. €
a.3\..ia1r¥:.€
. .f. :2}.

.£ 545.22:
4 .V List? (”a

 

.3? ,. r(l.«r.’a)t\(r£.).l(v (I)! If! 1 ‘1 . (t r {‘(Itl,i.~t(‘
. r13 x11 5})... 5.654;}? «f. rvlpv...‘ 59...} If it)?
if: ('15. I z (<r:.{(.~:(l.r/ Irv :2iirr a...

xiii; 3}}: 111.... .

‘ ((vllt‘rrr («till
(ti ; 4 u 1 ti :4... 5.12.... .- L513)? fr... strillz‘itvg... I” ill I (It
. ((5)5: 1.72.). .ct..: 1 ( (if fittircv; 5u§.xt.\<(.) 3.15153. (.9 u.) 4"...
(v 51:... ‘ c :‘r.(:{; . I c .l...: {ii (o.(t(rI.(t£¢l¢ol pills.
Crii‘tl: (i r. .2

131...}. «Lgic.).ln)\c:(.i<(f
3 .Cr \tr‘tlfip .317}. S) (...¢.r((l...: tliti‘tllflf‘rill‘)!
(r... c 3...: . r... 5.5.8.... ,jl‘lec/‘e. tilt; |¢r¢rltlzi é’gllilrlgtllog 1
E 7.5/5.2 . 1.1;}. _.,r€. ‘ ,1 31,115.31, .. ....,./a. .....J\(.}.¢..v..1\r v.1" -.<¢/r...v}lturn 1:1). 1!. ,}

 

 

 

, ,. wfim‘YAufih; ;.

$9 0 5 7‘3 0 @
"Hilllljllfllllll[lfllllllylljwlllllll I
* * *‘ Lid/BABY
Michigan State

University

 

 

 

 

This is to certify that the

dissertation entitled

THE GROWTH OF EMPLOYEE BENEFITS AND THE

DECLINE IN AMERICAN LABOR

presented by

GILBERT BRIAN DAVI S

has been accepted towards fulfillment
of the requirements for

Ph . D . degree in Economics

Odd/M [MtWt/rm

Major professor

Daniel S. Hamermesh
Date February 19, 1990

 

MSU is an Affirmative Action/Equal Opportunity Institution

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

DATE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU Is An Affirmative Action/Equal Opportunity Institution

THE GROWTH OF EMPLOYEE BENEFITS AND THE DECLINE IN

AMERICAN LABOR

by

Gilbert Brian Davis

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirement
for the degree of

DOCTOR OF PHILOSOPHY
Department of Economics

1989

 

(O o 5 C3" EEC) ‘3

ABSTRACT

TIE GROWTH OF EMPLOYEE BENEFITS AND THE DECLINE

IN AMERICAN LABOR

By

Gilbert B. Davis

Previous work examining the determinants of the union status of workers
ignores the effect that the provision of benefits by employers will have on the
willingness of workers to vote for union representation. This omission is important
for two aspects. First, unionized workers are found to have a greater percentage of
their total compensation in the form of benefits than do non-unionized workers.
Thus, it is likely that non-unionized workers will vote for union representation to
obtain benefits. Second, there is a debate in the industrial relations literature that
non-union employers attempt to remain nonunion by providing their current workers
with a compensation packages approximating those obtained by union negotiations,
the management substitution hypothesis. Similarly, the government may be providing

benefits to workers, thereby decreasing the demand for union representation by

 

Gilbert Brian Davis
workers, the government substitution hypothesis. Therefore, it is necessary to
determine whether benefits affect the willingness of workers to vote for union
representation.

To determine the effect benefits may have on the willingness of workers to
vote for union representation two empirical studies are done. The first is a cross-
section study. The results of the cross-section study show that benefits affect the
willingness of workers to vote for union representation. In particular, the desire for
medical benefits is found to have a strong effect. The second empirical exercise in
the dissertation is a simulation using the coefficients from the cross-section results
and the variation in the mean levels of benefits in the economy as a whole. The
purpose of the simulation exercise is to determine whether the growth in the level
of benefits since 1950 has resulted in a decline in the willingness to vote for union
representation. The findings of the simulation study suggest that the growth in the
level of benefits and earnings may account for as much as 22 percent of the decline
in the success rate of unions in representation elections.

It appears that the management substitution hypothesis may play a serious

role in further declines in the success of labor unions in representation elections.

ACIQVOWLEDGMENT S

I am indebted to many individuals for their assistance in the researching and
writing of this dissertation. First among these are my committee: Professors Daniel
Hamermesh (my chairperson), Harry Holzer, Stephen Woodbury, and Richard Block.
For Professor Hamermesh a thank you for the advice of "Just do it." For Professor
Holzer a special thank you. I know of no individual who has contributed so much
to a dissertation without being the chairperson.

In addition to my actual committee, Professors Charles P. Larrowe, Rowena
Pecchenino, and Bruce T. Allen were always there with words of wisdom and
encouragement in what seemed to be the darkest hours.

Harriet Dhanek deserves a note of thanks for making it so easy to obtain the
data used in this dissertation. The data utilized in this dissertation were made
available in part by the Inter-University Consortium for Political and Social
Research. The data were originally collected by the Institute for Social Research at
the University of Michigan. Neither the original source or collectors of the data nor
the Consortium bear any responsibility for the analysis or interpretations presented
here.

Finally, I would like to thank Brenda E. Turner for all of her support through

my too many years as a graduate student.

 

TABLE OF CONTENTS

LIST OF TABLES ........................................
LIST OF FIGURES .......................................

CHAPTER
ONE - INTRODUCTION .................................

Institutional Process of a Union Organization ............

Factors Leading to the Decline in Demand for Union
Representation .................................

Management as a Substitute Provider of Union Services and
the Decline in the Demand for Union Representation . . . .

Government as a Substitute Provider of Union Benefits and
the Decline in the Demand for Union Representation .

TWO - A TRUE TEST OF THE SUBSTITUTION HYPOTHESIS . . .
Conclusion ......................................

THREE - THE GROWTH OF TOTAL COMPENSATION AND THE
DECLINE IN THE WILLINGNESS TO VOTE FOR UNION
REPRESENTATION ...............................

Conclusion ......................................

FOUR - SUMMARY AND CONCLUSIONS ...................

APPENDICES
CHAPTER TWO
APPENDIX A ........................................
APPENDIX B ........................................
CHAPTER THREE
APPENDIX A ........................................
APPENDIX B ........................................

BIBLIOGRAPHY .........................................

iii

vi

H

20
29

38
69

72
89

92

97
100

101
104

107

 

LIST OF TABLES

Page
TABLE

1-1 Union Membership as a Percentage of Employees in

Nonagricultural Establishments ................... 2
2-1 Current Fringe Benefits by Union Affiliation ........... 46
2-2 Maximum Likelihood Estimates Current Fringe Benefits

and the Willingness to Vote for Union Representation . . 51
2-3 Maximum Likelihood Estimates Current Broad Based

Fringe Benefits and the Willingness to Vote for

Union Representation .......................... 57
2-4 Maximum Likelihood Estimates Desire For Broad Based

Fringe Benefits and the Willingness to Vote for

Union Representation .......................... 60
2-5 Maximum Likelihood Estimates Current Broad Based Fringe

Benefits Desire for Broad Based Fringe Benefits and

the Willingness to Vote for Union Representation ..... 64
3-1 Mean Medical Benefit Coverage and Union Victory Rates

in N.L.R.B. Representation Elections, 1950-1987 ....... 75

 

3-3

3—4

3-5

3-6

3-7

3-8

3-9

3-10

LIST OF TABLES (continued)
Page

Actual and Predicted Union Victory Rates in N.L.R.B.

Representation Elections, 1950-1987 ................ 79
The Effect of Changes in Total Compensation on the

Predicted Outcomes of N.L.R.B. Representation Elections 83
Simulated Decines in the Willingness to Vote for Union

Representation, 1950—1987 ....................... 86
Regression Results of the Effect of Changes in Total

Compensation on the Outcome of N.L.R.B. Representation

Elections ................................... 88
Regression Results of the Effect of Changes in Total

Compensation on the Outcome of N.L.R.B. Representation

Elections ................................... 90
Medical Benefits, Retirement Benefits, and Annual Earnings,

1950-1987 ................................... 103
Maximum Likelihood Estimates Current Fringe Benefits, and

the Willingness to Vote for Union Representation ..... 105
Actual and Predicted Union Victory Rates in N.L.R.B.

Representation Elections, 1950-1987 ................ 106
The Effects of Changes in Total Compensation on the

Predicted Outcomes of N.L.R.B. Representation Elections 107

V

 

LIST OF FIGURES

FIGURE Page
3-1 Actual and Predicted Union Victory Rates in N.L.R.B.
Representation Elections, 1950-1987 ................ 80
3-2 Total Compensation Changes and Actual and Predicted
Union Victory Rates in N.L.R.B. Representation

Elections, 1950-1987 ........................... 84

vi

 

CHAPTER ONE

INTRODUCTION

One of the more interesting social phenomena of recent years is the steady
decline in the United States of both the percentage of workers who are unionized
and the number of workers who are unionized. As Table 1.1 shows, the percentage
of nonagricultural private sector employees who are unionized reached a high point
in the early 19505 but has since declined. During this time period, the success of
labor unions in union representation elections has also declined. Whereas unions
won more than 70 percent of representation elections during the early 19505, at the
present time unions are winning less than 50 percent of the elections.

The reasons for the decline in the success rate of unions in representation
elections are many and may be represented by the forces of the supply and demand
for union services. The primary focus here will be the demand for union services.
Specifically, this dissertation will explore the possibility that the government and
employers have become major providers of benefits which were once supplied only
as a result of the collective bargaining process. With the provision of benefits by the
government and employers, the demand for union representation on the part of

unorganized workers has declined. The resulting decline in the demand for union

 

2
TABLE 1.1

Union Membership as a Percentage of Employees in
Nonagricultural Establishments‘

 

 

Year Percentage Year Percentage
1930 11.6 1960 31.4
1931 12.4 1961 30.2
1932 12.9 1962 29.8
1933 11.3 1963 29.1
1934 11.9 1964 28.9
1935 13.2 1965 28.4
1936 13.7 1966 28.1
1937 22.6 1967 27.9
1938 27.5 1968 27.8
1939 28.6 1969 27.0
1940 26.9 1970 27.3
1941 27.9 1971 27.0
1942 27.9 1972 26.4
1943 31.1 1973 25.8
1944 33.8 1974 25.8
1945 35.5 1975 25.5
1946 34.5 1976 24.7
1947 33.7 1977 24.8
1948 31.9 1978 23.6
1949 32.6 1979

1950 31.5 1980 23.0
1951 33.3 1981

1952 32.5 1982

1953 33.7 1983

1954 34.7 1984 18.8
1955 33.2 1985 17.5
1956 33.4 1986 18.0
1957 32.8 1987 17.3
1958 33.2 1988 16.0
1959 32.1

 

1Data for 1930 to 1978 are from Handbggk at m Statisfica Bullatia L019
(Washington, DC: US. Government Printing Office, 12980): 412. Data for 1987
and 1988 are from Emplgmant aad Earnings, 34 (January 1987(:, 21. Data for 1984
are from Emplgment aad Earnings, 33 (January 1986): 213.

3
representation may be responsible for the decline in the success rate of labor unions
in representation elections.

Focusing on the demand for union services as an explanation for the decline
in the success rate of unions is appropriate in light of the most recent work done on
the supply of union services. Labor unions have a choice between allocating scarce
resources for representing current members or spending funds to organize new
members.2 The spending of funds on organizing activities represents the supply of
union services to unorganized workers. A reduction in union organizing expenditures
would explain the decline in the success rate of unions in representation elections.
Recent studies of the supply of union services conclude that while organizing is a
very expensive undertaking for unions, it is economically rational for unions to
continue their organizational efforts.3 Additional work finds total organizing
expenditures on the part of unions have not declined in recent years.‘1 Furthermore,
there is little reason to believe that the concentration of unions into a few relatively
highly organized sectors has retarded union growth by leading unions to spend less

on organizing? The results of the studies of the supply of union services indicate

 

2Richard N. Block. "Union Organizing and the Allocation of Union Resources,"
Industrial m Labor Relations Review, 34 (October 1980): 101-113.

 

 

 

3Paula B. Voos. "Union Organizing: Costs and Benefits," Industrial fl Labor
Relations Review, 36 (July 1983): 582-584.

 

 

 

4Paula B. Voos. "Trends in Union Organizing Expenditures, 1953-1977," Industrial
m Labor Relations Review, 38 (October 1984): 60-63.

 

sPaula B. Voos, "Union Organizing Expenditures: Determinants and Their
Implication for Union Growth," Journal o_f Labor Research, 8 (Winter 1987): 23.

4
that it is unlikely that the declining success rate of unions in representation elections
may be traced to a reduction in the supply of union services.

The idea that actions of either the government or employers may lessen the
demand for union services by providing services which substitute for the services
normally thought to be provided by unions is not a new one. As discussed below,
the provision or control of the terms and conditions of employment by the
government is something organized labor has opposed in the past. The opposition
to action by the government stems from the fear that there will be a reduction in the
demand for union services. The movement of management into the areas of
providing unilateral improvements in the terms and conditions of employment,
thereby providing a substitute for union services, is an action organized labor
experienced with the "welfare capitalism" schemes of the 19203.

More recent concern over the substitution of union provided services by
either management or the government has its clearest explanation in the work of
Garbarino.° Garbarino argues that the American system of industrial relations may
no longer be thought of as a simple dichotomy between the unionized and
nonunionized sectors. In addition to the collective bargaining model of industrial
relations, there are three other major employee relations systems: the administrative
model; the civil service model; and the legal model. These four systems of industrial
relations exist simultaneously and act as possible substitutes for each other. In

particular, the legal model of industrial relations represents the provision of benefits

 

6Joseph W. Garbarino, "Unionism Without Unions: The New Industrial
Relations?", Industrial Relations 23 (Winter 1984): 40-51.

 

5
by substantive labor law. For example, the provisions of the Fair Labor Standards
Act, the Occupational Safety and Heal Act, the Employees Retirement Income
Security Act, Social Security legislation, and civil rights legislation all may act to
substitute for the actions of labor unions in the collective bargaining process, thereby
reducing the demand for the services of labor unions. Strauss points out that
substantive labor laws "have given many workers most of the benefits and protections
commonly provided by unionization" (and that) "their net impact has been to make

union organizing more difficult."7

Garbarino states the importance of the legal
system of industrial relations is likely to be greatest where the size of the employer
is small. The reason for the importance of the legal system for the small employer
is that small employers lack the internal management structure necessary for the
administrative model.8

With Garbarino’s administrative model of industrial relations the employer
is seen as unilaterally establishing the terms and conditions of employment. The
unilaterally established terms and conditions of employment include paid vacations
graded by length of service, paid holidays, insurance benefits, premium pay for
working unusual schedules, annual pay increases, allowances for changes in the

consumer price index, the uses of seniority in determining promotions, layoffs, and

the scheduling of work and vacations, a formal disciplinary procedure, and some

 

7George Strauss, "Industrial Relations: Time of Change," Industrial Relations 23
(Winter 1984): 5.

8Joseph W. Garbarino, "Unionism Without Unions: The New Industrial
Relations?", Industrial Relations, 23 (Winter 1984): 46.

 

6

effort to minimize the instability of employment? These items are normally provided
by an employer only after negotiations with a labor union and the signing of a
formal collective bargaining agreement. The substitution argument is that if the
employer provides terms and conditions of employment usually obtained through the
collective bargaining process, then workers will see less need for the organization of
a union and negotiating with an employer. As a result of the decline in the
perceived need for a union, workers will be less willing to vote for union
representation.

The possible result that the substitution effects of management and
government provision of services may have on the number of workers who are
unionized in the United States is illustrated by a simple stock flow model showing
how changes in union membership result from two sources.10 Changes in union
membership in a given year may be decomposed into gains and losses:

Changes in Union Membership = Gains - Losses.
Or more formally,
6Ul E O, + ¢,U, - ¢2U,, Eq. 1-1
where U[ is union membership at time t; 6U, is the change in union membership
from t-1 to time t; O[ is the net number of workers organized into unions in the
previous time period, that is, workers organized into new bargaining units minus

those lost through decertification elections; as, is the rate of automatic increase in

 

5'Joseph W. Garbarino, "Unionism Without Unions: The New Industrial
Relations?", Industrial Relations, 23 (Winter 1984): 43—44.

loMichael Goldfield, m Decline o_f Organized Labor i_n th_e United States,
(Chicago, The University of Chicago Press, 1987): pp. 78-81.

7

unionized jobs due to the expansion of the work force in union shops; and ¢2 is the
rate of automatic loss in unionized jobs due to layoffs, plant closures, or runaway
shops. Equation 1-1 shows that changes in the number of union members may result
from two factors. First are those factors which determine employment levels in the
already organized sectors of the economy, U,. For example, steel imports have
resulted in a decrease in the demand for domestically produced steel, which in turn
has resulted in a decline in the number of employed unionized steel workers.
Changes in union membership and employment levels in already organized industries
will be identical if a union security provision such as a union shop is assumed.

The second factor affecting the number of union members, and the concern
of this dissertation, is that which affects union organizing, 0,, the willingness of
unorganized workers to vote for union representation. In light of recent research,
concern with the factors affecting union organizing is especially important as changes
in neither the occupational mix of the labor force nor the industrial mix of the
economy are found to explain much of the decline in unionization.“ Before
reviewing the literature concerned with the management and government substitution
effects and how actions by employers and the government may decrease the demand

for union services, it is necessary to outline how workers obtain union representation

 

1‘William T. Dickens and Jonathan S. Leonard, "Accounting for the Decline in
1985): 333. _
and
Henry S. Farber, "The Recent Decline of Unionization in the United States,"
Science, 238 (November 1987): 916-917.

8
and to acknowledge other factors contribution to a decline in the demand for union

services.

THE INSTITUTIONAL PROCESS fl UNION ORGANIZING

There are three ways in which a worker may obtain union representation.
First, an employer may voluntarily agree to recognize a group of employees as a
bargaining unit and engage in collective bargaining with that group. The second and
most common mechanism for the formation of a union is an election under the
guidelines of the National Labor Relations Act. The success of labor unions in
these elections is determined largely by the demand for union services on the part
of the workers. The final and most infrequently utilized manner in which a union
may be organized is through a court order.

The voluntary recognition of a union by an employer is most commonly done
after a neutral third party compares the union authorization cards signed by the
employees with the payroll list of the employer. If a majority of the employees have
signed authorization cards, the employer may voluntarily recognize the union so as
to avoid the tribulation of a representation election.

The statutory procedures for an authorized election are set forth in Section
9 of the National Labor Relations Act. Prior to an election, the union must show
the Board there is a significant "showing of interest" on the part of the workers to
merit a representation election. The showing of interest is accomplished by union

authorization cards or by a petition signed by at least 30 percent of the employees

 

9
in the requested bargaining unit. Although the Act requires that the union seeking
representation only acquire signatures of 30 percent of the members of the
prospective unit, the union usually attempts to secure signatures from at least 50
percent of the prospective unit in order to be more assured that it’s success in the
actual representation election.

Upon receipt of the representation petition, a field examiner or attorney from
a regional office of the National Labor Relations Board, N .R.L.B., investigates the
petition to determine whether the Board has jurisdiction over the case, whether
there is a true question of representation in an appropriate bargaining unit, and if
the filing of the petition has been timely. The regional office will attempt to
encourage the parties to enter a "consent election" agreement. If an agreement can
be reached then the date, place, and necessary voter list is established and an
election will be held without the holding of a formal hearing.

When a consent election agreement cannot be reached, the regional office
of the Board will conduct a formal hearing. The object of the hearing is to develop
a full record on such issues as jurisdiction, the appropriateness of the unit, questions
of representation and timeliness. The hearing officer then forwards to the regional
director a report summarizing the evidence and analyzing the issues, but making no
recommended decision. The regional director makes a decision on the disputed
matters and either orders an election or dismisses the petition. In recent years, the
time period between the director’s orders and the election has increased
dramatically, contributing to the lack of success for unions in representation

elections.

