Journal of Economic Perspectives—Volume 2, Number 2—Spring 1988—Pages 63–88
Contraction and Expansion:
The Divergence of Private Sector and
Public Sector Unionism in
the United States
Richard B. Freeman
T
he institutional structure of the American labor market changed remarkably
from the 1950s and 1960s to the 1980s. In the '50s and '60s trade unions
seemed permanently established in the private sector of the economy: a third
of nonagricultural wage and salary workers and over half of blue-collar workers were
union members; hundreds of thousands of workers voted annually in National Labor
Relations Board (NLRB) representation elections to join unions; most large firms
sought stable collective bargaining relations with their unions. By contrast, in the
public sector only 10–12 percent of workers were union members; fewer were covered
by collective bargaining contracts; and most experts regarded public employees as
intrinsically nonorganizable. According to AFL-CIO president George Meany, it was
"impossible to bargain collectively with the government" (Kramer, 1962, p. 14).
The massive contraction of unionism in the private sector and expansion in the
public sector in the 1970s and 1980s (see Exhibit 1) has produced an utterly different
situation today. In the private sector, the proportion of wage and salary workers in
unions plummeted to 14 percent in 1986—a level comparable to that in the Great
Depression; only a minuscule number of workers joined unions through NLRB
elections; and national companies openly proclaimed their intent to establish a
"union-free environment." By contrast, in the public sector over a third of the work
force was unionized; some 40 percent were covered by collective contracts;1
and
1
Estimates of organization in the public sector differ somewhat among sources, though all data show greater
organization than in the private sector. See Freeman, Ichniowski, and Zax (forthcoming) for a detailed
analysis of the various statistics.
• Richard B. Freeman is Director of Labor Studies, the National Bureau of EconomicResearch,
Cambridge, Massachusetts, and Professor of Economics, Harvard University, Cambridge, Mas-
sachusetts.
64 Economic Perspectives
Exhibit 1
Changing percent of non-agricultural workers who are union, by sector.
public sector unions such as the American Federation of State, County and Municipal
Employees, National Education Association, Service Employees International (which
became largely public sector in the period), and the American Federation of Teachers
were among the largest in the country. With one in three union members working in
the public sector, women with master's degrees (largely schoolteachers) more highly
organized than male high school graduates, and police and firefighters the exemplars
of craft unionism, the union movement differed drastically from that headed by
Meany in the 1960s.
Because the private sector employs nearly 85 percent of nonagricultural wage
and salary workers in the United States,2
the contraction of unionization in the
private sector dominates the trends for the economy as a whole, with the result that
the union proportion of nonagricultural employees fell from 36 percent in 1956 to 18
2
U.S. employment data come from two basic sources: Current Population Survey data on individuals and
establishment data from the Bureau of Labor Statistics. The 85 percent figures comes from the establishment
survey.
Freeman 65
percent in 1986. This decline in union density was larger than that of the 1920s, and
thus arguably represents the most significant change in labor market institutions since
the Depression—the effective de-unionization of most of the U.S. labor force.
What explains the decline in union representation of private wage and salary
workers? Why have unions expanded in the public sector while contracting in the
private sector? Is the economy-wide fall in density a phenomenon common to
developed capitalist economies, or is it unique to the United States? To what extent
should economists alter their views about what unions do to the economy in light of
the fact that they increasingly do it in the public sector?
To answer these questions I examine a wide variety of evidence on the union
status of public and private workers. I contrast trends in unionization in the United
States with trends in other developed countries, particularly Canada, and use these
contrasts and the divergence between unions in the public and private sectors of the
United States to evaluate proposed explanations.
Because standard economic theory is not normally used to analyze massive
changes in the institutional structure of an economy, many may think that an
explanation of changing union density requires extensive reliance on social factors that
go beyond economics. I argue that this is not the case and base my analysis on
economic forces and behavior. The complexity of institutional change does, however,
dictate a catholic analytic approach, and I use several lines of argument and evidence
rather than a single econometric model or hypothesis test to reach conclusions. In
addition, I recognize that a significant residual remains that can and perhaps should
be attributed to noneconomic factors such as the abilities of union leaders, public
opinion, and the like.
The Dimensions of Change
"It is a capital mistake to theorize in advanceof the facts."
Arthur Conan Doyle, recording the words of S. Holmes
Three firm observations can be made about the change in union density in the
United States: first, that the decline in density in the private sector has been virtually
ubiquitous, encompassing workers in all industries, regions, and occupations during
the 1980s and earlier; second, that by contrast union expansion in the public sector
has been highly uneven, occurring rapidly in some states and occupations but not in
others; and third, that the overall drop in union density—the de-unionization of the
economy—is distinctly American, contrasting sharply with developments in most
Western countries.
