Content uploaded by Sanford M Jacoby
Author content
All content in this area was uploaded by Sanford M Jacoby on May 04, 2019
Content may be subject to copyright.
Enterprise Unions in the United States
SANFORD
M.
JACOBY
and
ANIL
VERMA*
Through
a
case study
of
a
large industrial company (TRW), this paper
examines the history and functioning
of
independent local unions
(It
Us).
TRW’s
ILU
plant wages were about the same
as
those at afiliated union
plants and higher than those at nonunion plants. The premium explains
why TRW and other companies discarded
ILUs
in
favor of
a
“new”
non-
union model. Yet
ILUs
can remedy some
of
the defects in that model and
thus belong
on
the policy and research agenda.
DURING
THE
LAST
TWENTY
YEARS,
employee involvement pro-
grams (EIPs) have made significant inroads in large American companies.
Employers tout EIPs as integrative approaches to employee relations that
offer workers new channels for participation in decision making and that
encourage cooperation with management in improving productivity. Over
one-half of the large firms surveyed by the New York Stock Exchange
(1982)
said that they had “taken a shift in their basic approach to manage-
ment” (p.
24)
by introducing various participative programs. These run the
gamut from quality circles to teams and more complex participation
schemes. EIPs initially took hold in the nonunion sector and are now
spreading to unionized firms as well. Their basic premise, however, runs
counter to that
of
the traditional adversarial model of collective bargain-
ing.
As
a result, making EIPs fit into unionized workplaces has not been
easy. Critics contend that
EIPs
sap worker interest in unionism, and for
this reason, suspicion of them lingers within the labor movement. Many
of
the most elaborate and far-reaching EIPs are found today at nonunion
companies in high-tech and other industries. Complementing the EIP phi-
*The authors’ affiliations are, respectively, Anderson Graduate School of Management, UCLA; and
Faculty
of
Management and Centre for Industrial Relations, University
of
Toronto. The authors wish
to
thank Stanley Engerman and Jeffrey Pfeffer
for
helpful comments and Maury
Pearl
for
research
assistance
on
this project. Funding was provided by the UCLA Institute
of
Industrial Relations.
INDUSTRIAL RELATIONS,
Vol.
31,
No.
1
(Winter
1992).
0
1992
Regents of the University
of
California
137
138
/
SANFORD
M.
JACOBY
AND
ANIL
VERMA
losophy are a host
of
personnel policies, such as career development,
employment security, and complaint systems (Foulkes, 1980).
Although the new nonunion model is a formidable achievement, it
is
not
without problems. Employee participation occurs chiefly at the
job
level;
more vertical forms
of
influence are, with
a
very few exceptions, limited to
voice mechanisms that leave hierarchical power relations unchanged. In-
deed, EIPs ignore power issues or individualize them. Moreover, employ-
ees
in new-model nonunion firms lack independent channels for articulating
their concerns and must rely on those provided by management. Although
managers often try to be fair, employees are unaccustomed to or fearful
of
expressing their concerns and dissatisfactions. Decisions that are supposed
to be made with employee involvement are still made unilaterally. Com-
plaint systems-even in the most progressive nonunion companies-fail to
protect against stigmatization and retaliation (Heckscher, 1988; Lewin and
Peterson, 1988).
Solving these problems requires some kind of autonomous representa-
tion structure to handle individual employee complaints, coordinate EIPs,
focus and articulate group concerns, and bring these to management’s
attention. Obviously, one form this could take is representation by a na-
tional union, such as an affiliate
of
the AFL-CIO. That option should
always be readily available to employees, but
so
far national unions have
failed to gain significant support from employees in progressive nonunion
firms. Neither industrial solidarity nor the conflictual ethos of national
unionism meshes with the communitarian culture and decentralized EIPs
found in these firms, and it appears unlikely that national unions will ever
gain significant support from employees in these firms. Given the fact that
nearly
87
percent
of
private-sector workers currently lack independent
representation, it is time to consider some alternatives.
Industrial relations scholars are beginning to study institutional forms
that could remedy defects in the progressive nonunion model. Most of
these involve some blend of the integrative and adversarial approaches,
hybrids designed to be grafted onto workplaces that make extensive use of
EIPs and related policies. Some are entirely new forms
of
representation,
such as Heckscher’s (1988) associational unionism, while others are bor-
rowed from abroad, including the works councils and enterprise unions
found in Western Europe and Japan (Wever, 1988; McCormick,
1987).
The search for new models is laudable and important, but it is also worth
looking more closely at domestic alternatives
,
such as independent local
unions (ILUs), which have existed for many years in the United States and
whose features overlap those of recently proposed models. ILUs may
be
a
Enterprise Unions in the United States
I
139
possible remedy for some of the problems that currently exist in American
workplaces, even, or especially, the most progressive ones.
Independent
Local
Unions
Independent local unions (ILUs) are single-employer, unaffiliated
unions that usually represent workers at a single plant or facility.’ ILUs are
often associated with employee representation plans started by employers
in the
1920s
and early
1930s.
A majority
of
the
ILUs
surveyed in
1967
were
organized between
1936
and
1945.
Many of these were representation
plans that reconstituted themselves because
of
section
8(2)
of the Wagner
Act, which made it an unfair labor practice for an employer to dominate,
interfere with, or provide financial assistance to a labor organization. Un-
der intense scrutiny from the National Labor Relations Board (NLRB),
these former company unions now became financially and otherwise inde-
pendent of employers and added other features “modeled after trade
union organization and procedure” (Saposs,
1936,
p.
803;
Rosenfarb,
1940).
In cases where these unions still received support from an employer
or had not established
a
clean break with an employer-supported predeces-
sor organization, the NLRB ordered their disestablishment and barred
them from appearing on election ballots (Rosenfarb,
1940;
Crager,
1942).
This was a severe remedy, one that the NLRB did not apply to tainted
locals of an affiliated union, a “sweetheart” situation. An affiliated union
that received illegal employer support could appear on the ballot
if
it could
be proven that the employer had ceased its support.
The rigid application
of
section
8(2)
to unaffiliated unions stemmed
from the early NLRB’s determination to hasten the demise of company
unions, even if it meant stigmatizing those ILUs that were bona fide inde-
pendent unions supported by their members. The NLRB thought that
stringency was necessary in order to protect workers who supported a
union out
of
a “mistaken continued belief that the company-dominated
organization affords a genuine agency for collective bargaining” (U.S.
Senate,
1947,
p.
1912).
It rationalized its more lenient treatment of affili-
ated union locals on the dubious grounds that an affiliated local-as part
of a large and powerful national union-could not actually be dominated
by the employer. During the
1940s,
the NLRB’s stance in these cases came
under growing criticism from Congress and the courts. As a result, around
‘In
1967,
92
percent
of
unaffiliated intrastate and single-employer unions had only one local
(BLS,
1967).
140
/ SANFORD M.
JACOBY
AND
ANIL
VERMA
1944
it began to treat ILUs more leniently (Millis and Brown,
1950).
