Organizational Structures and Job

Rewards in Sweden
Carl le Grand,Ryszard Szulkin and Michael Tåhlin
Department of Sociology, Stockholm University
In this article, we examine how the organizational structure affects workers’ job autonomy
and earnings in Sweden. The analysis is based on survey data from a probability sample
of Swedish employees and their work organizations. Four aspects of the organizational
structure are studied: establishment size, hierarchical structure, firm internal labour
market arrangements, and formalization of work routines. Our results indicate that the
structural features of the organization have a considerable impact on job autonomy and
earnings. Some of the findings are in line with conventional hypotheses Other results,
however, are not as might be expected from the mainstream theoretical literature For
example, one central result is that tall and lean hierarchies seem to enhance, rather
than restrict, job autonomy Another finding is that while structural features of the
organization seem to affect job autonomy in the same way for both men and women,
the organizational structure has an impact on male, but not female, earnings

Carl le Grand, Ryszard Szulkin and Michael Tåhlin, Department of Sociology, Stockholm
University, S-10691, Stockholm, Sweden

1. Introduction the wider context of the organization (such

as the national labour market).
Inequality in the labour market is deter- This article can be seen as part of the
mined by two kinds of factors, individual
and structural. According to a traditional
growing literature which investigates the
human capital approach (Becker 1964;
impact of structural factors on stratification
and inequality in the labour market. Our
Mincer 1974), the variation in rewards can
be sufficiently explained with individual
specific aim is to examine the influence of
determinants. The focus is on heterogeneity
organizational characteristics on outcomes
located at the level of jobs and individuals.
within the labour force. Workers differ In particular, we analyse how job autonomy
from each other in ability, skills, and other and individual earnings depend upon the
individual properties that influence their
opportunities created inside organizations.
productive capacity, which in turn deter- The focus on autonomy and earnings is
mines the level of individual rewards. Such motivated, first, by their importance for the
a narrowly individualistic approach has welfare of workers, and, secondly, by their
been criticized for neglecting the impact
of structural factors, on stratification and
dependence (at least m theory) on the struc-
tural conditions prevailing in organizations.
inequality (Kalleberg & Sorensen 1979; Consider a resourceful organization with
Baron & Bielby 1980; Kalleberg & Berg a fairly stable work-force. According to the
1987). The main argument of the structural typical image of such an organization, it
perspective is that individuals with similar would have at least the following charac-
productive capacities are often differently teristics : (a) a large number of positions (or
rewarded, due to variations in the structural jobs); (b) the positions are hierarchically
setting within which they work. This setting structured; (c) the positions are filled
consists of three main features: the position
mainly through internal recruitment; and
(or job) which the individual occupies, the (d) dailv work routines are formalized to a
organization where the job is located, and high degree. In short, powerful and stable
organizations tend to exhibit bureaucratic
231
internal structures. Variations in these for individual employees. Although wages
structural traits across organizations con- in Sweden have traditionally been deter-
stitute opportunities or restrictions con- mmed according to a ’solidanstic wage
fronting individual workers. policy’, with the explicit purpose of mini-
We intend to analyse the organizational mizing variation in wages between similar
impact on job rewards by treating each of jobs across firms or sectors, wage nego-
the aforementioned characteristics sep- tiations have been decentralized in recent
arately. In other words, we examine the years. The national federation of employers
effects of size, hierarchy, internal labour strongly prefers wages to be determined at
market arrangements, and formalization on the firm level, while the attitudes among
job autonomy and individual earnings. As the trade unions have been mixed. During
will be developed at length below, the dif- the 1980s, there has also been a substantial
ferent organizational traits might very well local wage drift. It is clear that the scope
have counteracting effects on the outcome for interorganizational variation in earnings
variables we distinguish. has increased in Sweden during the past
The paper is organized as follows: In decade.
the next section we present our theoretical
framework in a more precise manner, and
descnbe the properties of organizations, 2. A causal model
jobs and individuals we are interested in. In Figure 1, a simplified causal model is
We then discuss the mechanisms by which
the four organizational properties might shown that visualize,, our understanding of
influence autonomy and earnings. Fol- the processes of reward determination in
the labour market. This model will serve
lowing a description of our data and van- as a guideline for the empirical analyses
ables, we carry out empirical analyses of
the relationships discussed in the theo- below. The sets of variables involved in our
retical part of the article. Finally, we sum- analyses are:

marize and evaluate our findings.

Before we turn to a formulation of (1) Individual resources brought into the

hypotheses, a few comments on the case of labour market. This set of exogenous vari-
Sweden are in order. We believe that many ables includes sex, education, and general
of the conclusions drawn on the basis of experience (number of years in the labour
Swedish data are likely to be applicable to force).’ Individual characteristics are used
other developed countries. The oppor- in our analysis mainly as control variables.
tunities created in specific work organ- Their importance for individual rewards is
izations may, however, be more important well known from previous studies.
in the Swedish labour market context com- (2) The structure of the work organization.
pared to for instance the US, as Swedish As will be discussed below, size, hierarchy,
workers tend to spend a very long time with firm internal labour market arrangements,
a single employer. The average seniority and formalization are structural properties
(number of years spent with the present of an organization which may be expected
employer) in Sweden is almost ten years. to affect both job autonomy and individual
Since the average number of years in the earnings.
labour force is 18 years, the typical Swedish (3)Autonomy, and othernon-pecuniary job
worker spends more than half of his work- characteristics. In addition to job auto-
ing life with his or her present employer (le nomy, this set of variables includes edu-
Grand 1993). Moreover, the great majority cational requirements for the job, general
of managers usually try to fill vacancies or specific skills acquired at the present job

internally, before listing them externally or firm, supervisory responsibilities, and

(ibid.). Hence, the Swedish labour market seniority.-’ Several of these characteristics
to a high degree consists of firm internal may also be regarded as non-pecuniary job
labour markets. The opportunities, or lack rewards.
of opportunities, offered within work (4) Monetary rewards, measured as hourly
organizations are thus of great importance earnings.
232
Figure 1. Causal relations between organizational structures and Job rewards.

The arrows in Figure1 indicate the causal rence & Lorsch 1969). Of particular import-
pattern assumed to tie these fourtypes of ance in the present case is that the
variables together. Individual resources are relationship between the organizational
expected to influence (arrow a) the type of structure and the character of jobs is not
work organization that the individual will simply of a one-way kind, but reciprocal.
join, i.e., his or her access to organizational In other words, although we believe that
opportunity structures, as well as (b) eligi- structural features of an organization have
bility to various kinds of jobs within an an impact on the degree of job autonomy,

organization. In addition, individual traits as indicated in Figure 1, we should also

might (c) have a direct impact on the level take into account that the organizational
of monetary rewards. The characteristics of structure to a certain extent may be
jobs are (d) a major source of variations in designed to deal with control problems
earnings. Finally, the organizational struc- emanating precisely from the autonomous
ture has two types of effects: it influences character of the organization’s jobs. We
(e) the content of jobs; and it has (f) a will return to this issue in the following
direct impact on earnings.;3 theoretical discussion and in the empirical
This casual pattern is straightforward and analyses.
consistent with the usual arguments devel-
oped by proponents of the structural 3. The
approach to stratification (Baron 198~1; organizational impact
&
Colbiornsen 1986; Kalleberg Berg 1987; on autonomy and earnings -
England & Farkas 1988). The major under- hypotheses
lying assumption is that organizations have
an impact on the content of jobs and on 3.1. Size
individual rewards. However, the pattern Size is by many writers considered to be
in Figure 1 is too simplified. The structure one of the most crucial characteristics of
of organizations is not exogenous in relation organizations. There is a vast body of litera-
to environmental and internal conditions, ture where the relation of size to other
but dependent upon them (cf., e.g., Law- features of organizational structure is dis-