Exhibit 2 documents the claim that union density has fallen among virtually all
workers during the 1980s, when the rate of decline was exceptionally severe. The
industry figures show the union proportion of workers falling in the areas of traditional
union strength such as transportation and public utilities (including trucking,
which the Teamsters once dominated); construction (long the preserve of craft
66 Economic Perspectives
Exhibit 2
Percentage of private wage and salary workers who are members of unions,
by industry and occupation 1980–1986
unions); manufacturing (where industrial unions have held sway since the organizing
drives of the 1930s and 1940s); and services and trade (where proportions organized
were so low that even a modicum of union success would raise the union shares.) The
occupation data (which understate declines in private sector density because they
include public sector labor) reveal a similar trend, with sizeable drops in density
among operatives, craft workers, and laborers. As for the longer run, the percentage of
production workers in metropolitan areas with collective bargaining contracts fell
from 73 percent in 1960–61 to 51 percent in 1984, with massive declines in every
region of the country ranging from – 16 points in the South (48 percent to 32 percent)
to – 32 points in the West (80 percent to 48 percent, Goldfield, 1987). These declines
Contraction and Expansion of Unionism 67
show that much more is involved in the contraction of union density in the private
sector than changes in the regional mix or production worker share of employment.
Overall, the data indicate that the explanation of the decline in private sector
union density should focus on factors that affect all private employees as opposed to
factors that affect employees in some segments of the labor market rather than others.
The Rise in Public Sector Union Density
Measuring changes in unionism in the U.S. public sector is difficult. One
difficulty is that each American state (and the federal government) regulates public
sector labor in its own jurisdiction, producing labor organizations with different legal
rights. Whereas in the private sector union membership, collective bargaining representation,
and the right to strike are coterminous under the same legal code, in the
public sector a worker can be a union member but not be covered by a collective
contract in some states (because management can refuse to bargain or to sign an
agreement with a union) while being covered in others; be covered by a contract that
does not include wages and salaries, as in federal employment; be allowed to strike in
some jurisdictions but obligated to resolve disputes through arbitration elsewhere, and
so on. Historically, moreover, public sector worker organizations have ranged from
associations opposed to collective bargaining to acting as aggressive unions, with the
same organization serving as a union in some localities but not in others (the
American Association of University Professors is a case in point). Indicative of the
measurement problem, the figures on public sector density in Exhibit 1 show discontinuities
when the relevant surveys first included "associations" with unions in their
definition of collective organization. While it might seem misleading to compare union
membership in early years with union plus association membership in later years, this
comparison does in fact accurately measure the magnitude (though not the timing) of
change. This is because in the 1950s and early 1960s most associations—for instance,
the NEA or the various police and firefighters organization—did not engage in
collective bargaining, often rejecting it as "unprofessional," whereas in the 1980s they
embraced bargaining and the other attributes of traditional trade unions.
A final complexity is that the legal environment under which public sector unions
operate has changed over time. From the 1960s through the early 1970s, when public
sector workers joined unions or saw their associations turn into unions, most public
sector labor laws did not require employers to bargain/resolve impasses and outlawed
the strike weapon that unions use to force employers to come to agreement. As a result
many workers who joined unions did not work under collectively bargained contracts.
By contrast, from the mid-1970s to the 1980s more and more union workers (and
nonunion workers in organized workplaces) were able to gain contracts from their
employers (35 percent of state and local government workers were covered in 1982
compared to 26 percent in 1975).3
At the same time that contract coverage was
increasing, however, the proportion of public sector workers in unions stabilized or
3
These data are from the Survey of Governments, which differ from those in the Current Population
Survey, as described in Freeman, Ichniowski and Zax (forthcoming).
68 Economic Perspectives
Exhibit 3
Geographic variation in percentage of employees represented by bargaining units,
by occupation and state law, 1982
fell. In this case, statistics on membership and coverage tell different stories about the
trend in union strength. As collective bargaining resulting in a contract has been the
essence of American union activity, I stress increased contract convergence rather than
membership trends in assessing the 1980s.