This
was not enough to deter including the stipulation in the
1947
Taft-Hartley
Act that the NLRB was not to discriminate between affiliated and unaffili-
ated unions when deciding section 8(2)-now 8(a)(2)-cases.
After passage of the act, the NLRB took a more permissive approach to
unaffiliated unions. In section 8(a)(2) cases it now paid
no
attention to
affiliation status. Instead, the distinction applied was that between em-
ployer domination-the remedy for which continued to be disestablish-
ment-and lesser forms of illegal employer interference and support,
which, if withdrawn by the employer, permitted a union, including an ILU,
to appear
on
the ballot. NLRB chair Paul Herzog supported this approach
by noting, “This is
1947,
not
1935;
in the interim employees have learned
much about protecting their own rights and making their own choices with
the full facts before them” (Schwab,
1957,
p. 356).*
One result of this greater leniency was an increase
in
the formation
of
ILUs in the late
1940s
and
1950s
(Troy,
1961).
Now, however, ILUs ap-
peared at smaller firms: Those established between
1947
and
1960
ac-
counted for 52 percent of the total number of ILUs surveyed in
1967
but
only 26 percent of their total membership (see Table
1).
After
1960,
how-
ever, the pace
of
ILU formation slackened, then slowed to
a
trickle during
the
1970s
and
1980s.
ILUs organized in these years are usually the result of
a decertification or a local’s opting out
of
a union merger.
In
1983,479,000
workers belonged to some
1,500
ILUs, accounting for about
3
percent of
total U.S. union membership. Today, ILUs are found at a few large
multiunit companies (Du Pont, Texaco,
Exxon,
AT&T, Procter
&
Gam-
ble), several medium-size firms (Dow-Jones, Weirton Steel, Zenith), and
numerous small firms in a variety of industries.
Ironically, a spate
of
studies about ILUs was published in the early
1960s,
just when ILUs had begun to lose their organizational momentum (Brandt,
196O;Troy,
1961;
Marshall,
1961;
Shostak, 1962; Collins,
1964).
Their subse-
quent fading from the industrial relations research agenda was unfortunate
because ILUs have the potential to fill the representation gap that exists in
large nonunion companies in the United States. As enterprise-oriented
*The
NLRB
did not shift entirely to
a
position
of
letting employees make these choices
on
their
own.
In
a
1957
decision, an appeals court overturned the
NLRB’s
disestablishment
of
a shop committee and
then criticized the board by saying that the law “does not make
it
the duty
of
the employer,
nor
a
function of the board, to ‘baby’ along the employees in the direction
of
choosing an outside union as
their bargaining representative”
(Coppus
Engineering,
1957,
p.
2321).
This issue-whether workers
have the experience and sense to judge employer claims and other industrial relations issues-
resurfaced
in
the controversy over the
Law
and
Reality
study (Getman et al.,
1976;
Weiler,
1983)
and
in recent debates over the legality
of
EIPs.
Enterprise Unions
in
the United States
I
141
TABLE
1
PERIOD
OF
ORGANIZATION
OF
1967
ILUs
AND
ILU
MEMBERSHIP
IN
SELECTED YEARS
%
of
%
of
Membership
Members
in
1967
ILUs
in
1967
(in
thousands)
Pre-1930
8
1
1930-35 4 4
1936-39 32 17
1939 163
1940-45
28
19
1947 469
1946-60 26 52
1961 452
1961-67 2 7
1967 475
19x3 479
SOURCES:
U.S. Bureau
of
Labor Statistics. “Unaffiliated Intrastate and Single-Employer Unions,
1967.”
Bulletin
No.
1640
(Washington,
DC,
1969),
pp.
I,
14;
Leo
Troy.
“Local Independent Unions and
the
American
Labor
Move-
ment,”
Indusrrial and
Labor
Relations Review
14
(April
1961):339:
Leo Troy and Neil Sheflin,
U.S.
Union
Sourcebook
(West Orange, NJ:
IRDIS,
1985).
pp.
3-7.
organizations, they tend to fit with the insular and communitarian culture
found in these companies. At the same time, however, they are unions, with
their own officers and funds; unlike
EIPs,
they are autonomous, possess
independent resources, and are capable of providing influence at higher
levels than the job or shop floor. Recently, there has been growing interest
in new forms of employee representation. In
1989,
the AFL-CIO’s Commit-
tee on the Evolution
of
Work heard several proposals calling for state or
federal legislation that would require employees in almost all companies to
elect members of “Worker Representation Committees,” which would func-
tion like limited
ILUs.
Oregon and Washington presently have, and the
federal government is considering, legislation that establishes joint labor-
management health and safety committees in all workplaces, union as well
as nonunion. Given these developments, it is appropriate to again take a
closer look at ILUs-their rationale, achievements, and limitations.
The earlier literature debated whether ILUs were capable of aggressive
bargaining with management. Some argued that ILUs were “holes,” sham
unions started by employers who wished to keep their labor costs down by
giving employees the
form
but not the substance of unionism (Taft,
1956).
Others agreed that management preferred ILUs over national unions be-
cause ILUs were more cooperative and struck less often,
if
at all, but that
companies paid for this peace with wages and benefits that matched those
negotiated by national unions. This was what national union leaders called
the “golden handcuffs” technique: Keep workers satisfied with their ILUs
142
/
SANFORD
M.
JACOBY
AND
ANIL
VERMA
by reducing the incentive for them to join more adversarial, national
unions. On the other hand, it was asserted that not all ILUs were passive,
that some possessed significant bargaining power, exercised through peri-
odic threats to strike or to affiliate with a national union.
A
variant of this
was Troy’s
(1960)
claim that ILU members eschewed national unions be-
cause ILUs preserved skill or geographic wage differentials that a national
union might compress and standardize. In short, the debate revolved
around alternate hypotheses concerning the ILU wage effect: Did it exist
and
if
so,
why?
To test these hypotheses and to revive scholarship in this area,
we
stud-
ied ILUs at TRW Inc., which, before its merger with Ramo-Wooldridge in
1958,
was known as Thompson Products, an automotive and aircraft parts
company. In the
1940s
and
1950s
the company was a leading practitioner of
independent local unionism. But after
1960
its industrial relations strategy
changed, and today TRW is best known as an exemplar of the new non-
union model (Kochan et al.,
1986).
Its nonunion plants are state-of-the-art
workplaces, replete with team system, pay-for-knowledge plans, and em-
ployee participation and communication programs, such as attitude sur-
veys and “sensing” sessions (Ziskin,
1986;
Oates,
1973).
A
case study has obvious limitations, but we feel this approach is war-
ranted by the paucity
of
current research and aggregate data
on
ILUs.
Moreover, TRW is not just another company, but one of the leading develop-
ers
of
EIPs and other advanced nonunion workplace systems. It also still
has
some plants that are represented by national unions and others that are
represented by ILUs. Therefore, the company provides an unusual opportu-
nity to compare and contrast institutional forms (ILUs, national unions, and
the new union model) within the same organizational context. First, we give
a history
of
labor relations at TRW from
1930
to the present. Next, we
analyze pay differentials between the company’s national union, ILU, and
nonunion plants during the period
1979-83.
