Others build on the ideas of “high performance work
practices” or “human resource management systems”. Although we could interpret their
results as measuring organizational capital, they fail to disentangle the eﬀects of diﬀerent
practices, because they use (additive) indices or dummy variables that indicate similar
organizational systems3. us, these studies assume that all practices contribute in the
same way to performance (Ludewig (2006))4. Further, they focus on human resources,
and do not account for other sources of organizational capital.
3 Examples of this type of Literature are MacDuffie (1995); Huselid, Jackson, and Schuler (1997); Ichniowski,
Shaw, and Prennushi (1997); Ludewig (2001).
4 These points are only shortcomings from the organizational capital perspective. However, the respective authors
have a human resource management view.
SBR 61 October 2009 393412 397
Atkeson and Kehoe (2005) use macro data to calculate the payments received by organi-
zational capital owners. ese payments are deﬁned as the fraction of output that is not
accounted for by payments to labor, (physical) capital, and managers. ey estimate that
in the U.S.-manufacturing industry 4% of all payments go to the owners of organiza-
tional capital. According to their calculations this is more than 1/3 of the payments going
to owners of real capital.
Tomer (1981; 1987) suggests an approach similar to growth accounting models. A number
of recent papers (e.g. Lev and Radhakrishnan (2003); Ramirez and Hachiya (2006a;
2006b)) built on this idea. ey identify the (idiosyncratic) contributions of organiza
tional capital by analyzing the residuals of production function estimations (Schneider
(2008) provides an interesting overview of these eﬀorts). However, these contributions
do not investigate which organizational practices explain the idiosyncratic performance
diﬀerentials, in other words, which practices contribute to organizational capital (Bres-
3 Meas ur e Me nt O f Org ani zati Ona l Cap ital
We identify the establishment-speciﬁc performance diﬀerentials applying a production
function approach as well. We examine the plant or establishment level, because this level
is the locus of value-generating processes and decision making. In large corporations and
businesses, the diﬀerent divisions have substantial discretionary leeway (Bartel (2004);
Schmitt (2002)). Bartel (2004) postulates that the performance measurement of organiza-
tional practices “[...] can only be done through detailed analysis at the plant level, [...].”
We base our measurement method on the time structure of panel data. Such data makes
it possible to control and identify the “unobserved heterogeneity of single observations”,
i.e., the unobservable (or at least unobserved) establishment-speciﬁc characteristics that
might have a causal relation with the performance and that might be correlated with
other important variables. In conventional applications, such observation-speciﬁc eﬀects
would bias the estimates if uncontrolled. ere are various techniques to implement such
However, our prime goal is not to control the establishment-speciﬁc eﬀects, but to identify
and measure them. erefore, our ﬁrst step is to estimate the establishment-speciﬁc eﬀect
on performance. We do so by applying a within (ﬁxed eﬀect) estimator that bases the esti-
mation on diﬀerences between each variable and its average over time for each observa-
tion (Greene (2003)). We extract the ﬁxed eﬀect from the estimates.
Within a production function framework, this establishment-speciﬁc time ﬁxed eﬀect
reﬂects time-invariant output diﬀerentials between establishments. is establishment-
speciﬁc performance diﬀerential is in our interpretation and that of others (e.g., Lev and
Radhakrishnan (2003); Ramirez and Hachiya (2006a; 2006b)) generated to a substantial
extent by organizational capital.
O. LUDEWIG/D. SADOWSKI
SBR 61 October 2009 393412
In our second step we explain the establishment-speciﬁc contribution of organizational
capital to performance through organizational and personnel practices. Establishment-
speciﬁc output diﬀerentials serve as dependent variables, making it possible for us to
derive the contribution of diﬀerent practices to organizational capital. Finally, we combine
the estimated model with establishment-speciﬁc variable values to predict individual estab-
lishment performance. We do not interpret the whole ﬁxed eﬀect as organizational capi-
tal’s impact on performance, because the establishment-speciﬁc part of the residual might
contain some unmeasured impact on performance (Bresnahan (2005)). Instead, we deﬁne
only that portion of the ﬁxed eﬀect as organizational capital that can be explained by orga-
We augment the residual based approach in two ways. First, we control for the intangible
assets human capital and social capital. Second, we apply a two-step procedure in which
we decompose the establishment speciﬁc performance diﬀerential into the contributions
of organizational practices that constitute organizational capital.
4 est iMat iOn MOdel
In the ﬁrst step of our two-step method, we assume a Cobb-Douglas production function,
which we amend with a term for organizational capital (Ω):
Qi,t = AΩ
β3 . (1)
Q denotes the output, K the capital stock and L the labor input of establishment i in
period t. e constant A represents overall eﬃciency.