Given differences in state regulations of public sector collective bargaining, it is
perhaps not surprising to find that the extent and, by extension, expansion of
bargaining has varied across states and among occupations (see Exhibit 3). In states
with laws favorable to collective bargaining (arbitration or strike permitted laws, and
duty to bargain), proportionately more workers are represented by bargaining units
than in states that permit but do not require employers to bargain, while representation
by unions is low in states that have no collective bargaining provisions and lowest
in those that prohibit bargaining. Though these data relate to cross-section differences
at a point in time rather than changes over time, we can infer patterns of growth from
them. We can do this because contract coverage was sufficiently slight in the earlier
period that states/occupations with high density in the 1980s almost always had rapid
growth whereas those in which density was low could not have had such growth.
Interpreted in this way, the exhibit shows the uneven expansion of union representation,
with proportionately more workers gaining representation in states (and occupations)
having laws favorable to collective bargaining than in other states.
The evidence on geographic divergence implies that an explanation of union
growth in the public sector must have a geographic dimension related to state laws as
opposed to the "across-the-boards" explanation that seems appropriate for the private
sector.
Cross-Country Contrasts
Because the meaning and measurement of union membership differs across
countries, the figures in Exhibit 4 should be regarded as crude indicators of degrees of
Freeman 69
Exhibit 4
Levels and changes in union density as a percent of
nonagricultural wage and salary across countries, 1970–85
unionization. Even read cautiously, however, they clearly contradict the notion that
the decline in union density in the United States is part of a general collapse of unions
in the developed world; in most countries, union density increased in the 1970s and
stabilized or declined modestly in the 1980s at levels above those in earlier years. In
Japan, where union density fell in both decades, the rate of decline was half as large
as in the United States, and less significant in terms of the entire work force because
the proportion of Japanese working as nonagricultural wage and salary workers
increased, maintaining the union share of total employment. In Canada, where many
of the same companies and unions operate as in the United States and where living
standards and culture (exclusive of Quebec) are similar, union density went from
roughly the same to twice the U.S. level from 1970 to 1985. A persuasive explanation
of the decline in union density in the United States should also explain why density
did not decline in Canada in the same time period. A third country to which I would
70 Economic Perspectives
like to draw attention is the United Kingdom, where density fell in the 1980s,
probably at rates above those shown in the table (the union figures come from the
unions, who exaggerate membership in a period of decline); where organization in the
strongest union structure, the closed shop (which requires workers to be union), fell
sharply; where an increasing proportion of private manufacturing firms operated
nonunion; where employment grew more rapidly in low-union density than in
high-union density plants; and where proportions covered by collective contracts also
fell noticeably.4
Because union density increased in the 1970s in the United Kingdom,
however, these changes have not produced anything like an American-style drop
through 1985.
Overall, the patterns of change shown in Exhibit 4 highlight the fact that
de-unionization is largely, if not exclusively, a U.S. development.5
This fact has
important implications. First, it disallows any broad explanation of the decline, say,
that "unions have become obsolete in the modern 'post-industrial' market economy"
(whatever those grandiose terms mean). Second, it constitutes powerful evidence
against the structuralist argument that changes in the composition of the work force
and jobs from (traditionally union) male blue collar labor to (traditionally nonunion)
female, white collar, and service labor underlie the drop. If "post-industrial" or
structural changes inexorably reduce unionization, density would have fallen in
Canada and other developed capitalist countries, all of which have experienced
essentially the same structural changes as the United States, as well as in the United
States. Finally, the divergent country trends also cast doubt on any general macroeconomic
explanation of U.S. de-unionization: it is hard to argue that the economic
problems that followed the oil shock, the inflation of the 1970s, or ensuing deflation
caused unions to decline in the United States when the same forces did not reduce
union density elsewhere. Put differently, the international data direct attention away
from worldwide economic developments to the particulars of American labor rela-
tions.
How American Workers Organize
"No, no! The adventures first," said the Gryphon in an impatient tone: "explanations take such a
dreadful time."
As reported by Lewis Carroll
Before evaluating proposed explanations for the change in union density in the
United States, it is important to understand the way in which workers are organized.
4
Millward and Stevens (1986) report a rise in the percentage of manufacturing establishments operating
nonunion from 18 percent to 29 percent (pp. 58–59) and a drop in the percentage with a closed shop from
30 percent to 18 percent (p. 103).
5
This is not to say that unions throughout the West do not face serious problems in adjusting to the
changing economic climate. They do. The difference between the United States and most other countries is
that only in the United States have these problems taken the form of massive de-unionization of the private
sector.
Contraction and Expansion of Unionism 71
In contrast to some countries such as the United Kingdom where unionization is left
largely to private parties, governments in the United States regulate extensively the
process of union representation in both private and public sectors.