Finally, we discuss these results
in light
of
the earlier debate about ILUs as well as the more recent search for
new models
of
workplace representation.
ILUs
at
Thompson
Products
and
TRW
Founded in
1901
in Cleveland, Ohio, Thompson Products developed
during World War I1 into a major manufacturer
of
automotive and aircraft
parts. After the war, the company began to diversify through mergers and
acquisitions, most
of
them in the Midwest and Northeast. The process
accelerated in the
1960s
and
1970s,
as the company spread throughout the
Enterprise Unions in the United States
I
143
country and abroad. Today, TRW is a sizable conglomerate, with 75,000
employees worldwide and headquarters in Cle~eland.~
Prior to the 1930s Thompson Products was a relatively small firm that
had not yet adopted modern techniques of personnel management. In
1933 Frederick
C.
Crawford became Thompson’s president and took
steps to modernize personnel practices and make the firm impervious to
the national unions that had begun to stir in Cleveland. Considering
national unions anathema, Crawford pursued dramatic anti-union tactics
over the next twenty years (including numerous clashes with the NLRB
and the War Labor Board, especially over the free speech issue), which
made him a prominent figure among employers and brought him the
presidency
of
the National Association
of
Manufacturers. Crawford hired
Raymond
S.
Livingstone in 1933 to direct a new corporate personnel
department. Livingstone immediately established various employee wel-
fare programs and formalized the company’s personnel procedures. In
1934 he started employee representation plans at the company’s major
plants in Detroit and Cleveland. Until the late 1950s, company unions
continued to be established at all
of
Thompson’s new and existing domes-
tic plants. Combined with progressive personnel policies and an aggres-
sive management, these company unions proved resistant to national
unions; only two Thompson ILU plants were organized
by
national
unions during those years.
The largest
of
Thompson’s ILUs was the Aircraft Workers Alliance
(AWA), which represented workers at the firm’s two main Cleveland
plants until 1984, when the plants were sold. Although the AWA began in
1934
as
a
representation plan controlled by management, it gradually be-
came a more aggressive and independent organization. Impelling this evo-
lution were various factors: the NLRB’s enforcement
of
the Wagner Act,
competition with national unions (the UAW was a constant presence at the
Cleveland plants between 1937 and 1947, during which time it lost
six
elections), and a management that was willing to give the AWA real power
and influence in order to deter “outside” unionism. By 1947, when the
appellate courts held that the AWA was lawful and not a dominated or
illegally supported organization, it had become far more autonomous and
active than it was in the 1930s and continued in this mode through the
1970s. Although it made a point
of
its strike-free record, the AWA was
willing
to
pressure management during wage negotiations, and a strike
3This history
is
based
on
material at the Harvard Business School and at the Western Reserve
Historical Society; see Jacoby
(1989)
and Shore
(1966).
144
/
SANFORD
M.
JACOBY
AND
ANIL
VERMA
occurred over wage issues in
1979.
This is not to say that the AWA came to
resemble an affiliated local union in all respects. It was highly loyal to the
company and usually took a cooperative, problem-solving approach in its
dealings with management. A plant council held monthly meetings to
discuss issues and problems, and the AWA regularly participated in a
variety of union-management committees.
On
the other hand, AWA leaders were susceptible to manipulation
by
management. The AWA had few contacts with the company’s ILUs out-
side Cleveland and had to rely on a hired attorney for technical expertise.
Therefore, it was forced at times to turn to management for information
and other assistance, which management willingly provided. This blurred
the line between the company and the union, creating what the UAW
called “a cozy relationship.” But for a variety
of
reasons, workers pre-
ferred the AWA to the UAW. They had good pay and working conditions,
which they received for only a fraction of the cost
of
UAW membership.
Also, the AWA offered a less bureaucratic form
of
representation than a
national union. All its officers came from the plant and were known by the
members. Elections for union office were contested and publicized. Unlike
an affiliated union local, the AWA did not have to answer to a national
union or conform to policies it did not choose. Its small size and lack
of
staff made it less formal and legalistic than a national union.
The AWA’s organizational structure and formal mechanisms for dealing
with management were reproduced by the company’s other
ILUs,
al-
though the precise mixture of adversarialism and cooperation varied from
one ILU to another. Detailed data
on
industrial relations outcomes during
the
1970s
are available for three unionized TRW plants (see Table
2).
Two
were represented by
ILUs
organized in the
1950s,
and one was repre-
TABLE
2
INDUSTRIAL
RELATIONS
AT
THREE
TRW
PLANTS,
1971-80
Pennsylvania Ohio Connecticut
ILU ILU
UAW
Plant sizea
Strikes
Wildcats
Grievance rateh
1971-75
1976-80
Grievances
to
arbitration
965
0
0
1.3
1.9
1
878
2
1
6.3
4.2
80
68
1
2
1
10.4
24.8
21
aAverage
number
of
production
wcirkers.
bAverage number
of
grievances
filed
pcr
I00
workers
Enterprise Unions in the United States
I
145
sented by the UAW. The plants were similar in some respects. All had
blue-collar workforces, were roughly the same size, and were located in
small towns. But the similarities stopped there. The Pennsylvania ILU was
a relatively quiescent organization that never struck, filed few grievances,
and only once in ten years had contested a grievance
all
the way to arbitra-
tion. The Ohio ILU (not the AWA) was more aggressive and in most
respects resembled the UAW local in Connecticut. During the
1970s,
the
Ohio ILU and the UAW local had the same number of strikes and wild-
cats, and although the UAW local filed more grievances, the Ohio ILU
pressed a greater number through to arbitration. We do not know why the
Ohio and Pennsylvania ILUs differed, but others have reported similar
variations in ILUs: Some are strong and others are weak, both within and
across companies (Shostak,
1962).
TRW began to change its industrial relations strategy around
1960.
Al-
though ILUs were organized at seven out
of
the eight plants built or
acquired by the company during the
1950s,
not a single ILU was organized
at any
of
the
66
plants built or acquired during the
1960s
and
1970~.~
Many
of these new plants were located in the South or Southwest, and a majority
of them
(60%
of
the plants added in the
1960s
and
95%
of
those from the
1970s)
were unorgani~ed.~ It was in these plants that the company per-
fected its new nonunion model. The reasons for TRW’s shift away from
ILUs are not hard
to
find. First, there was always the danger that an ILU
would affiliate with a national union or become as adversarial as an affili-
ated local (like the Ohio ILU shown in Table
2).
Second, collective bargain-
ing and contract administration are time-consuming procedures that limit
management’s ability to make prompt and unquestioned decisions.
Em-
ployers were willing to shoulder these burdens when national unions posed
a threat, but by the early
1960s
that threat had begun to fade, and, at the
same time, a new nonunion workplace model became available, one that
combined some
of
the voice features
of
company unionism with EIPs and
other programs for group flexibility and participation (Kochan et al.,
1986).