10
When an employer is deemed to have committed such "outrageous" and
"pervasive" unfair labor practices that a fair representation election cannot be held,
the Board may certify the proposed bargaining unit without an election. In these
cases the Board will rely on "convincing evidence of majority support" such as a
union-called strike or strike vote or possession of authorization cards from a majority

of the employees in the bargaining unit when certifying the bargaining unit.12

FACTORS LEADING T_O THE DECLINE IA] DEIWAND FOR UNION

REPRESENT A TI ON

This section of the dissertation examines factors other than the provision of
benefits by either the government or by employers which have possibly led to a
decline in the demand for union services. These factors include: delays in the
election process by either the NLRB. or by the parties themselves, the hiring of
management consultants, and discrimination against union organizers by
management.

The first concern is with the election process itself. With regard to the
election process, there is the possibility that one of the reasons why unions have
experienced a steady decline in their success rate in representation elections is an

increase in the time from the petitioning of the NLRB. for an election to the

 

(1969).

11

holding of the election. Prosten states that pre-election time delays are "the most
likely source of labor’s problem."13

Concern over the timeliness of a representation election is very legitimate on
the part of unions. When elections are concluded within two months of petition,
unions are likely to be the winner, whereas when elections are concluded more than
two months after petitioning the N .L.R.B., unions are likely to lose.“ The research
of Cooke confirms the importance of a relatively short campaign on the probability
of a successful union election. In 1979, Cooke found the probability of a union
victory was fourteen percentage points higher in a consent election than in the more
lengthy stipulated election.15

Roomkin and Block examine the possibility that delays in the election process
may be attributable to the actions of the NLRB. They point out the NLRB. is
committed to improving case processing time, as the Board recognizes "unnecessary
delays are incompatible with the rights and obligations created by the National

Labor Relations Act."16 Even thought the Board is committed to the timely

 

l3Richard Prosten, "The Longest Season: Union Organizing in the Last Decade,
a/k/a How Come One Team Has to Play With Its Shoelaces Tied Together?",
Proceeding o_f t_l_1§ Thug-First Annual Meeting, Madison, Industrial Relations
Research Association, 1978, p. 241.

1“Myron Roomkin and Richard N. Block, "Case Processing Time and The
Outcomes of Representation Election: Some Empirical Evidence." Universig o_f
Illinois Law Review, 1981: 75-97.

 

15William N. Cooke, "Determinants of the Outcomes of Union Certification

16Myron Roomkin and Richard N. Block, "Case Processing Time and The
Outcomes of Representation Election: Some Empirical Evidence." University o_f
Illinois Law Raview, 1981: 75

 

 

12

processing of representation elections, there is no legally specified time period within
which an election must be completed.'7

Concern over increased delays in representation elections led the NLRB.
to adopt a new managerial process in 1961. This new approach delegated authority
to the regional directors of the NLRB. and is credited with shortening the time
necessary to process a representation case. During the period 1973 through 1978
"the overwhelming portion of elections took place in a relatively short time: 42%
within one month of petition, 83% within two months, 92% within three months, and
94.5% within four months."18 As a result of the research of Roomkin and Block, the
suggestion that the increased time period between the petition of an election and the
holding of an election is due to the actions of the Board itself may be rejected.

An alternative explanation for the increase in the delay rests with the actions
of the parties to the election, the employer and the union. One of the key areas
where the actions of either the employer or the union may result in a delay in the
election process is the increase in the number of stipulated elections. Stipulated
elections result in an increase in the amount of time from union petition to the
election due to the time necessary for a hearing.‘9

With the stipulated election, employers are able to question the

appropriateness of the proposed bargaining unit. The question of what constitutes

 

laMyron Roomkin and Richard N. Block, "Case Processing Time and The
Outcomes of Representation Election: Some Empirical Evidence." University o_f
Illinois Law Review, 1981: 79.

 

”M, p. 80.

13
an appropriate bargaining unit is one which is extremely important, often difficult
to decide, and therefore time consuming.

The importance of the bargaining unit is twofold. First, the bargaining unit
determination delineates which workers will be able to vote in the representation
election. Second, the bargaining unit determines the group of workers covered by
the collective bargaining contract resulting from negotiations with the certified labor
organization. Those workers who are excluded from the unit are not represented
by the labor union and must adjust their employment problems on an individual
basis, or they may constitute a separate bargaining unit under certain conditions.20

The National Labor Relations Act empowers the NLRB. to decide unit
determination issues. The Act gives the Board only the most modest guidance in
doing so. Section 9(b) of the Act begins:

The Board shall decide in each case whether, in order to assure

to employees the fullest freedom in exercising the rights guaranteed

by this Act, the unit appropriate for the purposes of collective

bargaining shall be the employer unit, craft unit plant unit, or

subdivision thereof...
With this language the Board possesses a considerable amount of discretion in the
establishment of a bargaining unit. When exercising its discretion the Board utilizes
a series of guidelines when establishing a bargaining unit. These guidelines have a
common denominator, the establishment of a unit which will have a "community of

interest" in the exercise of the collective bargaining process.

 

”Black nd Decker Mfg. Q, 147 NLRB. 825 (1964), 70.

 

 

14

The vagueness of Section 9 had led to an increase in the number of
stipulated elections. Employers, knowing the vagueness of the law, and also knowing
the law is designed to protect the interests of differing groups of workers as well as
themselves, may request a Board hearing prior to the election. The employer may
be seeking to establish a proper bargaining unit out of his or her own concern,
concern over the rights of the employees in the bargaining unit, or simply as a
delaying tactic. The use of the hearing as a delaying tactic may have as its goal the
defeat of the union in the representation election.21

Cooke makes an additional comment regarding the difference between
consent and stipulated elections, which proves to be important when modeling
worker choice for union representation.” With a consent election, employers are
"signalling" to the workers in the proposed unit that they do not strongly oppose
collective bargaining. Therefore, workers will tend to increase their expected utility
from unionization. With a stipulated election, a "signal" is given to the workers that
the employer is opposed to unionization and will more actively campaign against the
union. In addition, increased management opposition may be expected during
subsequent negotiations. As a result, workers facing a stipulated election will have
a lower expected utility from unionization.

In addition to the effect that employers may have on the outcome of

representation elections by delaying the election, there are other employer actions

 

2‘William N. Cooke, "Determinants of the Outcomes of Union Certification

”Ib__id., p. 406.

 

15
which may have a negative impact on election outcomes. The actions by employers,
resulting in a decline in the success rate of unions, are summarized by Lawler and
West in the following manner:
First, employers may attempt to bar prounion influences (filtering);
discrimination against union supporters, for example, eliminates
sources of agitation. Supervising training often emphasizes techniques
for enforcing nonsolicitation and nondistribution rules, helping to
insulate the organization from disquieting influence. Another external
substrategy consists of activities which modify the election process and
may prevent the expression of worker’s true preferences. Extreme
(and presumably rare) examples include stuffing ballot boxes, outright
bribery, and intimidation of voters; more common examples include
interference with the union’s campaign by failing to provide an
accurate Excelsior address list, delaying the election through
procedural challenges, and attempting to alter the composition of the
election unit. A third substrategy involves altering contextual
constraints directly; for example, increasing pay or implementing an
employee grievance procedure during an organizing campaign.23
The effect of these external strategies on the outcome of union representation

elections has been a matter of heated debate since the publication of Union

 

Representation Elections: Law m Reality, by Getman, Goldberg, and Herman in

 

z’John J. Lawler and Robin West, "Impact of Union Avoidance Strategy in
Representation Elections." Industrial Relations, 24 (Fall 1985): 409.

16
1976. In that study of several campaign tactics in N .L.R.B. representation elections,
the authors concluded that: (1) Workers pay little attention to either union or
company campaigns; (2) Most worker’s minds are made up far in advance of the
major union and company campaigning; (3) Unlawful campaigning has no greater
effect on workers’ votes than does legal campaigning; and (4) It is impossible to
discriminate between "successful" and "unsuccessful" campaigns.“

In contrast to the findings of German, Goldberg, and Herman, there is an
increasing body of research arguing that external strategic behavior has played a
major role in the declining success rate of unions in representation elections.
Indeed, the results of one study concluded that "as for the decline in union election
success in the last 20 years, the salient determinant appears to be stepped-up
employer resistance.”

One form of management behavior, alleged to result in a decline in the union
success rate in representation elections, is the practice of hiring management
consultants or lawyers prior to the election in the hopes that they may be able to
provide assistance in defeating the union’s campaign. The use of consultants is a
relatively recent phenomenon but becoming increasingly widespread. It is estimated

that 80 percent of all employers engaged in a representation campaign are now

 

“Julius G. Getman, Stephen B. Goldberg, and Jeanne B. Herman, Union
Representation Election: m m Realig (New York: Russell Sage Foundation,
1976) pp. 139-146.

 

2’Ronald L. Seeber and William N. Cooke, "The Decline in Union Success in
N.L.R.B. Representation Elections." Industrial Relations, 22 (Winter 1983): 43.

17

using consultants.” Although the effect of consultants on the outcomes of
representation elections has not been covered by much research, the following
conclusions may be drawn. First, the presence of a consultant may reduce the
probability of union support by 28 percent.” Second, the presence of a consultant
may result in approximately a 6 percent loss of union support at the time of an
election.” Both of the factors show that the effect of a consultant is not large. This
is not to say that the use of a consultant will not affect the outcome of an election.
As Roomkin and Block point out, the typical election is decided by between six and
ten votes. Therefore, the effect of the consultant does not have to be large in terms
of either percentage or number of voters influenced to have a significant effect on
the outcome of an election.29 Third, the effect of a consultant is greater when the
level of union density within an industry is low.30

Discrimination against employees advocating unionization is another

management practice leading to the decline in the success rate of unions in

 

26James A. Craft and Martin Extejt, "New Strategies in Union Organizing,"
Journal Qf Labor Research, 4 (Winter 1983): 29.

”John J. Lawler, "The Influence of Management Consultants on the Outcome

(October 1984): 46.

2"John J. Lawler and Robin West, "Impact of Union Avoidance Strategy in
Representation Elections," Industrial Relations, 24 (Fall 1985): 416.

z"’Myron Roomkin and Richard N. Block, "Case Processing Time and The
Outcomes of Representation Election: Some Empirical Evidence," University o_f
Illinois Law Review, 1981: 92.

 

 

30John J. Lawler and Robin West, "Impact of Union Avoidance Strategy in
Representation Elections," Industrial Relations, 24 (Fall 1985): 416—418.

 

18
representation elections. Here violations of Section 8(a) (3) of the National Labor
Relations Act are encountered. Under Section 8(a) (3) it is an unfair labor practice
for an employer to discriminate "in regard to hire or tenure of employment or any
term or condition of employment to encourage or discourage membership in any

I

labor organization...’ This clause was included in the Act to eliminate the most
effective weapon that employers possessed in their fight against union organization
drives, the firing of those employees who either supported a drive for unionization
or who were leading the drive. Section 8(a) (3) has been in the Act since its
passage in 1935 and substantial case law has been developed so that the meaning of
this section should be clear to all employers. Yet employers continue to dismiss
employees who choose to exercise their right to organize under the Act.

The reason behind the dismissal of employees for their union activities is
simple. Those employees who are dismissed are unable to vote in the
representation election. As noted above, the majority of elections are decided by
only a few votes; therefore the employer need only dismiss a few workers in order
to affect the outcome of many elections. In addition, the dismissal of key union
advocates will give a "chilling" message to the remaining workers, reducing their
participation in the union campaign, or, in other words, reducing the demand for
representation by a labor union.31

The number of 8(a) (3) violations was high immediately after the passage of

the National Labor Relations Act, but as employers became aware of the law, the

 

31Morris M. Kleiner, "Unionism and Employer Discrimination: Analysis of 8(a)
(3) Violations," Industrial Relations, 23 (Spring 1984): 235.

 

19

number of such violations declined. Beginning in the late 19505 the number of such
violations began to increase dramatically. The number of 8(a) (3) unfair labor
practice charges has increased approximately 300 percent since 1960. It may be
alleged that such charges may be a "sour grapes" reaction on the part of labor
unions to the decreasing success rate in certification elections. However, with the
number of backpay awards also increasing 500 percent it appears that the concern
over the increase in 8(a) (3) violations is legitimate. Also the number of violations
being found to be meritorious has not varied widely over time even though the
political makeup of the NLRB. has varied from being predominately Democratic
to Republican in nature. The political makeup of the NLRB. has been found to
lead to bias in the decisions of the Board, so it is surprising that there has not been
a wide range in the percentage of 8(a) (3) cases which have been found to be
meritorious.”

A review of the literature dealing with union representation elections,
compels the following conclusions. Management behavior has an effect on the
outcome of union representation elections. Management may reduce the likelihood
that a union will be successful in a representation election by demanding a
stipulated election rather than a consent election. Raising of spurious unit
determination issues by management results in delay in the election process. The
hiring of consultants appears to enable employers to reduce the success rate of

unions in elections. Finally, the willingness of employers to violate the National

 

32William N. Cooke and Frederick H. Gautschi 111, "Political Bias in NLRB.

1982): 548.

 

20
Labor Relations Act by discriminating against employees engaged in union activity

contributes to a decline in the demand for union representation

The possibility that the provision of benefits by management services as a
substitute for those negotiated by a union and thereby results in a decline in the
demand for union representation is a topic receiving some attention in the literature.
One of the first researchers to examine the possibility that non-union firms would
attempt to match the compensation levels of union firms in order to avoid
unionization was Rosen.33 Rosen argues that once an industry begins to be
organized by a union, the unorganized firms within the industry are faced with the
dilemma of either risking unionization or attempting to forestall unionization by
providing a compensation package approximating that of the unionized firms. In
Rosen’s work, the payment of a compensation package by non-union firms
approximating that paid by the unionized firms so as to avoid unionization is known
as the "threat effect."

More recent work in area of the management substitution hypothesis
considers the provision of a compensation package approximating that paid by

unionized firms as a strategic choice by management. The provision of benefits by

 

33Sherwin Rosen, "Trade Union Power, Threat Effect and the Extent of
Organization," Review o_f Economic Studies, 36 (April 1969): 192.

 

21

employers represents Garbarino’s second alternative to the traditional collective
bargaining model and is termed the "administrative model" of industrial relations.
Under the administrative model the terms and conditions of employment are
established unilaterally by the employer and are administered through the
hierarchical structure of the organization.“ With the administrative model, the rules
and procedures of the employer are either formally established in the form of
personnel policies or handbooks, or are the result of customs or practices known to
the employees and a part of their expectations of future behavior. The informal
rules of the administrative model form an implicit contract, where the various
characteristics of the wage-employment relationship are not specifically negotiated
but are generally accepted to exist by both parties.” The administrative model is in
contrast to the typical collective bargaining model of employee relations systems,
where the terms and conditions of employment are determined by means of a
bilateral negotiation between the employer and the elected representative of the
employees. Under the collective bargaining model, implicit contracts are kept to a
minimum with most of the terms and conditions of employment being established
in a formal contract signed by the representative of both parties.

With the administrative model of industrial relations, management often
takes the initiative in bargaining demands and personnel practices. In unorganized

firms, personnel policies often go beyond the goal of matching provisions in the

 

3"Joseph W. Garbarino, "Unionism Without Unions: The New Industrial
Relations?", Industrial Relations, 23 (Winter 1984): 43.

35M, p. 43.

 

22

unionized firms, in an effort to keep unions out.36 The introduction of personnel
practices designed to keep a firm unorganized represents a "strategic choice" on the
part of management. In recent years there has been a fundamental change in the
industrial relations system.37 As one group of researchers state:

Not since the twenties has it been as socially and politically acceptable

for American management to embrace publicly a "union-free"

approach. Many companies now make union avoidance or union

containment a very high priority. The pluralistic assumption of

industrial relations researchers that independent worker organizations

have a legitimate role seemingly are not shared by the majority of

American employers. This change in management’s view has

coincided with a rearranging of many firm’s industrial relations/ human

resources function, transferring power from labor relations staffs to

those human resource functions associated with union avoidance.38

Initial work on the effectiveness of human resource policies as a strategic
option for management to reduce the demand for unionization began with case
studies of several firms. Kochan, McKersie, and Cappelli, not that in the case of

Emerson Electric, where 10 percent of division managers’ bonuses were tied to

 

3"Thomas A. Kochan, Robert B. McKersie, and Peter Cappelli, "Strategic Choice
and Industrial Relations Theory," Industrial Relations, 23 (Winter 1984): 19.

”M1,, p. 23.
38M» p' 18'

 

23
keeping plants union free, the firm had lost only one of thirty-four representation
election campaigns in the last decade.

Foulkes’ survey of twenty-six large nonunion companies expands the case
study-interview type research into the area of the strategic behavior of large
nonunion firms.39 Foulkes finds management making many strategic choices in an
effort to keep their firms nonunion. Important for the work here, all of the
nonunion companies surveyed had a stated policy of compensating their employees
at average or above-average levels for the industry and plant location. In addition
to paying as liberally or more liberally than their unionized competitors, the
compensation plans at the companies studied had several other features in common.
First, management usually set wages for production and maintenance workers
according to local labor markets rather than adopting a national standard. Second,
the companies closely monitored union settlements in order to maintain equal or
superior compensation. Third, these companies communicated their pay and benefit
policies well, and in many cases had been doing so before the passage of the
Employees Retirement and Income Security Act of 1974.40 The conscious effort to
pay wages equal to or in excess of those paid by unionized firms represents a clear
example of the substitution effect of management.

Being aware that employees are concerned not only with wages but with the

total compensation package, the nonunion companies attempted to be leaders, or as

 

39Fred K Foulkes, "Large Nonunionized Employers," in US. Industrial Relations
1950-1980: A Critical Assessment, Industrial Relations Research Association Series,
1981, pp. 129-158.

 

”mid, pp. 148-150.

 

24
least competitive in the benefits they offered. Foulkes finds that many of the
nonunion companies have actually been the first in their industry with innovative
profit-sharing, medical, or other benefit plans.“

The final strategic practice by management to keep their firms nonunion is
effective communications. Foulkes finds that management devised a number of
communications programs to disseminate information to its workforce and to elicit
employees’ attitudes about the company’s treatment of its workforce. The purpose
of these communication programs is to give employees the opportunity to voice their
complaints and to crate a climate in which it is legitimate to request and receive
help.42 The existence of the communication programs in the nonunion firms presents
an interesting counter argument to the exit-voice hypothesis of Freeman and Medoff,
and represents yet another effort by management to provide similar terms and
conditions of employment in their nonunion setting as could be found in a unionized
setting.

Kochan, McKersie, and Chalykoff extend the research into how corporate
decision-making can bring about changes in industrial relations at the workplace
with an analysis of data from the 1983 Conference Board Survey.“3 This survey

elicited information from leading companies in the United States on a broad

 

“Fred K. Foulkes, "Large Nonunionized Employers," in US. Industrial Relations
1950-1980: A Critical Assessment, Madison, Industrial Relations Research
Association Series, 1981, p. 150.

"Ib_id., pp. 151-152.
“Thomas A. Kochan, Robert B. McKersie, and John Chalykoff, "The Effects of

Corporate Strategy and Workplace Innovations on Union Representation," Industrial
m1 Labor Relations Review, 39 (July 1986): 487—501.

 

 

25

spectrum of human resource policies, including employee participation programs,
profit-sharing, employer-employee communication, grievance systems, and flexible
work schedules. The respondents to the survey were also asked to indicate how
many certification elections were held in the company during the preceding five
years and how many of these elections were won by unions. Where the firms placed
a high priority on union avoidance, the effect of the implementation of the practices
reduced the likelihood of the plant being organized by more than twenty percentage
points. These results are very important in light of the fact that this means that
union election victories added relatively few new members compared to the number
of union members lost due to management strategies.“

Their analysis led Kochan, McKersie, and Chalyoff to draw the following
conclusions:

management innovation and union avoidance strategies substantially

reduced the probability of organization of new plants; union

representation election victories have not offset these management

strategies; and larger union that were able to obtain either neutrality

pledges or voluntary recognition from management have had a more

substantial offsetting effect than unions that have attempted to gain

recognition via the election process.“

 

“Thomas A. Kochan, Robert B. McKersie, and John Chalykoff, "The Effects of
Corporate Strategy and Workplace Innovations on Union Representation," Industrial
m Labor Relations Review, 39 (July 1986): 496.