To obtain a linear expression and to account for random measurement errors and
stochastic shocks, we take the natural logarithm and we add a random error εi,t:
lnQi,t = lnA + β1 lnΩi,t + β2 lnKi,t + β3 lnLi,t + εi,t. (2)
However, Ω is unknown. us, we must reformulate the model so as to approximate the
inﬂuence of an idiosyncratic and intangible resource on the outcome. Since we are only
interested in the idiosyncratic component of organizational capital, it is suﬃcient to derive
the establishment-speciﬁc variation of performance that is not explained by the remaining
factor endowment. Using a ﬁxed eﬀect approach similar to Lev and Radhakrishnan (2003)
or Ramirez and Hachiya (2006a; 2006b) yields:
lnQi,t = lnA + νi + β2 lnKi,t + β3 lnLi,t + εi,t, (3)
lnΩ are the establishment-speciﬁc ﬁxed eﬀects. It gives the organizational
capital’s impact on Q. is positive (negative) impact shifts the base production function
AKβ 2Lβ3 outward (inward). As noted, we augment the basic function by using variables
for human capital and social capital.
SBR 61 October 2009 393412 399
e estimation procedure imposes the standard distributional assumptions of the esti-
mator on the ﬁxed eﬀect. erefore the average and median are around zero and there
are rather extreme values at both ends of the distribution. is distributional eﬀect has
two important implications. First, that we must not interpret the absolute values derived
for establishments and their signs. e zero point is arbitrarily imposed by the estima-
tion procedure. e estimates are interval measures. Second, that we must account for the
distributional features in the second step estimation.
In our second step we use ν
as the dependent variable and regress it on organizational
practices, here summarized in the vector OrgVar:
νi = α0 + α1 OrgVari,t + µi,t. (4)
Equation (4) implies that time-variant variables explain the time-invariant variable νi.
Although we assume that organizational capital is to some extent time invariant, we do
not believe that it constant during the whole observation period, 1997-2005. It seems to
be more plausible to assume a stable organizational capital only for an intermediate time
interval (e.g., as do Lev and Radhakrishnan (2003)). Additionally, some of our core vari-
ables cover periods of two years. Due to the time structure of these variables, their possible
lagged eﬀect on performance, and the expected intermediate durability of organizational
capital, we divide the total observation period into overlapping intervals of three years. We
apply the ﬁrst step of our analysis to each of these intervals. We then assign the resulting
ﬁxed eﬀects for each interval to the respective establishments for the year in the middle of
the interval under consideration. Doing so yields equation (5):
νi,[t – 1, t + 1] = α0 + α1 OrgVari,t + µi,t. (5)
We extract the establishment-speciﬁc performance diﬀerential for the year 2000 from
the ﬁxed eﬀects model that we apply to the interval 1999-2001. We obtain the impact
of organizational capital for the year 1998 from the ﬁrst-step estimation of the inter-
vals 1997-1999, and so on. However, the data set covers several important variables that
aﬀect organizational and personnel practices only in irregular intervals, so we cannot use
every possible three-year interval. Table 1 shows the four intervals that we can use for the
Table 1: Interval structure
First step interval of three years [t – 1, t + 1] Year assigned
1997, 1998, 1999 1998
1999, 2000, 2001 2000
2000, 2001, 2002 2001
2003, 2004, 2005 2004
O. LUDEWIG/D. SADOWSKI
SBR 61 October 2009 393412
We use the IAB Establishment Panel (IAB-Betriebspanel; see Bellmann (1997) and Kölling
(2000) for details) as our database. is large-scale, general-purpose survey, which was
collected by the Institute for Employment Research (IAB), is a stratiﬁed random sample.
It is a longitudinal data set that provides information for annual waves from 1993 onward
for West Germany, and from 1996 for East Germany. It is primarily interviewer based,
but supplemented in some regions by mailed surveys that comprise identical question-
naires containing a broad range of variables regarded as important in economic theory.
e sample includes establishments companies of all sizes. ese basic structural elements
correspond to some of the principles of an ideal set of panel data suggested by Hamer-
mesh (1993). An establishment as it is comprised in the panel might be identical to the
entire ﬁrm or it might be the local division of a corporation.
Starting in 1993 with 4,300 units, the sample size of the survey was extended in several
steps and now covers almost 16,000 establishments. e sample ﬁrms are drawn from the
so-called Betriebsdatei of the Federal Labor Oﬃce. is database contains the base popu-
lation, which consists of all establishments with at least one employee who is covered by
the compulsory social security system. Over 80% of German establishments fulﬁll this
condition. e stratiﬁcation of the sample implies a sampling probability that is increasing
with establishment size. Industry is the second sampling criterion.
Since the survey is supported by the German employers’ association and Federal Employ-
ment Agency (Bundesagentur für Arbeit), there is a response rate of around 70% for initial
contacts and about 80% for repeated contacts. e data provide general information on
the companies, such as organizational practices, total sales, employment, or the indus-
trial relations within the establishment. e IAB panel is unbalanced because of panel
mortality, the replacement of closed or nonresponding establishments and a more or less
continuous increase in sample size5.
e available data cover the period from 1993-2006. Due to some data restrictions, we
could use only the waves 1997 to 2005 for our ﬁrst-step estimates. However, we also inte-
grate information from the 1995 and 2006 surveys, because some variables are deﬁned
for previous as well as following years.