Finally, there was the possibility that TRW could cut its labor costs
by using nonunion labor if, as some alleged, it paid wage premiums to
JAn
ILU
organized by the AWA in a warehouse operation in 1962 was more an extension of
AWA
operations in Cleveland than a fully independent and separate ILU.
5If
we
exclude the acquired plants whose union status was determined prior to acquisition by
TRW,
only two
of
25 plants opened by the company during the period 1960-83 were ever unionized. An
Illinois plant was organized in 1967 by the Seafarers’ International Union, whose influence at the time
extended
to
areas around the Great Lakes. A second plant producing electronic components in Texas
was organized in 1975 by the International Union of Electrical Workers
(IUE).
The company opposed
both these attempts, with the Texas campaign being made a priority by the company. As a
result,
the
union lost the first election, and it took more than one campaign to get certified.
146
/
SANFORD
M.
JACOBY
AND
ANIL
VERMA
workers represented by
ILUs.
This was the central issue in the earlier
debate on ILUs-whether there was an ILU wage effect.
Analysis
of
Compensation
Costs
For our analysis, we collected data on wage rates and benefits for all
84
of
TRW’s domestic plants for the period 1979-83 (see Table 3)6;
of
the
84
plants,
56
were nonunion, 9 had independent unions, and the remaining 19
were organized by affiliated unions. The four measures of compensation,
according to union status, for the years 1979-83 were: (1) Average Rate:
the average
of
all wage rates paid in the plant weighted by the number
of
TABLE
3
AVERAGE COMPENSATION
AND
PLANT
CHARACTERISTICS
NU’
IU
AU
NU
IU
AU
NU
IU AU
(1979) (1983)
(%
change
1979-83)
Compensationb
Average Rate
Low Rate
High Rate
Total Benefits
nance workers
No.
of
production and mainte-
Plant Age
Skill Mix
Total
no.
of
plants
(yrs.1
(%
skilled)
(by region)
Midwest
Northeast
Southwest
Southeast
Total no.
of
plants
(by industry)
Electronics
Machinery
Metals
Transportation
Other
6.04 8.26 7.25 8.34 10.94 10.02 38.1 27.0 38.2
4.28 7.77 5.97 5.42 9.77 7.49 26.6 25.7 25.5
8.41 9.62 8.81 11.85 12.31 11.63 40.9 28.0 32.0
2.20 3.82 2.88 3.38 6.35 4.65 53.6 66.2 61.5
250 786 591 197 535 278 -21.2 -31.9 -44.5
17.3 36.2
50.1
15.2 13.5 14.3
56
9
19
16
5
7
6 4 8
17
0
2
17
0
2
21
1
6
11 2 8
11
1
2
10
4 3
3 1
0
aNU
=
56
nonunion plants;
IU
=
9
independent unions;
AU
=
19
affiliated union
plants.
hWeighted across plants
by
plant size, expressed
in
the
number
of
production and maintenance
workers
per
plant
6Data
for
later years are not available as the company launched a major divestment program in
1984,
and data from two plants that were
sold
during the study period are excluded.
Enterprise
Unions
in the United States
I
147
employees at each wage rate; (2) Low Rate: the wage rate for the lowest
paid occupation in the plant;
(3)
High Rate: the wage rate for the highest
paid occupation in the plant; and
(4)
Total Benefits: the cost
of
health and
pension benefits, other company benefits, legally required benefits, and
pay for time not worked. Benefit costs and wage rates are expressed in
dollars per hour, a figure obtained by dividing actual benefits for the year
by a standardized work year of 2,080 hours.
In 1979 and 1983, compensation in ILU plants was higher than in
either affiliated union or nonunion plants. Perhaps because
of
their al-
ready high levels, Average Rate and High Rate rose more slowly in the
ILU plants over the period as compared with the increase in affiliated
union plants and in nonunion plants. The increase in Low Rate was
roughly the same in all three sectors, while benefits rose most sharply in
the ILU plants.
We also examined wage differentials, using multivariate tests, to see if
the differences were significant after controlling for other effects. Our
wage model (see Table
4)
is specified on the basis of factors identified in
past studies. Evidence shows that Plant Size, an indicator of employer
ability to pay, has a positive impact
on
wages (Lester, 1967; Masters, 1969;
Personick and Barsky, 1982); it is measured by the number
of
production
and maintenance workers. Plant Age is used as a proxy for the age and
experience profile of the workforce; older plants are likely to have an older
workforce, which is compensated at a higher level. The age coefficient,
therefore, is expected to be positive.
ILU
plants were older than nonunion
plants, but younger than affiliated union plants. (Plant Age is measured in
1983 from date
of
construction for both company-opened and acquired
plants.) Following human capital theory,
a
measure of skill level in the
plant is entered as a control-Skill Mix, defined as the percentage of
skilled workers in the plant. This was similar across the three sectors; the
observed differences are not statistically significant.
Some studies have also shown that union-nonunion wage differentials
vary across regions (Personick, 1974),
so
dummies are included in the
specification to control
for
regional variations in living costs and labor
markets; the Northeast is the reference group. The geographic distribution
of the ILU plants was similar to the affiliated union plants; both were
concentrated in the Northeast and Midwest. Wages also vary across indus-
tries (Kochan and Block, 1977),
so
industry dummies are also included;
they are specified with transportation as the reference group.
To
measure
differential union effects, dummy variables are entered separately for Af-
filiated Union plants and Nonunion plants, with ILU plants as the refer-
ence group.
148
/
SANFORD
M.