 

 

“rm, pp. 496—497.

26

Also using the Conference Board Survey, Fiorito, Lowman, and Nelson
expand the research of management substitution effects.“6 In particular, their
utilization of a logistic quantal choice model and estimation by multivariate
maximum likelihood techniques provides additional evidence in support of the
substitution hypothesis. Employee participation and communication policies emerge
as effective means to reduce unionization. Work sharing programs actually enhance
the probability of union success. Grievance systems and pay-for-knowledge schemes
inhibit union success, whereas, all-salaried schemes have a positive impact on the
success rate of unions.“7

Of the control variables utilized by Fiorito, Lowman, and Nelson, only
employer opposition to unions gave consistent results. The size of the bargaining
unit, whether or not there had been wage reductions, the change in the wage rate,
and the wage rate itself surprisingly had no statistically significant effect on the
results. The insignificance of the size of the bargaining unit may be due to the fact
that size is highly correlated with the implementation of human resource policies so
that its effect cannot be separated.

Fiorito, Lowman, and Nelson reach the conclusion that "human resource

 

“Jack Fiorito, Christopher Lowman, and Forrest D. Nelson, "The Impact of
Human Resource Policies on Union Organizing," Industrial Relations, 26 (Spring
1987): 113-126.

"Lid" pp. 122-123.

27
policies which are often part of employers’ union substitution efforts do inhibit
unionization.“8

In summary, the following strategic choices are found to contribute to a
decline in the success rate of unions in organization elections: the size of the plant,
the geographic location of the plant, the presence of a form of employment security
scheme, the promotion of employees from within, the presence of training and
educational opportunities, effective communication between management and
workers, employee participation plans, the presence of an employee grievance
systems, pay-for-knowledge programs, wage payment on a salary basis, fringe benefits
either comparable to or in excess of those paid in the unionized sector, and a wage
level either comparable to or in excess of that paid in the unionized sector. It is
important to note that none of the studies of the management policies attempted to
determine the effect of such policies on the actual willingness of workers to vote for
union representation.

The adoption of many of the current human resource practices appears to
parallel closely the practices of the 19205. During the 19205, managers of large
corporations maintained a sturdy opposition to trade unionism in areas where it had
not yet effectively penetrated. In accordance with the "enlightened" precepts of the
"new economic era" they instituted employment policies in which welfare work,

company unions, and employee-representation plans, played a large part. The

 

“Jack Fiorito, Christopher Lowman, and Forrest D. Nelson, "The Impact of
Human Resource Policies on Union Organizing," Industrial Relations, 26 (Spring
1987): 124.

 

28
increase in the real income of wage earners during the 19205 also contributed to the
decline in trade union organization.

Other changes in the 19205 parallel those today. One of the marked trends
of the 19205 was that fewer persons were attached to manufacturing and agriculture
as the years passed, while a larger proportion of the employed workers were found
in the transportation, communication, clerical, and service occupations.” This
occupational shift resulted in a decline in number of those workers most likely to
seek union representation.

The development of "welfare capitalism" in the 19205 also served to lessen
the appeal for unionism. Here the provision by management of higher real wages,
and policies of direct benefits to employees in the form of improvements in the
physical conditions of employment, insurance against sickness and accidents and
death benefits, pension plans, social, recreational, and educational facilities, company
housing projects, and health and safety work were rapidly expanded.so These policies
are remarkably similar to the ones adopted by firms since the 19705 to avoid the
unionization of their work force.

A further similarity with current developments was the advent of "scientific"
management, resulting in the development of human resource policies or personnel
administration. These policies introduced a formalization of personnel policies

restricting the sometimes arbitrary power of foremen by entitling workers to a

 

49Harry A. Millis and Royal E. Montgomery, Th_e Economics o_f Labor, Volume
_3_, Organized Labor (New York: McGraw-Hill Book Company, 1945), p. 156.

”Ib_id., p. 158.

29
hearing before an allegedly impartial individual or committee. As a result of these
policies, "the leaders of organized labor found themselves without defensive formula

"51

or effective mode of procedure. This is the apparent situation many labor leaders

find themselves in today.

The possibility that the government serves as a provider of benefits, thereby
lessening the demand for union representation, is something the leadership of the
American labor movement has debated for many years. For example, organized
labor was initially cool to the idea of the Fair Labor Standards Act. The reason for
the opposition to the Act was labor "probably feared that government regulation of
wages and hours would remove these two components from collective bargaining."52
Even John L. Lewis objected to the Act on the grounds that it would interfere with

collective bargaining.53 Eventually Lewis supported the Act, but the leadership of

the American Federation of Labor continued to express fear that the provisions of

 

“Harry A. Millis and Royal E. Montgomery, me Economics o_f Labor, Volume
3, Organized Labor (New York: McGraw-Hill Book Company, 1945), p. 160.

 

”Horst Brand, "The Evolution of Fair Labor Standards: A Study in Class
Conflict," Monthly Labor Raview, 106 (August 1983): 27.

 

”ma, p. 27.

 

30
the Act could interfere with the collective bargaining of its affiliated unions.“

As mentioned above, the government as a substitute provider of union
negotiated benefits is termed the legal model of industrial relations by Garbarino.”S
The legal model of industrial relations has two variants. The first variant is the law
of labor-management relations, for example the National Labor Relations Act. This
law is primarily procedural as it establishes and regulates the framework within
which union-management relations are conducted. The second variation of the legal
model deals with substantive regulation of the employee-employer relationship.
Examples of substantive legislation are the Fair Labor Standards Act, the
Occupational Safety and Health Act, the Employees Retirement Income Security
Act, the Civil Rights Act, and Social Security legislation.“ The legal system of
employee relations applies to all workers, whether they be union or nonunion, and
to all employers, with some exceptions by industry.

Substantive labor law has expanded tremendously since the end of the Second
World War. If the government substitution hypothesis is correct, then substantive
labor legislation may have become an alternative for union representation.
Garbarino states the "legal system’s expansion in the sixties and seventies into the
areas of discrimination, health and safety, income protection and other phases of

employment relations may be the most serious threat to the labor movement’s future

 

“Harry A. Millis and Royal E. Montgomery, m Economics o_f Labor, Volume
34 Organized Labor (New York: McGraw-Hill Book Company, 1945) p. 320.

 

”Joseph W. Garbarino, "Unionism With Unions: The New Industrial
Relations?", Industrial Relations, 23 (Winter 1984): 44-45.

”mg, p. 45.

 

31

growth.”7 The threat of the legal model to the grth of the labor movement is a
particular possibility in smaller firms, firms with fewer than 500 employees, because
they are less likely to have an administrative type of industrial relations system than
are the larger firms. Without the existence of an administrative system, or the
presence of a collective bargaining system of industrial relations, the legal system is
likely to be the only functioning employee relations system. With the average size
of the unit in representation elections being fifty employees, if the substitution
hypothesis has validity it is likely its effects will be greatest where unions are
attempting the majority of their organizing efforts.’8

Empirical tests of the government substitution hypothesis are few, with the
major work being that of Neumann and Rissman.”9 Neumann and Rissman focus
their work on two possible substitutes for union services: social welfare expenditures
and judicial limitations to the employment-at-will doctrine.

Turning first to social welfare expenditures as a substitute for union services,
Neumann and Rissman consider government spending on unemployment
compensation, worker’s compensation, educational expenditures, and welfare benefits
as their measure of social welfare expenditures. The argument is that these funds

represent benefits which unions could provide their workers. With their provision

 

”Joseph W. Garbarino, "Unionism With Unions: The New Industrial
Relations?", Industrial Relations, 23 (Winter 1984): 49.

”Herbert G. Heneman III and Marcus Sandver, "Predicting the Outcome of
Union Certification Elections: A Review of the Literature," Industrial and Labor
Relations Review, 36 (July 1983): 537.

S’George R. Neumann and Ellen R. Rissman, "Where Have All The Union
Members Gone?", Journal gf Labor Economics, 2 (April 1984): 175-192.

32
by government there will be a decline in the demand for union services. A time-
series analysis of the period 1904 through 1970 shows a slight substitution effect for
these expenditures, and therefore some support for the substitution hypothesis.

In a critique of Neumann and Rissman, Freeman finds that variations in
social welfare expenditures across countries have no effect on union organizing.60
Freeman also looks at cross-state variations in social welfare spending to see if there
was any effect on change in union density. Again, no effect was found. Freeman
concludes his critique of the findings of Neumann and Rissman by stating, "my
reading of their calculations is that the negative welfare effect is sufficiently frail that
almost any change in model specification, period covered, or measure of social
welfare would yield the insignificant results found in the shorter period."61

It is important to note two things about the studies of Neumann and Rissman
and of Freeman. First, neither study attempts to determine whether or not the
social welfare expenditures considered are factors which actually affect a worker’s
willingness to vote for union representation. The social welfare expenditures
examined do not include the legislation which Garbarino hypothesized would result
in a substitution effect. Second, the social welfare expenditures may not be the sort
of: benefits for which unions negotiate. As a result, they may not be substitutes for
union representation. That these social welfare expenditures are not substitutes for

union negotiated benefits may account for the lack of effect on union organizing.

 

“Richard B. Freeman, "Unionism and Protective Labor Legislation," Proceedings
o_f th_e Thirty-Ninth Annual Meeting, Barbara D. Dennis, ed., Madison, Industrial
Relations Research Association Series, 1987, p. 260-267.

 

6‘I_b_id., p. 266.

 

33

The second possible government substitute examined in the literature are
judicial changes in the employment-at-will doctrine. The basic statement of this
may dismiss their employee(s)-at-will, be they many or few, for good cause, for no
cause or even for cause morally wrong without being thereby guilty of legal wrong."

In recent times the employment-at-will doctrine has been eroded on several
fronts by state court decisions and it is this erosion which may be a possible
substitute for the employment security provisions found normally only with a union
contract. The state court decisions limiting the employment-at-will doctrine are
based on one of three theories: the violation of public policy theory, the existence
of an implied contract theory, and the covenant of good faith and fair dealing
theory.

The public policy exception to the employment-at-will doctrine is the most
widely accepted. Under this exception, an employer may not fire an employee who
is following public policy. Typical cases involve the firing of an employee for
refusing to give false testimony at a trial or administrative hearing, serving on a jury,
reporting illegal conduct by an employer, "whistle-blowing," refusing to violate a
professional code of ethics, filing a worker’s compensation claim, or refusing to take
a polygraph test."2

The implied contract exception to the employment-at-will doctrine exists if

the courts in a state recognize an exception stemming from an employer’s policy

 

“Peterman L International Brotherhood o_f Teamsters Local fl, 174, C. A. 2d

184, (1959), Trombetta y, Detroit Toledo aad Ironton & Q 81 Mich. App. 489
(1978).

 

 

 

 

34

manual, handbook, or other representation. This exception is also referred to as the
implicit contract exception.63

The covenant of good faith exception occurs if courts have based a decision
on good faith and fair dealing in employment contracts. Thus, an employee may be
discharged at will but in doing so the employer must act in good faith.“

Neumann and Rissman utilize cross-sectional analysis to test for the effect
of the court limitations across states. Their results indicate a slight but significant
negative effect on union organizing success."5 In another study of the impact of
employment-at-will judicial decisions on the outcomes of representation elections,
Block, Mahoney, and Corbitt found no impact of the judicial decisions during the
period January 1973 through August 1985."6 They concede that in drawing their
conclusions two caveats must be noted. First, it is unclear if any of the voters in the
representation elections are covered by the exceptions to the employment-at-will
doctrine. Second, they are not sure how many voters in the representation elections

are aware of the exclusions to the employment-at-will doctrine.

 

Pugh v. See Candies m 116 Cal. App. 3d 311 (1981). Weiner L McGraw-Hill, 83

AD 71%) (1981).

“Foussant v. Blue Cross and Blue Shield of Michigan, 408 Mich. 579, (1980).

 

 

. 2
“Fortune a National Cash Register, 373 Mass. 96 (1977). Cleary L American

 

 

Mont. Sup Ct. No. 83-468, (1980).

(”George R. Neumann and Ellen R. Rissman, "Where Have All The Union
Members Gone?", Journal o_f Labor Economics, 2 (April 1984): 186-190.

66Richard N. Block, Christine L. Mahoney, and Leslie Fay Corbitt, "The Impact
of Employment—At-Will Judicial Decisions on the Outcomes of NLRB Elections,"
Proceedings _(_)_f t_h§ Thirty-Ninth Annual Meeting. Barbara D. Dennis, ed., Madison,
Industrial Relations Research Association Series, 1987, pp. 268-275.

 

 

35

The caveats noted by Block, Mahoney, and Corbitt are important in light of
the fact the exceptions to the employment-at-will doctrine apply to only a small
proportion of the individuals who are discharged each year. As Stieber points out:

The overwhelming majority of discharged employees are fired for

such everyday occurrences as: excessive absenteeism or tardiness,

sleeping on the job, fighting in the workplace, horseplay,

insubordination, using abusive or profane language, falsifying company

records or application forms, dishonesty, theft, disloyalty to their

employer, negligence, incompetence, refusal to accept a job

assignment, refusal to work overtime, and possession or use of

intoxicants or drugs. In more than half of the discharges for the above

reasons, arbitrators selected under union-management agreements

have found insufficient evidence to support the discharge penalty and

have reinstated the employee with full, partial, or no backpay

depending on the circumstances in each case. Yet none of these

discharges would qualify as an exception to the employment-at-will

doctrine if they occurred in a non-unionized company.‘57

The current exceptions to the employment-at-will doctrine are very limited
and would not be relevant in the vast majority of discharge cases. The current
exceptions are limited effectively as they are: "used almost exclusively by executives,

managerial, and higher-level employees, who constitute a small minority of all

 

“Jack Stieber, "Most U.S. Workers Still May Be Fired Under the Employment-
At-Will Doctrine," Monthly Labor Review, 107 (May 1984): 36.

 

 

 

36

employees...Typical job titles of plaintiffs in wrongful discharge cases are: company
vice presidents, sales managers, marketing directors, foremen, physicians, sales
representatives, pharmacists, and department managers.""‘3 Stieber argues that this
limitation to higher-level employees is due to several factors. First, such employees
are more likely to consult attorneys and pursue their claims because of higher
expectations of successful legal action. Second, there is an inherent bias in the
nature of the public policy and implied contract exceptions. Lower-level employees
are unlikely to have access to company records nor are they usually in a position to
detect dangerous or illegal acts. Finally, the implied contract exception has little
relevance to lower-level employees because they are rarely in a position to inquire
about future job security when they apply for a job. "Nor are they likely to read
carefully an employee handbook which may give rise to an implied contract
obligation. Even if they were aware of such a handbook provision, most employees
would not realize it could be used to bring a court suit for wrongful discharge.“

Given the limitations noted by Stieber, it is not surprising that the work by
Neumann and Rissman as well as that by Block, Mahoney, and Corbitt finds little,
if any, impact of the employment-at-will judicial decisions on the outcomes of
representation elections. What the current exceptions give is protection for many
of those individuals who are not even covered by the National Labor Relations Act.

Furthermore, the individuals to whom exemptions apply have not been found to vote

 

6"Jack Stieber, "Most U.S. Workers Still May Be Fired Under the Employment-
At-Will Doctrine," Monthly Labor Review, 107 (May 1984): 36.

 

 

“ma, p. 36.

37
for union representation. Neither study attempts to control for the type of workers,
thereby making their results suspect. Therefore, to disregard the union substitution
hypothesis on the results of these studies would be unwise.

In summary, past research into the questions of the government providing
benefits to workers and thereby decreasing the demand for union representation is
lacking in two aspects. First, the items considered as possible substitutes, social
welfare expenditures and exemptions to the employment-at—will doctrine, are unlikely
to be substitutes for benefits negotiated by labor unions. Second, none of the
studies develops a model to show how the existence of benefits provided by the
government actually reduces the demand for union representation by testing the
effect of government provided benefits on the willingness of workers to vote for

union representation.

CHAPTER TWO

A TRUE TEST OF THE SUBSTITUTION HYPOTIESES

The review of the literature points out that there are two major weaknesses
in the previous research examining the substitution hypotheses. The first of these
weaknesses is the lack of an economic model showing how the provision of benefits
by either the government or employers will affect the willingness of workers to vote
for union representation. The second weakness is absence of a test of the effect
benefits have on the willingness of workers to vote for union representation. In this
chapter both of these shortcomings will be corrected.

In previous work, the decision whether or not to vote for union representation
is analyzed in terms of the benefits and costs of doing so. If the expected benefits
from voting for union representation exceed the costs, it is economically rational for
an individual to vote for union representation.70 In this work, the discussion of the
expected benefits of unionization is limited primarily to higher levels of wages and

implied benefits such as protection from discriminatory practices by management for

 

70Henry S. Farber and Daniel H. Saks. "Why Workers Want Unions: The Role
of Relative Wages and Job Characteristics." Journal o_f Political Economy 88 (April
1980):349-369.

 

38

39
black and other minority workers.71 To determine the effect wages and benefits may
have on the willingness of workers to vote for union representation, a model of the
economic decision to vote for union representation is modified to consider not only
wages but also benefits in the compensation package of the worker. The economic
model to be modified is Oswald’s.72
Oswald begins by stating workers attempt to maximize the following indirect

utility function:

v(w) = Max[u(c,h) | c = wh], Eq. 2-1
h

where c represents the consumption of goods and services, w the wage rate, and h
the number of hours worked. Oswald assumes unions control the wage level, while
individuals choose the number of hours they wish to work, and firms determine the
number of individuals who will be employed. The number of individuals employed
by the firms within the industry depends on a continuous labor demand function,
N = N(w), where N is the number of individuals employed and w is the wage rate.

Oswald proposes the following example. Suppose a union is to be formed
within an industry by M workers who are currently receiving some non-union wage
rate WNU. The question becomes how will the workers benefit from unionization and
the expected higher union wage level, WU, they believe they will negotiate? An

economically rational worker knows with a higher wage level, WU, fewer workers will

 

"ma, pp. 365-366.

72Andrew J. Oswald. "The Microeconomic Theory of the Trade Union." Economic
Journal 92 (Septembeer 1982):577-589.

 

 

40
be employed by the firms within the industry. In fact, workers can expect with a
probability q to be employed at the higher union wage rate after unionization and
to be unemployed with a probability of l-q, where

= N(wu)_
M

q

 

Following Oswald and assuming an unemployed worker is able to find another
job at the non-union wage rate, WNU, the expected utility of each worker from the
formation of a union is

E(Wu) = (MWU) + (l-q)V(WNU)- Eq- 2-2
To determine the individual worker’s expected benefit from unionization, Oswald
subtracts v(WNU) from equation 2-2 leaving

B(Wu) = q{V(Wu) - V(Wnu)]- EQ- 2-3
According to Oswald’s model, workers faced with voting in a union representation
election union will compare their expected union wage, WU, and their nonunion
wage, WNU. If the workers believe they will be employed at the higher union wage,
they will vote for union representation.

Oswald’s model is based on the assumptions that unions control the wage
level, individuals choose the number of hours they wish to work, and firms determine
the number of workers. These assumptions are overly rigid, as there are constraints
on the union’s ability to fix the wage rate. Individuals do not have complete ability
to choose the number of hours they wish to work due to standardized work-week
requirements of most employers. Finally, certain union contracts limit the

employer’s ability to determine the number of workers employed.