233
cussed (Pugh et al. 1969; lilau 19/U; tilau l hus, according to researchers who
& Schoenherr 1971 ; Mintzberg 1979; Miller adhere to the ’small is beautiful’ argument,
1987; Aldnch & Marsden 1988; Szulkin one should expect that small firms create
1993). According to both theoretical argu- positive conditions for employees to exer-
ments and empirical evidence, large organ- cise autonomy and discretion at work.
izations develop hierarchies, formalized Large establishments, by contrast, are
and standardized work routines, and a high charactenzed by more controlled and alien-
degree of subdivision of labour, i.e., ating conditions of work.
bureaucratic systems of coordination of the This kind of argument has been chal-
work process. Thus, one can expect that lenged, however. The proponents of the
organizational size has several important opposite view claim that small size facili-
structural correlates. The importance of tates direct, unsophisticated and oppressive
these structural features for the autonomy forms of control (Edwards 1979; Rainnie
of employees is discussed in subsequent 1985, 1989). A small unit may be controlled
paragraphs of this article. The subject of by one person who can exercise personal
this subsection is whether the sheer size power over the employees. The power of
of an organization could be expected to the executive staff in a small firm can be
influence the content of jobs. relatively uncircumscribed, harsh and arbi-
According to a conventional ’small is trary.~ Thus, the merit of face-to-face
beautiful’ argument, small firms in com- relationships between employees and staff
’

parison with large are characterized by ’har- could be questioned.

monious’ industnal relations, a large scope Futhermore, it has been argued in the
for individual initiative, and more attention economic literature that job autonomy in
small firms may be strongly limited.
being paid to the human aspects of work
(Rainnie 1989). Arguments for this view According to this argument, monitonng dif-
were developed (and broadly accepted) in
ficulties increase with the size of the organ-
an early influential article by Worthy
ization (cf., e.g., Williamson 1967). Small
size is seen as a precondition for effective
(1950). According to Worthy, work in a control. Managers in small establishments
small establishment is relatively meaningful
who exercise personal power over workers
for the workers because they can easily see
can therefore put severe constraints on the
the importance of their work for the whole
work unit, and because there are many job activities of employees, i.e., limit their
autonomy.
spontaneous face-to-face interactions with In contrast to the few previous studies of
other employees and with the top man-
the association between size and autonomy,
agement. The work of employees m small there is a substantial and growing literature
firms is also usually evaluated by examining
output rather than behaviour.
concerning the size-wage effect. A robust
The 1980s flexibility debate marked a
empirical generalization - which seems to
be valid for industrialized countries across
renewed interest for the conditions pre- both time and space - is that workers
vailing in small firms. According to Piore employed in large organizations receive
& Sabel (1984), we are approaching the end
of an era when extensive, homogeneous
higher pay than those in small organiz-
ations. Furthermore, empmcal research
markets and mass production dominated indicates that a large part of this wage dif-
economic life. Old forms of manufacturing ferential remains after taking account of
are successively replaced by more flexible size differences in individual employee
production systems, characterized by short attributes and job characteristics (Stolzen-
production runs and more flexible work berg 1978 ;Kalleberg et al. 1981;Hashimoto
arrangements which are dependent on & Raisan 1985 ; Colbjørnsen 1986 ; Villemez
workers’ composite skills and control over & Bridges 1988; Brown & Medoff 1989).
the pace and flow of work. For the pro- A number of different explanations of
ponents of flexible specialization, the small why organizational size affects earnings
firm is the proper site to establish a modern have been proposed - such as size-related
version of craft production. differences in unmeasured individual pro-

234
ductivity, rent-sharing, problems of con- Previous results published in Swedish, how-
trolling and coordinating work, costs of ever, show that there is a positive associ-
labour turnover, capital intensity, the ation between the degree of wage inequality
impact of unions and collective action, and and the relative wage level of the estab-
the presence of firm internal labour market lishment (le Grand et al. 1993). In organ-
arrangements. So far, however, none of izations with a large variation in earnings,
these explanations seem convincingly to the average earnings of workers thus tend
account for the size-wage effect, at least to be high. If bottom wages are more or
not in the US (Brown & Medoff 1989).5 less fixed across establishments, this finding
is entirely straightforward. If bottom wages
3.2. Hierarchy are flexible, however, the positive associ-

According to the traditional burenucratic- ation between the dispersion and level of
alienatioti model, hierarchical structures of wages may be explained with reference to
authority imply close supervision, tight theories of equity and relative deprivation
discipline and little leeway for individual (Stouffer et al. 1949 ; Levine 1993). To tear
initiative (Kohn 1971 ; Child 1984; Perrow an established structure of internal pay dif-
1986; Lincoln & Kalleberg 1990). Tall and ferentials apart by increasmg inequality is
lean hierarchies imply small spans of con- detrimental for the morale of workers at
trol and therefore make monitoring of sub- low organizational levels, resulting in
ordinates by supenors effective. decreased work effort (Akerlof 1982; Aker-
It is seldom acknowledged, however, that lof & Yellen 1990). To compensate these
this line of criticism of bureaucracy is hard workers for the perceived injustice when
to reconcile with the ’small is beautiful’ they compare their relative earnings with
argument.b Since the implication of a tall other workers within the firm, they receive
hierarchy is a small span of control, the an extra wage bonus in order to make their

average size of a subunit in a pyramidal comparison with similar workers in other

organization should be relatively small. (By firms more favourable.
a subunit we mean a superior and his or her
immediate subordinates.) If one accepts 3.3. Formalization
that small organizations create positive con- Formalization is a means of regulating and
ditions for employees to exercise autonomy controlling behaviour inside organizations,
and discretion at work, one should there- in order to reduce the vanability of behav-
fore also accept that tall hierarchies repro- iour and make it more predictable (Mintz-
duce this positive effect of ’smallness’ at the berg 1979). According to the conventional
level of subunits.’ cnticism of bureaucracy, highly formalized
Hence, a tall and lean hierarchical struc- organizations prescribe the behaviour of
ture does not necessarily mean an oppress- employees in a narrow way. Formalization
ive system of work control. As several of everyday rules and procedures reduces
writers have argued, a small span of control the leeway for personal initiative and flex-
could instead indicate close cooperation ible action, and thereby routinizes the work
between superiors and subordinates, com- activities of employees (Mintzberg 1979;
mon problem solving and team work, i.e., Kalleberg & Leicht 1986; Lincoln & Kal-
forms of coordination of the work process leberg 1990). Formalization is seen by
which are quite the opposite to harsh and critics of bureaucracy as one of the most
authoritarian discipline (Woodward 1965; prominent features of ’mechanical’ (as
Blau 1968; Perrow 1986; Lincoln & kalle- opposed to ’organical’) structures, which
berg 1990). undermine the conditions for self-fulfilment
While there both theoretical argu-
are and autonomous action of the employees
ments and empirical support for the hypoth- (Burns & Stalker 1961).
esis that the span of control and the Again, however, this line of reasoning
hierarchical structure of the organization has been questioned. Formalization is seen,
will affect wage dispersion within organ- by several authors, as a structural device
izations (Hedstrom 1991 ), their impact on which can serve the interests of employees
the level of wages is a more dubious issue. (Crozier 1964; Kohn 1971; Perrow 1986;
235
Lincoln & Kalleberg 1990). According to ization. Other positions are filled from
this argument, formalized power is less within.
oppressive than unformalized. Written (3) Skill accumulation. The firm invests in
rules and strictly defined procedures may the new worker by extensive training and
reduce the power of superiors to issue arbi- within-firm education, which results in an
trary orders. Thus, formalization means accumulation of skills and gradually more
constraints not only for rank-and-file demanding job tasks.
employees, but for persons in positions of (4) Job ladders. Workers employed within
authority as well. Whether the net effect FILMs are offered upward mobility chances
of these constraints on the autonomy of within the firm, parallel to the development
employees is negative or positive is an open of skills.
question.8 (5) Job competition instead of wage com-
There are few strong arguments for sup- petition. Wages are tied to jobs, not to
posing that formalization will affect earn- individuals. Employees acquire higher
ings. If formalization implies a higher level wages through careers in a well-defined
of control, it can be expected to have a structure of positions, rather than through
negative effect on earnings. The argument pay raises within jobs. ,