JACOBY
AND
ANIL VERMA
TABLE
4
WEIGHTED LEAST SQUARES REGRESSIONS
ON
AVERAGE
RATE
(standard
errors
in
parentheses)
Pooleda
Variable
1979 1980 1981 1982 1983 1979-83
Plant Sizeb
Skill Mix
Plant Age
Affiliated Union
Nonunion
Region
Southeast
Southwest
Midwest
Industry
Metals
Machinery
Electronics
Other
Constant
R2
N
F-test
0.301 7
*
(0.0292)
0.0023
(0.0153)
0.0114
(0.0085)
(0.5298)
(0.5282)
0.2804
(0.5823)
1.5966*
(0.6505)
0.6376
(0.4141)
0.1592
(0.4722)
(0.3708)
-2.5897*
(0.4375)
(
1
.O
169)
14.9604
*
(
1.3403)
0.9508
62
78.854*
-
0.0599
-
1.1744*
-0.2403
-0.2272
0.1909*
(0.031 7)
0.0079
(0.0131)
0.0079
(0.0082)
(0.5421)
-
1.4976*
(0.5016)
0.2900
(0.5392)
1.6563*
(0.6085)
0.7512
(0.4047)
-0.6227
-0.4521
(0.4932)
-0.9225"
(0.4226)
-
3.2620'
(0.4202)
(0.9073)
13.4986*
(1.0664)
0.8400
62
21.440*
-0.6777
0.3358*
(0.0351)
0.0157
(0.0126)
0.0071
(0.0082)
(0.5375)
(0.4825)
-
0.2824
-
1.3630*
-0.0926
(0.5576)
1.7042"
(0.6054)
0.6537
(0.423
1)
-
0.6907
(0.5157)
(0.4479)
(0.4435)
(0.7611)
14.1507*
(1.1
116)
0.8963
65
37.434'
-0.7819
-
3.5760*
-0.2027
0.5622*
(0.0509)
0.0075
(0.0173)
0.0099
(0.0099)
0.3188
(0.6925)
(0.6090)
-
1.3549%
-
0.2302
(0.7126)
2.5717*
0.7459
(0.4963)
(0.7454)
-0.3;253
(0.6448)
0.0837
(0.5152)
(0.5877)
(0.9415)
10.3659*
(0.9682)
0.9571
72
107.850*
-3.2537*
-0.0268
0.5857*
(0.0598)
0.0157
(0.0
186)
0.0110
(0.01 04)
0.5703
(0.70
I
4)
(0.6352)
-1.3488*
-0.2406
(0.7373)
2.6898%
(0.7972)
0.6070
(0.5358)
-0.4716
(0.6939)
(0.5604)
(0.6054)
(1.0171)
10.3871*
(0.9845)
0.9621
72
124.68 1
*
-0.0796
-3.5522*
-0.7323
0.3722*
(0.0169)
0.0218'
(0,0068)
0.0102*
(0.0041)
0.0888
(0.27 12)
(0.2515)
0.0622
(0.2895)
2.0619*
(0.3
1
12)
0.6812*
(0.209
1)
-1.2888*
-
0.209
1
~ 0.0559
-
3.0463*
(0.2589)
(0.2108)
(0.2330)
0.4737
(0.3925)
11.3731
(0.49 82)
0.9337
333
277.447*
*Indicates significance
at
.05
or
lower error level.
aDummy variables were entered to control for year-specific effects; all four dummies were statistically significant.
bNumber of production and maintenance workers
x
100.
The weighted least squares method was used to estimate the effect of
union status
on
wage rates; the results are shown in Table
4
for the Aver-
age Rate. Results were also obtained in a cross-sectional regression esti-
mated separately for each year, allowing us to observe the variability in
these effects over the five-year period. A separate regression was also
estimated with data pooled over the five years. All the regressions were
significant, with the
R2
ranging from .84 to
.96.
The number of observa-
Enterprise
Unions
in
the United States
J
149
tions varied because annual data were not available for all
plant^.^
These
results clearly establish that Average Rate, a good indicator
of
compensa-
tion costs, was significantly higher in ILU plants than in nonunion plants.
There was, however, no significant difference between Affiliated Union
and ILU average plant rates. That is, ILU plants were earning roughly the
same premiums over nonunion plants as were Affiliated Union plants. This
result is consistent across the five years.
Among our other findings, Plant Size had a consistently positive and
significant effect on average rate. The coefficient for Skill Mix was positive
as expected, but it was significant only in the pooled regression. Such
variation can result from varying degrees of wage compression in individ-
ual years. Similarly, the coefficient for Plant Age was significant only in the
pooled regression and, as expected, had a positive sign. At the regional
level, Southeast wages were similar to the reference group, the Northeast.
Southwestern plants (including the high-technology sector on the West
Coast) had wages that were consistently higher than other regions. The
coefficient for the Midwest was occasionally significant, but its sign was
consistently positive across the years. A similar pattern is observed for
Industry coefficients.
To further explore these findings, we ran the same weighted regressions
for High Rate, Low Rate, and Total Benefits. Table
5
shows the nonunion
and affiliated union dummy coefficients from weighted ordinary least
squares regressions for each dependent variable. The results show that the
differences between ILU and nonunion plants were consistently large and
significant
for
Low Rate and Total Benefits and for High Rate in the
pooled regression. However, the gap between the ILU and the affiliated
union plants was much smaller than the ILUJnonunion gap. In several of
the year regressions (Low Rate, Total Benefits) and pooled regressions
(High Rate), this gap was insignificant. In short,
ILU
members are paid as
’On the basis
of
the Park-Glejser and Breusch-Pagan tests (Judge
et
at..
1982).
we found that mild
heteroscedasticity existed in
our
data with respect
to
Plant Size (as measured by employment,
PMEMP).
We
corrected for this by assuming that the correct error variance model was:
Ln (var),
=
a,
+
b (PMEMP),
+
ui
We estimated this model by regressing on employment the squared residuals from the original ordinary
least squares model. From this we derived a weighting variable,
where
P
is the predicted value from the preceding regression. This gives us the estimated error
variance by which to weight the data. We tried different error variance models from the above, but the
results were
poor.
Our
correction for heteroscedasticity improved the significance of several coeffi-
cients without changing their magnitudes
or
signs. Most importantly, there was no change in either the
Affiliated Union or Nonunion coefficients.
150
/
SANFORD
M.
JACOBY
AND
ANIL
VERMA
TABLE
5
NONUNION
AND
AFFILIATED UNION PREMIUMS
FOR
HIGH
RATE,
Low RATE,
AND
TOTAL BENEFITS
IN
ILU PLANTS~
Dependent Variable
High Rate Low Rate Total Benefits
Year
NUb AU NU AU NU AU
1979 -1.1827 -0.7917 -2.1886* -0.9371*
-
1.1390* -0.8991*
(0.6662) (0.7173) (0.4196) (0.4638) (0.4206) (0.441
5)
(0.7427) (0.8070) (0.5026) (0.5514) (0.4795) (0.5246)
(0.8538) (0.9582) (0.5791) (0.6507) (0.4263) (0.4755)
1980 -1.4232 -1.0871 -2.351
1*
-0.9488 -1.5163*
-
1.3048*
1981
-
1.0442 -1.2370
-
2.4544*
-
1.6484*
-
1.4401
*
-0.9685*
1982
-
0.789
1
-0.6781 -2.7642* -1.5513* -1.4094* -0.5766
(0.8759) (0.9646) (0.6653) (0.7263) (0.4927) (0.5433)
1983 -0.9408 -0.4931 -3.4455*
-
1.6296* -1.5550*
-
0.6397
(0.8758) (0.9587) (0.5530) (0.6059) (0.4969) (0.5424)
Pooled
-0.9602* -0.7297
-
2.6828* -1.3553*
-
1.3811' -0.8535*
(0.3473) (0.3806) (0.2327) (0.2553) (0.20 12) (0.2207)
*The coefficients are significant at the
.05
error level.
aNonunion and affiliatcd union dummy coefficients are from weighted ordinary least squares regressions
on
selected
dependent variables. Other independent variables entered are Plant Age, Plant Size, Skill Mix, and region and
industry dummies.
bNU
=
nonunion plants; AU
=
affiliated union plants.
well as or slightly better than workers in affiliated union plants and signifi-
cantly better than those in nonunion plants.