The National Labor-Management Relations Act (NLRA) of 1935, amended in
1947 (Taft-Hartley Act) and afterwards, establishes secret ballot representation elections
for private sector workers to choose whether or not to become union and
regulates the labor relations conduct of employers and unions in several ways. The Act
forbids employers to discriminate against workers for union activity and forbids unions
to engage in certain tactics deemed unfair as well, such as secondary boycotts.6
Under
the National Labor Relations Board process, unionization currently takes the following
form: a union petitions for an election by producing the signatures of 30 percent
or more (in practice a union will petition for an election only if two-thirds of the
employees are willing to request an election). The NLRB decides the set of workers
eligible to vote, and, subject to legal objections and appeals that can delay the actual
election for several months, supervises the voting. If a majority votes for the union the
NLRB will certify it as the collective bargaining representative of the work force,
which obligates the employer to bargain, though not to come to agreement. To resolve
impasses, workers can strike while employers can lock out employees or hire
strikebreakers. The NLRB also supervises secret-ballot "decertification" elections for
unionized workers who wish to reconsider their status.
The key to understanding the process is to recognize that although only workers
vote in a representation election, management plays an active role in most campaigns,
generally trying to convince workers to reject unions. One study of some 200
organizing campaigns in 1982–83, for example, estimated that the direct supervisors
of workers played a "sizeable" or "extreme" role in two-third of the elections, while
abstaining in just 6 percent (AFL-CIO, Appendix, p. 37). Many managements
employ labor-management consulting experts ("union-busters") to convince workers
to vote against the union. Determining union status through an adversarial election
framework which accords management a major role is a distinctly American process
which has contributed, I will argue, to the observed decline of private sector union
density.
The fact that each state has its own public sector labor law means that there is no
uniform method by which workers unionize in the public sector. Indeed, while some
states enacted NLRA-type laws that create secret ballot election procedures for
workers to decide union status and that require employers to negotiate with the
unions, often mandating compulsory arbitration to resolve disputes, other states
outlaw unions or collective bargaining.7
Because state laws often mimic the NLRA,
however, the situation is not as chaotic as might first appear. A reasonable generalization
is that in states with comprehensive public sector labor laws, the de jure process of
unionization in the public sector is much like that in the private sector, with workers
6
That is, situations where a union tries to pressure an employer by placing economic pressure on a third
party, often a customer or supplier.
7
The changes in law are reviewed in Freeman (1986) and in Valletta and Freeman (forthcoming) in greater
detail. The NBER has a computer file on regulations across states, departments, and time, which is
available to researchers on request.
72 Economic Perspectives
deciding union status through government-sponsored voting procedures. As the method
for unionization is comparable between the public and private sectors, an explanation
of divergent sectoral trends must explain why comparable processes produce different
outcomes in the two sectors.
Finally, U.S. labor law establishes representation elections for workers to choose
union status but does not outlaw private agreements between workers and management
regarding unionization. Unions and employers can agree that a given workplace
will be organized (as GM, Toyota, and UAW have done for the Fremont, California
plant) with an understanding that if the workers or another union seek an election,
management will remain neutral or encourage workers to choose the preferred union.
Unions can also sign up workers and demand recognition, threatening a strike should
management refuse. In some sectors, moreover, employers are union because they hire
from union hiring halls. While non-NLRB modes of organization are likely to become
more important in the future, representation elections have been the predominant
mode of organization in the past thirty years, making the outcome of the election
process critical to union growth in the period under study.
Relating New Organization to Density
The discussion thus far has dealt with the process by which workers choose to
unionize and hence with the flow of new union members. As our ultimate concern is
with changes in union density it is essential to relate the flows to the "stock" of union
members. The following identity, which makes total union membership a function of
the changes in existing membership and of "investment" through organizing activity,
relates the flows to the stocks:
UNION(t) = UNION(t – 1) rUNION(t – 1) + NEW(t).
UNION is the number of union members (persons covered by collective
bargaining contracts) in a given year t.
-r is the rate of change in membership due to changes in employment in
organized establishments. It may be negative or positive, though the "natural"
process by which some union plants close yearly (while new plants are born
nonunion) suggests that -r will be negative, implying a natural depreciation in
union membership, even in a growing economy.
NEW is the number of new members obtained through organization of new
workplaces or lost through decertification of unions at existing workplaces in the
period from t – 1 to t. For reasons of data availability NEW is best thought of
as net members gained through NLRB elections, though in principal one would
like to include the result of all new organizing activity.