is that the employer chooses either to pay

more or intensify the control, in order to The link between FILM arrangements and
increase workers’ effort. It seems reason- the job autonomy of workers is not rig-
able, however, that a lack of formalization orously discussed in the literature. We
does not imply low, but rather unformalized distinguish between two competing hypoth-
control, for instance through direct per- eses. The first is based on the assumption
sonal supervision. Our hypothesis therefore that the interdependency of jobs within
is that, while formalization certainly may FILMs also entails an interdependency of
diminish autonomy, it should not have any work tasks between employees, which
significant direct impact on earnings. should result in a comparatively small scope
for autonomous individual decisions (Kal-
3.4. Firm internal labour markets leberg & Van Buren 1993). The assumption
Modern research on how organizational is not self-evident, since jobs might be con-
contexts influence work-related outcomes nected in a promotional structure which
has often focused on the functioning of does not necessarily involve strong mter-
internal labour markets (Kerr 1955; Doer- dependencies in everyday work tasks. To
inger & Piore 1971 ;Althauser & Kalleberg the extent that the assumption holds, how-
1981). Although many different arrange- ever, the hypothesis of a negative effect of
ments are possible, the type of internal FILMs on autonomy is not unreasonable.
labour market which has received most An alternative hypothesis is that FILMs
interest in recent literature is without doubt create incentives, which diminishes the
the firm internal labour market (FILM). need for overt control. Long expected firm
According to many writers, FILMs are tenure, promotion chances, and wage pro-
characterized by the following traits (for gression may be devices not only for skill
a conceptual discussion, see Althauser & development in the firm, but also for cre-
Kalleberg 1981): ating loyalty and commitment to the organ-
ization (Lincoln & Kalleberg 1990;
(1) Firm-specific skill requirements. The Kalleberg & Mastekaasa, this issue). An
skills and qualifications needed at work are important purpose of FILMs is thus to
to a high degree developed within the firm, increase workers’ attachment to and depen-
and cannot be bought on the external dency on the employer. The need for overt
labour market. control is therefore small, and consequently
(2) Low-level ports of entry. Because job autonomy may be expected to be larger
specific skills cannot be transferred between for employees within a FILM than for other
firms, the firm normally prefers to hire workers.
young, untrained workers, often directly With regard to the effects of FILMs on
from school, to a low position in the organ- earnings, two hypotheses may again be for-
236
mulated. The first hypothesis is based on expected to experience steeper seniority-
conventional behavioural premises in wage profiles. In contrast, however, the
labour economics. It states that the lifetime second hypothesis states that earnings
earnings of a worker are not affected by within FILMs will exceed the externally
whether he or she takes part in a FILM or prevailing rate, even in a lifetime perspec-
not, since such premia would not survive tive. The main argument for this is that
on a competitive market (Becker 1964; Par- most workers have a shorter time per-
sons 1986). However, the starting pay is spective than their lifetime earnings when
lower and the upward slope of the senior- looking for a job. If firms with a FILM offer
ity-wage profile is steeper for workers a starting wage which is substantially below
within FILMs. The argument is that an the external rate, they will find it difficult
employer who makes costly investments m to recruit high qualty workers. These firms
his workers’ skills takes a nsk, as the invest- will therefore offer the same - or at least
ment is lost if the worker quits. The rational a not much lower - starting pay as that
solution for such an employer is therefore prevailing m the external market. Thus, the
to lower the starting pay for newly recruited steeper wage profile is supplemented with
workers and, instead, offer them higher a pure wage bonus.

wages, if they remain with the firm after

acquiring firm-specific skills. According to 3.5. of
this kind of reasoning. the main effect of Summary hypotheses
FILM on earnings should be negative, while 3.5.1. Autonomy
there should be a positive interaction effect Hla: According to the ’small is beautiful’
between FILM and senionty on earnings.~ argument, one should expect that the
Another implication is that the wage effects degree of job autonomy in small estab-
of FILM arrangements can be attributed to lishments is higher than m large.
the specific skills that workers within FILMs Hlb: According to the opposite argument,
acquire. In other words, once we control small firms are charactenzed by oppressive,
for differences in skill specificity between arbitrary and uncircum>cribed forms of
employees within and outside FILMs, their control. Thus, job autonomy in small
respective wage profiles should be similar. organizations is severely limited.
The second hypothesis, by contrast, is
based on efficiency wage theory (for an H2a: Proponents of the bureaucratic-alien-
overview, see Akerlof & Yellen 1986) and ation model claim that a hierarchical struc-
modern sociological class theory (Gold- ture of authority in organizations means
thorpe 1982 ; Wright 1985; Erikson & Gold- close supervision and little leeway for
thorpe 1992). It states that job conditions autonomous action of employees.
in general, including earnings, are more
H2b: The ’small is beautiful’ argument can,
favourable within FILMs than elsewhere.
however, be reproduced at the level of sub-
FILM employers find it profitable to pay a units. Extensive hierarchy implies a small
higher wage than the rate prevailing on span of control and therefore a small aver-
the external market, for two main reasons
related to why they developed a FILM in age size of subunits. Hence, the association
between autonomy and hierarchy should be
the first place. The first reason is that some
jobs are difficult to monitor, and therefore positive.
pose a special control problem. The second
reason is that workers in jobs which require H3a: The behaviour of employees in for-
a large amount of internal training are malized organizations is narrowly circum-
costly to replace. Monitoring problems and scribed, i.e., the degree of autonomy is low.
turnover costs thus give nse to long-term H3b: Formalization does not just imply
employment relationships, which typically constraints for subordinate employees, it
take the form of FILMs with a prospective also reduces the power of superiors to issue
reward structure. Consequently (just like arbitrary orders. The net effect of for-
the proponents of the neoclassical per- malization on autonomy cannot be estab-
spective argue) workers within FILMs are lished on the basis of theoretical arguments.

237
H4a: The effect of FILMs on autonomy is organizational structure is not of a one-way
negative. This is due to the interdependent nature, but reciprocal. The character of the
character of jobs withm FILMs. work to be done in an organization is, of
H4b: The association between FILMs and course, an important precondition for the
autonomy is positive. The reason is the specific structural solutions adopted.
relatively limited need for, or even dys- Hence, not only may it be the case that the
functionality of, overt control of employees autonomy of employees is affected by the
inside FILMs. structure of the work organization, but the
reverse is probably also true. In fact, as
’