Plant Size, measured by the number
of
production and maintenance
workers, was largest in
ILU
plants, followed by affiliated union plants and
nonunion plants. During the deep recession of the early
1980s,
employ-
ment decreased in all plants. However, the decline was most pronounced
in the affiliated union plants, followed by
ILU
plants and nonunion plants
(see Table
3).
These relative employment declines do not fit the pattern of
compensation costs observed earlier. Given that
ILU
compensation was as
high as in affiliated union plants and that the industry profile of the two
sectors was largely the same, one would expect
ILU
and affiliated employ-
ment declines to be roughly the same. The lesser shrinkage of
ILU
plants
suggests that these plants may have received preferential treatment from
management as compared with the affiliated union plants.* On the other
8Data
on
capital expenditures are available
for
a sample
of
six plants
for
the period
1977-80.
Decisions on how
to
allocate these funds were made at the top corporate level. Previous research with
these data found that
TRW
management followed a strategy of favoring investment in its nonunion
plants, although that research did not distinguish between affiliated and independent union plants
(Verma,
1985).
A
reexamination
of
the data shows that average annual expenditures on plant and
Enterprise
Unions
in
the United States
I
151
hand, however, relatively high wage rates in the ILU plants may have
played a role in the
1984
divestiture plan, which eventually led to the sale
of many
of
the company’s ILU plants (Giffiths and Phillips,
1988).’
Sensitivity analyses.
A number of sensitivity tests were carried out to
examine the effect
of
alternate specifications on the main finding that ILU
members received relatively greater compensation. First, since inclusion
of Skill Mix reduces our sample size,
we
estimated the regressions without
Skill Mix. The
R2
dropped by approximately
.015
on average, but the
union dummy coefficients and their statistical significance remained un-
changed. Second, as noted earlier, there
were
no ILU plants in the South-
east or the Southwest. We tested the hypothesis that regional, rather than
representational, differences were driving our results by excluding South-
east and Southwest plants
from
the sample. For Average, Low, and High
Rates, a slight decrease in ILU premiums over affiliated union plants was
observed, but the coefficients remained insignificant; ILU premiums over
nonunion plants increased slightly. Thus, the substance of our findings
remains the same; the absence
of
ILU plants in the Southeast and South-
west is not responsible for the results we observed.
Discussion.
The hypothesis that ILUs are a device to pare compensa-
tion costs must be rejected, although it
is
likely that ILUs allow employers
to reduce other costs, such as those related to dispute resolution and
contract administration. Indeed, during the
1950s
managers at Thompson
Products and other companies said that they preferred ILUs over national
unions because
of
perceived industrial relations economies. In the oil indus-
try, for example, managers “agreed that [ILUs] had cost their company
money,
. .
. but they quickly added that the additional cost
of
having an
independent union was much less than the cost of a few strikes’’ (Brandt,
1960,
p.
46).
These savings, however, were relevant if ILUs were the only
machinery per employee was considerably higher in the nonunion plants ($4,685) than the affiliated
one ($928), with the ILU plants ($2,034) situated in the middle. Because of the small sample size and
categorical variance (one of the nonunion plants received expenditures of only $1,501 per year), it is
impossible to say anything more than that the data are consistent with a corporate strategy
of
favoring
ILU
over
affiliated union plants despite the fact that the average hourly wage rate at the ILU plants in
1979-80 was considerably higher ($8.77) than at the affiliated union plant ($7.79).
”We attempted
to
test the hypothesis that the relative employment declines are significantly different
in the three sectors even after effects
of
plant age, plant size,
skill
mix. region, and industry are
removed. Unfortunately,
our
results were inconclusive because the regressions showed no statistical
significance and explained little of the variance in employment levels.
Our
sense is that other contribut-
ing factors-notably, changes in product markets-account for much
of
the variance. Thus, even
if
plant union status was a factor in making employment decisions, that effect was overwhelmed
in
our
data by other influences, such as product market factors,
for
which no data were available.
152
/
SANFORD
M.
JACOBY
AND
ANIL
VERMA
perceived alternative to omnipresent national unions. But after
1960,
as
national unions began to shrink and the new nonunion model took root,
TRW and other firms shifted away
from
ILUs as they discovered that
nonunion plants could provide similar advantages at a lower cost.
We must also reject-at least
in
the case of TRW-Troy’s hypothesis
that ILU compensation premiums stem from preexisting skill or geo-
graphic wage differentials. Troy
(1960)
had argued that workers in high-
wage labor markets or in skilled occupations were drawn to ILUs “in order
to maintain differentials that national unions would attempt to eliminate”
(p.
489)
by policies of geographic standardization and skill compression.
As
proof, he cited examples drawn from the Bell System-where ILUs
were found in high-wage cities such as New York and Chicago, while the
Communications Workers (CWA) had its locals in lower-wage cities such
as Atlanta, Cleveland, and Philadelphia-and from various industrial
unions such as the Transit Workers (TWU) and the Automobile Workers
(UAW), which periodically had experienced defections of their skilled
trades into ILUs. Troy’s argument is a variant of the view that union wage
premiums are endogenous, that
is,
that they exist prior
to
unionization
(Johnson,
1975).
While his explanation makes sense in some cases,
we
do
not think it applies to all ILUs, including those at TRW.
Although some ILUs can be found in high-wage regions, others are
located in small towns and rural areas that have relatively low wages.10
Although the data are
30
years old, a BLS
(1962)
report shows no notice-
able difference-at least at the state level-in the geographic distribution
of affiliated and unaffiliated unions. At TRW in the
1960s
and
1970s,
ILU
plants were situated in small towns in rural areas and in large cities such as
Cleveland. Among large Midwestern industrial cities, however, Cleveland
did not pay particularly high wages. Nor did TRW (Thompson) workers in
Cleveland receive high pay prior to the formation of the AWA. In the late
1930s
and early
1940s,
wages at the Cleveland plants were less than prevail-
ing rates for other local automotive and aircraft parts firms (Shore,
1966).
It is also unlikely that TRW workers joined and supported ILUs in order
to preserve skill differentials. First, TRW’s ILU plants employed roughly
the same proportion of skilled workers as its affiliated union plants (see
Table
3).
Moreover, although the results in Table
5
show that the TLU
plants paid more to their highest-paid workers than did the affiliated plants
(the coefficients are not statistically significant, however), they also show
that the ILU plants paid more to their lowest-paid workers (here the
’OSeveral federations
of
ILUs still exist, including the Midwest Independent Union Council in St.
Louis and the National Federation
of
Independent Unions
in
Philadelphia.
Enterprise
Unions
in
the United States
I
153
pooled coefficient is significant and larger than in the high-pay regres-
sions). That is, when skill, region, and other variables are controlled for,
high-paid and (especially) low-paid
ILU
members received premiums rela-
tive to affiliated union members, a pattern at variance with Troy’s exam-
ples of
ILUs
formed to maintain upper-tail skill differentials.