Dividing both sides of the identity by the number of employees in year t to
measure union density (UDENS), and manipulating, one obtains
UDENS(t) = [1/(1 + g)][(1 r)UDENS(t – 1) + PCTNEW(t)]
(1 – r – g)UDENS(t – 1) + (1 – g)PCTNEW(t)
Freeman 73
PCTNEW is the ratio of workers organized to the work force in t – 1; g is the
rate of growth of total employment; and – (r + g) is the net depreciation or
appreciation of union density.
The steady state level of union density implied by this difference equation is
UDENS = PCTNEW/(r + g), which shows that the permanent level of union
density changes whenever r + g or PCTNEW change; that is, whenever economic
conditions cause union plants to contract/expand relative to the growth of total
employment or when the rate of union organization changes over time. For example,
if unions organize 1 percent of the work force per annum, as they did in the 1950s,
and suffer a net depreciation of density of 4 percent, the union share of employment
would stabilize at 25 percent. If, alternatively, new organization fell to .7 percent of
the work force, as it did in the 1960s, and r + g remained the same, the union share
would drop to 17.5 percent in the long run.
How important are changes in new organization versus changes in the rate of
depreciation of existing union membership in the recent collapse of private sector
union density? Crude calculations applying our difference equation to private sector
union membership and number of workers won through NLRB elections show a rising
trend in the depreciation rate in recent years: estimated compound annual rates of
r + g for the 1950s of 4.0 percent; for the 1960s, 3.4 percent; for the 1970s, 4.7
percent; and for 1980–85, 6.1 percent.8
Two factors are likely to underlie the increase
in r + g in the 1970s and 1980s: the concentration of unionism in slow-growth areas
and potential increased death of union plants due to the rising union wage premiums
of the seventies. The crudeness of the data underlying the estimates of r + g makes
one leery, however, of taking estimated changes too seriously. What should be taken
seriously is the negative value of – (r + g), which implies that unions, like the Red
Queen in Through theLooking Glassfor whom "it takes all the running you can do, to
keep in the same place," must organize large numbers of workers each year to
maintain private sector density.
That unions have failed to do this through the NLRB electoral process is evident
from the statistics summarized in Exhibit 5. Columns 1 and 2 measure the success of
unions in organizing workers through NLRB representation elections in selected years
from 1950 to 1983: column 1 gives the absolute number of workers won by unions in
representation elections while column 2 gives the number of workers won relative to
the work force. These data show that union success in representation elections has
declined to the point where virtually no workers are organized through NLRB
8
These estimates are based on the private sector union membership numbers of Troy and Sheflin for 1950,
1960, 1970, 1980, and a projected 1985 figure based on his 1983 (updated using CPS figures on the
percentage change in membership from 1983 to 1985); estimates of private nonagricultural wage and salary
workers exclusive of private household workers from the BLS; and the number of members won by unions
in NLRB elections in 1951–1960, 1961–1970, 1971–1980, 1981–1985 (with 1984 and 1985 assumed to be
the Same as 1983). I calculated the compound annual rate of depreciation in union membership by
comparing actual membership in, say 1980, with membership in 1970 plus the number of workers won from
1971–1980, and then subtracted the compound annual rate of growth of the relevant employment.
74 Economic Perspectives
Exhibit 5
Organizing success of unions through NLRB elections
procedures. Whereas in the early 1950s unions organized 1 percent to 2 percent of the
work force via government-sponsored elections, in the 1960s they organized about .7
percent; in the 1970s, about .5 percent; and in 1983,just .1 percent: 91,000 workers in
a work force of some 90 million! Even these figures, moreover, understate union
inability to gain members through NLRB elections for a union electoral victory does
not guarantee that workers will obtain a collective bargaining contract. In recent
years, in fact, workers voting to unionize failed to gain a contract approximately a
third of the time (Weiler, 1985), and thus worked under management-determined
rather than collectively bargained conditions. The implication is that in 1983 perhaps
only 60,000 workers gained union representation through the electoral process. When
account is taken of the 25,000 or so workers who chose to decertify unions in elections,
it is apparent that the legally established mode of organizing labor in the private
sector of the United States has run dry for trade unions.
Columns 3 and 4 of Exhibit 5 decompose the number of workers won by unions
in representation elections relative to the work force into: a measure of the union
success rate—the number of workers in elections won by unions relative to the number
of workers eligible to vote (column 3); and a measure of the extent of electoral activity—the
number of workers eligible to vote in elections relative to the work force.