Because of nonresponse, panel mortality, and missing variable values, some ﬁrms generate
few observations, even over the whole observation period. We analyze only observations
that are not missing any values for the variables we use in estimation. We use only the
manufacturing sector. We exclude establishments with less than ﬁve employees, because
at this threshold legal requirements might inﬂuence organizational decisions. We use only
units with at least two observations. We end up with between 278 and 537 establish-
ments per three-year period, which gives us about 570 to 1,210 observations per obser-
5 Hartmann and Kohaut (2000) show that the panel mortality, which contributes substantially to the unbalanced
nature of the panel, is not systematically correlated with corporate characteristics.
SBR 61 October 2009 393412 401
e panel has a core of questions that are repeated every year. ese questions address
general business development and strategy as well as personnel policies. Other issues are
raised biannually, triannually, or only once. By using the collected information, we can
check the impact of several organizational practices that might constitute organizational
capital, even though the set of practices is certainly not exhaustive.
First estimation step: dependent variable
Diﬀerent studies that analyze the productive impact of organizational practices use several
measures to operationalize performance. e most prominent measure, the market value
of the ﬁrm, is in our view not feasible. First, given the current worldwide economic crisis,
it is questionable if ﬁnancial markets are really providing correct estimates of ﬁrm values
(Bresnahan (2005)). Second, and even more important there are many ﬁrms that are not
listed in share markets. is observation is especially true in Germany (Vitols (2004))
Value added is available for most ﬁrms and it is strongly linked to the real value of a busi-
ness. erefore, we use value added as the dependent variable. Taking logarithms leads to
the exclusion of 27 observations.
First estimation step: explanatory and control variables
e IAB Establishment Panel does not directly gather information on the physical capital
stock. We approximate it by summing up the investment of the current and previous
year7. Investment is in many cases zero. us, we add one before taking the logarithm of
this variable to avoid drop outs. We approximate labor input by the logarithm of the total
number of employees.
We introduce proxy variables for the intangible human and social capital into our spec-
iﬁcation of the production function. ese variables are designed to indicate the impor-
tance of these intangible assets relative to the industry average. We derive a skill ration as
approximation of human capital. We calculate the average share of skilled employees on
a two-digit industry level (NACE equivalent) and divide the share of each ﬁrm by this
average. e resulting “skill ratio” gives us the industry-adjusted human capital intensity
of each company.
As discussed we include a proxy variable for social capital. We assume that a business that
relies on social capital must keep this capital within the establishment, so the ﬁrm must
aim at low quit rates. Based on this assumption, we put the relative rate of voluntary quits
into the function to control for the productive impact of social capital.
We also include the share of part-time employees in the production function. By doing
so, we can control for the eﬀects of labor input variation due to diﬀerences in the number
of monthly working hours of employees. We must rely on this crude proxy because exact
working time of part-time employees is not collected in all waves.
6 Although there were about 2,445 public corporations in 1996, these were only 0.09% of all firms subject to VAT
(DAI (1999)). Additionally, only 802 of those firms were listed on a stock exchange (DAI (1999)). In 2001 there
were about 1,075 firms listed, with a total value of €1,200 billion, which corresponds to 60% of GDP. In the
U.S. and the UK, this relation was well above 100% (Vitols (2004)).
7 See for example Möller (2007), Bellmann and Büchel (2001), or Bellmann, Bender, and Schank (1999).
O. LUDEWIG/D. SADOWSKI
SBR 61 October 2009 393412
Second estimation step: dependent variable
We estimate the production function by using a ﬁxed eﬀects model for each of the inter-
vals given in Table 1. We extract the performance diﬀerential of each establishment by
predicting the added value of each unit by using the estimation results twice, ﬁrst, by
making this prediction with ﬁxed eﬀects, then without. After taking the antilogarithm,
the establishment-speciﬁc performance diﬀerential is then the diﬀerence between these
two values. We calculate per capita values by dividing the performance diﬀerentials by
the number of employees (full-time equivalents). We perform this calculation for each
interval and assign the resulting values to the respective ﬁrms as described above. ese
ﬁxed eﬀects are centered, due to the distribution imposed on the residual. e ﬁxed eﬀects
represent the establishment-speciﬁc diﬀerences in value added per employee for a given
factor endowment. We analyze these diﬀerentials in the second step.
Second estimation step: explanatory and control variables
As we explained above, we propose that a substantial part of these establishment-speciﬁc
performance diﬀerentials is due to the diﬀerences in the ﬁrm’s ability to organize the
production process eﬃciently. It is this part of the ﬁxed eﬀect that we deﬁne as organiza-
tional capital. In our second step we estimate the contribution of speciﬁc organizational
practices to the performance diﬀerential.
e wage structure is one of the major inﬂuences on employee behavior. According to
the eﬃciency wage literature, a wage premium can improve productivity by, for example,
reducing shirking and increasing commitment8. e eﬃciency wage mechanism depends
on wages that are above the average market wages. We construct an industry-adjusted wage
ratio by dividing the average per capita wages at the ﬁrm level by the average industry per
capita wage. However, some researchers argue that high wages that are aimed at increasing
external motivation crowd out intrinsic motivation9. Consequently, the positive eﬀect of
the wage premium may be oﬀset by a negative eﬀect, and the net eﬀect is unknown.