 

41

For the work here, Oswald makes yet another simplifying assumption. He
considers only wages and not wages and benefits in the expected utility function.
The omission of benefits is important for the testing of the substitution hypotheses.
Furthermore, the omission of benefits must be corrected for three additional reasons.

First, there is the legal perspective. Under Section 8(d) of the National
Labor Relations Act employers and unions are required to bargain in good faith
over "wages, hours and other terms and conditions of employment." The phrase
"terms and conditions of employment" is interpreted broadly by the National Labor
Relations Board and by the Supreme Court. With the decision in N.L.R.B. y,

Wooster Division o_f Borg-Warner Corporation, 356 US. 342 (1958), the National

 

Labor Relations Board and later the Supreme Court established certain "mandatory
items" of collective bargaining. These mandatory items include such things as
subcontracting,” stock purchase plans,74 profit sharing plans,” pension and employee
welfare plans,“ work loads and production standards,77 and plant rules.78 Thus, the
mandatory items of collective bargaining include not only wages but benefits as well.

The second reason why other terms and conditions of employment cannot be

ignored is the evidence from earlier research on the attitudes of workers towards

 

73Fibreboard hpfi! MS 2. N.L.R.B., 379 US. 203 (1964).
7"Richfield Qi_l_ Company, 110 N.L.R.B. 356 (1954).

75M Q Midi Manufacturing, 129 N.L.R.B. 112 (1960).

“mama E Company y. N.L.R.B., 336 US. 960 (1949).

”mp Lecp pm a Finishing Company, 121 N.L.R.B. 953 (1958).
”Miller BLwing Company, 116 N.L.R.B. 90 (1967).

 

42

unions and the effect of unions on the provision of fringe benefits. The work of
Kochan shows that the major reason why workers prefer unionization is that unions
improve both wages and fringe benefits.79 In addition, Woodbury finds some fringe
benefits and wages are substitutes in the utility functions of workers."0

The findings of Freeman show that in organized firms the fringe benefit share
of total compensation is raised. In particular, unions have their greatest effect on
pensions, on life, accident, and health insurance, and on vacation and holiday pay.81

Therefore, when modeling why workers choose to vote for union
representation, the expected utility function should include not only wages but

benefits as well. The expected utility of each worker will no longer be

E(Wu) = qV(WU) + (l'q)V(WNU) Eq. 2-2
but rather,
E(WU,BU,) = qV(WU,BU) + (l-q)V(WNU,BNU). Eq. 2-2a

In equation 2-2a WU represents the union wage, 13,, a vector of benefits a worker will
receive if the employer becomes unionized, WNU the non-union wage level, and BM,

the vector of benefits the worker will receive if the firm remains non-union. In a

N(WuBU)
similar fashion, q = , and the demand for labor in the industry is
M

 

7”Thomas A. Kochan. "How Amewrican Workers View Labor Unions," Monthly
Labor Review 102 (April 1979):24.

 

 

”Stephen A. Woodbury. "Substituion Between Wage and Nonwage Benefits,"
American Economic Review 73 (March 1983):174-179.

 

81Richard B. Freeman. "The Effect of Unionism on Fringe Benefits," Industrial
@ Labor Relations Review 34 (July 1981):502.

 

 

43
N = N(WU,BU), reflecting the cost of total compensation of the workers to the
employers.

With the inclusion of benefits in the analysis the expected benefit from
unionization for the individual worker is

B(Wu,3u) = (llV(Wu,Bu) ' V(W~u,l3nu)]- Eq. 2-3a

Equation 2-3a allows one to see how the provision of benefits by either
management or government may affect workers’ willingness to vote for union
representation. What Neumann and Rissmann argue in their discussion of the
employment-at-will doctrine is that one of the benefits in the union benefit vector
is a form of protection against unjust discharge and that the various state court
rulings have the effect of giving non-union workers the same level of protection
against unjust discharge as union workers. Although one of the benefits unions
negotiate for their members is some form of progressive discipline procedure,
Stieber’s argument that the employment-at-will remedies are used primarily by
management employees make this an inadequate test of the government substitution
hypothesis as management employees are not usually union members.

The second test of the government substitution hypothesis of Neumann and
Rissman is that the level of government social welfare expenditures affects the
willingness to vote for union representation. Again, these expenditures are unlikely
to be found in the vector of benefits negotiated by a union, BU. Thus, the
conclusions of Neumann and Rissmann are not likely to be valid.

The goal of the management substitution hypothesis, as articulated by

Foulkes, is to compensate employees at average or above average levels for the

 

44
industry and plant location.82 In terms of the union choice model of equation 2-3a,
the goal of this managerial policy will be to equate the workers’ expected utility from
the nonunion compensation package, v(WNU,BNU), with the expected utility of the
union compensation package, v(WU,BU), thereby reducing the likelihood that workers
will find the expected benefits of unionization large enough to vote for union
representation.

To test the validity of the substitution hypotheses, two basic pieces of
information are needed. The first is a measure of total compensation received by
the workers. Data are needed which provide information on both the level of
earnings and the extent of coverage by various fringe benefits. This data will allow
a comparison between the compensation packages of union and nonunion workers.
Second, there must be a measure of the willingness of the worker to vote for union
representation. By combining the information on total compensation and the
willingness to vote for union representation, we overcome the major problems of the
earlier work.

Data for the testing of the substitution hypotheses are available from the 1977

Cross Section o_f 111; Quality pf Employment Survey, QES. The QES provides

 

 

measures of employee compensation for earnings and benefit coverage. In addition,
the QES contains information on the willingness of an individual to vote for union
representation. (For a discussion of the variables utilized in the analysis see

Appendix A of Chapter 2.) Of the fringe benefits workers receive, only three,

 

”Fred K. Foulkes. "Large N onunionized Employers," in U.U. Industrial Relations
1950:1980; A Critical Assessment (Madison, Industrial Relations Research
Association Series, 1981): 148-150.

45
medical insurance, retirement benefits, and paid vacations, occur in enough cases to
allow further analysis. Table 2—1 compares the means of earnings and the relative
frequency of the fringe benefits between unionized and nonunionized workers.

Table 2-1 has only one result not in keeping with the idea that unionized
workers have greater total compensation. First, the earnings of the workers covered
by a union contract are significantly greater than those of the non-union workers.
Second, unionized workers are more apt to have medical insurance, 94 percent, than
are non-union workers, 72 percent. Third, unionized workers are more likely to have
retirement benefits, 91 percent, than are non-union workers, 59 percent. Fourth,
there is no statistical difference between union and non-union workers regarding
vacations, with each reporting the presence of vacation benefits in approximately 83
percent of all cases.

Due to the binary nature of the response, the results should be interpreted
with caution. Although the zero-one nature of the data gives a narrow measure of
each benefit, a preferred measure would include the dollar value of such benefits.
For example, if the less frequently occurring medical insurance plans in the
non-union sector are all available with no co-payment, whereas all of the medical
benefits for the unionized workers require a large co-payment, it may be argued that
although unionized workers are more likely to receive medical insurance, the
non-union workers have a better medical insurance benefit. Another example of the
weakness of the binary data is provided by the vacation variable. Although there is
no statistical difference between the availability of vacation benefits, the data provide

no information as to the length of time allotted for vacations.

 

46
Table 2-1

Current Fringe Benefits by Union Affiliation

 

 

Benefit Mean Std. Dev. Cases
Union Workers

Earnings 15133. 7409. 323
Medical Benefits .241 418
Retirement Benefits .907 .289 413
Vacation Benefits .828 .377 420
Super Medical Benefits .949 .219 434
Super Retirement Benefits .935 .245 434
Want Super Medical Benefits .345 .476 434
Want Super Retirement Benefits .737E-01 .261 434
Want Super Vacation Benefits .322E-01 .176 434
Nonunion Workers

Earnings 13882. 11283. 640
Medical Benefits . .447 794
Retirement Benefits .588 .492 792
Vacation Benefits .825 .379 793
Super Medical Benefits .671 .469 983
Super Retirement Benefits .609 .488 983
Want Super Medical Benefits .342 .474 983
Want Super Retirement Benefits .157 .364 983
Want Super Vacation Benefits .467E-01 .211 983

47

Table 2-1 also shows the means for five additional variables: super medical
benefits, super retirement benefits, want super medical benefits, want super
retirement benefits, and want super vacation benefits. These variables are
constructed to provide broader measures of benefits. The super medical benefit
variable, SMED, has a value of one if the respondent has either medical, surgical,
or hospital insurance that covers any illness or injury which may occur while off the
job, or sick leave with full pay, or dental benefits, or eyeglass or eyecare benefits.
The super retirement benefits variable, SRET, has a value of one if the respondent
has either a retirement program, a profit-sharing plan, stock options, or a thrift or
savings plan. Of these two variables, SMED is of particular importance for the
government substitution hypothesis as it is an indicator of the type of benefits to be
provided if a national medical insurance program were to enacted.

The results in Table 2-1 show broad medical coverage, SMED, is much more
likely with a unionized employer, 95 percent, than a non-union employer, 67 percent.
The presence of broad retirement benefits is also found to be more likely with union
employers, 94 percent of all cases, as opposed to 61 percent of the non-union cases.

The final set of variables in Table 2-1, want super medical benefits, WSMED,
want super retirement benefits, WSRET, and want super vacation benefits, WSVAC,
represents measures of the desire for broad medical, retirement, and vacation
benefits on the part of those workers who do not have such benefits. The mean
desire for the broad measure of medical benefits is the same for both union and
non-union workers: 34 percent. The desire for broad retirement benefits is twice as

prevalent for non-union workers, 15 percent, as for union workers, 7 percent. The

48
desire for improved vacation benefits is also larger for the non-union workers, 5
percent, as opposed to 3 percent for union workers.

Given the comparisons of the means of the primary fringe benefits as well as
the earnings of the union and non-union workers, we offer the following hypotheses.
First, the willingness of a worker to vote for union representation will be affected
by the presence of medical benefits. Those workers who do not have medical
benefits are more likely to vote for union representation than are those who have
medical benefits. Second, the workers who do not have retirement benefits will be
more likely to vote for union representation. Third, it is unlikely that vacation
benefits will have any effect on the willingness of a worker to vote for union
representation. Fourth, the lower the level of earnings the more likely an individual
will be to vote for union representation. The inverse relationship between earnings
and the willingness to vote for union representation represents a straight forward
example of the management substitution hypothesis. If non-union employers are
willing to pay a level of wages approximating the higher wages of the union
employer, then they will reduce the demand by workers for union representation.

To test the hypothesis that the availability of fringe benefits will affect the
willingness of a worker to vote for union representation a probit analysis is utilized.
The hypothesis is: the more extensive the compensation package, the less likely
workers are to wish to vote for union representation.

In recognition of the fact that certain personal characteristics of workers are
found to influence an individual’s willingness to vote for union representation, four

control variables are included in the initial analysis. The first control variable is

49

BONW, indicating whether or not a worker is black or other non-white race. This
variable is included, as non-white workers are more willing to vote for union
representation than white workers."3 This control variable should have a positive
coefficient. Second, is a control variable for the sex of the respondent, FEML. This
control in necessary as past research indicates women are more likely to vote for
union representation than are men.“ The third control variable included is for the
region of the country. Prior research indicates workers in the South are less likely
to vote for union representation than are workers in other regions of the country.“
The fourth control variable is the age of the respondent. Older workers are found
to be less likely to vote for union representation.“

In addition to the control variables for the personal characteristics of the
workers, three control variables are included to provide measures of employment
stability. The first of these control variables, LAYOFF, is whether or not the worker
experienced a period of layoff in the last year. The second control variable, XLO,
is whether or not the worker expects to be laid off in the forthcoming year. A

positive response to either of these variables is expected to be related with the

 

”Henry S. Farber. "The Determinants of the Union Status of Workers,"
Econometrica 51 (September 1983):1431.

“Henry S. Farber. "Trends in Worker Demand for Union Representasentation."
American Economic Review 79 (May 1989):167.

8’Henry S. Farber. "The Determinants of the Union Status of Workers."
Econometrica 51 (September 1983):1431.

“Henry S. Farber and Daniel H. Saks. "Why Workers Want Unions: The Role
of Relative Wages and Job Characteristics," Journal o_f Political Economy 88 (April
1980):362.

 

 

 

50

willingness to vote for union representation. Finally, a control variable, FOJ, is
included as a measure of the workers belief in their ability to obtain another job
with wages and benefits similar to the one they presently have. This control variable
provides a measure of q in equation 2-3a and should have a positive coefficient.

The dependent variable in the analysis is the response by workers who are
neither self-employed, members of a labor union, nor covered by a collective
bargaining agreement to the following question in the E: "If an election were held
with secret ballots, would you vote for or against having a union or employees’
association represent you?" Thus, the QES provides a direct measure of the
willingness of the respondent to vote for union representation. The dependent
variable is indicated as VFU in the results. It is important to note, that unlike
previous studies on the management and government substitution effects, a direct
measure of the willingness of a worker to vote for union representation is utilized.

The results of the initial analysis, shown in column 1 of Table 2-2, indicate
that the level of earnings, ERN, is related negatively to the willingness to vote for
union representation. Second, the presence of medical insurance, MED, has the
expected negative coefficient but is not statistically significant. The coefficient for
retirement benefits, RET, and vacation benefits, VAC, are positive, contrary to
expectations, but neither is satistically significant.

The coefficient for black or non-white workers, BONW, is positive and
significant. The coefficient for sex, FEML., is positive. Contrary to expectations, the
coefficients for region, SOUTH, and age, AGE, are positive. The coefficient for

region is statistically significant.

51
Table 2-2
Maximum Likelihood Estimates

Current Fringe Benefits and the Willingness to
Vote for Union Representation

 

 

Variable 1 2 3
CONSTANT -.696* ‘ " -1.446* " ‘ -1.300"‘ * * *
(.258) (.441) (.429)
ERN -.224E-04* * ‘ * -.109E-04* “‘ * -.120E-04* * "‘
(.833E-05) (.901E-05) (.899E-05)
MED -.329E-01 .475E-01 .465E-01
(.175) (.185) (.186)
RET .438E-01 .159 .113
(.146) (.163) (.166)
VAC .940E-01 .578E-01 .473E—01
(.196) (.205) (.209)
BONW 898““ 1001"" 1006“”
(.210) (.222) (.223)
FEML .255 * * * .249 * * .201 * *
(.139) (.158) (.152)
SOUTH 248*“ 337*“ 281*"
(.133) (.141) (.142)
AGE .914E-03 .136E-02 .875E-03
(.177E-02) (.398E-02) (.288E-02)
LAYOFF 293 .106 .132
(.443) (.453) (.454)
FOJ - 246 .261E—01 .192E-01
(.130) (.139) (.137)
XLO .616 617”" 662""
(.199) (.207) (.208)
< 12YED .786E-01 .381‘
(.265) (.254)
12YED .179 .447“ " *
(.221) (.217)
12—16YED -.237 -.535E-01
(.229) (.229)
CDGR -.136 -.355E-01*
(.245) (.256)
<1YT 376*“ .442***
(.188) (.185)
1-5YT .333* * * .376" * *
(.159) (.156)

 

52

Table 2-2
(Continued)
Variable l 2 3
< 50 .179E-01 .494E-01
(. 176) (.183)
50-499 -.422E-01 .345E-01
(.178) (. 179)
CLR . 1 17
(.203)
CFT -.660E-01
(.228)
OPR .593 * "‘ * *
(.240)
TOP .800" * *
(.4 18)
SER .471 * * *
(.233)
MIN -.958* *
(.592)
CON -.259
(.368)
MFG .155
(.208)
TRN -. 166
(.308)
WHL -.535 *
(.397)
RTL -.293 ‘
(.225)
FIN -. 169)
( .256)
BRS -.115
(.359)
PSR -.102
(.450)
PUB -.207
(.278)
Log-L -283.76 -269.55 -272.34
Chi-Sq. 56.60 86.01 80.43
Sample Size 494 494 494
Mean of VF U .328 .328 .328

Standard errors in parentheses.
‘ Significant at the 0.15 level

"* Significant at the 0.10 level
*** Significant at the 0.05 level
”** Significant at the 0.01 level

53

The control variables for the layoff, LAYOFF, and expected layoff, XLO,
have the expected positive signs, with the coefficient for expected layoff being
significant. The control for the ability of workers to find another job has a negative
coefficient, contrary to what is expected.

This initial analysis gives mixed support for the possibility of either a
government substitution hypothesis or a management substitution hypothesis. The
negative and significant results for earnings and medical benefits support the
substitution hypotheses. The negative and significant coefficient for retirement
benefits does not support to the substitution hypotheses.

In recognition that other worker characteristics have an effect on the
willingness of workers to vote for union representation, the analysis is expanded to
include controls for occupation, industry, educational attainment, and tenure with the
employer. Occupational controls are included for clerical and kindred workers,
CLR, craftsmen and kindred workers, CFT, operatives, except transport, OPR,
transport equipment operative, TOP, and service workers, SER. Occupational
controls are included because of the wide variation in the level of unionization
across occupations.37 Industrial controls are included for mining, MIN, construction,
CON, manufacturing, MAN, transportation, communications, and other public
utilities, TRN, wholesale trades, WHL, retail trades, RTL, finance, insurance, and

real estate, FIN, business repair services, BRS, personal services, PSR, and public

 

37Richard B. Freeman and James L. Medoff. "New Estimates of Private Sector

1979):162-165.

 

54

administration, PUB. These controls are included due to the variation in
unionization across industries."8

Educational controls are included for those workers with less than twelve
years of education, < 12YED, workers who had completed a high-school education,
12YED, workers with some college, 12-16YED, and finally those workers who had
graduated from college, CDGR. Those workers with an educational attainment of
a high-school degree or less, <12YED or 12YED, are likely to vote for union
representation, while those workers with some college, 12-16YED, and college
graduates, CDGR, are unlikely to vote for union representation. In addition, control
variables are included for the size of the employer. Controlling for the size of the
employer is necessary for two reasons. First, as noted above, most of the recent
union elections have taken place in a bargaining unit of fifty or less workers.
Therefore, a control variable, <50, is included for employers of a size of fifty or less.
Second, because the administrative model of industrial relations usually does not
develop until an employer size of at least 500 employees, another control variable,
50-499, for employer size of between fifty and 499 employees is included.
Furthermore, with an employer size of less than 500 employees the legal model of
industrial relations is likely to be the dominant form of industrial relations system.
With the dominance of the legal model of industrial relations the possibility of the
government substitution hypothesis is larger. The coefficients for both of the size

variables are expected to be negative under the government substitution hypothesis.

 

”lb—id, pp. 155-161.

55

A second set of variables is included to control for the tenure of the employee
with the firm. Here two variables are added. The first of the two control variables
is for those individuals who have less than one year with the employer, < 1YT, and
the second is for those employees who have between one and five years of tenure,
1-5YT, with the employer. The reason for the inclusion of these additional control
variables is twofold. First, the availability of some benefits is related to the seniority
an employee has with a firm. Some benefits are not available to employees until
they accumulate a contractually stipulated amount of seniority. Second, familiarity
with the terms and conditions of employment, either in terms of available benefits
or actual working conditions is a learning process and those workers with the least
amount of tenure are less likely to be fully aware of the true conditions. As tenure
with an employer increases, both knowledge and availability of benefits increases.
Therefore, increases in tenure are related negatively to the willingness to vote for
union representation. The two measures of tenure included, less than one year,
<1YT, and between one and five years of tenure, 1-5YT, should have positive
coefficients. Column 2 of Table 2-2 presents results for the additional controls of
age, education, occupation, tenure with the employer, and employer size. Column
3 of Table 2-2 presents results for industrial controls. In both cases, earnings, ERN,
remains related negatively to the willingness to vote for union representation. None
of the coefficients for the benefit variables, MED, RET, or, VAC, have the negative
signs necessary to offer support to the substitution hypotheses. The control for sex
is positive and significant. The regional control variable remains positive and

significant, counter to what is expected. All of the controls for job stability have the

 

56

expected positive coefficients, with the coefficient for expected layoff, XLO, being
statistically significant. The controls for education produce the expected results.
Workers with lower levels of educational attainment, <12YED and 12YED, are
found to be willing to vote for union representation. The workers with higher levels
of educational attainment, 12-16YED and CDGR, are not found willing to vote for
union representation. The controls for tenure of the employees have the expected
positive coefficients, and in the case of the occupational controls the coefficients are
significant. The controls for the size of the employer produced mixed results. In the
case of the occupational controls the coefficient for firms of between fifty and 499
employees is negative as expected, but in all other cases the coefficients are positive,
running counter to expectations. The results of occupational controls show that
clerical, CLR, operatives, OPR, transport operatives, TOP, and service workers,
SER, are all likely to vote for union representation. The results for operatives,
transport operatives, and service workers are statistically significant. The coefficient
for craftsmen, CFT, is negative but is not significant. Of the industrial controls only
manufacturing, MFG, has a positive coefficient. The negative coefficients for
mining, MIN, wholesale, WHL, and retail, RTL, are significant.