The figures in column 3 show that one factor in the decline in workers organized
through representation elections has been a fall in the proportion of workers in
elections won by unions (which reflects declines in union win rates and in the average
size of union victories).9
The figures in column 4 show, however, an even greater
proportionate drop in the extent of electoral activity: in 1960 some 6000 NLRB
elections covered 1.2 percent of the work force; in 1983 4400 elections covered a bare
9
One can use various identities to decompose the decline in members won/employment (Freeman, 1985).
Contraction and Expansion of Unionism 75
.3 percent of the work force. Had unions won all of these elections they would not
have gained enough members to increase density.
Turning to the public sector, there are good reasons to believe that both terms in
our equation for union density—the net depreciation of existing density – (r + g) and
success in organizing new workers—played a role in the speed with which unionization
and/or collective bargaining representation grew in the 1970s and 1980s. First,
because few government departments, union or nonunion, go out of business, the loss
of density due to closure of union workplaces (reflected in r) will be small. Second, the
relatively moderate growth of public sector employment into new areas implies that
existing unions do not have to appeal to workers outside their traditional jurisdictions
to maintain their share of employment, which also should make r + g smaller in the
public than in the private sector. Third, organizing workers into unions has undoubtedly
been much easier in the public sector. In part this is because many workers
have been historically organized into employee associations, which were readily
transformed into unions without massive organizing campaign. More importantly,
however, organization is easier in the public sector because politicians and public
sector managers do not contest unionization to the extent that private employers do,
for reasons to be laid out shortly. Finally, state laws that require employers to
negotiate and mandate impasse procedures guarantee that workers who vote union
ultimately receive union representation. Hence, in the public sector, once workers
choose union representation, they get what they choose.
With this and the information in the preceding section as background, let us now
turn to proposed explanations for the changed union density.
Potential Causes of Change
"It's rather hard to understand. . . Somehow it seems to fill my head with ideas—only I don't
know exactly what they are! However, somebody killed something: that's clear, at any rate."
Alice, as reported by Lewis Carroll
Researchers have proposed a wide variety of factors to explain the changes in
unionization in the United States. Among the hypothesized causes for the decline in
private sector density are: structural shifts in the composition of the work force and
mix of jobs; changes in public attitudes toward unions reflected in opinion polls;
increased governmental regulation of the labor market substituting for union protection;
"positive labor relations" by nonunion firms; the performance of unions in
representing workers and in allocating resources to organizing drives; antiunion
policies of the Reagan administration evinced in the air traffic controllers' strike and
selection of members to the NLRB; antiunion campaigns by managements. Among
the factors said to cause the spurt in public sector unionism are: extension of NLRA
type laws to public sector workers in many states; extension of some union rights to
federal workers by executive order; pent-up demand for unions by workers whose
organization level was exceptionally low by U.S. and world standards.
76 Economic Perspectives
By examining the impact of factors on the changes in the private and public
sectors at once, and requiring that the factors have a consistent effect, I eliminate some
hypotheses and highlight the role of others.
Structural Changes
Because the proportion of the work force in traditionally nonunion occupations
(white collar), demographic groups (females, college graduates), industries (service
and trade) and regions (the South) has increased rapidly, various analysts have
explored the possibility that the changing structure of the work force underlies the
drop in union density. The tool for this exploration is a fixed coefficient model which
decomposes the workforce into a number of groups with varying degrees of unionization
at some base time, and explores the impact of changes in the relative size of the
groups on aggregate union density under the assumption that the density of each
group is fixed at its base level.
Fixed coefficient analyses covering from the 1960s through 1980 attribute 50
percent to 70 percent of the decline in private sector union density to compositional
factors with the increase in white collar employment having a particularly sizeable
depressant effect on density. While these calculations would appear to go far toward
explaining the decline in unionization, I believe the structuralist analysis is misleading
and should be rejected. There are three reasons for rejecting it.
First, the structuralist hypothesis is inconsistent with the rise in union density in
other countries (notably Canada) which had structural changes in the work force
similar to those in the United States. Second, surveys of worker desire for unionism
show that structural changes cannot explain the decline in union success in NLRB
elections, since the groups whose proportion of the work force increased (such as
women and young workers) have as great or greater desire to unionize as do white
prime-age male workers (Freeman and Medoff, 1984, p. 228). Third, I reject the
structuralist hypothesis because it assumes that the union share of workers in a sector
should remain fixed over time. That assumption is inconsistent with the history of
union growth, which is one of expansion into nonunion areas, as occurred in the
public sector in the period under study. The claim that public sector workers
organized because of pent-up demand for unionization, indeed, implies that unionism
"naturally" grows in new sectors over time. Fixed coefficient calculations sidestep the
key issue in the decline in U.S. unionization, which is why unions failed to organize
historically nonunion workers in the private sector while doing so in the public sector
and in other countries.