In the German regulatory setting, the impact of works councils is ambiguous. e existence
of a works council can reﬂect a policy aimed at employee involvement which is expected
to increase performance. In contrast to other participatory practices (e.g., teamwork or
open-door policies) the works council is not under the discretion of management. e
existence of this institution depends on the willingness and eﬀort of employees to establish
one (Addison et al. (2000)). If there is the willingness and the legally required minimum
number of ﬁve employees, workers have the right to establish a works council. Although
a works council can hardly be regarded as a managerial practice, managers can either
encourage or discourage its formation. Even if a council is formed despite management
resistance, it can support more cooperative candidates. To determine their empirical rele-
vance, we treat works councils as an organizational practice despite its ambiguous nature.
8 See for example Akerlof (1982); Shapiro and Stiglitz (1984); Groshen (1994); Stiglitz (1987).
9 See for example Osterloh and Frey (2000), Osterloh, Frey, and Frost (2001), Frey and Jegen (2001), Akerlof and
Kranton (2005; 2008), or Prendergast (2008).
SBR 61 October 2009 393412 403
One section of the IAB questionnaire examines ﬁrms’ organizational practices. A set of
dummy variables indicates the implementation of certain practices either at a speciﬁc
point in time or up to two years before. By using these variables we can create dummies
to distinguish whether or not these practices are implemented in the respective establish-
ment during the observation period. We note the presence or absence of the following
practices: quality control systems and quality circles (“quality control”), proﬁt or cost
centers (“proﬁt center”), teamwork (“teamwork”), delegation of decision-making power to
line workers (“empowerment”), and increased supply by external vendors or outsourcing
(“outsourcing”). us we cover a broad set of practices ranging from HRM (“empower-
ment” and “teamwork”) over controlling (“proﬁt center”) to quality management covering
many dimensions of organizational capital.
We noted above that self-binding rules, such as job guarantees, can also enhance labor
productivity. erefore, we include the rate of layoﬀs in our estimation function. Again,
we relate the establishment level to the industry by dividing the establishment layoﬀ rate
by the industry average. However, this variable might suﬀer from a problem of reversed
causality. It is unclear whether establishments have a better performance due to job guar-
antees or if they lay oﬀ fewer employees because of better performance. To avoid this
problem, we also include lagged values of this variable in our estimation function. Past
layoﬀs should not depend on current performance; however, a low number of layoﬀs
in the past still reﬂects a policy of job guarantees
. However, in economic crises layoﬀs
might reduce costs and raise net return in future periods. Following Schneider (2008),
we account for a U-shaped relation by inserting the (lagged) squared values of the relative
layoﬀ rate as control variables into the estimation function.
Although information and communication technology (IT) is physical capital, it can have
a substantial impact on organizational structures and the outcome of organizational prac-
tices11. us, the eﬃcient utilization of IT equipment can be a major source of organiza-
tional capital. In the second-step equation, we use a dummy that shows the inﬂuence of
IT investments in a given period12.
10 One might suspect that employer-generated layoffs and voluntary employee quits might be correlated. The cor-
relation coefficient for both variables is (highly) significant but very small, with values around 0.05. Thus, we
can ignore the co-movements of both variables.
11 See for example Black and Lynch (2005), Lev and Radhakrishnan (2005), or Brynjolfsson, Hitt, and Yang
12 Schneider (2008) is even using the presence of CNC-equipment in a company as a major indicator for organi-
O. LUDEWIG/D. SADOWSKI
SBR 61 October 2009 393412
We add 19 industry dummies
, which control for industry-speciﬁc circumstances such
as market structure, industrial relations, market concentration, or demand conditions
ree time dummies control time ﬁxed eﬀects. Exposure to the world market is covered
by the export rate.
6 res ults
We present our ﬁrst-step results in the appendix. e conventional statistical tests for
ﬁxed eﬀects models are satisfactory. e ﬁxed eﬀects of all within-estimations are jointly
diﬀerent from zero at high signiﬁcance levels. Table 2 shows their distribution. For inter-
pretation purposes an artiﬁcial zero point is generated by subtracting the minimum (here:
-300,769). e ﬁrm-speciﬁc performance diﬀerential has a range of almost a million
euros per employee. is high range is to some extent due to the residual analysis. We
note especially the high overall range and the long tails of the distribution result from the
econometric construction of the ﬁxed eﬀects. When we compare only the 5
percentiles, the range shrinks to about €215,000. e interquartile range is about €66,800
per employee. ere are substantial diﬀerences in the productivity of establishments even
after we control for four types of input.
13 The industy classification is not taken from the Establishment Panel but from its extension the so called LIAB
(http://fdz.iab.de/en/Integrated_Establishment_and_Individual_Data.aspx (2008-11-04)). The LIAB is the em-
ployer-employee panel of the IAB. However, we only use the industry information on establishment level. The
LIAB classification is more differentiated than the original establishment-panel variable. In both classifications
there are two structural breaks due to general overhauls of these classifications. We try to compensate for these
14 The industry dummies and the effects they control do not change very much over time. Thus, the dummies
would be eliminated by the within-estimator. These variables are sometimes described as quasi fixed, because
they can change over time but do so very rarely (e.g., Zwick (2004)). Hence, they should be included in the sec-
ond step. Given our data, we do not provide a detailed discussion of corporate parent and industry effects. A
recent account of this issue in the strategic management literature recommends “focussing toward the business
unit” (Misangyi et al. (2006, 587)), which we have done.