An expansion of the analysis to include the broader measures of medical
insurance, SMED, and retirement, SRET, gives similar results. As shown in columns
1, 2, and 3 of Table 2-3, earnings, ERN continues to be related negatively to the
desire for union representation. The broad measures of employee benefits offer
mixed support for the substitution hypotheses. The coefficient for broad medical

coverage, SMED, is negative when neither the occupational or industrial controls are

 

Maximum Likelihood Estimates

57
Table 2-3

Current Broad Based Fringe Benefits and the Willingness to

Vote for Union Representation

 

 

Variable 1 2 3
CONSTANT -.767‘**“ —1.463*"* -1.331"‘***
(.268) (.430) (.423)
ERN -.126E-04* * ‘ -.474E-05 -.410E-05
(.721E-05) (.795E-05) (.788E—05)
SMED -.854E-01 .174 .195
( .233) (.249) (.206)
SRET -. 128 -.309 -.922E-01
(.189) (.206) (.206)
VAC .721E-01 .458E-01 .843E-02
(.203) (.211) (.215)
BONW 913"“ 1.004**** 1.004"**
(.202) (.212) (.213)
FEML .312* * * .290* * .243“ *
(.136) (. 155) (.149)
SOUTH .219 309* * * .258*
(. 132) (. 140) (.140)
AGE .5 01E-03 .895E-03 .421E-03
(.163E-02) (.314E-02) (.250E-02)
LAYOFF .625 * * .3 11 .402
(.416) (.433) (.429)
FOJ -.244 .1 18E-01 .143E-01
(.128) (. 137) (.135)
XLO .581t** .589**** .638****
(.194) (.201) (.202)
< 12YED —.857E-01 .387* *
(.261) (.251)
12YED . 195 .469“ * “
(.218) (.213)
12-16YED - 222 -.388E—01
(.224) (.224)
CDGR - 917 -.546E—02
(.241) (.250)
<1YT 390*" 457"“
(. 179) (.177)
1-5YT .355" * * .408“ * * *
(. 153) (.151)
< 50 -.386E-01 .460E-02
(.170) (. 177)

 

58
Table 2-3 (Continued)

Variable 1 2 3
50-499 -.571E-01 -.129E-01
(. 176) (.176)
CLR . 135
(.199)
CFT -. 1 12
(.223)
OPR .542“ * *
(.236)
TOP .929* * *
(-404)
SER .413 * * *
(.232)
MIN -.96 1 * *
(.589)
CON -.367
(361)
MFG -.299
(.203)
TRN -.104
(.297)
WHL -.558 *
(.387)
RTL -.333 *
(.217)
FIN -. 154)
(.254)
BRS -. 191
(.351)
PSR .128
(.449)
PUB -. 15 1
(.271)
Log-L -292 -277.33 -288.21
Chi-Sq. 55.941 86.52 80.76
Sample Size 507 507 507
Mean of VFU .327 .327 .327

Standard errors in parentheses.
* Significant at the 0.15 level

** Significant at the 0.10 level
*** Significant at the 0.05 level
**** Significant at the 0.01 level

59

included, thereby supporting the substitution hypothesis. The results for the
retirement variable are negative in each case, but are not statistically significant.
The results for the vacation variable, VAC, do not support the substitution
hypothesis as the coefficients are positive. The control for race, BONW, is positive
and significant in all cases. The coefficient for sex, FEML, is positive and the
coefficient for region, South is positive, contrary to expectations.

The controls for employment stability all have the hypothesized positive signs, with
the results for the control for expected layoff, XLO, being significant. The controls
for tenure show that workers with little tenure, either less than one year, <1YT, or
between one and five years of tenure, 1-5YT, are likely to vote for union
representation. The results being statistically significant. The results of the
employer size continue to support the notion of the government substitution
hypothesis, as the coefficients for both variables, <50 and 50-499, are negative. The
coefficient for employers of size fifty to 499 employees being significant. In the
results for the occupational controls, clerical workers, CLR, operatives, OPR,
tranport operatives, TOP, and service workers, SER, are all positive. The positive
and significant effect for service workers is surprising, as it has been argued that the
growth of the service sector, an occupation thought not to be receptive to
unionization, is responsible for part of the relative decline in unionization.89 The
control for craftsmen, CFT, is negative but not significant. With the industrial

controls little is revealed. With the exception of personal services, PSR, and public

 

”Richard B. Freeman. "Contraction and Expansion: The Divergence of Private
Sector and Public Sector Unionism in the United States." Journal o_f Economic
Perspectives 2 (Spring 1988):66.

 

Maximum Likelihood Estimates

60
Table 2-4

Desire for Broad Based Fringe Benefits and

The Willingness to Vote for Union Representation

 

 

Variable 1 2 3
CONSTANT -.927* * * * -1.340* * * * -1.277* * *
(.209) (.356) (.343)
ERN -.128E-04* “ ‘ -.486E-05 * * * -.453E-05 * ‘ *
(.713E-5) (.802E-05) (.793E-05)
SMED -.307“ ‘ ‘ * .282‘ * * .303" * *
(.125) (.130) (.131)
SRET .211 .153 .144
(.159) (.167) (.167)
VAC .383* * .334 .388“
(.286) (.303) (.301)
BONW .929* * * ‘ .996‘ * * ‘ .997‘ * * *
(.205) (.216) (.215)
FEML .337" * * * .294 * * * .260“ * *
(.137) (.155) (.150)
SOUTH .224" 299*" 257*"
(.132) (.140) (.141)
AGE .663E-03 .127E-02 .597E-03
(.184E-02) (.399E-02) (.274E-02)
LAYOFF .461 .162 .223
(.424) ( .443) (.440)
FOJ -.527E-01 -. 176 -.655E—02
(. 130) (.138) (.138)
XLO .537**** 536"" 588”"
(.196) (.204) (.204)
< 12YED .528E-02 .327*
(.257) (.264)
12YED .170 .459* * * *
(.217) (.213)
12—16YED -.235 -.444E-01
(.223) (.224)
CDGR -.102 .248E-02
(.240) (.250)
< 1YT .285 * * .346 ‘ * “
(.182) (.179)
1-5YT .294" "‘ * .343 * * *
.157) (.154)
< 50 -.744E-01 -.755E-02
(.165) (.174)

61
Table 2-4 (Continued)

Variable 1 2 3
50-499 -.859 -. 13 1
(. 175) (.177)
CLR .134
(.198)
CFT -. 128
(.224)
OPR 532* * *
(.238)
TOP .895
(.412)
SER .425 ‘ * *
(.232)
MIN -.921 * “‘
(.598)
CON -.397
.361)
MFG .609E-01
(.204)
TRN -. 159
(.302)
WHL -.633 “ * *
(.382)
RTL -.398* * *
(.219)
FIN -.230
(.256)
BRS -.352
(.256)
PSR -. 162
(.447)
PUB -. 149
(.268)
Log-L -286.69 -273.52 -275 .79
Chi-Sq. 67.80 94.15 89.611
Sample Size 507 507 507
Mean of VFU .327 .327 .327

Standard errors in parentheses.
* Significant at the 0.15 level

** Significant at the 0.10 level
"” Significant at the 0.05 level
**** Significant at the 0.01 level

62

employment, PUB, all of the coefficients are negative the coefficients for mining,
MIN, retail trades, RTL, and wholesale trades, WHL, being significant. The Q_E_S
also provides information on fringe benefits which the respondents do not currently
have but they want, WSMED, WSRET, WSVAC. The next step in the analysis is
to determine if the desire for these benefits has an effect on the willingness of the
respondents to vote for union representation. With these variables, it is expected
that non-union workers who want the benefits are willing to vote for union
representation, as unionized workers are more apt to have the benefits. With the
variables WSMED, WSRET, WSVAC, a positive coefficient will lend support to the
substitution hypotheses. The results in columns 1, 2, and 3 of Table 2-4 provide
support for the substitution hypotheses. The coefficients for earnings, ERN, are
negative, thereby supporting the substitution hypotheses. The coefficients for desired
broad medical benefits, WSMED, have the hypothesized positive sign and are
significant. The coefficients for the measure of desired retirement benefits, WSRET,
is positive, supporting the substitution hypotheses. The results for desired retirement
benefits are significant only when neither the occupational nor industrial controls are
included. The coefficients for the measure of desired vacation, WSVAC, is positive,
supporting the substitution hypotheses, but is only significant when industrial controls
are included.

The next step in the testing of the substitution hypotheses is to include both
the broad measures of currently held fringe benefits, SMED, SRET, and VAC, and
the broad measures of desired benefits, WSMED, WSRET, WSVAC. This analysis
will show the joint effects of currently held fringe benefits as well as the desire for

fringe benefits on the willingness to vote for union representation.

63

The results of the analysis with the six measures of fringe benefits are shown
in columns 1, 2, and 3 of Table 2-5. Earnings, ERN, continue to be related
negatively and significantly to the willingness to vote for union representation. The
desire for broad medical coverage, WSMED, has a positive and significant effect on
the willingness to vote for union representation. Therefore, medical benefits appear
to be an important determinant in an individual’s willingness to vote for union
representation. The effect of the broad measure of retirement benefits on an
individual’s willingness to vote for union representation is less clear. The presence
of retirement benefits, SRET, is related negatively to voting for union representation
and the results are significant in two of the three cases, supporting the substitution
hypotheses. On the other hand, with no controls or with the occupational controls,
the desire for retirement benefits, WSRET, is found to be related positively to the
willingness to vote for union representation, supporting the substitution hypotheses.
These results are not significant.

With the finding the desire for broad medical coverage, WSMED, is the key
fringe benefit in determining the willingness to vote for union representation,
attention is turned to an estimation of the desire for broad medical benefit coverage.
This estimation will show where the provision of medical benefits may have their
greatest effect on the willingness to vote for union representation. To ascertain the
factors which determine the demand for medical benefit coverage it is necessary to

control for personal attributes of the workers, characteristics of the workplace, and

64
Table 2-5
Maximum Likelihood Estimates Current Broad Based Fringe

Benefits, Desire for Broad Based Fringe Benefits and the
Willingness to Vote for Union Representation

 

 

Variable l 2 3
CONSTANT 1159"" 1778"" 1629""
(.298) (.454) (.440)
ERN -.133E-04* * ‘ -.608E-05 * "‘ * -.5 10E-05 * ‘ *
(.733E-05) (.810E-05) (.802E-05)
SMED .160 .257 .267*
(.237) (.253) (.249)
SRET -.357* * .522E-01 -.122E-01*
(.197) (.212) (.211)
VAC .129 .962E-01 .717E-01
(.210) (.220) (.222)
WSMED .309“ * * * .290" * ‘ .308" * *
(.125) (.131) (.132)
WSRET .219E-01** .181 .156
(.166) (.173) (.173)
WSVAC .496' * .474 .508
(.304) (.321) (.318)
BONW 933"" 1013”“ 1015”"
(.206) (.217) (.216)
FEML .359“ * “ ’ .328" * * t .280" * *
(.138) (.158) (.151)
SOUTH .230‘ * * .315“ * * .274‘ * *
(.133) (.141) (.141)
AGE .502E-03 ~946E-03 .402E-03
(.183E-02) (.367E-02) (.277E-02)
LAYOFF .481 .201 .250
-.447E-01 (.445) (.442)
F OJ (.421E-01) -.797E-02 .304E-02
.192 (.139) (.138)
XLO (.197) 560“” 609”"
(.205) (.206)
< 12YED -.857E-01 .395 * *
(.264) (.253)
12YED -.213 .497“ * *
(.220) (.215)
12-16YED -.207 -.189
(.225) (.225)
CDGR -.808E-01 .332E-01
(.242) (.252)

 

65

Table 2-5 (Continued)

Variable 1 2 3
< 1YT .334 -.383
(. 185) (. 182)
1-5YT -.295 -.341
(. 158) (.155)
< 50 -.237E-01 -.150E-01
(.172) (.178)
50—499 -.698E-01 -.93 1E-O2
(.177) (.177)
CLR .900E-01
(.201)
CFI‘ -.149
(.225)
OPR .495 ‘ * "
(.239)
TOP .879* * “
(.410
SER .45 1 * * *
(.233)
MIN -.936
(.605)
CON -.386
.363)
MFG -.964E-01
(.207)
TRN -. 180
(.302)
WHL -.656“ * *
(.384)
RTL -.388* * *
(.220)
FIN -.256
(.258)
BRS .277
(.357)
PSR -. 138
(.444)
PUB -. 180
(.271)
Log—L -285.98 -272.28 —274.79
Chi-Sq. 69.22 96.63 91.60
Sample ,Size 507 507 507
Mean of VFU .327 .327 .327

Standard errors in parentheses.

‘ Significant at the 0.15 level *" Significant at the 0.10 level
*" Significant at the 0.10 level **"‘* Significant at the 0.01 level

66

the industry and occupation of the worker. The personal attributes used as controls
are the earnings of the individual, ERN, the age of the worker, AGE, the worker’s
race, BONW, the worker’s sex, FEML, the number of children in the worker’s
household, CHLDRN, whether or not the worker is married, MARR, and the level
of educational attainment of the worker, <12YED, 12YED, 12-16YED, CDGR.
Controls are also included for tenure with the worker’s employer, less than one year
of tenure, <1YT, and between one and five years of tenure, 1-5YT. Earnings, the
respondent’s age, the sex of the respondent, the number of children, and higher
levels of educational attainment, are expected to be related positively to the desire
for medical benefits, WSMED. The controls used for the employer are, first, the
size of the employer, <50, and 50—499. The size controls are included as smaller
sized firms are possibly less likely to provide medical benefits and the workers in
these firms will have a greater desire for such benefits. Second, industrial controls
are included. A third employer control is an indicator of possible exposure to
hazards at the workplace, DWC. If a worker is exposed to dangerous working
conditions, it is likely the worker would desire medical coverage due to the increased
risk of injury. An explanation for the construction of this control variable is in the
Appendix B of Chapter 2.

Because the dependent variable, WSMED, is a zero-one variable, a probit
analysis is utilized. The level of earnings, ERN, the age of the respondent, AGE,
the length of tenure on the job, both less that one year, <1YT, and between one and
five years, 1-5YT, as well as the relatively small firm size, of between fifty and 499

employees, 50-499, are all related positively and significantly to the demand for

Maximum Likelihood Estimates
Desire for Broad Medical Coverage

67

TABLE 2-6

 

 

Variable 1 2 3
Constant -.363 * * -.440‘ * * -.496* *
(.264) (.268) (.279)
ERN -.662E-05 * -.553E-05 -.798E-05 * *
(.550E-05) (.559E-05) (.571E-05)
AGE .103E-03 .229E-03" .852E-04“
( 740E-03) (.753E-03) (.740E-03)
BONW .604E-01 .478E-01 .734E-01
(.198) (.199) (.199)
FEML -.358**** -.403’**“ -.314
(.131) (.142) (.138)
CHLDRN .277E-01 .313E-01 .208E-01
(.419E-01) (.426E-01) (.199)
MARR .727 .936E-01 844E-01
(.143) (.144) ( 144)
< 12YED -.150 - 266 .229
(.197) (.211) (.208)
12YED -.199 -.176 -.945E-01
(.165) (.180) (.175)
12-16YED - 213 -.103 .951E-01
(.178) (.182) (.186)
CDGR -.499 -.658E-01 -.127
(.199) (.200) (.208)
<1YT .780*"* .757**** 824“"
(.147) (.149) (.151)
1-5YT .509**** .466**** 522“"
(.129) (.131) (.131)
<50 -.451***‘ -.421 -.417
(.145) (.147) (.158)
50—499 .156E-01 .299E-01 .199
(.158) (.160) (.162)
DWC .101 .115 .124
(.122) (.127) (.125)
CLR .410" * *
(.180)
CFT .297“ *
(.188)
OPR .200

Variable 1

TOP

SER

MIN

CON

MFG

TRN

WHL

RTL

FIN

BRS

PSR

PUB

Log-L -370.57
Chi-Sq. 60.95
Sample Size 610

Mean WSMED .367
Standard errors in parentheses.
* Significant at the 0.15 level,
** Significant at the 0.10 level,

**" Significant at the 0.05 level,
**** Significant at the 0.01 level.

68

Table 2-6 (Continued)

2
.802E-01

:664E-01
(.275)

-367.14
67.81
610
.367

69

broad medical benefits, WSMED. None of the other variables has a significant
effect on the demand for medical benefits. It is of interest that the broad measure
of exposure to dangerous working conditions, DWC, although having a positive
coefficient, is not statistically significant. The addition of occupational controls and
industrial controls, presented in columns 2 and 3 of Table 2-6, did not change the
results of the analysis.

In conclusion, the analysis of the demand for broad medical coverage,
WSMED, shows the level of earnings, ERN, the age of the worker, AGE, minimal
tenure, either less than one year, < 1YT, or between one and five years of tenure,
1-5YT, and an employer’s size of between fifty and 499 employees, 50-499, are
important determinants of the demand for medical coverage. Neither the number
of children of the worker, the sex of the respondent, the respondent’s marital status,
nor the level of educational attainment has any statistical effect on the demand for
medical benefits. Also, the controls for hazardous working conditions add little to
the analysis. The only control variable proving to be significant is for a dangerous
employer. These results are important for the smaller employer, employers of less
than 500 employees. By providing medical benefits to their employees these
employers are likely to reduce the demand for unionization on the part of their

workers allowing them to remain union free.

70
CONCLUSION

The results from the 1977 cross-section of the E provide the first true test
of the substitution hypotheses. The results of the analysis may be summarized in the
following manner. In all of the analysis the level of earnings is negatively related
to the willingness to vote for union representation. The negative relationship
between the level of earnings and the willingness to vote for union representation
represents clear support for the management substitution hypothesis. That is, if
employers are willing to raise the level of earnings of their employees, it is very
likely that the desire to seek union representation on the part of the employer’s
work force would be significantly reduced.

Of the benefits employers may provide, medical benefits play a key role in
supporting the theory of a substitution effect. When employees have extensive
medical coverage they are less likely to vote for union representation. Medical
benefits are important for both the management substitution hypothesis and the
government substitution hypothesis. For the management substitution hypothesis,
employers will find that if they are willing to provide broad medical coverage, then
the desire for union representation on the part of their employees will be reduced.
In terms of the government substitution hypothesis, the institution of a broad
national health insurance plan would reduce the demand for union representation,
as medical benefits would then be available to all workers regardless of their union
membership. This point is especially important, as the A.F.L.-C.I.O. has been one

of the major forces in lobbying the Congress for the passage of such legislation. If

71
successful in its future lobbying of the Congress for a national medical insurance
plan, the A.F.L.-C.I.O. may find that it has reduced one of the factors contributing
to the demand for union services. A decline in the demand for union services would
then contribute to an even further decline in success rate of unions in representation
elections.

The mixed findings regarding retirement benefits may, in part, be explained
by the presence of government "retirement" program already covering workers,
OASI. That is, through the OASI program the government has already established
a program which meets some of the needs retirement benefits would provide.