Public Opinion of Unions
The decrease in favorable attitudes toward unions shown in public opinion polls
offers another possible explanation for the decline of unionism in the private sector
(Lipset, 1986). While public opinion data should not be dismissed out of hand, I do
not find this hypothesis persuasive. The timing of the change in opinions and union
density are at best weakly related, with public support of unions as measured by the
Gallup Poll climbing from 62 percent in 1949 to 75 percent in 1953, falling to 64
Freeman 77
percent in 1962, rising to 70 percent in 1965, falling to 55 percent in 1981, and
rebounding slightly to 58 percent in 1985. Public favorableness toward unions was 59
percent in 1973 and 58 percent in 1985, but union density fell sharply during that
time. More damaging to this thesis, perhaps, are responses to the behaviorally more
meaningful question of whether workers would vote for a union at their work place,
which show no decline in worker desire for unions: in both 1977 and 1984 about a
third of nonunion workers said they wanted unions at their workplace.10
Finally, a
public opinion explanation of union decline is inconsistent with unionization of the
public sector, where public opinion could be expected to be especially important.
Substitutes for Union Protection
Workers join unions for protection against unfair treatment by management,
including low wages. This motivation suggests that the development of substitute
modes of protection in the form of welfare state interventions in markets (Neumann
and Rissman, 1984) or of better personnel practices by management might reduce
worker desire for union representation. If government activities substitute for union
protection, one would expect: unionism to decline most in countries with the greatest
welfare state and restrictions on management, in American states with extensive
protective labor legislation, and in periods of declining governmental interventions.
Further, unions should appeal least to workers who enjoy special legal protections,
such as blacks, women, and public sector employees protected by civil service rules.
Changes in union density by country, by states, and over time, and the desire for
unions by blacks, women, and public sector workers are uniformly inconsistent with
these implications (Freeman, 1986b). Hence, I reject the view that governmental
regulations and growth of the welfare state underlie the decline in density.
The possibility that "positive labor relations" practices adopted by many major
companies (but too expensive for smaller ones)—paying union level wages and
instituting union-style personnel practices such as seniority, job bidding and posting,
grievance systems (Foulkes, 1980)—has reduced worker desires for unions at these
workplaces finds better support in the data. Studies of organizing drives show that
companies with good personnel practices are more successful than others in defeating
unions (AFL-CIO, 1984; Kochan, McKersie, and Chalykoff, 1986), although the
largest effect of such practices must be in deterring drives in the first place. The
increased wage premium associated with size of establishment in the 1970s (Brown
and Medoff, 1986), which could be expected to make unionization less attractive to
those workers, is, moreover, consistent with the declining success of unions in NLRB
elections involving large establishments.11
Still, I doubt that positive labor relations
can explain much of the overall decline in union density. Large employers account for
10
The figures for 1977 are from the Quality of Employment Survey. Those for 1985 are from the Harris
Poll.
11
In the 1950s, unions had a better record in large units than in smaller ones, but this pattern of success was
reversed in the 1970s. The small number of elections in large units in recent years makes the 1980s figures
spotty. For data on union win rates by size of unit, see Goldfield (1987).
78 Economic Perspectives
a decreasing share of jobs in the U.S. and unions could increase density greatly
without organizing the IBMs of the world.
Government Industrial Relations Policies
Government industrial relations policies have been cited as causing unions to
decline in the private sector and to expand in the public sector. For the private sector,
some blame (credit) the Reagan Administration's operation of the NLRB and
destruction of the Professional Air Traffic Controllers Organization (PATCO) as
inducing an antiunion climate in the business community. The fact that union density
began falling in the private sector before the Reagan Administration and that
collective bargaining coverage has increased in the public sector during the 1980s
would seem to rule out this hypothesis, though it is possible that administration actions
may have contributed to the acceleration of the decline in the 1980s.