SBR 61 October 2009 393412 405
Table 2: Distribution of the establishment-specific performance differential
Percentile Centile 95% Conf. Interval With artificial
0 –300 769 –300 769 –300 769 0
5 –101 665 –115 767 –86 868 199 104
10 –69 677 –76 411 –63 491 231 092
15 –56 587 –61 646 –52 032 244 182
20 –44 049 –49 866 –37 575 256 720
25 –33 427 –37 441 –29 652 267 342
30 –25 407 –30 110 –22 435 275 362
35 –20 044 –22 801 –16 709 280 725
40 –14 028 –17 564 –10 786 286 741
45 –8 816 –11 678 –4 744 291 953
50 –2 713 –5 887 1 388 298 056
55 4 521 253 7 634 305 290
60 10 192 6 600 13 674 310 961
65 17 015 12 498 20 883 317 784
70 25 152 19 912 30 012 325 921
75 33 397 29 728 36 515 334 166
80 42 705 36 530 47 008 343 474
85 54 654 49 383 61 961 355 423
90 74 146 65 569 85 654 374 915
95 113 879 100 372 124 003 414 648
100 653 601 653 601 653 601 954 370
Obs.: 1092 Interquartile range 66 824
e second-step equation explains the variation of the ﬁrms’ speciﬁc productivity diﬀer-
ential by their organizational characteristics. We use two diﬀerent estimators.
e ﬁrst estimator accounts for the problem of heteroskedasticity. e second speciﬁca-
tion deals with the fact that the ﬁrst-step estimation produces some extreme values for
the establishment speciﬁc eﬀect. e heteroskedasticity-robust OLS estimator provides
corrected standard errors via a modiﬁed covariance matrix.
e outlier robust regression ﬁrst ﬁts the usual regression, then calculates Cook’s D and
excludes all observations with a D > 1 from further calculations. ereafter, we deter-
mine weights in an iterative procedure for which we use the absolute residuals. Finally,
we use these weights to estimate a weighted regression. is outlier-robust estimator is
O. LUDEWIG/D. SADOWSKI
SBR 61 October 2009 393412
especially well suited to deal with the long tails of the dependent variable’s distribution.
us, our interpretation rests mainly on those results, while the OLS estimator provides
a robustness check. Generally, the coeﬃcients of both estimates have the same sign and a
similar magnitude. However, the signiﬁcance levels vary. us, it seems plausible that at
least the direction of inﬂuence is stable.
Table 3: Results of the second step
OLS estimates with hetero-
skedasticity-robust standard errors
Coefficient t-value Coefficient t-value
Wage ratio 50 370*** 8.15 48 687*** 10.06
Works council 36 897*** 6.59 27 296*** 6.33
Empowerment 6 042 1.53 3 246 0.94
Team work –2 116 –0.60 –1 768 –0.56
Profit center 6 872* 2.03 7 644** 2.55
Quality control –2 163 –0.42 3 112 0.78
Outsourcing 1 947 0.62 2 692 0.94
Lag 0 52 0.02 –1 348 –0.76
Lag 1 –2 817*** –1.65 –2 885** –2.11
Lag 2 –3 890** –2.34 –3 043** –2.05
Fluctuation ratio squared
Lag 0 –143 –0.73 –38 –0.24
Lag 1 70 0.61 115 1.21
Lag 2 185** 2.28 178* 1.77
Information Technology 18 524*** 3.91 15 729*** 4.38
Export 315*** 4.17 351*** 5.81
Time dummies Three
Industry dummies 19
Constant –99 247*** –8.59 –101 225*** –12.1
Number of obs. 963 963
F-test F(37, 925) = 12.42*** F(37, 925) = 16.45***
Significance levels: * p < 0.1, ** p < 0.05, *** p < 0.01.
SBR 61 October 2009 393412 407
e positive, signiﬁcant eﬀect of the wage variable indicates that higher wages are accom-
panied by a higher productivity diﬀerential (Table 3 ). is ﬁnding is consistent with the
basic idea of eﬃciency wages. However, there are other explanations. For example, wages
usually rise with experience, but we are not able to control for this eﬀect.
e productivity of establishments with works councils is €27,000 per employee higher
than those of establishments without works council. However, as indicated above, it is a
matter of debate whether a works council can be regarded as a practice or not.
Proﬁt centers and investment in information and communication technology have a posi-
tive, signiﬁcant inﬂuence on the establishment-speciﬁc performance diﬀerential. Empower-
ment and outsourcing have positive but nonsigniﬁcant eﬀects. All these practices appear
to contribute to organizational capital.
e coeﬃcient pattern for the involuntary quit rate is more complex. For the ﬁrst and
second lags, the linear variable has signiﬁcant negative coeﬃcients. us, a higher quit rate
leads to less value added. Self-binding rules appear to generate economic value. However,
the second lag of the squared value of the layoﬀ ratio has a positive and (weakly) signiﬁ-
cant sign. us, in some instances, layoﬀs improve performance. is ﬁnding underlines
an important point: whether or not a speciﬁc practice is generating organizational capital
depends to a certain degree on the circumstances15.