Another possible explanation of these premiums is that they result from
higher productivity levels that exist in
ILU
plants as a result of greater
work rule flexibility and more flexible production methods of the sort that
have become popular in recent years (Piore and Sabel, 1984). We do not
have reliable data on plant productivity levels, but we do know something
about work practices, such as the number of job and wage-grade classifica-
tions and the use of pay-for-knowledge compensation systems (see Table
6).
Although these data are based on a very small subsample of
TRW
TABLE
6
WORKPLACE PRACTICES,
1983
Affiliated
ILU
Old Nonunion New Nonunion
Union Plant Plants Plants Plants
(N= 1)
(N=2)
(N=2)
(N=3)
Ave.
no.
of
job
classifications 12.5 82
6.5
6
No.
of
wage grades
15
13
11
7
Joint
QWL
committees
(%)
0 100 50 100
Ave. index
of
employee involve-
Pay-for-knowledge
(%)
0
0 0
66
2 2
3
Sensing sessions
for
production
0 100
100
66
-
ment
in
QWL
(l=low,
4=high)
workers
(%)
Employee meetings
(%)
0
100
100 100
plants, they show that differences between ILU and affiliated union plants
were relatively small in this area as compared with those between older
and newer plants. (Some
of
the nonunion and all of the unionized plants
were built before 1960.) Older plants-regardless
of
union status-had
a
finer division
of
labor and a more complex job/wage system than was found
in newer plants, which suggests that observed wage premiums had little to
do
with work rule flexibility. It also suggests that plant age (correlated with
capital vintage and state
of
technology), rather than union status, was the
main impediment in TRW’s shift to more flexible production and work
methods. It simply was easier for the firm to adopt these methods in a
“greenfield” setting than in an existing plant.
At the same time, Table 6 shows that various voice and job-level
EIPs
154
/
SANFORD
M.
JACOBY
AND
ANIL
VERMA
were equally prevalent in nonunion and ILU plants, but were not found in
the affiliated union plant, such things as Quality of Work Life (QWL)
committees, sensing sessions for production workers, and periodic em-
ployee meetings with management. While the sample is too small for
definitive conclusions, the data are consistent with the view that employee
relations in ILU plants were favorable to the sort of organizational culture
that fosters employee involvement. Therefore, the company’s shift away
from ILUs after 1960 is all the more striking. It reinforces our claim that
management received benefits from ILUs that were similar to those avail-
able from the new nonunion model (avoidance
of
national unions and
cooperative employee relations) but that the costs, especially compensa-
tion costs, were lower under the new model.
How then can we explain compensation patterns at TRW’s ILU plants?
One possibility
is
that the premiums were voluntarily paid by management
as a national union avoidance tactic (a bribe), what Troy (1961) called “the
employers’ desire to maintain differentials that will discourage affiliation.”
Troy claimed that little evidence “has been adduced to support or contra-
dict this widely held belief” (p. 337), and then presented anecdotal evi-
dence from General Motors that contradicted it. While
we
also have noth-
ing better to offer than anecdotal evidence, in this case
it
supports the
bribe or “golden handcuffs” explanation
of
ILU wage premiums. During
the 1950s and 1960s, TRW management carefully tracked compensation
developments at major automotive and aircraft firms with UAW contracts
to ensure that AWA members kept up with or ahead
of
UAW members.
When, for example, it was discovered in 1957 that the GM-UAW COLA
formula paid two cents more than the AWA’s, an internal management
memorandum noted that the company should consider adopting the
GM
index (Jacoby, 1989). Moreover, our data show a slight decline in the ILU
wage premium over affiliated union plants during 1982 and 1983. These
were peak concession bargaining years, therefore a time when national
unions posed less of an organizational threat to TRW’s
ILU
plants.
It is reasonable to ask why TRW did not pay a similar “bribe” to its
nonunion workers. There is a simple answer: The likelihood of affiliation
with a national union is far higher for an
ILU
plant than for a nonunion
one. ILU members already are a collectivity familiar with the arcana of
unionism: contracts, grievance procedures, officers, elections, and the
like. Affiliation with a national union would bring a relatively small change
in their work lives as compared with employees in nonunion plants. For
that reason, the UAW and other unions regularly made organizing raids at
TRW’s ILU plants; the CWA and Oil, Chemical, and Atomic Workers
(OCAW) did the same with the ILUs in the Bell System and at Exxon.
Enterprise
Unions
in
the United States
I
155
If the company paid a “bribe” to ILU members to keep them from
affiliating with a national union, that would account for the premium that
they earned relative to workers in nonunion plants. But what about pay
differences between ILU and affiliated union members? It may be that
these also were a “bribe.” We think that some
of
that differential (as well
as some
of
the TLUhonunion differential) resulted from the bargaining
power possessed by ILU members. One obvious example
of
that power
is
the strike. Aggregate strike data for TRW’s ILUs are unavailable, but
we
do know that some
of
them (e.g., the AWA and the Ohio ILU shown in
Table 2) occasionally did go on strike, not as often as some affiliated union
locals but certainly more than unorganized workers.
In addition to strikes, ILUs have other tactics that they can and do use to
pressure management. First, even if a strike does not occur, ILU members
can threaten to strike. This happened at TRW as well as at
Esso,
where ILU
officers told management during negotiations in 1959 that “[we] have always
preferred the peaceful route in negotiations but must confess that our nerves
are on edge” (Marshall, 1961, p. 829). Second, ILU members and officers
can influence management’s perception
of
the likelihood
of
a national union
takeover and
so
extract higher pay. Contract rejections, which were a peri-
odic occurrence with the AWA, signal to management (and to national
union organizers) that workers are dissatisfied with the negotiated terms.
Also, ILU officers at TRW,
Esso,
and other companies sometimes made
thinly veiled threats to management that they would affiliate with a national
union if better contract terms were not negotiated (Brandt, 1960; Jacoby,
1989). Finally, because ILUs are relatively democratic and decentralized
unions-there is direct ratification,
no
intermediate bodies, and little
hierarchy-they may be especially effective in bargaining, more
so
than
even affiliated union locals. Other research has shown a link between union
characteristics and bargaining outcomes (Fiorito and Hendricks, 1987, al-
though not specifically concerned with ILUs)
.I1
ILUs
and
Public
Policy
When compared with national unions, ILUs are not without defects, as
critics have been pointing out for a long time (Douglas, 1921). First, ILUs
“The same logic also applies to Japanese enterprise unions. Although often viewed
as
powerless
pawns of management, two recent studies refute that stereotype. Based on attitude surveys, Lincoln
and Kalleberg
(1985)
find “no support for the ‘strong corporatist’ hypothesis that Japanese enterprise
unions function directly to strengthen the bond between the employee and the firm” (p.
753).
Freeman
and Rebick
(1989)
find that Japanese enterprise unions have substantial positive effects on women’s
pay, bonus pay, severance pay, and leisure hours, and that newly organized Japanese unions raise
wages in about half the cases studied.
156
/
SANFORD
M.