3.5.2. Earnings indicated in the previous section, the

H5: In accordance with previous empirical character of work with regard to monitoring
research, we expect organizational size to difficulties is widely held to be one of the
have a positive effect on earnings. most important determinants of organ-
izational structure. One may in principle
H6: Span of control will have a negative distinguish between two main employer
effect on earnings, even when controlling strategies m solving the problem of control
for industry and job characteristics. over employees in autonomous jobs. The
more simple strategy is to strengthen per-
H7a: According to mainstream economic sonal supervision by narrowing the span of
theory, FILM arrangements are expected control. In many cases, however, this is not
to result in steeper seniority-wage profiles. a viable solution. The autonomy of a certain
The lifetime earnings of a worker, however, kind of job might stem precisely from the
are not affected by whether he or she takes fact that the job cannot be adequately per-
part in a FILM or not. Therefore, the start- formed under close personal supervision.
ing pay is lower for workers within FILMs The alternative strategy is then to solve
compared to other workers. the control problem not by increased overt
control, but, rather, by undermining the
H7b: Just like proponents of the neo- conflict of interest between employer and
classical perspective, those who favour employees, i.e., by attacking the very root
efficiency wage theory, or sociological class of the control problem. According to mod-
theory, argue that workers within FILMs ern class theory (Goldthorpe 198?; Wright
experience a steeper seniority-wage 1985, Erikson & Goldthorpe 1992), jobs
profile. In contrast to hypothesis 7a, which are difficult to monitor give nse to
however, membership within a FILM, even settings designed to increase effort and loy-
in a lifetime
perspective, is assumed to be alty among employees through other means
advantageous. In other words the starting than personal supervision, chiefly by a sys-
pay is not lower for workers within FILMs. tem of prospective rewards such as sub-
stantial wage increases with seniority and
H7c: Proponents of human capital theory opportunities for promotion. This solution
state that FILMs can be explained by the thus resembles the firm internal labour mar-
necessity of investments in specific human ket, and also what Edwards (1979) calls
capital within the firm. If this is an impor- bureaucratic control.
tant explanation, the wage effects of FILM When examining the influence of organ-
should substantially decrease when con- izational structure on job autonomy, it is
trolling for the workers’ specific skill acqw- therefore important to distinguish between
sitions in the firm. this impact and the possibly powerful
effects running in the reverse causal direc-
3.6. Reciprocal relations between tion. In the empirical analyses below, we
organizational structure and job content attempt to do this by including different
Our assumption in the preceding discussion measures of the character of work carried
has been that the structure of organizations out m the organization as control variables
influences the content of jobs. As pointed in the regression models. Effects which
out in the introduction, however, the remain after these controls we interpret as
relationship between job content and operating from the organizational level to
238
the level of jobs rather than the other way most establishments data have only
m our
around. one employee in our individual sample. It
is evident that the employee who happens
to be included in our sample need not be
4. Data and variables
typical of the personnel of the estab-
The empirical analyses in this paper are lishment in question. Although we believe
based on two sets of data. The information that we to some extent have been able to
on individuals and their jobs has been col- solve these problems, the estimated impact
lected in the Level of Living Survey 1991. of organizational charactenstics on indi-
The sample consists of approximately 6,000 vidual outcomes is likely to he downward-
individuals representative of the Swedish biased. This bias might to some extent be
adult (age 18-75) population. The response counteracted by the impact of unobserved
rate was 79 per cent. Around 3,500 persons individual productive characteristics. If
in the survey were wage-earners at the time workers are sorted across establishments
of the interview. with different organizational structures on
Respondents employed m organizations the basis of such unmeasured traits, the
with at least ten employees gave infor- estimates of organizational effects will be
mation on the name, address, and tele- upward-biased. Without access to panel
phone number of their place of work. This data, it is hardly possible to correct for this.
procedure generated a sample of 2,135 (For a further discussion, see, e.g.,Krueger
organizations. The top managers of these & Summers 1988).
establishments were contacted for tele- Since it is the job autonomy and earnings
phone interviews in the Swedish Estab- of individual employees we attempt to pre-
lishment Survey 1991. The interviews dict, the unit of observation is individuals,
contained questions on organizational not establishments. A number of large
structure, personnel policies, promotion organizations employ more than one Level
systems, training programmes, market of Living Survey respondent. The number
situation, and various other organizational of observations (individuals) is therefore
features. The response rate was 93 per cent. larger than the number of establishments. &dquo;’
Usable information was received from A description of all variables used in the
1,988 establishments. Data from the Estab- empirical analyses is given below. All addi-
lishments Survey are the source of organ- tive indices are standardized, i.e., trans-
izational-level vanables used in this paper. formed into having a mean of zero and a
The estimated effects of the organ- vanance of one, and in turn are based upon
izational variables are probably conserva- standardized items. Exact formulations of
tive, due to the design of the survey data interview questions are given in the Appen-
used in the analyses. First, it is difficult to dix, which also shows means and standard
define the borders of an organization. For deviations of all variables.
practical reasons, we have chosen to focus
on establishments, although most of these 4.1. Organizational variables
belong to larger organizational entities with Establishment size is the total number of
properties which ideally should be taken employees of the establishment. We follow
into account. Second, the internal con- the common practice of using the log-
ditions of a workplace might be far from arithmic transformation of this measure in
homogeneous across departments and our analyses. ’ ’ As stated above, the Estab-

organizational levels. It is therefore not lishment Survey includes only organizations

always easy for managers to give infor- with ten or more employees. The impli-
mation on workplace conditions which is cation of this restriction is that the size
valid for the entire establishment, or for effects in our study are truncated. Our data
entire personnel categories within estab- thus do not allow for any generalizations
lishment. What is more, the top managers applicable to the smallest establishments.
might in some cases not be very familiar Hierarchy is the (log) average span of
with the conditions prevailing far away or control within the establishment, stan-
further down in the organization. Third, dardized for the total size of the organ-

239
lzation. unaer tne simpiitying assumpnon l:-xperœnce is measurea Dy years In tne
that organizational subunit has only one
an labour force. A square term of experience is
supervisor, the formula for calculation is also used in the models, to take curvilinear
x =
(a - b)/b, where x is span of control, a effects into account.
is the total number of employees in the Female is an indicator variable taking the
organization, and b is the total number of value 1 for women and 0 for men.
supervisors.’~2 Autonomv is an additive index based on
Formalization is an additive index based two interview questions indicating the
on two interview questions indicating the degree to which the job incumbent has
extent to which the day-to-day work activi- influence (a) on what work tasks to
ties in the establishment are governed by
standardized rules and routines.
perform, and (b) on the choice of method
to accomplish the tasks.
Firm internal labour market (FILM) is an Educational requirements is the number
additive index of three components: (a) of years of post-compulsory schooling nor-
rates of promotion and internal recruit-
ment ; (b) investments in on-the-job train-
mally required in the present job.
Subordinates is the (log) number of
ing ; and (c) positional vs. individual wage employees that the job incumbent super-
setting. (Each component is based on vises.
several interview questions.) In most other
General job skills is the extent to which
studies, FILMs have mainly been skills have been acquired in the present job
operationalized by only the first of these which would be useful when working for
dimensions (promotions and internal
another employer.
recruitment). Although we believe that all
three dimensions are relevant, the first is Specific job skills is the amount of time
given a double weight in our measure, in necessary for learning to perform the cur-
rent job reasonably well. Our assumption
order to take the central importance of job
ladders into account. In sum, organizations here is that this measure represents firm
with high values of FILM are chiefly charac- specific skills, when the level of general
terized by high rates of upward mobility skills has been controlled for.
and internal recruitment, but also by large Sentorin, is the number of years spent
investments in internal training, and by job with the current employer. As with general
competition rather than individual wage experience, a square term of seniority is
competition.l33 also mcluded in the models, to allow for
Thus, while the concept of FILM has curvilinear effects.
been treated as a binary variable in the Earnings is the (log) pre-tax hourly wage.
theoretical literature, we follow Pfeffer & Since most employees do not receive a fixed
Cohen (1984) in operationalizing it as con- hourly wage, other kinds of pay - such as
tinuous, i.e., as the degree of FILM monthly, weekly and daily earnings,
arrangements in the organization. bonuses, piece-rate payments, com-
Industry is a classification of 37 categories missions, and additional compensation for
indicating the main kind of production or overtime or inconvenient working hours -
service carried out in the organization.l’~ have been recalculated into earnings per
Class composition is an additive index hour.
based on two items: the proportion of
higher-level white-collar employees, and 5. Results
the proportion of manual workers (reverse
coded), of all employees in the organ- In this section we present results of multi-
ization. variate analyses, with autonomy and earn-
For profit is an indicator variable taking ings, respectively, as dependent variables.
the value 1 for profit-seeking organizations The method used is Ordinary Least Squares
and 0 for other establishments.
regression. Earlier research shows that the
4.2. Job characteristics, individual wage determination process differs be-
resources, and earnings tween the sexes (Treiman & Roos 1983;
Education is measured by years of school- Rosenfeld & Kalleberg 1990; le Grand
ing. 1991). Consequently, the wage equations
240
Table I. Regression analyses of autonomy using characteristics of individuals and organizations as
predictors.

Note: Model B also includes 36 dummies for industry.