The continued negative result for the size of the employer does not bode well
for organized labor. Having already organized many of the larger employers,
organized labor is faced with attempting to organize the smaller employers. The
results of this analysis indicate success in such efforts is likely to be remote. A lack
of success in such firms is especially likely to be true if the employers in these firms
provide their employees with medical benefit coverage. As discussed above, many
of the remaining large non-union employers have adopted employee relations plans,
which provide benefits approximating those provided by unionized firms, in an
attempt to forestall unionization. The results of this work support the belief these
employee relations system will be effective in limiting the growth of unions in large

establishments.

CHAPTER THREE

TIE GROWTH OF TOTAL COMPENSATION AND TIE DECLINE IN THE

WILLINGNESS TO VOTE FOR UNION REPRESENTATION

With the findings in Chapter Two that the substitution hypotheses have
validity, the next issue to be addressed is whether or not changes in the levels of
benefit provision affect the willingness of workers to vote for union representation.
In particular, the cross-section results indicate that medical benefits are likely to
have an effect on the willingness of workers to vote for union representation. If
increased numbers of workers have medical benefit coverage, the provision of
medical benefits by employers may lessen the demand for union representation.
This decrease in demand for union representation may explain some of the decline
in the success rate of unions in representation elections. Such a finding will be a
valuable addition to the labor relations literature, as it offers an alternative
explanation for the decline.

To determine whether the presence of medical benefits leads to a decline in
the willingness of workers to vote for union representation the period from 1950 to
1987 is examined. The reasons for selecting this time period are twofold. First, it

covers the time period of a significant decline in the victory rate for labor unions in

72

73

representation elections administered by the National Labor Relations Board.
Second, this is the period of the most rapid growth of fringe benefits. For example,
while medical benefits were introduced by some firms prior to the Second World
War, it was not until after the Circuit Court decision in 1949 in m M gm
Company L N.L.R.B. that significant growth in medical benefits occurred. The
decision in Cmsa resulted in medical benefits becoming a subject of bargaining for
which the parties to a collective bargaining agreement are required to bargain to
impasse.89 This is because the m decision recognized total compensation rather
than just wages in collective bargaining.

The word "wages," following the phrase "rates of pay" in the Act must

have been intended to comprehend more than the amount of

renumeration per unit of time worked or per unit of work produced.

We think it must have been meant to comprehend emoluments

resulting from employment in addition to actual "rates of pay."°°
Similarly I_nl_a_ng Stag], required employers to bargain in good faith over pension
and retirement plans." With these decisions collective bargaining began to embrace
the notion of total compensation.

By beginning the analysis in 1950, the entire time period of the possible effect

of the growth of fringe benefits on the willingness to vote for union representation

 

89

2

 

 

.W. Cross ang Company L N.L.R.B., 174, F. 2nd 875 (lst Cir. 1949).

90

2

.W. Cross m Company L N.L.R.B., 174, F. 2nd 875 (lst Cir. 1949).

 

 

”Inland Steel Company y N.L.R.B., 170 no. 247 (7th Cir. 1948), cert. denied,
336 US. 960 (1949).

 

 

74
is covered. The analysis ends with 1987, as this is the last year for which published
data are available for the outcomes of representation elections.

A casual observation of the changes in success rate of unions in
representation elections and the percentage of workers covered by medical benefits
suggests the existence of a negative relationship between workers being covered by
medical benefits and their willingness to vote for union representation.” The data
in Table 3.1 show that while the percentage of workers receiving medical benefits
increased from less than 20 percent in 1950 to approximately 60 percent in the mid-
19805, the victory rate of unions in representation elections declined from more than
70 percent to less than 45 percent.93 While the percentage of workers receiving
medical benefits was experiencing its most rapid growth, during the period from
1950 to 1965, the success rate of unions in representation elections was falling from
nearly 75 percent to 60 percent. Furthermore, during the 19805 when the
percentage of workers receiving medical benefits did not vary much, the union
victory rate in representation elections remained relatively constant.

To go beyond the casual observation that medical benefit coverage and the
willingness to vote for union representation appear to be related negatively, it is
necessary to utilize the cross-section results from Chapter Two to construct a
probability of workers voting for union representation. Once this probability is

determined it is possible to ascertain how variations in total compensation will affect

 

”The data for the success rate of unions in N.L.R.B. elections are from the
Annual Reports of the N.L.R.B.

93For a discussion of the data that were used in this analysis see the Appendix
A to this Chapter.

75

Table 3.1
Mean Medical Benefit Coverage and Union Victory Rates in N.L.R.B.
Representation Elections, 1950-1987

 

 

Medical Union Victory
Benefit Rate In N.L.R.B.
Year Coverage Representation
Elections
1950 16.4% 74.5%
1951 20.2% 74.0%
1952 23.8% 72.9%
1953 28.9% 71.9%
1954 32.7% 65.6%
1955 36.6% 67.6%
1956 38.6% 65.3%
1957 41.6% 62.2%
1958 43.6% 60.8%
1959 45.1% 62.8%
1960 47.7% 58.6%
1961 50.4% 56.1%
1962 51.5% 59.5%
1963 53.9% 59.0%
1964 58.8% 57.1%
1965 53.0% 60.8%
1966 58.1% 60.8%
1967 60.6% 59.0%
1968 59.9% 57.2%
1969 61.5% 54.6%
1970 63.9% 55.2%
1971 64.1% 53.2%
1972 63.6% 53.6%
1973 66.3% 51.1%
1974 66.5% 50.0%
1975 64.5% 48.2%
1976 64.3% 48.1%
1977 64.1% 46.0%
1978 63.7% 46.0%
1979 61.2% 45.0%
1980 61.9% 45.7%
1981 62.0% 43.1%
1982 61.7% 40.3%
1983 61.0% 43.0%
1984 59.8% 42.0%
1985 60.2% 42.4%
1986 60.1% 43.2%

1987 56.6% 43.9%

76
the probability an individual will vote for union representation. Total compensation
is defined as annual earnings, medical benefit coverage, and retirement benefit
coverage.

The probability of an individual’s willingness to vote for union representation
is obtained from the cross-section results in the following manner. The estimate
coefficients for each of the variables from the cross-section are multiplied by their
mean values, and the resulting products are summed over all of the variables in
equation, with the probability determined from this sum.“ In algebraic terms:

n A
PROB VFU = F(2 BX), Eq.3-1
i-

where 8 represents the estimated coefficient from the cross-section results, X
represents the mean of each of the 11 variables in the cross-section results.
Without either the occupational or industrial controls the probability of the
respondents in the cross-section sample being willing to vote for union representation
is 41 percent. The estimated probability contrasts with the union victory rate in
representation elections held by the N.L.R.B. during 1977 of 64.1 percent. The
inclusion of occupational and industrial controls from the cross-section results in an
estimated probability of approximately 30 percent in the case of industrial controls
and 38 percent in the case of occupational controls for 1977. The difference in the

estimated probability of voting for union representation is due to differences in the

 

“G. S. Maddala, Limited-Dependents Qualitative Variables i_n Econometrics (New
York: Cambridge University Press, 1983): 18-27.

77
sum of the coefficients for each estimation in Table 2.5. In each case the estimated
willingness to vote for union representation is between 20 and 30 percentage points
lower than the actual union victory rate for 1977.

To test the effect of the change in the level of medical benefit coverage on
the willingness of an individual to vote for union representation, the basic procedure
for determining the probability of voting for union representation is repeated.
However, the mean value for medical benefits is not the cross-sectional sample mean
but the mean medical benefit coverage for the work force as a whole and the mean
desire for medical benefits. It is noted that while the mean of medical benefits,
SMED, for the cross—section sample is nearly 81 percent, the mean level for all
workers was significantly less at 64 percent. The difference between the two means
is probably explained by the all inclusive manner in which the medical benefit
variable is constructed for the cross-section work.” This probability simulation is
shown algebraically in Equation 3-2, where PROB VFU, represents the estimated
probability of voting for union representation in year t, NED, represents the mean
level of medical benefits for all workers in year t, and CONS, a constant is included
to equate the simulated result in 1950 with the actual union victory rate in N.L.R.B.
elections.

PROB VFU, = FQMEDNED, + Emma-Min.) Eq. 3_2

II A-
+2 (ax) + CONS
174 MED

 

9"Ihe analysis in the remained of this chapter is replicated in Appendix B to this
Chapter utilizing the narrowest measures of benefits. This replication is necessary
due to uncertainty of comparison in the definitions of benefit coverage with the data
from the OE and the data utilized in the simulations. By providing results from the
narrowest and the broadest definitions the entire range of outcomes is covered.

 

78
By substituting the actual mean of the medical benefits and the desire for medical
benefits for each year in the sample from 1950 through 1987, a simulation of the
effect of the changing levels of medical benefits on the willingness to vote for union
representation is obtained. This type of procedure was used by Parsons to simulate
trends in the labor force participation of older men.96

The ability of a cross-sectional model to explain time series trends may be
subject to criticism. The criticism stems from the fact one is assuming the
coefficients for the variables will remain constant over time. The stability of the
coefficients is unlikely, especially in the case of the compensation variables.
Instability of the coefficients for the compensation variables is likely as the time
period is from the introduction of fringe benefits to the time of their wide
acceptance. It is probable the demand for medical benefit coverage was higher
during the 19505, soon after medical benefits became a mandatory subject of
bargaining, than it was from 1970, during which time the percentage of workers
covered by medical benefits remained relatively constant.

The results of the initial simulation are presented in column 2 of Table 3.2.
During the time period of the sample, the predicted union victory rate declines from
approximately 74.5 percent to 69 percent due to the increase in the percentage of
workers receiving medical benefits. The expansion of medical benefits explains 5.5
percentage points of the 30 percentage point decline in the willingness to vote for

union representation. Column 3 of Table 3.2 presents simulation results with the

 

96Donald 0. Parsons, "The Decline in Male Labor Force Participation," Journal
o_f Political Economy, 88 (February 1980): 128-131.

79
TABLE 3.2

Actual and Predicted Union Victory Rates in N.L.R.B.
Represented Elections, 1950-1987

 

 

(1) (2) (3) (4)
Actual Predicted
Union Victory Union Victory

Year Rate Rate

1950 74.5% 74.5% 74.5% 74.5%
1951 74.0% 74.0% 74.5% 71.4%
1952 72.9% 73.5% 74.4% 71.4%
1953 71.9% 72.9% 74.4% 71.3%
1954 65.6% 72.4% 74.3% 71.2%
1955 67.6% 71.9% 74.3% 71.2%
1956 65.3% 71.6% 74.3% 71.1%
1957 62.2% 71.3% 74.2% 71.1%
1958 60.8% 71.0% 74.2% 71.1%
1959 62.8% 70.8% 74.2% 71.0%
1960 58.6% 70.5% 74.2% 71.0%
1961 56.1% 70.2% 74.1% 71.0%
1962 59.5% 70.2% 74.1% 70.9%
1963 59.0% 69.7% 74.1% 70.9%
1964 57.1% 69.5% 74.1% 70.9%
1965 60.8% 69.2% 74.0% 70.8%
1966 60.8% 69.2% 74.0% 70.8%
1967 59.0% 69.0% 74.0% 70.8%
1968 57.2% 69.0% 74.0% 70.8%
1969 54.6% 68.8% 74.0% 70.8%
1970 55.2% 68.5% 74.0% 70.7%
1971 53.2% 68.5% 74.0% 70.7%
1972 53.6% 68.5% 74.0% 70.7%
1973 51.1% 68.2% 74.0% 70.7%
1974 50.0% 68.2% 74.0% 70.7%
1975 48.2% 68.4% 74.0% 70.7%
1976 48.1% 68.5% 74.0% 70.7%
1977 46.0% 68.5% 74.0% 70.7%
1978 46.0% 68.5% 74.0% 70.7%
1979 45.0% 68.8% 74.0% 70.8%
1980 45.7% 68.8% 74.0% 70.8%
1981 43.1% 68.7% 74.0% 70.8%
1982 40.3% 68.8% 74.0% 70.8%
1983 43.0% 68.9% 74.0% 70.8%
1984 42.0% 69.0% 74.0% 70.8%
1985 42.4% 69.0% 74.0% 70.8%
1986 43.2% 69.0% 74.0% 70.8%
1987 43.9% 69.5% 74.1% 70.8%

80

 

80.00
75.00 \
70.00-
65.00—
60.00-
55.00—
50.00-
45.00-
40.00-
35.00—

30. 00 I I I I I l I
1950 1955 1960 1965 1970 1975 1980 1985 1990

YEAR

 

 

 

UNION VICTORY RATE

 

 

 

 

-— Actual Rate ----------- No Controls — Occupational ----- Industrial Co

 

FIGURE 3-1

Actual and Predicted Union Victory Rates in N.L.R.B.
Representation Elections, 1950-1987

81

inclusion of occupational controls. With the inclusion of the occupational controls,
the predicted union victory rate fell from 74.5 percent in 1950 to approximately 74
percent in the 19805. The simulation implies that the growth of medical benefits
accounts for little if any of the decline in the willingness to vote for union
representation. Column 4 of Table 3.2 presents the simulation results with the
inclusion of the industrial controls. The predicted willingness to vote for union
representation declined from 74.5 percent in 1950 to 70.8 percent in 1987. Figure 3-1
shows a graph of the actual union victory rate in representation elections as well as
the three predictions of the union victory rate for the 1950 through 1987 time period.

The initial simulation suggests that up to 16 percentage points of the decline
in the willingness to vote for union representation may be attributable to the growth
of medical benefits. This offers support for the management substitution hypothesis.

Next retirement benefits and earnings are examined to determine how total
compensation influences the willingness of workers to vote for union representation.
Retirement benefits are included as the cross-section results indicate they are related
positively to the willingness of workers to vote for union representation. Earnings
are included as the cross—section results indicate they are related negatively to
willingness to vote for union representation. Algebraically, the simulation of the
effect of changes in the level of total compensation is shown in Equation 3-3, where
PROB VFU, is the estimated probability of voting for union representation in time
period t, M E D,, R E T, and E R N, are the mean levels of medical benefits,

retirement benefits, and earnings in time period t.

82

PROB VFU, = mammal), + fawn-MED) + famRI-n: +
II

Baum-RET) + 13mm + 2 (23. X) + CONS Eq. 3-3

i7£SMED
iaéSRED
i7€WSMED
ifWSRET
i¢ERN

The results of the second simulation are in Table 3.3. Column 2 of Table 3-3
has the results without either occupational or industrial controls. The results
indicate the willingness of workers to vote for union representation falls 7 percentage
points from 74.5 percent to 67.5 percent during the time period under consideration.
When occupational controls are included, column 3, the willingness to vote for union
representation falls from approximately 2 percentage points. With industrial
controls, column 4, the willingness to vote for union representation declines 2.5
percentage points.

The simulated declines in the willingness to vote for union representation are
shown in Figure 3-2 along with the actual union victory rate in representation
elections. The actual union victory rate decreases almost continuously from 75
percent in 1950 to 45 percent in 1980 and then remains fairly constant at 43 percent.
The simulated union victory rates all decline from 1950 to 1974 reaching a low of

about 64.7 percent. Since 1974, the predicted union victory rate increased

83

TABLE 3.3
The Effect of Changes in Total Compensation on the Predicted
Outcomes of N.L.R.B. Representation Elections

 

 

(1) (2) (3) (4)
Actual Predicted
Union Victory Union Victory

Year Rate Rates

1950 74.5% 74.5% 74.5% 74.5%
1951 74.0% 73.9% 74.4% 74.3%
1952 72.9% 73.2% 74.2% 74.2%
1953 71.9% 72.2% 74.0% 73.8%
1954 65.6% 71.7% 73.8% 73.6%
1955 67.6% 70.8% 73.6% 73.3%
1956 65.3% 70.4% 73.4% 73.1%
1957 62.2% 69.9% 73.2% 72.9%
1958 60.8% 69.6% 73.1% 72.7%
1959 62.8% 69.2% 73.0% 72.5%
1960 58.6% 68.8% 72.9% 72.4%
1961 56.1% 68.3% 72.8% 72.3%
1962 59.5% 68.1% 72.7% 72.2%
1963 59.0% 67.5% 72.5% 72.0%
1964 57.1% 67.1% 72.5% 72.0%
1965 60.8% 66.7% 72.3% 71.7%
1966 60.8% 66.6% 72.2% 71.7%
1967 59.0% 66.4% 72.2% 71.6%
1968 57.2% 66.3% 72.2% 71.6%
1969 54.6% 66.0% 72.1% 71.5%
1970 55.2% 65.8% 72.1% 71.5%
1971 53.2% 65.6% 72.0% 71.4%
1972 53.6% 65.4% 71.9% 71.2%
1973 51.1% 65.0% 71.8% 71.1%
1974 50.0% 64.7% 71.9% 71.3%
1975 48.2% 65.8% 72.0% 71.4%
1976 48.1% 65.7% 72.0% 71.4%
1977 46.0% 65.6% 72.0% 71.3%
1978 46.0% 65.7% 72.0% 71.3%
1979 45.0% 66.2% 72.1% 71.4%
1980 45.7% 66.5% 72.2% 71.6%
1981 43.1% 66.6% 72.3% 71.7%
1982 40.3% 66.8% 72.4% 71.7%
1983 43.0% 66.8% 72.4% 71.7%
1984 42.0% 66.9% 72.4% 71.8%
1985 42.4% 67.0% 72.4% 71.8%
1986 43.2% 67.0% 72.4% 71.8%

1987 43.9% 67.5% 72.6% 72.0%

84

 

 

UNION VICTORY RATE

 

 

 

 

 

 

, C I I l l I l l
30 01950 1955 1960 1965 1970 1975 1980 1985 1990
YEAR
[— Actual Rate """"""" No Controls — Occupational ------ Industrial Go
Figure 3-2

Total Compensation Changes and
Actual and Predicted Union Victory Rates in
N.L.R.B. Representation Elections, 1950-1987

85
approximately 3 percentage points during the rest of the 19708 and then remains
relatively constant during the 19805.

In each of the simulations allowing for changes in total compensation, the
willingness to vote for union representation declinesduring the period for which the
data are available. The largest decline is with neither the occupational nor industrial
controls, where the decline is 7 percentage points. This decline of 7 percentage
points represents approximately 22 percent of the total decline in the willingness to
vote for union representation. The 2.2 percentage point decline with the
occupational controls represents 7 percent of the total decline and the 2.5 percentage
point decline with industrial controls accounts for 8 percent of the total decline in
the willingness to vote for union representation. These results support the idea that
the changes in employee compensation play a major role in reducing the likelihood
of a union success in representation elections. Table 3.4 provides a summary of the
six simulations. In Table 3.4 the predicted decline in the probability of voting for
union representation is shown along with the 30.4 actual percentage decline in the
union victory rate in the N.L.R.B. supervised representation elections. With the
changes in medical benefits alone, the simulated willingness to vote for union
representation falls between 0.4 and 5 percentage points. This decline represents
approximately 16 percent of the actual 30 percentage point decline. When total
compensation is considered the simulation results indicate that between a 2.2 and
7 percentage point decline the willingness to vote for union representation may be
explained by changes in total compensation as measured by earnings, medical

coverage, and coverage by retirement benefits. Thus, about 22 percent of the decline

86
TABLE 3.4

Simulated Declines in the Willingness to Vote For
Union Representation, 1950-1987

 

Percentage of

 

Predicted Actual Decline
Percentage Predicted By
Point Simulation
Decline Results
Variation in Medical 5.0% 16.3%
Coverage, N0 Controls
Variation in Medical 0.4% 1.3%
Coverage, Medical
Controls
Variation in Medical 3.7% 12.1%
Coverage, Industrial
Controls
Variation in Total 7.0% 22.9%
Compensation, No
Controls
Variation in Total 2.2% 7.2%
Compensation,
Occupational Controls
Variation in Total 2.5% 8.2%

Compensation,
Industrial Controls

87
in the willingness to vote for union representation may be explained by changes in
compensation.