JACOBY
AND
ANIL
VERMA
have little influence at the top levels of a corporation because they are
typically single-local unions. Therefore, they lack the ability or vision to
coordinate policies to affect strategic corporate decisions, an issue of grow-
ing importance for today’s labor organizations.
n
Second, the logic
of
com-
pany unionism is at variance with the traditional national union objective
of
taking wages and work standards “out of competition” by standardizing
them throughout
an
industry or occupation. Third, ILUs reinforce, rather
than break down, the secretive insularity that pervades many
of
today’s
new-model nonunion firms (Heckscher, 1988). Finally,
ILUs
have the
potential-as in the
1920s
and early 1930s-to be used by employers
as
a
potent substitute for national unions. Indeed, it was these defects, espe-
cially the last, that guided the early NLRB’s policy to wipe out unaffiliated
single-employer labor organizations, and it nearly succeeded in doing
so.
As
a result, today the United States is unique-as compared with Western
Europe and Japan-in lacking this form
of
representation.
National unions are still necessary, but they are not the solution to all of
today’s representation problems. Most national unions have not appealed
to workers in nonunion firms with
EIPs
and other progressive personnel
policies. Moreover, forces such as deregulation and international competi-
tion are making it increasingly difficult for national unions
to
adhere to the
old ideal
of
wage standardization. Industrywide bargaining is breaking
down and giving way to company-oriented wage policies. Mechanisms
such as Employee Stock Ownership Plans
(ESOPs)
and board representa-
tion give nonunion employees as much strategic influence as is exercised
by some national union members.
In
addition, our study raises two other important points. First, some
of
the purported defects of ILUs-such as their lack
of
bargaining power or
their complacence and inability to be critical of management-are not
inherent in this form
of
organization. Even within a single company
like
TRW, there was considerable variation among ILUs. Some were aggres-
sive and independent, others less
so.
Second, what have traditionally been
viewed as ILU defects may turn out to be virtues when it comes to repre-
senting workers in today’s new-model nonunion firms. Not only do ILUs
mesh with the
EIPs,
communications programs, and communitarian cul-
ture found inside these firms, but they are also more likely than national
unions to appeal to college-educated workers, who are increasingly em-
ployed there. These employees have a high degree of corporate loyalty,
are able to run organizations without much outside assistance, and may
’*Unlike
some
of
TRW’s
other
ILUs, this
was not
a
problem
for
the
AWA,
which was based
in
Cleveland, the
site
of
company headquarters.
Enterprise
Unions
in
the United States
i
157
dislike national unions because of their male, blue-collar image, however
inaccurate that stereotype may be. Although the most recent data are from
1967, they show that among ILUs, the proportion of white-collar workers
was more than twice
as
high, and the proportion of women nearly twice as
high, as it was within national unions
(BLS,
1969).
The TRW experience also suggests that new-model nonunion employers
are unlikely to welcome or encourage ILUs because labor costs are likely
to be higher in ILU facilities than in those lacking collective representa-
tion. Even without a significant threat from outside unionism, there still
will be some positive ILUhonunion differential, as at TRW and as in
Japan today (Freeman and Rebick, 1989). However, because ILUs have
the potential to remedy various voice defects in the new nonunion model,
they may raise employee satisfaction and productivity. Whether this would
be sufficient to overcome employer preferences for the nonunion model is
an open question. Employers who deal with national unions clearly do not
give much weight to the productivity offset, whereas ILUs are different
from affiliated unions and
so
employers might be more inclined to value
any positive effects. Indeed, some firms practicing the new nonunion
model have gone
so
far as to create EIPs whose representational structures
resemble ILUs in many respects (Beer, 1985). Whether these firms are
reluctant to go further because of section 8(a)(2) or because they fear
creating genuine independent unions is not known.
At this point, public policy may have a role to play. Coordinated market
economies such as Japan’s and West Germany’s exhibit high levels of
innovation and an ability to pursue economic restructuring, which some
experts attribute
to
the network of consultative relationships that exist at
various levels in these societies, including plant-level works councils and
enterprise unions (Wever, 1988; Soskice, 1990). To boost national competi-
tiveness, public policy in the United States could be designed to encourage
more industrial participation and consultation (including ILUs) because
these would generate positive social externalities. The carrot for employ-
ers would come from the recent and growing tendency of courts and legisla-
tures to regulate the workplace-everything from dismissal restrictions to
mandated employee benefits.
As
that process unfolds, employers may
prefer ILUs
to
standardized, top-down legislation and costly dismissal
awards, especially if new legislation specifically exempts employers who
negotiate private, customized terms with a national or company union.
Before that could happen, however, the law regulating company unions
would have to be relaxed. In a piecemeal fashion, the courts have already
taken a more permissive approach to the legality
of
EIPs.
Their stance has
been criticized by some legal scholars as contrary to the intent of the
158
/
SANFORD
M.
JACOBY
AND
ANIL
VERMA
Wagner Act (Kohler, 1986) and praised by others for opening up new
avenues
of
choice for employees (Jackson, 1977). Somewhere between
these positions are those who argue that the courts are steering the right
course but for the wrong reasons: It is argued that all nonunion EIPs
(presumably including employer-supported ILUs) should be legalized, not
because they offer choices, although they might, but because they do no
harm to anyone and pose no greater threat to national unions than benign
outcomes such as high wages or good benefits,
so
long as these are not
adopted during the course
of
a union campaign (Weiler,
1990).
But ILUs would not be appropriate for all workers. National unions still
make sense for those sectors where EIPs have not taken hold, where
personnel policies are not benign, and where management and workers
expect the worst from each other. Yet there is a danger that a revised
section 8(a)(2) might make it more difficult to organize or maintain na-
tional unions in these sectors. The purpose
of
any legal reform should be
to expand the realm of representation, rather than change its composition.
Therefore, a comprehensive revision of section 8(a)(2) might be offered to
employers as a quid pro quo for changing those parts of the law that
govern organizing campaigns and elections.
Conclusions
These are ambitious proposals to tack on to an empirical analysis of one
company’s industrial relations experience. We offer them in the hope
of
stimulating others to pursue more research on this topic. Ironically, Ameri-
can scholars with an interest in industrial relations reform probably know
more about European works councils than they do about ILUs in the United
States. Future research should occur along several vectors. First, we need to
identify factors that cause organizational characteristics of ILUs to vary
across and within firms. We know that
TRW’s
ILUs differed from one plant
to another, but what caused those variations? How and why do ILUs differ
from firm to firm? Second, we need more information on the attitudes of
workers, managers, and union leaders toward this kind of unionism, espe-
cially more case studies. Finally,
we
need more research that compares
different institutional models for handling industrial relations-both within
the United States (e.g., ILUs versus
EIPs)
and across national boundaries
(e.g., EIPs versus works councils)-in terms of their impact on workers,
firms, and nations. Policy-oriented research on ILUs holds great promise
for those interested in improving workplace relations and economic perfor-
mance in the United States.