.**
p < 0.01 : **p < 0.05; * p < 0.10.

are estimated separately for men and In the second model (Table 1, model B),
women. For autonomy, however, no dif- we estimate the impact of organizational
ferences were found between the sexes. structure on autonomy. As pointed out
These models are therefore estimated above, one can expect reciprocal effects
jointly for men and women. between the degree of job autonomy and
several of the structural traits which are
5.1. Autonomy and organizational investigated here. In order to distinguish
the effects running from the organizational
structure
We level, we include two measures of the gen-
begin the empirical analysis by esti- eral character of work carried out in the
mating the relations between individual organization as control vanables: industry
resources, organizational structure, and job
and class composition of the establishment.
autonomy. Table 1 shows the results from These measures may be seen as proxies for
two multivariate analyses. In the first
(model A), we include only human capital
organizational production tasks. One can
assume that technological requirements
indicators and sex as predictors. These indi-
vidual variables all have significant effects vary according to industrial sector and that
task complexity vanes with class compo-
on autonomy. Respondents with higher
sition. s
education, persons with longer labour force The individual characteristics studied in
experience, and males tend to have rela- model A are also included in model B as
tively autonomous jobs. The increase in control variables. Thus, the effects of
autonomy with labour force experience organizational structure on job autonomy
diminishes over time (the squared term in in model B may be interpreted as estimates
the regression equation has a negative net of the effects of individual resources
sign). The total impact of individual and the general nature of production tasks
resources on autonomy is moderate; in the establishment.’6
around 9 per cent of the variation is From the results in model B we conclude
accounted for. that all structural variables appear to influ-

241
ence the degree of job autonomy. Estab- Finally, FILMs are structural arrange-
lishment size, the average span of control ments which appear to have a positive
within the organization, and the degree of impact on the job autonomy of employees.
formalization of everyday work activities The need (or scope) for overt and rigid
are all negatively related to the amount of forms of control which reduce autonomy
autonomy, whereas the association be- seems to be limited inside FILMs. Hypoth-
tween FILMs and autonomy is positive.&dquo; esis 4b is thus supported, while hypothesis
The predictors applied in the model alto- 4a, stating that the interdependency of jobs
gether explain around 15 per cent of the within FILMs results in a low degree of
variation in job autonomy. autonomy, is not. This result is the opposite
Our results support the ’small is beauti- of what Kalleberg & Van Buren (1993)
ful’ view of organizations. The effect of found in their analyses of data from the US.
establishment size is a net effect after check- The character of FILMs thus seems to differ
ing for other structural features of the between the two countries, which is an issue
organization. This effect could thus be for future comparative research.
interpreted as a ’sheer effect of large num-
bers’ (see Kalleberg & Van Buren 1993, 5.2. Earnings and organizational structure
who also report a negative effect of size on Tables 2 and 3 present results from a series
autonomy in the US). Hence, the leeway of regression analyses with hourly earnings
for autonomous action seems to be rela- as the dependent variable. As discussed

tively limited in large organizations. above, there are good reasons to assume
that the mechanisms which generate wage
Hypothesis la is thereby supported, while
hypothesis lb is rejected. inequality are gender-specific. The analyses
Our findings with regard to the hier- are therefore made separately for men and
women. In the first model (A), we include
archical shape of organizations can be inter-
preted along the same lines. The negative organizational-level variables together with
individual resources as predictors. We also
relationship between the average span of control for industry (37 categories), and
control (size of the work unit) and job
include a dummy variable indicating
autonomy indicates that a tall hierarchy whether the organization is profit-seeking.
does not, pnman’ly, mean oppressive super-
Previous research has shown that
visory control. Elaborate pyramidal hier-
archies rather seem to be associated with employees of profit-seeking organizations
are relatively well paid in Sweden, and that
teamwork, cooperation and informal coor-
dination of work activities. Thus, the posi- wages differ between industries (Arai 1990;
le Grand et al. 1993).
tive effect of ’smallness’ appears to be
In Model B, the interaction effect
reproduced at the level of organizational between FILM and seniority is tested. This
subunits. The conventional intuition that is done by including seniority, seniority
hierarchy reduces the scope for auton- squared, and two multiplicative terms,
omous action is therefore not confirmed by
the analysis. In terms of the hypotheses
(FILM*Seniority) and (FILM*Seniority
formulated above, hypothesis 2b is
squared). The interaction terms are
intended to capture differences between
supported, while hypothesis 2a is workers in semonty-wage gradients con-
rejected. &dquo; tingent on membership in a FILM. Since
Formalization of everyday work activities we are interested in the total effect of
in the organization affects job autonomy FILM, the other three indicators of organ-
in the way assumed by the conventional izational structure (size, hierarchy, for-
criticism of bureaucracy. Formalized struc- malization) - which may partially capture
tures thus seem to reduce the scope for some of the FILM effects - are not included
autonomous action. This result is m line in the model (see also note 22).
with previous empirical evidence for the In model C, finally, the analysis is ex-
US (Kalleberg & Leicht 1986; Lincoln & panded by including job-related variables.
Kalleberg 1990), and lends support to These measures include supervisory
hypothesis 3a. responsibilities, educational requirements,
242
Tab!e 2. Regression analyses of (log) hourly earnings iisitig characteriStics of mdmduals, jobs, and
organizations as predictors : Women

Note: All models include 36 dummies for mdustrry

***
p < 0.01; **p < 0.05; *p < 0.10.

seniority, specific job skills, general job have significant effects on earnings. In
skills, and job autonomy. By these controls accordance with previous research, male
we are able to investigate the degree to workers in large establishments are paid
which the wage effects of the organizational more than male workers in small organ-
structure are mediated by the character of izations. The effect of the average span of
jobs. control is negative. Thus, male workers in
The results of the three models for organizations characterized by a tall and
women (Table 2) can be quickly summa- lean hierarchical structure earn more than
rized : none of the four measures of the those working in flat organizations. As
organizational structure has any significant expected, formalization does not affect
effect on earnings.&dquo; We will return to this earnings.
finding in the concluding discussion. The FILM arrangements have no net effect
effects of individual attnbutes and job on male wages in model A. However, as can
characteristicsare generally in accordance be seen from model B, the FILM-seniority
with previous research and conventional interaction terms are significant for men
expectations. while the main effect of FILM is negative,
For men, the results are quite different but not significant. This result is in accord-
(see Table 3).’-° Model A shows that organ- ance with both hypotheses 7a and 7b, since
izational size and average span of control the wage effect of seniority is stronger for

243
Table 3. Regression analyses of (log) hourly earnings using characteristics of individuals, jobs, and
organizations as predictors.- Men.

Note: All models include 36 dummies for industry.

*&dquo;p<O.Ul.’&dquo;p<0.05.*p<0.10.

workers within FILMs. Hypothesis 7b When controlling for job and skill charac-
seems more plausible, however, since the teristics in model C, the FILM-seniority
starting wages of workers within FILMs are interactions are still significant. This result
not significantly lower than those of other is unexpected from the point of view of
workers. Thus, FILMs appear to pay pure human capital theory (hypothesis 7c),
wage premia. according to which the FILM effects mainly
The inclusion of positional characteristics (or even entirely) reflect strategies to
in model C (Table 3) only marginally affects ensure the enhancement of specific skills.
the size-wage relation as compared to In other words, when we control for dif-
model A. Hypothesis 5 is thereby ferences in the character of job skills
supported. The impact of span of control, between FILMs and other establishments,
however, is decreased by almost four the impact of FILM on wage determination
tenths, although it is still significant. A sub- should disappear, or at least decrease sub-
stantial proportion of the total association stantially. Since this does not happen,
between span of control and earnings is hypothesis 7c is not supported. The wage
apparently due to vanations m the charac- profiles of FILMs are apparently not simply
ter of work between organizations with related to the development of specific skills,
differing hierarchical structures. The but have other causes as well.
significant direct effect, however, also lends It is difficult to evaluate the composite
some support to hypothesis 6, stating that wage effect of seniority and FILM by look-
the wider wage dispersion in lean and tall ing at the results in Table 3, as the model
hierarchies must be compensated for by a specifications are rather complex. Based
wage bonus also to workers on lower levels on the coefficients of model B, Figure 2
in the organization. 21 graphically illustrates how the seniority-
244
Figure 2. Senioritywage effects, conditioned by FILM arrangements. Male workers.
Note: The wage profiles are calculated from model B, Table 3 as the sum of: -0.0195
FILM +0.0058 Senionty -0.000123 Seniority2 +0.0058 (FILM*Senionty) -0.000135
(FILM.Seniority2), where Low FILM = -
1, Medium FILM 0, and High FILM
=
1.
=

wage effects vary for men according to cessively higher earnings with increasing
FILM arrangements. 22 Seniority-wage pro- seniority. These wage premia are in line
files are shown for three levels of FILM: with hypothesis 7b, but not with 7a.
High FILM = one standard deviation
above the mean; medium FILM = the
mean level; Low FILM one standard
= 6. Conclusions .