To provide a further test of the substitution hypotheses a regression analysis
of the effect of the coverage of medical benefits, retirement benefits, and earnings
on the union victory rate in representation elections is the dependent variable as
itserves as a proxy for the willingness of workers to vote for union representation.
That is, the greater the willingness of workers to vote for union representation, the
greater the percentage of representation elections won by unions should be. The
results of this analysis are in Table 3.5. Table 3.5 contains two sets of results. The
first are estimates made using ordinary least squares, whereas the second estimation
uses the Cochrane-Orcutt iterative technique to correct for the serial correlation
found in the ordinary least squares estimation. The results of the regression analysis
do not provide much support for the substitution hypothesis as either the cross—
section or simulation results. Support for the substitution effects is unlikely given
that many of the crucial control variables, such as occupational and tenure controls,
are not included in this analysis. The ordinary least squares results did contain a
negative and significant result for retirement benefits, which supports the substitution
hypotheses. The coefficients for medical benefit coverage and earnings are positive,
which runs counter to the substitution hypotheses. Because of the problem of serial
correlation the Cochrane-Orcutt estimation results are more meaningful. In these
results retirement benefit coverage is related negatively to the union victory rate.
The coefficients for medical benefits and earnings should also be negative to support

the substitution hypotheses.

88
TABLE 3.5

Regression Results of the Effect of Detrended Total Compensation
on the Outcome of N.L.R.B. Representation Elections

Dependent Variable - Percentage of N.L.R.B. Representation Elections Won

 

 

OLS Cochrane-Orcutt
Independent Variables Coefficient' Coefficient"
Constant 86.00 87.38
(9.10) (9.10)
Time —.51 -.52
(.06) (.06)
Percentage of Workers .28 .33
Covered by Medical Benefits (.19) (.23)
Percentage of Workers -1.50 -1.54
Covered by Retirement (.29) (.34)
Benefits
Annual Earnings .002 .001
(.0007) (.0008)
R2 .98 .98
Durbin Watson 1.47 1.97
Rho .27

8
Standard Errors shown in parentheses.

89

Because the data in this regression analysis are likely to be related by a time
trend, the analysis was repeated after the time trend was removed from the
variables. The time trend was removed by first regressing the logarithms of the
variables on time and time squared. The residuals from these regressions were then
used to estimate the effects of the change in the level of benefit coverage on the
union victory rate. The results of these estimations are shown in Table 3.6 and
provide much stronger support for the substitution hypotheses. The coefficients for
both medical and retirement benefit coverage are negative and significant, thereby
supporting the substitution hypotheses. The results for earnings were positive, which
runs counter to the substitution hypotheses. The results imply that changes in the
provision of medical and retirement benefits may account for some of the decline

in the union victory rate since 1950.

W

The results of the analysis in this chapter provide a new insight into the
possibility of a substitution effect explaining a significant proportion of the decline
in the success rate of unions in representation elections. If one is willing to accept
the rather heroic assumption upon which the simulation analysis is based, that the
coefficients from the results from the cross-section work remain constant over time,
then new insight into the decline of the success of unions in representation elections
is obtained. Rather than dismissing the substitution hypotheses as has been done in

the past, new concern must be raised by organized labor, as a significant proportion

90
TABLE 3.6

Regression Results of the Effect of Total Compensation
on the Outcome of N.L.R.B. Representation Elections

Dependent Variable - Percentage of N.L.R.B. Representation Elections Won

 

 

OLS Cochrane-Orcutt
Independent Variables Coefficient' Coefficient“
Constant -.16 -.25

(9.10) (9.10)
Percentage of Workers —9.17 -14.36
Covered By Medical Benefits (3.21) (4.63)
Percentage of Workers -17.99 -24.21
Covered By Retirement (4.76) (5.99)
Benefits
Annual Earnings .00004 .00002

(.000005) (.00008)

R2 .35 .35
Durbin Watson 1.04 2.33
Rho .57

*
Standard Errors shown in parentheses.

91
of the decline in the willingness of unorganized workers to vote for union
representation is explained by increases in benefits made available to workers.
Further increases in compensation by those firms which are currently non-union are
seen as being an effective means to forestall a desire for union representation on the

part of workers.

CHAPTER FOUR

SUMMARY AND CONCLUSIONS

The American labor movement has been in a state of continued decline for
more than forty years. As evidence of the decline, the number of workers who are
represented by labor unions continues to decline. Furthermore, labor unions are
winning fewer representation elections than at any time since the passage of the
National Labor Relations Act. Various factors are offered as explanations for the
decline, from the changing industrial makeup of the American economy to violations
of labor law by management. The work of this dissertation offers a new insight
into the decline of the labor movement.

This dissertation focuses on the effect of total compensation on the
willingness of workers to vote for union representation. Previous work advanced the
notion that if employers or the government provide benefits to workers, which
previously had only been available through the collective bargaining process, the
willingness of workers to vote for union representation is likely to decline. The
effect of the provision of benefits by the government is the government substitution
hypothesis and the effect of the provision of benefits by employer is the management

substitution hypothesis. Prior research on these hypotheses has two shortcomings.

92

93
First, none of the research considers benefits traditionally bargained for by labor
unions. Second, there is no direct test of the provision of benefits on the willingness
of workers to vote for union representation. Both of these shortcomings are
corrected in this dissertation.

The substitution hypotheses are tested with data from the 1M Cross-Section
of the Quality o_f Employment Sumey. This data set provides two major
improvements over the earlier research on the substitution hypotheses. The data set
has a direct measure of the willingness of an individual to vote for union
representation. In addition, the data set contains measures not only of earnings but
also coverage by a wide variety of benefits. For benefit coverage, the QES provides
two pieces of information for each of the benefits. The first is whether or not the
respondent has the benefit in question and the second is if the respondent does not
have the benefit does the respondent want the benefit. With this information the
data from the @S allows for testing of the effect total compensation may have on
the willingness of workers to vote for union representation.

Of the benefits included in the @, medical benefits, retirement plans, and
vacation benefits occurred in large enough numbers to allow for further statistical
analysis. A simple comparison of the mean level of earnings shows union workers
received higher earnings than their non-union counterparts. Furthermore, union
workers are more likely to have medical and retirement benefit coverage than
workers not covered by a collective bargaining agreement. No difference is found
in vacation benefit coverage. The comparisons of the level of earnings and fringe

benefit coverage allows the formulation of the following hypothesis: If workers

94
have low levels of earnings and do not have medical or retirement benefits they
should be more willing to vote for union representation.

To test the hypothesis a probit analysis is utilized. The results of the probit
analysis confirm earnings are related negatively to the willingness of workers to vote
for union representation. The novel result of the analysis is the effect medical
benefits have on the willingness of workers to vote for union representation.
Workers who have medical benefit coverage are less likely to vote for union
representation than are those who did not have coverage. The workers who did not
have medical benefits and who wanted them are found to be more willing to vote
for union representation than the workers who did not want medical benefits. The
effect of medical benefits held regardless of the variable specification used or the
control variables included in the analysis. In terms of the management substitution
hypothesis, if employers are willing to provide medical benefits to their non-union
workers the probability of the workers voting for union representation will be
reduced. For the government substitution hypothesis the results imply if the
government were to enact a national health insurance plan workers would be less
willing to vote for union representation.

The findings for the coverage by retirement benefits runs counter to the
substitution hypotheses with all of the measures for the retirement benefits having
positive coefficients. A possible explanation for this result is the presence of social
security retirement benefits, which workers may regard as a sufficient retirement

benefit. Vacation benefits never produce any statistically significant results.

95

In the cross-section analysis of the QES data a major weakness is
acknowledged. The data on fringe benefits is not the dollar value of the benefits but
only zero—one indicators of whether or not the worker receive the benefit. Future
work should replicate the cross-section study using dollar-value data on fringe
benefits if they become available.

With the results of the cross-section study showing benefit coverage affecting
the willingness of workers to vote for union representation a simulation was
undertaken to determine if variations in benefits coverage may explain the decline
in the success rate of unions in representation elections. The results of the
simulation provide the second important result of this dissertation. When medical
benefit coverage is allowed to vary, as it has over the period from 1950 to 1987, up
to 16 percent of the actual decline in the willingness to vote for union representation
is explained. An expansion of the simulation allowing for changes in medical
benefits, retirement benefits, and earnings explains up to 23 percent of the decline
in the willingness to vote for union representation. The results provide an
alternative explanation for the decline in the success rate of unions in representation
elections.

The findings of this dissertation do not bode well for organized labor in the
United States. The management of firms currently not organized may be able to
reduce substantially the probability of their firms becoming unionized by providing
workers with medical benefits and by paying higher wages. Such actions will result

in a further decline in the number of workers represented by unions.

96

The potential impact the government may have on the outcomes of future
representation elections is large. The institution of a national medical insurance
program may mute the provision of medical benefits through the collective
bargaining process. A national medical insurance program would remove one of the
key determinants of the demand for unionization. This would lead to a further
decline in the success rate of unions in representation elections. The A.F.L.-C.I.O.
may well be advised to eliminate its lobbying of the Congress for national health

insurance as such efforts may jeopardize the existence of organized labor.

APPENDICES

CHAPTER TWO
APPENDIX A

The M m m o_f th_e QM o_f Emploment Su_rw;y, Q_E_S_, provides
several measures of employee compensation. Beginning with earnings, the variable
chosen was "How much do your total earnings from your job figure out to be a year,
before taxes and other deductions?" This was chosen as the broadest measure of
monetary compensation in the survey.

To expand beyond simple monetary compensation to total compensation
"Fringe benefits" are included in the analysis. In the QE the information on fringe
benefits is of two types. First, respondents to the survey are asked to answer the
following questions "I’ll read a list of fringe benefits that workers sometimes get in
addition to their wages. Please tell me whether each fringe benefit is available to

you." Respondents were then given a list of fringe benefits to respond to. This list
of fringe benefits includes the following" "medical, surgical, or hospital insurance

that covers any illness or injury that might occur to you while off the job, sickleave
with full pay," "dental benefits," "eyeglass or eyecare benefits," "life insurance that
would cover a death occurring for reasons m); connected with your job," "a

retirement program," "profit sharing," "stock options," "thrift or savings plan," and

97

98
"paid vacation." This series of questions gives zero-one data on the availability of
fringe benefits for workers.
After answering the initial question on fringe benefits, the respondents to the
QES are asked the following question: "Are there any fringe benefits that you’re n_o_t
getting that you’d lifi to be getting?" The respondents are then given another list
of fringe benefits to respond to. This list included: "medical, surgical, or hospital

insurance that covers a_ny illness or injury that might occur to you while fo the job."

 

 

H H

eyeglass or eyecare benefits," "fie

 

"sickleave with full pay," "dental benefits,
insurance that would cover a death occurring for reasons mt connected with your

job," "a retirement program," "profit sharing," "stock options," "thrift g savings plan,"

 

 

"paid vacation," "time _ff for personal business; death in family; jury duty; voting;

doctor’s appointments, paid holidays," "medical, surgical g hospital insurance that
covers illness or injury occurring while on the job," "prescription plan, discount on

prescriptions," "more pay; pay differentials; full pay throughout the year,”

 

" ealth medical insurance--NA whether on or off job." This series of questions
gives zero-one data on the desire for fringe benefits for those workers who do not
currently receive these fringe benefits.

As noted above, the @ also gives information for those individuals who did
not have fringe benefits but either wanted the fringe benefit or who did not want the
fringe benefit. The number of individuals who did not have a fringe benefit and who
did not want the benefit was so small as to prevent any further analysis. Those
individuals who did not have fringe benefits but who wanted them provided

additional insight into the substitution hypothesis. Due to the small number of

99
respondents who stated that they did not have one of the specific medical,
retirement, or vacation benefits, broader measures of medical insurance, retirement,
and vacation benefits are used. The measure of medical benefits, WSMED, was
constructed in the following manner: If the respondent did not have but wanted

either "medical, surgical, or hospital insurance that covers fly illness or injury that

 

 

might occur to you while o_ff the job," or "sickleave" with full pay," or "d_ent_a_l
benefits," or "eyeglass or eyecare benefits," or “medical, surgical g hospital insurance
that covers illness or injury occurring while on the job," or "prescription pl_ap;
discount on prescriptions," then WSMED was coded one. If not, then WSMED was
coded zero.

The broader measure of retirement benefits, WSRET, the variable is
constructed in the following manner. If the respondent did not have but wanted
either "@ insurance that would cover a death occurring for reasons n_o_t connected
with your job," or "a retirement program" or "profl m," or "m QM" or

"thrift g saving plan," then WSRET is coded one; if not, then WSRET is coded

 

zero. The broad measure of vacation benefits, WSVAC, is constructed as follows:
If the respondent does not have either "paid vacation" or "ti e _ff m personal

business; death in family; jury duty; voting; doctor’s appointments" but wants these

 

benefits then WSVAC is coded one; if not WSVAC is coded zero.

With these two series of questions, the Qlfi provides two pieces of
information concerning fringe benefits. First, if a worker has a fringe benefit.
Second, if a worker does not have a fringe benefit did the worker want the fringe
benefit? The piece of information that the QES does not give is whether a worker

has a fringe benefit but does not want the benefit.

100

CPMPTER TWO
APPENDIX B

The Qfl provides several measures of dangers to which workers may be
exposed to on their jobs. None of these measures occurs in enough of the cases to
allow for further statistical analysis. Therefore, it is necessary to create broader
measures of dangerous work conditions to test their effects on the demand for
medical benefits.

The broadest measure of the measures of danger on the job, DWC, is a
measure of exposure to any of the dangerous conditions covered in the QE.
Respondents to the ggs were asked the following question, "Does your job ever
expose you to..." "dangerous chemicals; radiation," "dangers from fire; burns or

shock," "air pollution from dust, smoke, gas, fumes, fibers, or other things, working
outside in bad weather," "extremes of temperature or humidity," "dirty or badly
maintained areas at your workplace," "things that are placed or stored dangerously,"
"too much noise," "dangerous tools, machinery, or equipment," "risk of catching
diseases," "risk of traffic accidents while working," "risks of personal attack by people
or animals," or "dangerous work methods." If the respondents said yes to any one

of these on-the-job dangers then the variable DWC is given the value of one; if not,

then DWC is given the value of zero.

CHAPTER THREE

APPENDIX A

The data for medical benefits were available from two sources. For 1950 to
1974 data were obtained from the Health Insurance Institute as reported by Skolnik

in the Social Security Bulletin.‘ The data for 1979 through 1987 are unpublished

 

 

data from the Census Bureau and were made available by the Employee Benefits
Research Association. For the four-year gap between 1975 and 1978 a cubic time
trend equation was utilized to provide estimates of the extent of medical benefits
coverage. Data for retirement benefits were obtained from the same sources. The
same procedure was utilized to overcome the four-year gap from 1975 through 1978.

The data for real annual earnings were constructed in the following manner;
average weekly hours for production or nonsupervisory workers on private
nonagricultural payrolls were multiplied by the average weekly wage which was then
multiplied by fifty-two. This product was then deflated by the "all items" measure
of the consumer price index to obtain real annual earnings. Prior to its adjustment
by the CPI the 1977 earnings of workers were $9828.00, which compares

favorably with the cross-section sample mean of $9974.07.

\_—_

.‘Alfred M. Skolnik. "Twenty-Five Years of Employee Benefit Plans." Social
My Bulletin, 39 (September 1976): 6.

 

101

102

The data for medical benefits, retirement benefits, and annual earnings used
in the simulation are shown in Table 3-7. The means of medical benefit coverage
and retirement benefit coverage in the cross-section results were much higher, 81
and 72 percent respectively, than the level of medical and retirement benefit
coverage for the economy as a whole, 64.1 and 42.1 percent. The difference between
the cross-section means and the observed coverage may be explained by the manner
in which the medical and retirement benefit variables were constructed in the QES

survey.

103

 

 

Table 3-7
Medical Benefits, Retirement Benefits, and Annual Earnings: 1950-1987
Percent Percent
Coverage Coverage Real
Medical Retirement Annual
Year Benefits Benefits Earnings
1950 16.4 22.5 $41386.18
1951 20.2 23.5 $38361.81
1952 23.8 24.2 $37638.00
1953 28.9 26.4 $37356.08
1954 32.7 29.0 $37078.33
1955 36.6 29.6 353721668
1956 38.6 31.4 $36669.38
1957 41.6 33.7 $35494.91
1958 43.6 35.8 $34512.35
1959 45.1 36.7 $34275.15
1960 47.7 37.2 $33696.18
1961 50.4 38.3 $33358.09
1962 51.5 38.3 $33026.72
1963 53.9 38.9 $32595.00
1964 55.8 39.2 $32174.42
1965 58.0 39.2 $31663.71
1966 58.1 39.3 $30784.17
1967 60.0 41.2 $29862.48
1968 59.9 41.0 $28661.12
1969 61.5 41.7 $27177.30
1970 63.9 42.1 $25706.37
1971 64.1 42.6 $24627.33
1972 63.6 43.1 $23861.41
1973 66.3 43.7 $22464.12
1974 66.5 44.0 $20231.38
1975 64.5 44.0 $18539.16
1976 64.3 44.0 551752912
1977 64.1 44.0 $16458.86
1978 63.7 43.9 $15297.65
1979 61.2 44.9 $13738.39
1980 61.9 44.9 $12104.45
1981 62.0 44.9 $10972.57
1982 61.7 44.3 $10335.82
1983 61.0 43.4 $10014.13
1984 59.8 42.0 $9599.68
1985 60.2 42.8 $9269.58
1986 60.1 42.6 $9100.43
1987 56.6 40.8 $8779.99

CHAPTER THREE
APPENDIX B

Due to the fact the definitions of broad measures of medical and retirment
benefits in the cross—section results may overstate the presence of the coverage of
these benefits the simulation procedures are repeated using the narrowest measures
of medical and retirement benefits. Table 3-8 contains the cross-setion results used
in the simulations. Table 3-9 shows the simulation results when only medical
benefits are allowed to vary. Column 2 of Table 3-9 contains the results with neither
occupation or industry controls. Column 3 presents the results with industrial
controls and column 4 the results with occupational controls. Allowing only medical
benefits to vary accounts for a 4 to 5 percentage point decline in the willingness to
vote for union representation. Table 3-10 shows the effect on the willingness to
vote for union representation when medical benefit coverage, retirement benefits
coverage, and annual earnings are allowed to vary. These simulations show
variations in total compensation may account for between 4 and 7 percent of the
decline in the willingness to vote for union representation. Thus, even with the
narrower definitions of the medical and retirement benefits the simulation results
continue to support the previous findings that the growth in the total compensation
received by workers contributes to the decline in the willingness of workers to vote

for union representation.

104

105

Table 3-8
Maximum Likelihood Estimates Current Fringe Benefits,
Desire for Fringe Benefits and the Willingness to
Vote for Union Representation

 

 

Variable 1 2 3
CONSTANT -1.019**** -1.860**** -1.695***
(.302) (.496) (.473)
ERN -.222E—04* * * * -.102-O4* * * -.111E04"*
(.839E-05) (.911E-05) (.909E-05)
MED .154 -.260* .262“
( .213) (.228) (.230)
RET .918E-01 .205* .148
(.167) (.185) (.190)
VAC .173 .162 .142
( .213) ( .222) (.226)
WMED .431* * .492* * .489“ * *
(.273) (.291) (.288)
WRET .939E-01 .341E-01 -.125
(.230) (.245) (.241)
WVAC 539* .653* * .594"
(.463) (.503) (.504)
BONW .897**** .997**** 1013““
(.211) (.224) (.224)
FEML .280* * * -.282 .238’“ *
(.141) (.161) (.154)
SOUTH .259* * * .346* * * * .288* *
(.135) (.143) (.143)
AGE .741E-03 .152E-02 .835E-03
(.182E-02) (.450E-02) (.329E-02)
LAYOFF .221 ‘.290E-02 .272E-01
(.441) (.449) (.450)
FOJ -.203E-01 .299 .217E-01
(.131) (.140) (.140)
XLO .576**** .587**** .632****
(.202) (.211) (.211)
< 12YED .778E-01 .386* *
(.269) (.257)
12YED -.207 .470* * *
(.222) (.218)
12—16YED -.233* -.495E-01
(.229) (.229)
CDGR - 120 -.198E-01
(.246) (.257)
<1YT .390*** .453****
(.193) (.187)

106
TABLE 3-8 (continued)

1-5YT .329* * *
(. 163)

< 50 -.543E—01
(.177)

50—499 -.423E-01
(.179)

CLR .965E-01
(.204)

CFT -.424E-01
(.230)

OPR .583 * * * *
(.242)

TOP .776* * *
(.423)

SER .506 * * * *
(.236)

MIN

CON

MFG

TRN

WHL

RTL

FIN

BRS

PSR

PUB

Log-L -281.38 -266.92

Chi.Sq. 62.36 91.27

Sample Size 494 494

Mean of VFU .328 .418

Standard errors in parentheses.