In Table 4 we estimate the economic value of practices that enhance organizational capital.
We do so by inserting the variable values for each ﬁrm into the estimated function. We
omit variables with nonsigniﬁcant coeﬃcients and control variables, because they do not
contribute to organizational capital.
For a robustness check we also estimate the part of the performance diﬀerential that is
explained by all practices, including the nonsigniﬁcant coeﬃcients. e diﬀerence between
the two approaches is small. We use the coeﬃcient of the zero lag ﬂuctuation ratio for
the estimates even though its coeﬃcient is nonsigniﬁcant. We believe that in this case, it
makes more sense to use the whole set to reﬂect the time structure of the ﬂuctuation vari-
able. However, the squared values are only control variables and thus they are not part of
the estimates provided in Table 4, regardless of their signiﬁcance level.
15 Similar results were found by Schneider (2008).
O. LUDEWIG/D. SADOWSKI
SBR 61 October 2009 393412
Table 4: Estimated contribution of organizational practices to the establishment-
specific performance differentials
Percentile Centile 95% Conf. Interval Artifical zero point
0 –1 191 –1 191 –1 191 0
5 42 543 38 781 46 177 43 734
10 56 896 51 350 59 429 58 087
15 63 795 60 450 67 018 64 986
20 70 410 67 182 73 396 71 601
25 75 647 72 977 77 823 76 838
30 79 913 77 361 82 371 81 104
35 84 010 81 786 86 229 85 201
40 87 456 85 411 89 315 88 647
45 90 242 88 469 91 741 91 433
50 92 687 91 010 94 159 93 878
55 94 980 93 563 96 161 96 171
60 96 747 95 629 98 256 97 938
65 98 728 97 766 100 319 99 919
70 100 829 100 025 102 164 102 020
75 103 274 101 870 104 418 104 465
80 105 683 104 096 107 281 106 874
85 109 738 107 407 111 611 110 929
90 114 520 112 231 117 704 115 711
95 123 981 121 317 126 499 125 172
100 156 450 156 450 156 450 157 641
Obs.: 936 Interquantile range: 27 627
Table 4 gives the distribution of the resulting estimates and the artiﬁcial zero point. e
portion of the range of the ﬁrm-speciﬁc performance diﬀerential that can be explained
by the practices constituting organizational capital, including the works council variable,
is about €157,600 per employee. e interquartile range is €27,600 per employee. is
value is considerable. In our data set the average gross wage sum per employee is about
€26,400, and the average added value is €60,000.
7 COn Cl us i On
n this paper we discuss the theoretical concept of organizational capital and empirically iden-
tify its productive impact and its value. We deﬁne organizational capital as sets of organiza-
tional practices, processes, and designs that make it possible for companies to extract a higher
level of returns from a given resource endowment. Such organizational capital is intangible,
not tradable, and idiosyncratic. Due to these characteristics, it is hard
SBR 61 October 2009 393412 409
We measure organizational capital, or, more precisely, its impact on establishment
performance, by using establishment level data and ﬁxed eﬀects models in a two-step
procedure. Retrieving the ﬁxed eﬀects in the ﬁrst step provides establishment-speciﬁc
performance diﬀerentials. ese diﬀerentials are quite large. In our view, some portion
of these diﬀerentials is due to the diﬀerent and mostly idiosyncratic ways to organize
To determine the share of the performance diﬀerential that is attributable to organiza-
tional practices, we regress the variation of the ﬁxed eﬀects on organizational practices. If
these practices contribute signiﬁcantly to the idiosyncratic performance diﬀerential, then
they contribute to organizational capital. Such practices include wage premia, self-binding
rules, proﬁt centers, and IT - investments. Using the estimates for the contribution of
these practices, we derive the share of the performance diﬀerential that can be attributed
to their use. We ﬁnd that this share is substantial. In other words, organizational capital
can generate substantial value.
app en d ix
Table 5: Results of the relevant first step estimates
1997-1999 1999-2001 2000-2002 2003-2005
Number of obs. 571 670 869 1210
Number of groups/
Obs. per group 278
F-test F(8, 285) = 8.59 F(8, 327) = 2.27 F(8, 372) = 3.29 F(8, 665) = 5.34
Prob > F: 0.000 Prob > F: 0.022 Prob > F: 0.001 Prob > F: 0.000
R20.19 0.05 0.066 0.060
Coefficient t-value Coefficient t-value Coefficient t-value Coefficient t-value
lnEmployment 1.113 7.10 0.800 3.54 0.678 3.76 0.588 4.13
Skilled ratio –0.086 –1.97 –0.018 –0.62 –0.016 –0.59 0.037 1.75
lnCapital 0.119 3.18 –0.024 –1.37 –0.029 –2.25 0.024 2.06
Lag 1 –0.019 –0.96 0.015 0.70 0.040 1.69 0.018 1.87
Lag 2 –0.010 –0.50 –0.012 –0.62 –0.009 –0.37 0.016 1.47
Share of part time 0.002 0.47 –0.008 –1.50 –0.003 –0.46 –0.009 –2.86
Constant 8.681 8.33 12.630 9.95 13.223 13.08 12.920 17.00
O. LUDEWIG/D. SADOWSKI
SBR 61 October 2009 393412
ref er e nC es
Addison, John T., W. Stanley Siebert, Joachim Wagner, and Xiangdong Wei (2000), Worker Participation and Firm
Performance: Evidence from Germany and Britain, British Journal of Industrial Relations 38, 7-48.