deviation below the mean. In accordance In this article we have shown that organ-
with both hypotheses 7a and 7b, well-devel- izational characteristics are important for
oped FILMs seem to result in steeper the content of jobs and for individual
seniority-wage gradients than those found rewards. More specifically, we find that
elsewhere. The starting pay is almost 4 per organizational structures have a consider-
cent lower in high FILMs compared to low able impact on job autonomy and individual
FILMs. Among workers with five years’ earnings in Sweden. Four organizational
employment, however, wages are higher features are distinguished: establishment
in more developed FILMs..A comparison size, hierarchical structure, firm internal
between employees who have stayed 20 labour market arrangements, and for-
years with the same firm reveals that malization of daily work routines. Although
workers in high FILMs receive almost 9 these features are commonly conceived to
per cent higher earnings than comparable be strongly interrelated, they turn out to
workers in low FILMs. This pattern implies vary considerably from one another in their
that the lifetime earnings of workers within influence on the content of jobs and on
FILMs are higher than those of comparable individual monetary rewards.
workers outside FILMs. Although the earn- Organizational size has a negative effect
ings of a worker within a FILM may be on autonomy, but a positive effect on earn-
somewhat lower in the first few years, this ings. Tall and lean hierarchies have a posi-
is more than compensated for by the suc- tive impact on both job autonomy and

245
earnings. Establishments with well-devel- make sure that the work is done in an
oped firm internal labour markets (FILMs) appropriate manner. Stringent discipline
tend to have relatively autonomous jobs, and cooperation could then be seen as
as well as steep seniority-wage gradients, alternative devices for achieving this goal.
and probably higher expected lifetime earn- Small work units appear to constitute rela-
ings. Finally, formalization affects auto- tively favourable conditions for non-auth-
nomy negatively, but is apparently oritarian managerial practices.
unrelated to the level of earnings. The The four organizational characteristics
results for autonomy apply to both men and studied here affect job autonomy in the
women, while the effects on earnings are same way for both men and women. A

present only among males. central finding is, however, that those
In concluding, we would like to indicate aspects of the structure of organizations
the more general implications of some of examined here only seem to affect the
our results. Students of organizations often earnings of men, but not of women. This
take for granted that bureaucratic struc- result is in line with the arguments of those
tures suppress initiative and reduce worker who claim that segregation by sex in the
autonomy. This assumption has the status labour market is the principal source of
of a conventional wisdom rather than of a gender-wage differentials. The implication
well-substantiated empirical generalization. is, however, that even though segregation
More rigorous research programmes have, by occupation, industry and firm may be
to our knowledge, very seldom been devel- important factors for explaining sex dif-
oped to systematically investigate the con- ferences in pay, they are not sufficient. As
ditions prevailing in a bureaucracy. The findings based on American data show
results of the present study may be seen as (Baron et al. 1986; Bielby & Baron 1986),
a contribution to an understanding of the even when men and women work in the
ways bureaucracies work. same kind of job within the same enter-
Some of our results support the con- prise, there is a sex segregation in the access
ventional view. The ’small is beautiful’ to job ladders and therefore a tendency for
argument is, for instance, substantiated. men to dominate opportunities for career
Other results, however, point in a direction
mobility. The conclusions drawn by Bielby
opposite to expectations developed by pro- & Baron (1986: 761) are also in accordance
ponents of the bureaucratic-alienation with the results reported here, namely that
model. In the theoretical part of the paper, ’... sex segregation is built into the hier-
we reviewed the arguments which may be
used in understanding these findings. There
archy of organizational positions and is sus-
tained by sex stereotypes and workplace
is no need to repeat this discussion here. social relations’. Further analyses are
Instead, we offer some comments on what needed to throw more light on this issue.
might be seen as the most unexpected result In sum, the results reported in this paper
of the present study; namely that hier-
indicate that organizational structures are
archical structures seem to enhance rather
of considerable importance for the con-
than restrict the scope for exercising initiat-
ditions of individual employees. Despite
ive and autonomous action.
the formidable difficulties involved in large-
Perhaps the idea of a hierarchical organ- scale empirical analyses of the impact of
ization needs some reconceptualization.
Conventional arguments about hierarchical organizational structures on individual out-
comes, several significant effects emerge.
organizations have been rather one-sided. Efforts to develop an organizational per-
If hierarchy means a small span of control,
there is good reason to believe that the spective on stratification are evidently a
well-motivated enterprise.
potential for close supervision, which
undoubtedly exists in small units, could be
outweighed by cooperation and mutual Acknowledgements
adjustment. The ultimate goal of super- We thank Peter Hedstrom, Ame Kalleberg, and
vision is, after all, not to exercise control two anonymous referees for useful comments on
and suppress subordinates but instead to an earher draft of this paper.

246
 First version received March 1994 10
For 110 establishments our data include two
Final version accepted May 1994 respondents and for a further 106 more than two.
11
Organizational size can be measured in dif-
ferent ways (for an overview, see Kimberly
1976). The alternative size measures are often
Notes based on indicators of the economic scale of the
1
Since experience, especially for men, is establishment, such as assets, profits, or volume
highly correlated with age, it will also capture of organizational transactions (Kalleberg & Van
possible age effects. Buren 1993). The logarithmic transformation of
2
Although skills and seniority might plausibly the number of employees reduces the skewness
be considered as individual resources, we regard of the distribution of the variable, and reflects
them as primarily tied to the job since they have the obvious idea that the effect on other variables
been acquired there. of a move from, for instance. 100 to 200
3
Figure 1 does not include any interaction employees is likely to be much larger than a
effects. It can, however, be argued that some move from 1,000 to 1,100 employees.
12
organizational variables may modify the effects Our measure is identical with the average
of some individual resources, or of some posi- subunit size minus 1 (i.e., minus the supervisor).
tional characteristics. One such interaction effect Supervisors’ average span of control is commonly
will be discussed below, namely how firm internal taken to indicate the hierarchical shape of the
labour markets condition the seniority-wage establishment. Ideally, the measure of hierarchy
effect. should also include information on the tallness
4
The uncircumscribed power of management of the organization, i.e., the number of levels of
is partly due to the relatively low ability of trade authonty. It is very difficult, however, to get
unions to organize employees in small estab- reliable estimates of the latter. The establishment
lishments (Amin 1991; Johnson 1991). The survey nevertheless contained a question on the
absence of trade unions allows employers to keep number of supervisory levels, and this variable
wages low and working hours long. correlates 0.47 (Pearson’s r) with span of control
5
Some of these explanations could be tested (holding establishment size constant). Thus,
with our data. However, this would be a subject even if there are some flat
organizations with a
for a separate article. small average span of control, and vice versa,
6
The proponents of the ’small-is-beautiful’ they are exceptions rather than the rule. It there-
argument and the bureaucratic-alienation model fore seems defendable to use span of control as
are, indeed, quite often the same persons. an indicator of hierarchical shape.
7 13
The terms ’hierarchies’, ’spans of control’ The three parts of the FILM measure belong
and ’size of subunits’ will be used interchangeably to a single dimension in a principal component
in the remainder of this article. factor analysis (using an eigenvalue of 1.0 as the
8
One should notice that formalization of dif- distinguishing criterion).
14
ferent aspects of organizational behaviour may The classification is a collapsed version of
have different impacts on the conditions of the official Swedish classification of industries,
employees. Formalization of everyday rules and SNI.
15
routines can reduce the autonomy of employees, In addition, holding these variables constant
as critics postulate Formalization of the employ- should control for spurious effects, i.e., cor-
ment relationship and of promotion schemes relations between organizational structure and
might, however, have positive effects on the autonomy which are due to common causes
working conditions of employees (Lmcoln & Kal- rather than one causing the other
16
leberg 1990). Littler (1982) distinguishes In accordance with the causal model in
between bureaucratization of the system of con- Figure 1, job characteristics are not included as
trol, which he equates with Taylorism, and predictors of autonomy. Although autonomy is
bureaucratization of the career structure, which correlated with several other job characteristics,
he sees as part of an internal labour market it makes no real sense to say that these cor-
9
That is, the wage effect of FILM is specified relations reflect causal relations.
17
as consisting of two elements in a regression We have information on establishment size
equation: b 1 FILM, the main effect, and b, for all workers (including those in organizations
(Seniority* FILM), the interaction effect, where with less than ten employees), but not on the
1 and b
b 2 are regression coefficients and FILM is other organization measures. Therefore, we re-
a measure of the degree of FILM arrangements estimated Table 1, model B, minus the other
in the organization. It then follows that the whole four organizational characteristics, for all
wage effect of FILM for newly recruited workers workers. The results show that the negative effect
is simply b , since seniority (years with the
1 of (log) size on autonomy is somewhat streng-
present employer) is zero. thened, from -0.054 in the restncted sample to