* Significant at the 0.15 level,

** Significant at the 0.10 level,
"** Significant at the 0.05 level,
**** Significant at the 0.01 level.

'377* * t *
(.158)

.828E-01
(.185)

3351301
(.180)

Actual and Predicted Union Victory Rates in N.L.R.B.

107

Table 3-9

Representation Elections, 1950-1987

 

(1)

Actual Union

(2)

(3)

(4)

Predicted Union Victory

 

Victory Rate Rates
1950 74.5% 74.5% 74.5% 74.5%
1951 74.0% 74.1% 74.2% 74.1%
1952 72.9% 73.7% 73.8% 73.8%
1953 71.9% 74.1% 73.3% 73.3%
1954 65.6% 72.8% 73.1% 73.0%
1955 67.6% 72.3% 72.7% 72.7%
1956 65.3% 72.1% 72.5 % 72.5 %
1957 62.2% 71.8% 72.3% 72.5 %
1958 60.8% 71.6% 72.1% 72.1%
1959 62.8% 71.4% 72.0% 72.0%
1960 58.6% 71.2% 71.8% 71.8%
1961 56.1% 70.9% 71.5% 71.6%
1962 59.5% 70.8% 71.4% 71.5%
1963 59.0% 70.5% 71.2% 71.3%
1964 57.1% 70.3% 71.1% 71.1%
1965 60.8% 70.1% 70.9% 71.0%
1966 60.8% 70.1% 70.9% 70.9%
1967 59.0% 69.8% 70.7% 70.8%
1968 57.2% 69.1% 70.1% 70.8%
1969 54.6% 69.7% 70.6% 70.7%
1970 55 .2% 69.5% 70.4% 70.5 %
1971 53.2% 69.5 % 70.3% 70.4%
1972 53.6% 69.5 % 70.4% 70.5 %
1973 51.1% 69.2% 70.2% 70.3%
1974 50.0% 69.2% 70.1% 70.3%
1975 48.2% 69.4% 70.3% 70.4%
1976 48.1% 69.4% 70.3% 70.4%
1977 46.0% 69.5 % 70.3% 70.4%
1978 46.0% 69.5 % 70.4% 70.5 %
1979 45.0% 69.8% 70.6% 70.7%
1980 45.7% 69.7% 70.5% 70.6%
1981 43.1% 69.7% 70.5% 70.6%
1982 40.3% 69.7% 70.5 % 70.6%
1983 43.0% 69.8% 70.6% 70.7%
1984 42.0% 69.9% 70.6% 70.8%
1985 42.4% 69.9% 70.7% 70.8%
1986 43.2% 69.8% 70.7% 70.8%
1987 43.9% 70.2% 71.0% 71.0%

108
Table 3-10

The Effect and Changes in Total Compensation on the Predicted

Outcomes of N.L.R.B. Representation Elections

 

(1)

Actual Union

(2)

(3)

(4)

Predicted Union Victory

 

Victory Rate Rates
1950 74.5% 74.5% 74.5% 74.5%
1951 74.0% 72.0% 74.2% 74.2%
1952 72.9% 72.1% 73.8% 73.8%
1953 71.9% 71.1% 73.2% 73.2%
1954 65.6% 71.9% 73.1% 73.1%
1955 67.6% 70.9% 72.6% 72.6%
1956 65.3% 70.3% 72.4% 72.4%
1957 62.2% 70.0% 72.2% 72.5%
1958 60.8% 69.8% 72.2% 72.2%
1959 62.8% 69.2% 71.9% 71.9%
1960 58.6% 68.8% 71.7% 71.7%
1961 56.1% 68.4% 71.5% 71.5%
1962 59.5 % 67.9% 71.2% 74.2%
1963 59.0% 67.5% 71.0% 71.0%
1964 57.1% 67.1% 70.7% 70.7%
1965 60.8% 66.5% 70.5% 70.5%
1966 60.8% 66.4% 70.4% 70.4%
1967 59.0% 66.2% 70.3% 70.3%
1968 57.2% 66.0% 70.2% 70.2%
1969 54.6% 65 8% 70.1% 70.0%
1970 55.2% 65.7% 69.9% 69.9%
1971 53.2% 65.4% 69.9% 69.8%
1972 53.6% 64.9% 69.7% 69.7%
1973 51.1% 64.6% 60.5% 69.5%
1974 50.0% 65.2% 69.8% 69.8%
1975 48.2% 65.8% 70.1% 70.1%
1976 48.1% 65.6% 70.0% 70.0%
1977 46.0% 65.5% 70.0% 70.0%
1978 46.0% 65.5% 70.0% 70.0%
1979 45 .0% 66.2% 70.4% 70.4%
1980 45 .7% 66.8% 70.7% 70.7%
1981 43.1% 67.0% 70.7% 70.8%
1982 40.3% 67.2% 70.8% 70.9%
1983 43.0% 67.1% 70.9% 70.8%
1984 42.0% 67.1% 70.8% 70.8%
1985 42.4% 67.3% 70.9% 70.9%
1986 43.2% 67.2% 70.9% 70.9%
1987 43.9% 67.5% 71.1% 71.1%

BIBLIOGRAPHY

BIBLIOGRAPHY

Anderson, John C., O’Reilley, Charles A. III, and Bushman, Gloria. "Union
Decertification in the U.S.:1947-1977."
1980):100-107.

Industrial Relations 19 (Winter

, , and . "The Decertification Process: Evidence

from California." Industrial Relations 21 (Spring 1982):178-195.

Ashenfelter, Orley and Pencavel, John H. "American Trade Union Growth:

 

1900-1960." Quarterly Journal o_f Economics 83 (August 1968):434-448.

Barbash, Jack. American Unions

 

 

 

Structure Government m Politics. New York:
Random House, 1962.

. "Theories of the Labor Movement in an Institutional Setting." Journal
o_f Economic Issues 15 (June 1981):299-303.

Berkowitz, Monroe. "The Economics of Trade Union Organization and

Blair, Douglas H. and Crawford, David. "Labor Union Objectives and Collective

 

Bargaining." TIE Quarterly Journal 9_f Economics 99 (August 1984):547-566.

Block, Richard N. "Union Organizing and the Allocation of Union Resources."

109

110
and Wolkinson, Benjamin W. "Delay in the Union Election Campaign
Revisited: A Theoretical and Empirical Analysis." Advances i_n Industrial m

Labor Relations. Vol. 3, New York: JAI Press, 1986.

 

, Christine L. Mahoney, and Leslie Fay Corbitt. "The Impact of
Employment-At-Will Judicial Decisions on the Outcomes of NLRB Elections."
Proceedings o_f me Thirg-Ninth Anial Meeting. Barbara D. Dennis, ed.,
Madison: Industrial Relations Research Association Series, 1987.

Bok, Derek C. and Dunlop, John T. Lflifl m th_e American Community. New
York: Simon and Schuster, 1970.

Brand, Horst. "The Evolution of Fair Labor Standards: A Study in Class Conflict."

 

My Q1119; Review 106 (August 1983):25-28.

Brody, David, Workers i_n Industrial America: flay; g th_e Twentieth may
Struggle. New York: Oxford University Press, 1980.

Carroll, Marjorie Smith. "Private Health Insurance Plans In 1976: An Evaluation."

Social Security Bulletin 41 (September 1978):3-16.

 

 

Chafetz, I. and Fraser, C. R. P. "Union Decertification: An Exploratory Analysis."
Industrial Relations 18 (Winter 1979):59-69.

Chaison, Gary N. "Union Growth and Union Mergers." Industrial Relations 20
(Winter 1981):98—108.

Commons, John R. and Andrews, John B. Principles of Labor Legislation. New

 

York: Harper & Brothers, 1920.
Cooke, William N. "Determinants of the Outcomes of Union Certification Elections."

Industrial and Labor Relations Review 36 (April 1983):402-414.

111
Cooke, William N. "The Rising Toll of Discrimination Against Union Activists."
Industrial Relations 24 (Fall 1985): 421-442.
Cooke, William N. and Gautschi, Frederick H. III. "OSHA, Plant Safety Programs
and Injury Reduction." Industrial Relations 20 (Fall 1981):245—257.
Dickens, William T. and Leonard, Jonathan S. "Accounting for the Decline in Union
1985):323-334.
, Wholey, Douglas, and Robinson, James C. "Correlates of Union
Supporting NLRB Elections." Industrial Relations 26 (Fall 1987):240-252.

Dulles, Foster Rhea. Labor i_n_ America: A Histog. Third Edition. Northbrook,

 

Illinois: AHM Publishing Corporation, 1966.
Dunlop, John T. Industrial Relations Systems. Carbondale, Illinois: Southern
Illinois University Press, 1958.

Elkouri, Frank and Elkouri, Edna Asper. How Arbitration Works. Third Edition.

 

Washington, DC: Bureau of National Affairs, 1977.
F arber, Henry S. "The Determinants of the Union Status of Workers." Econometrica
51 (September 1983):1417-1437.

. "The Recent Decline of Unionization in the United States." Science

 

238 (November 1987):915-920.
. "Trends in Worker Demand for Union Representation." American

Economic Review 79 (May 1989):166-171.

 

112
and Saks, Daniel H. "Why Workers Want Unions: The Role of Relative

Wages and Job Characteristics." Journal o_f Political Economy 88 (April

 

1980):349-369.

Fink, Gary M. "The Rejection of Voluntarism." Industrial m Labor Relations

 

Review 26 (January 1973): 805-819.

 

Fiorito, Jack and Greer, Charles R. "Determinants of US. Unionism: Past Research

and Future Needs." Industrial Relations 21 (Winter 1982):1-32.
, Lowman, Christopher, and Nelson, Forrest D. "The Impact of Human

Resource Policies on Union Organizing." Industrial Relations 26 (Spring
1987):113-126.

Flanagan, Robert J. "Workplace Public Goods and Union Organization." Industrial
Relations 22 (Spring 1983): 224—237.

Freeman, Richard B. "The Effect of Unionism on Fringe Benefits." Industrial fl

Labor Relations Review 34 (July 1981):489-509.

 

. "Unionism and Protective Labor Legislation." Proceedings o_f th_e

Thirty-Ninth Annual Meeting, Barbara D. Dennis, ed., Madison: Industrial

 

Relations Research Association Series, 1987, 260-267.
. "Contraction and Expansion: The Divergence of Private Sector and

Public Sector Unionism in the United States." Th_e_ Journal o_f Economic

 

Perspectives 2 (Spring 1988):63-88.

and Medoff, James L. "New Estimates of Private Sector Unionism."

113

and . What m Unions Do? New York: Basic Books, 1984.

 

 

Frumkin, R, and Wiatrowski, W. "BLS Takes A New Look At Employee Benefits."

 

 

My Labor Review 105 (August 1982): 41-45.

Garbarino, Joseph W. "Unionism Without Unions: The New Industrial Relations?"
Industrial Relations 23 (Winter 1984):40-51.

Getman, Julius G., Goldberg, Stephan B., and Herman, Jeanne B. Un_ion
Representation Election: m m Re_ality New York: Russell Sage
Foundation, 1976.

 

 

The University of Chicago Press, 1987.

Greenstone, J. David. Labor i_n American Politics. Second Edition. Chicago:

 

University of Chicago Press, 1977.
Heneman, Herbert G. 111 and Sandver, Marcus. "Predicting the Outcome of Union
Certification Elections: A Review of the Literature." Industrial and Labor

Relations Review 36 (July 1983):537-559.

 

 

and Co., 1917.

Kau, James B. and Rubin, Paul H. "The Impact of Labor Unions on the Passage of

 

 

Economic Legislation." Journal o_f Labor Research 2 (Spring 1981):133-146.
Kleiner, Morris M. "Unionism and Employer Discrimination: Analysis of 8(a)(3)

Violations." Industrial Relations 23 (Spring 1984):234-243.

114
Kochan, Thomas A. "How American Workers View Labor Unions." My m
m 102 (April 1979):23-31.
and Richard N. Block. "An Interindustry Analysis of Bargaining
Outcomes: Preliminary Evidence From Two-Digit Industries." m Quarterly
Jo_urn_a1 o_f Economics 91 (August 1977):431-452.
Kochan, Thomas A. and Katz, Harry C. Collective Bargaining and Industrial

Relations From Theogy t_o Poligy m Practice. Second Edition. Homewood,

 

Illinois: Irwin, 1988.

Kokkelenberg, Edward C. and Sockell, Donna R. "Union Membership in the United
States, 1973-1981." I_nd_ustr_ia_1 an_d Lam Relations Re_view 38 (July
1985):497-543.

Lawler, John. "Wage Spillover: The Impact of Landrum-Griffen." Industrial
Relations 20 (Winter 1981):85-94.

and Hundley, Craig. "Determinants of Certification and Decertification
Activity." Industrial Relations 22 (Fall 1983):335-348.

and West, Robin . "Impact of Union Avoidance Strategy in

Representation Elections." Industrial Relations 24 (Fall 1985):406-420.

Unionism. Princeton, New Jersey: Princeton University Press, 1958.
Leigh, Duane E. "The Determinants of Worker’s Union Status: Evidence from the
National Longitudinal Surveys." Journal o_f Human Resources 20 (Fall

1985):555-566.

115
"Union Preferences, Job Satisfaction, and the Union-Voice Hypothesis."
Industrial Relations 25 (Winter 1986):65-71.
and Hills, Stephen M. "Male-Female Differences in the Potential for

Union Growth Outside Traditionally Unionized Industries." Journal o_f Labor

 

Research 8 (Spring 1987):131-142.
MaCurdy, Thomas E. and Pencavel, John H. "Testing Between Competing Models

of Wage and Employment Determination in Unionized Markets." Journal __f

 

Political Economy 94 (June 1986):S3-S39.

 

Maddala, G.S. Limited-Dependent MQualitative Variables in Econometrics. New
York: Cambridge University Press, 1983.

Masters, Marick F. and Zardkoohi, Asghar. "The Determinants of Labor PAC
Allocations to Legislators." Industrial Relations 25 (Fall 1986):328-338.

McCaffery, David P. "An Assessment of OSHA’s Recent Effects on Injury Rates."
Mal _o_f Hit—mam Resources 18 (Spring 1983):144-145.

Millis, Harry A. and Montgomery, Royal E. M15. Progress ad Spm_e Ba_sic m
Problems. New York: McGraw-Hill Book Company, Inc., 1938.

and . The Economics o_f Labor. Vol. 1: Organized Labor.

 

 

 

New York: McGraw-Hill Book Company, Inc., 1945.

Neumann, George R. and Rissman, Ellen R. "Where Have All the Union Members
Gone?" Mal o_f m Economics 2 (April 1984):175-192.

Oswald, Andrew J. "The Microeconomic Theory of the Trade Union." Economic

Journal 92 (September 1982):577—589.

 

116
. "New Research on the Economics of Trade Unions and Labor Contracts."
Industrial Relations 26 (Winter 1987):30-45.

Parsons, Donald 0. "Trends in Male Labor Force Participation." Journal o_f Political

 

Economy 88 (February 1980): 117-134.
Pencavel, John H. "The Tradeoff Between Wages and Employment in Trade Union

Objectives." The Quarterly Journal o_f Economics 99 (May 1984):215-232.

 

Perlrnan, Selig. A Theoty o_f th_e Labor Movement. New York: The Macmillan
Company, 1928.
Price, Daniel N. "Private Industry Health Plans: Type of Administration and Insurer

in 1974." Social Security Bulletin 40 (March 1977):13-27 & 42-43.

 

 

Prosten, Richard. "The Longest Season: Union Organizing in the Last Decade, a/ k/ a
How Come One Team Has to Play with Its Shoelaces Tied Together?"

Proceedings o_f me Thirty-First Annual Meeting, Madison: Industrial Relations

 

Research Association, 1978.

 

University of Chicago Press, 1977.
Rogin, Michael. "Voluntarism: The Political Functions of an Antipolitical Doctrine."
Roomkin, Myron and Block, Richard N. "Case Processing Time and The Outcomes

of Representation Election: Some Empirical Evidence." University o_f llinois

 

Law Review 1981: 75-97.
Rossiter, Louis F. and Taylor, Amy K. "Union Effects on the Provision of Health

Insurance." Industrial Relations 21 (Spring 1982):167-178.

117
Sandver, Marcus and Heneman, Herbert G. HI. "Union Growth Through the
Election Process." Industrial Relations 20 (Winter 1981):109-115.
Shister, Joseph. "Unresolved Problems and New Paths for American Labor."
Seeber, Ronald L. and Cooke William N. "The Decline in Union Success in N .L.R.B.
Representation Elections." Industrial Relations 22 (Winter 1983):34-44.
Skolnik, Alfred M. "Pension Reform Legislation of 1974." Saga; m I_3u_llati_n
37 (December 1974):35-41.

. "Twenty-Five Years of Employee-Benefit Plans." Spc'al Security

 

Bulletin 39 (September 1976):3-21.
Smith, Robert Stuart. "The Impact of OSHA Inspections on Manufacturing Injury

Rates." Journal _(_)_f Human Resources 14 (Spring 1979):145-170.

 

 

Solomon, Benjamin. "Dimensions of Union Growth, 1900-1950." Indastrial m

Labor Relations Review 9 (July 1956): 544-561.

 

 

Stepina, Lee P. and Fiorito, Jack. "Toward a Comprehensive Theory of Union
Growth and Decline." Industrial Relations 25 (Fall 1986):248-264.
Stieber, Jack. "Most U.S. Workers Still May Be Fired Under the Employment-at-

Will Doctrine." Monthly Labor Review 107 (May 1984):34-38.

 

Strauss, George. "Industrial Relations: Time of Change." ndustrial Ralations 23
(Winter 1984):1-15.

Sundquist, James L. Politim and Policy; 1115; Eisanhower, Kennady gala mam
lea—Ls. Washington, DC: The Brookings Institution, 1968.

1 18
Brothers, 1957.

The Structure and Government of Labor Unions. Cambridge,

 

Massachusetts: Harvard University Press, 1962.

. Organized Labor i_n American Histoty. New York: Harper & Row

 

Publishers, 1964.

 

Harvard University Press, 1958.
Viscusi, W. Kip. "The Impact of Occupational Safety and Health Regulation." _B_efl

Journal o_f Economics 10 (Spring 1979):117-140.

 

Voos, Paula B. "Union Organizing: Costs and Benefits." Industrial m Labor

 

Relations Review 36 (July 1983): 576-591.

 

. "Trends in Union Organizing Expenditures, 1953-1977." Industrial

amd Labor Relations Review 38 (October 1984): 52-63.

 

 

. "Union Organizing Expenditures: Determinants and Their Implication

 

 

for Union Growth." Journal o_f Labor Research 8 (Winter 1987):19-30.

Witte, Edwin E. "The New Federation and Political Action." Industrial fl Labor

 

Relations Review 9 (April 1956):406-418.

 

Wolkinson, Benjamin W. Blacks Unions ad th_e EEOC: A Study o_f Administrative

 

 

Futility. Lexington, Mass: Lexington Books, 1973.
Woodbury, Stephen A. "Substitution Between Wage and Nonwage Benefits."

American Economic Review 73 (March 1983): 166-182.

 

 

UNIV LIBRARIES

H Hiiniiiu ii iii ii j