Akerlof, George A. (1982), Labor contracts as partial gift exchange, Quarterly Journal of Economics 97, 543-569.
Akerlof, George A. and Rachel Kranton (2008), Identity, Supervision, and Work Groups, American Economic Review
Akerlof, George A. and Rachel Kranton (2005), Identity and the Economics of Organizations, Journal of Economic
Perspectives 19, 9-32.
Atkeson, Andrew and Patrick J. Kehoe (2005), Modelling and Measuring Organizational Capital, Journal of Political
Economy 113, 1026-1053.
Barney, Jay, Mike Wright, and David J. Ketchen (2001), The Resource Based View of the Firm: Ten Years After 1991,
Journal of Management 27, 625-641.
Bartel, Ann P. (2004), Human Resource Management and Organizational Performance: Evidence from Retail Bank-
ing, Industrial and Labor Relations Review 57, 181-203.
Bellmann, Lutz (1997), Das Betriebspanel des IAB, in: Reinhard Hujer, Ulrich Rendtel, and Gert Wagner (eds.),
Wirtschafts- und Sozialwissenschaftliche Panel-Studien, Göttingen: Vandenhoeck und Ruprecht, 169-182.
Bellmann, Lutz, Stefan Bender, and Thorsten Schank (1999), Flexibilität der Qualifikationsstruktur aus betrieblicher
Sicht: Substitutionalität oder Komplementarität, Jahrbücher für Nationalökonomie und Statistik 219, 109-126.
Bellmann, Lutz and Felix Büchel (2001), Betrieblich finanzierte Weiterbildung und Unternehmenserfolg. Eine Ana-
lyse für West- und Ostdeutschland unter besonderer Berücksichtigung von Selektionseffekten, in: Uschi Backes-
Gellner and Petra Moog (eds.), Bildungssystem und betriebliche Beschäftigungsstrategien. Beiträge der Jubiläums-
tagung 25 Jahre Bildungsökonomischer Ausschuß, Schriften des Vereins für Socialpolitik Bd. 279, Berlin: Duncker
und Humblot, 75-92.
Black, Sandra E. and Lisa M. Lynch (2005), Measuring Organisational Capital in the New Economy, in: Carol
Corrado, John Haltiwanger, and Daniel Sichel (eds.), Measuring Capital in the New Economy, Chicago et al.:
University of Chicago Press, 205-236.
Bounfour, Ahmed (2009), IT, organizational capital and the reporting (measurement) issue, in: Ahmed Bounfour (ed.),
Organisational Capital – Modelling, measuring and contextualizing, London, New York: Routledge, 268-288.
Bresnahan, Timothy F. (2005), Comment, in: Carol Corrado, John Haltiwanger, and Daniel Sichel (eds.), Measuring
Capital in the New Economy, Chicago et al.: University of Chicago Press, 99-110.
Brynjolfsson, Erik, Lorin M. Hitt, and Shinkyu Yang (2002), Intangible Assets: Computers and Organizational
Capital, Brookings Papers on Economic Activity, 137-198.
Coleman, James S. (1988), Social Capital in the Creation of Human Capital, American Journal of Sociology 94,
DAI – Deutsches Aktieninstitut (1999), DAI-Factbook 1999. Statistiken, Analysen und Graphiken zu Aktionären,
Aktiengesellschaften und Börsen, Frankfurt am Main.
Frey, Bruno and Reto Jegen (2001), Motivation Crowding Theory, Journal of Economic Surveys 15, 589-611.
Greene, William H. (2003), Econometric Analysis, 5th Edition, Upper Saddle River et al.: Pearson Prentice Hall.
Groshen, Erica L. (1994), Five Reasons Why Wages Vary Among Employers, Industrial Relations 30, 350-381.
Hamermesh, Daniel S. (1993), Labor Demand, Princeton et al.: Princeton University Press.
Hardin, Russel (1999), Social Capital, in: James Alt and Margaret Levi (eds.), Competition and Cooperation: Conversa-
tions with Nobelists about Economics and Political Science, New York: Russell Sage Foundation, 170-189.
Hartmann, Josef and Susanne Kohaut (2000), Analysen zu Ausfällen (Unit-Nonresponse) im IAB-Betriebspanel,
Mitteilungen aus der Arbeitsmarkt- und Berufsforschung 33, 609-618.
SBR 61 October 2009 393412 411
Huselid, Mark A., Susan E. Jackson, and Randall S. Schuler (1997), Technical and Strategic Human Resource Man-
agement Effectiveness as Determinants of Firm Performance, Academy of Management Journal 40, 171-188.
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 87, 291-313.
Knyphausen, Dodo zu (1993), “Why are Firms Different?” Der „Ressourcenbasierte Ansatz“ im Mittelpunkt einer
aktuellen Kontroverse im Strategischen Management, Die Betriebswirtschaft 53, 771-792.