247
-0.061, when all workers are included. Hence, References
the size truncation does not affect the size effect Akerlof, G. A. 1982. Labor Contracts as Partial
on autonomy very much. Gift Journal Econ-
18
Exchange. Quarterly of
It is conceivable that this result is simply due omics 97, 543-569
to supervisors having more autonomy than non- Akerlof, G. A. & Yellen, J. L. 1986. Efficiency
supervisors. Organizations with a relatively small Wage Models of the Labor Market
average span of control have a relatively high Cambridge: Cambridge University Press.
proportion of supervisors, which increases the Akerlof, G. A. & Yellen, J. L. 1990. The Fair
likelihood that the individual-level respondent is Wage-Effort Hypothesis and Unemployment.
a supervisor. However, when the hierarchical
Quarterly Journal of Economics 105, 255-283.
position (number of subordinates) of the indi- Aldrich, H. E. & Marsden, P. A. 1988. Environ-
vidual employee is included in the regression and In N J. Smelse
ments Organizations.
equation, the effect of span of control is only (ed.), Handbook of Sociology, pp. 361-392
marginally reduced.
19
However, when re-estimating model C in
Newbury Park, CA: Sage
Althauser, R. P. & Kalleberg, A. L. 1981. Firms
Table 2 (minus span of control, formalization,
FILM, FILM*Seniority, and FILM*Seniority
)
2 Occupations, and the Structure of Labor Mar
kets: A Conceptual Analysis. In I. Berg (ed.)
for all female workers, including those employed
Sociological Perspectives on Labor Markets
at establishments with less than ten employees,
the size-wage coefficient increases from 0.007 to New York: Academic Press.
0.012, and becomes significant at the 0.05-level. Amin. A 1991. Flexible Specialization and Small
Firms in Italy: Myths and Realities. In A
Thus, the exclusion of the smallest organizations
tends to substantially decrease the size-wage Pollert (ed.), Farewell to Flexibility?. pp. 119-
effect for women. (For men the truncation does 137 Oxford: Basil Blackwell.
not matter ) It should also be noted that industry Arai, M. 1990. Essays on Non-Competitive Wage
and the dummy for profit-seeking organization - Differentials. Stockholm: Almqvist & Wiksell
which are both organizational-level variables - International.
have significant effects on women’s earnings Baron, J N. 1984. Organizational Perspectives
20
The following coefficients are significantly on Stratification Annual Review of Sociology

different between men and women at least at the 10, 37-69

10 per cent level Education (model A, B and C), Baron, J. N & Bielby, W T 1980 Bringing the
Firms Back In: Stratification, Segmentation,
Experience (model A and B), Span of Control
(model A and C), Seniority and the Organization of Work. American
2 (model C), and
Subordinates (model C). Sociological Review 45, 737-765.
21
This conclusion is also supported by results, Baron, J. N., Davis-Blake, A. & Bielby, W.
not presented here, which show that there is no T. 1986. The Structure of Opportunity: How
interaction effect between span of control and Promotion Ladders Vary Within and Among
number of subordinates. Such an effect would Organizations Administrative Science Quar-
have been expected if a tall hierarchical structure terly31, 248-273.
only benefited those on higher levels in the organ- Becker, G. S 1964 Human Capital. New York:
ization. Columbia University Press
22
FILM has a relatively strong (negative) cor- Bielby, W. T. & Baron, J. N. 1986. Men and
relation with span of control and a moderate Women at Work: Sex Segregation and Stat-
(positive) correlation with establishment size. As istical Discrimination. American Journal of
already indicated, however, we prefer to evalu- Sociology 91, 759-799.
ate the total wage effect of FILMs on the basis Blau, P M. 1968. The Hierarchy of Authonty in
of the results of model B, that is, without controls Organizations. American Journal of Sociology
for other organizational traits and job charac- 73, 453-467.
teristics. In our view, it makes no sense to control Blau, P. M. 1970. A Formal Theory of Dif-
for these characteristics, when discussing the two ferentiation in Organizations. American Socio-
hypotheses on how FILMs affect earnings, even logical Review 35, 201-219.
if some of the total wage effects of FILMs may Blau, P. M. & Schoenherr, R. A. 1971. The
be explained by FILMs being more prevalent in Structure of Organizations. New York: Basic
large organizations and in organizations with a Books.
tall and lean hierarchical structure. In a similar Brown, C. & Medoff, J. 1989. The Employer
manner of reasoning, one should not control for Size-Wage Effect. Journal of Political Econ-
positional attributes, such as the possession of omy 97, 1027-1059.
specific job skills, if they are an outcome of FILM Bums, T. & Stalker, G. M. 1961 The Manage-
arrangements. ment of Innovation. London: Tavistock.

248
Child, J. 1984. Organization. London: Harper Power, and Income Inequality American
and Row. Journalof Sociology 87, 651-683.
T. 1986. Dividers in the Labor
Colbj&oslash;rnsen, Kalleberg, A. L. & Berg, I. 1987. Work and
Market. Oslo: Norwegian University Press Industry. Structures, Markets, and Processes.
Crozier, M. 1964. The Bureaucratic Phenom- New York: Plenum Press.
enon. University of Chicago Press. Kalleberg, A. L. & Leicht. K. T. 1986. Jobs and
Doeringer, P. B. & Piore, M. J. 1971. Internal Skills: A Multivariate Structural Approach.
Labor Markets and Manpower Analysis. Social Science Research 15, 269-296.
Lexington, Mass.: Heath Lexington. Kalleberg, A. L. & S&oslash;rensen, A. B. 1979 The
Edwards, R. C. 1979. Contested Terrain. The Sociology of Labor Markets. Annual Review
Transformation of the Workplace in the Twen- of Sociology 5, 351-379.
tieth Century. New York: Basic Books. Kalleberg, A. L. & Van Buren, M. E. 1993. Is
England, P. & Farkas, G. 1988. Economic and Bigger Better?: Explaining the Relationship
Sociological Views of Industries, Firms, and between Organization Size and Job Rewards.
Jobs. In G. Farkas & P. England (eds.), Indus- Paper presented at the 1993 meeting of the
tries, Firms, and Jobs, pp. 331-346. New York: Amencan Sociological Association. Miami.
Plenum Press. Kerr, C. 1955. The Balkanization of Labor Mar-
Erikson, R. & Goldthorpe, J. H. 1992. The Con- kets. In C. Kerr (ed.), Labor Markets and
stant Flux: A Study of Class Mobility in Indus- Wage Determination. Berkeley: University of
trial Societies. Oxford: Clarendon. California Press.
Goldthorpe, J. H. 1982. On the Service Class, Kimberly, J R 1976, Organizational Size and
its Formation and Future. In A. Giddens & the Structuralist Perspective: A Review. Cnt-
G. Mackenzie (eds ), Social Class and the Div- ique, and Proposal Administrative Science
ision of Labour. Cambridge University Press. Quarterly 21, 577-597.
le Grand, C 1991. Explaining the Male-Female Kohn, M. L. 1971. Bureaucratic Man: A Portrait
Wage Gap. Job Segregation and Solidarity and an Interpretation. American Sociological
Wage Bargaining in Sweden. Acta Sociologica Review 36, 461-474
34, 261-278. Krueger, A. B. & Summers, L H. 1988.
le Grand, C. 1993. Karnar- och utvecklings- Efficiency Wages and the Inter-industry Wage
mojligheter p&aring; de interna arbetsmarknaderna Structure Econometrica 56, 259-293.
(Career Opportunities and Job Enlargement Lawrence, P. R & Lorsch, J. W. 1969 Organ-
Within Internal Labour Markets) In C. le ization and Environment. Homewood,
Grand. R Szulkin & M. T&aring;hlin (eds.), Sveri- Illinois: Richard D Irwin, Inc.
ges Arbetsplatser Organisation, personalut- Levine, D I. 1993. What Do Wages Buy?.
veckling, styrning (Swedish Establishments Administrative Science Quarterly 38, 462-483.
Organization, Personnel Development, Con- Lincoln, J R. & Kalleberg, A. L. 1990 Culture,
trol), pp. 48-83 Stockholm: SNS. Control and Commitment A Study of Work
le Grand, C., Szulkin, R. & T&aring;hlin, M. 1993. Organization and Work Attitudes in the United
Arbetsplatsstrukturens inverkan p&aring; indi- States and Japan. Cambridge: Cambndge Uni-
vidernas loner (The Impact of Organizational versity Press.
Structure on Individual Earnings) In C le Littler, C. R. 1982 The Development of the
Grand, R. Szulkin & M T&aring;hlin (eds.), Sveri- Labour Process in Capitalist Societies.
ges Arbetsplatser Organisation, personalut- London: Heinemann.
veckling, styrning (Swedish Establishments. Miller, G. A. 1987. Meta-analysis and the Cul-
Organization, Personnel Development, Con- ture-free Hypothesis Organization Studies 8,
trol), pp. 196-221 Stockholm: SNS. 309-326.
Hashimoto, M. & Raisan. J. 1985. Employment Mincer. J. 1974 Schooling, Experience and
Tenure and Earnings Profiles in Japan and the Earnings. Columbia University Press.
United States. American Economic Review 75, Mintzberg, H. 1979. The Structuring of Organ-
721-735. izations. Englewood Cliffs, New Jersey: Pren-
Hedstrom, P. 1991 Organizational Differenti- tice-Hall.
ation and Earnings Dispersion American Osterman, P. 1985 Introduction: The Nature
Journal of Sociology 97, 96-113 and Importance of Internal Labor Markets. In
Johnson, S. 1991. The Small Firm and the UK P. Osterman (ed.), Internal Labor Markets
Labour Market in the 1980s. In A. Pollert Cambridge, Mass.: MIT Press.
(ed.), Farewell to Flexibility? pp. 239-258. Parsons, D. O. 1986. The Employment Relation-
Oxford. Basil Blackwell. ship : Job Attachment, Work Effort, and the
Kalleberg, A. L., Wallace, M. & Althauser, Nature of Contracts. In O. C. Ashenfelter &
R. P. 1981. Economic Segmentation, Worker R. Layard (eds.), Handbook of Labor Eco-

249
 nomics, Vol. II, pp 789-848. Amsterdam: Adjustment During Army Life Princeton NJ
North Holland Princeton University Press.
Perrow, C 1986. Complex Organizations. A Szulkin, R 1993 Organisationers struktur (The
Critical Essay. New York: Random House Structure of Organizations). In C. le Grand
Pfeffer, J. & Cohen, Y. 1984. Determinants of R Szulkin & M. T&aring;hlin (eds.), Sveriges
Internal Labor Markets in Organizations. Arbetsplatser Organisation, personalutveck
.
Administrative Science Quarterly 29, 550-572. ling, styrning (Swedish Establishments.
Piore, M. J. & Sabel, C. F. 1984. The Second Organization, Personnel Development, Con-
Industrial Divide Possibilities for Prosperit
y trol), pp 17-47. Stockholm- SNS
New York Basic Books. Treiman, D J & Roos, P. A 1983. Sex and
Pugh, D. S et al 1969. The Context of Organ- Earnings in Industrial Society. A Nine-Nation
ization Structures. Administrative Science Comparison American Journal of Sociology
Quarterly 14, 91-114. 89, 612-050
Rainnie, A. 1985 Is Small Beautiful ? Industrial Villemez, W J & Bridges, W P 1988 When
Relations in Small Clothing Firms. Sociolog
y Bigger is Better:Differences in the Individual-
19, 213-224. Level Effect of Firm and Establishment Size
Rainnie, A. 1989. Industrial Relations in Small American Sociological Review 53, 237-255.
Firms. Small isn’t beautiful London, New Williamson. O E 1967 Hierarchical Control
York: Routledge. and Optimum Firm Size Journal of Political
Rosenfeld, R. A. & Kalleberg, A. L. 1990, A Economy 75, 123-138.
Cross-National Comparison of the Gender Woodward, J. 1965. Industrial Organization
Gap in Income American Journal of Sociology Theory and Practice. London: Oxford Uni-
96 , 69-106 versity Press.
Stolzenberg, R. M. 1978 Bringing the Boss Back Worthy, J C. 1950 Organizational Structure
in: Employer Size, Employee Schooling, and and Employee Morale. American Sociological
Socioeconomic Achievement American Review 15, 169-179.
Review 43, 813-828.
Sociological Wright, E O. 1985. Classes. London Verso.
Stouffer, S. A. et al. 1949 The American Soldier

Appendix: Interview questions and

descriptive statistics
Forma/¡zarlOfI Additive index based on answers (2) Do you usually try to fill vacancies mternally.
’

of the workplace manager to the following ques- before advertising m any other way?
tions : (0 = No, 1 = Yes)

(1) To what extent are daily of the

activities Investments m on-the-job tramng.
workplace governed by wntten rules’? (0 Not =
(1) How common is it for employees (manual
at all, 1
To a =
small extent, ? 2 = To some extent, workers, low or mid-level white-collar
3 To a large extent).
=
employees, higher whUe-collar employeesl
(2) To what extent are the work tasks specified gradually to achieve more and respon-
qualified
in advance according to rules and regulations sible tasks within the framework of their jobs.
which must be followed‘’ (0 Not at all, 1
=
To =
i.e., without being promoted? (0 Very uncom-
=

a small extent, 2 To some extent, 3 To a

= =
mon, 1 Rather
=
uncommon, 2 = Rather
large extent). common, 3 Very common)
=

(2) Does it happen that previously unskilled

Firm Internal Labour Market (FILM) Additive (manual workers, white-collar employees)
index based on answers of the workplace man- through mternal tramng get jobs that require
ager or the personnel manager to the hollowing skills? (0 =
Never, 1 Seldom, 2 Yes, some-
= =

questions times, 3 =
Yes, often).

Rates of promotion and internal recruitment Positional vs individual wage setting:

(1) How many (manual workers, low or mid- ( 1 ) Apart from the employee’spersonal charac-
level white-collar employees, higher white-collar tenstics, to what extent do wages vary between
employees) at this workplace have been pro- (manual workers, low or mid-level white-collar
moted m the past three Vears’~ (This number is employees, higher white-collar employees),
divided by the total number of employees m the dependmg only on:
organization,.) (a) job complexity ?

250
(b) supervisory responsibilities? (1)what tasks to carry out?
(c) autonomy of work tasks? (2) how you carry
out your tasks9
(0 Not at all, 1 To a small
= =
extent, 2 To =
(0 Not at all, 1
=
To a small extent, 2
= =
To
some extent, 3 To a large extent)
=
some extent, 3 To a large extent).
=

(2) To what extent do wages vary between

(manual workers, low or mid-level white-collar General
job skills. Do you know of any other
employees, higher white-collar employees) dep- employers where you would have good use for
ending on: what you’ve learnt in your present job? (0 =

(a) experience acquired within the firm/ No, 1 =

Yes, a few 2 =
Yes, some 3 =
Yes,
workplace? many).
(b) general occupational expenence’?
(0 Not at all, 1 To a small extent,
= =
2 =
To Specific Job skills. In addition to skill demands at
some extent, 3 =
To a large extent)> job recruitment, how long does it take to learn
to perform your job reasonably well? (0: ; 1 day,
Alllollomy. Additive index based on two ques- 1: 2-5 days, 2: 1-4 weeks, 3: 1-3 months, 4: 3
tions to individual employees months-1 year, 5: 1-2 years, 6: more than 2
To what extent do you have influence over ... years).

Means and Standard Deviations

251