Kölling, Arnd (2000), The IAB Establishment Panel, Schmollers Jahrbuch, Journal of Applied Social Science Studies
Lev, Baruch and Suresh Radhakrishnan (2005), The Valuation of Organisation Capital, in: Carol Corrado, John
Haltiwanger, and Daniel Sichel (eds.), Measuring Capital in the New Economy, Chicago et al.: University of
Chicago Press, 73-99.
Lev, Baruch and Suresh Radhakrishnan (2003), The Measurement of Firm-specific Organization Capital, NBER
Working Paper No. 9581, Cambridge, NBER.
Ludewig, Oliver (2006), Komplementäre Organisationspraktiken und strategisches Organisationsdesign – Mythen und
Fakten, München et al.: Hampp.
Ludewig, Oliver (2001), Personalpolitische Systeme in Deutschland, in: Hugo Kossbiel (ed.), Modellgestützte Personal-
entscheidungen 5, München et al.: Hampp, 9-26.
MacDuffie, John P. (1995), Human Resource Bundles and Manufacturing Performance: Organizational Logic and
Flexible Production Systems in the World Auto Industry, Industrial and Labor Relations Review 48, 197-221.
Matiaske, Wenzel (1999), Soziales Kapital in Organisationen. Eine tauschtheoretische Studie, München et al.: Hampp.
Misangyi, Vilmos F., Heather Elms, Thomas Greckhammer, and Jeffrey A. Lepine (2006), A New Perspective on a
Fundamental Debate: A Multilevel Approach to Industry, Corporate, and Business Unit Effects, Strategic Manage-
ment Journal 27, 571-590.
Möller, Iris (2007), Organisation, Personalpolitik und betriebliche Performance, München et al.: Hampp.
Osterloh, Margit and Bruno S. Frey (2000), Motivation, Knowledge Transfer, and Organizational Forms, Organiza-
tion Science 11, 538-550.
Osterloh, Margit, Bruno S. Frey, and Jetta Frost (2001), Managing Motivation, Organization and Governance, Journal
of Management and Governance 5, 231-239.
Prendergast, Canice (2008), Intrinsic Motivation and Incentives, American Economic Review 98, 201-205.
Prescott, Edward C. and Michael Visscher (1980), Organization capital, Journal of Political Economy 88, 446-461.
Ramirez, Pablo G. and Toyohiko Hachiya (2006a), Measuring Firm-Specific Organizational Capital and its Impact on
Value and Productivity: Evidence from Japan, Review of Pacific Basin Financial Markets and Policies 9, 549-573.
Ramirez, Pablo G. and Toyohiko Hachiya (2006b), How Do Firm-Specific Organizational Capital and other Intan-
gibles Affect Sales, Value and Productivity? Evidence From Japanese Firm-Level Data, International Journal of
Innovation and Technology Management 3, 265-282.
Sadowski, Dieter (2002), Personalökonomie und Arbeitspolitik, Stuttgart: Schaeffer Poeschel.
Sadowski, Dieter and Oliver Ludewig (2004), Organisationskapital, in: Georg Schreyögg and Axel von Werder (eds.),
Handwörterbuch Unternehmensführung und Organisation, 4th Edition, Stuttgart: Schaeffer Poeschel, 1017-1025.
Schauenberg, Bernd (1983), Arbeitsorganisation, Unternehmung und ökonomische Institutionen, Habilitations-
schrift, Universität Frankfurt.
Schmitt, Matthias (2002), Der grenzüberschreitende Transfer von Personalpraktiken innerhalb multinationaler Un-
ternehmen, München et al.: Hampp.
Schneider, Martin (2008), Organisationskapital und spezifisches Humankapital als strategische Ressourcen, Zeitschrift
für Personalforschung 22, 12-34.
Shapiro, Carl and Joseph E. Stiglitz (1984), Equilibrium Unemployment as a Worker Discipline Device, The Ameri-
can Economic Review 74, 433-444.
O. LUDEWIG/D. SADOWSKI
SBR 61 October 2009 393412
Stiglitz, Joseph E. (1987), The Causes and Consequences of the Dependence of Quality on Price, Journal of Economic
Literature 25, 1-48.
Teece, David J., Gary Pisano, and Amy Shuen (1997), Dynamic Capabilities and Strategic Management, Strategic
Management Journal 18, 509-533.
Tomer, John F. (1987), Organizational Capital. The Path to Higher Productivity and Well-being, New York et al.:
Tomer, John F. (1986), Productivity and Organisational Behavior: Where Human Capital Fails, in: Benjamin Gilad
and Staney Kaish (eds.), Handbook of Behavioral Economics, Vol. A, Greenwich (Connecticut) et al.: Jai Press Inc.,
Tomer, John F. (1981), Organizational Change, Organization Capital and Economic Growth, Eastern Economic
Journal 7, 1-14.
Vitols, Sigurt (2004), Changes in Germany’s Bank-Based Financial System: A Varieties of Capitalism Perspective.
WZB Discussion Paper SP II 2004-03, Berlin.
Zwick, Thomas (2004), Weiterbildung und betriebliche Produktivität, Zeitschrift für Betriebswirtschaft 74, 651-668.