The argument that state policies toward unionism have played a major role in
the expansion of collective bargaining in the public sector has, on the other hand,
considerable research support. Numerous studies have found that public sector
unionization and collective bargaining contracts are more likely in states with favorable
labor laws than in other states (Freeman, 1986). Other studies—perhaps more
persuasive—have shown that unionism has spurted in states following passage of
favorable laws (Ichniowski, forthcoming; Saltzman, 1985, forthcoming). Even within
a city, departments for which state laws are more favorable toward collective
bargaining end up with contracts more frequently than "brother" departments
operating under less favorable state law (Freeman and Valletta, forthcoming). At the
minimum, we have an empirical regularity: public sector collective bargaining
coverage and union density increase markedly in the presence of laws favorable to
collective bargaining. This finding raises, however, two further questions: why comparable
laws induce different outcomes between the public and private sectors, with
which I will deal in detail shortly; and why many states enacted laws favorable to
unions while " labor law reform" failed at the national level, which I believe is due in
part to the simple fact that a minority of senators, representing states unfavorably
disposed to unionism, can stop national legislation but cannot, of course, undo
legislation in states that are favorably disposed to unionism.
Union Performance
Union performance may have contributed to the decline in union density in three
ways. First, unions may have represented members poorly, discouraging nonunion
workers from organizing. This hypothesis has little empirical support; it is inconsistent
with union success in obtaining wage and benefit increases exceeding those of other
workers in the 1970s and with opinion poll data showing union members to be
reasonably satisfied with their unions (Kochan, 1979). Second, unions may have failed
to allocate sufficient resources to organizing activity in the 1960s and 1970s. This
claim has some a priori validity, as organizing expenditures deflated by wages
(organizing is labor-intensive) have failed to keep pace with the growth of the
Contraction and Expansion of Unionism 79
increasing nonunion labor force and have been concentrated in sectors in which
unions were already strong rather than in new and growing industries (Paula Voos,
1983; 1984). Because decisions to allocate resources to organizing depend in part on
the organizing environment and the perceived benefits and costs of organizing
campaigns, however, I am leery of attributing a large independent role to union
organizing efforts. Union failure to embark on major organizing campaigns may
simply reflect rational responses to the low expected rates of success. Third, and
paradoxically from the perspective of analyses that treat unionization as a worker
decision, the success of unions in raising union wage premiums in the 1970s may have
contributed to the decline in density by raising the cost of unionization to firms and
thereby intensifying the antiunion activity of management—to which I turn next.
Aggressive Antiunion Management
Given that the NLRB electoral process allows management to influence results,
and that management is likely to prefer to operate nonunion, it is logical to seek an
explanation for declining union success in the behavior of management. Did antiunion
activity by management increase in the private sector in the period studied? How have
public sector managers behaved?
Exhibit 6 shows graphically that one indicator of management antiunion activity—"unfair"
labor practices under the law—has skyrocketed. While the NLRB
data in the exhibit measure charges of unfair activity rather than actions found illegal
in court (raising the possibility that unions may simply be filing more unfair charges),
the proportion of charges upheld has been roughly constant at 30 to 40 percent over
time (Weiler, 1983), and NLRB statistics on numbers of workers discharged for union
activity and ordered reinstated by the NLRB and courts show increases comparable to
those in the figure. As for legal management opposition, a Conference Board survey
reveals that 45 percent of firms in their Personnel Practices Forum had "operating
union free" as a labor policy goal in 1983 compared to 31 percent in 1977, indicating
that even over a short period management opposition to unionization has grown
substantially (Kochan, McKersie and Chalykoff, 1986).
In the public sector, by contrast, there has been no outburst of antiunion activity
by management. Charges of unfair labor practices concern interpretation of state
bargaining laws—whether a particular topic is subject to collective bargaining or is a
management prerogative—not opposition to the existence of unionism per se. Public
sector managers rarely hire union-preventing firms to discourage organization by their
workers. Because of changes in state laws, moreover, the trend in the public sector is
toward less rather than more management opposition to collective bargaining.
On the basis of these facts, I argue that the antiunion management offensive in
the private sector is the key to de-unionization of the United States, and that its
absence from the public sector explains the successful organization of public employees.
Moreover, I claim that the differential behavior of management in the public
and private sectors is explicable by the incentives facing them.
80 Economic Perspectives
Exhibit 6
Indices of charges of unfair labor practices against management and NLRB
elections (1960 = 100)
Management Offensive, Union Wage Premiums,
and Cost of Opposition
My proposed explanation can be most easily represented with a simple schematic
model (Exhibit 7), which diverges from many models of union organization by
stressing the role of management in unionization and the endogeneity of both
management opposition and union organizing. I postulate three basic relations.
The first relation is a production function for organizing success. It relates the
number of workers newly organized relative to the labor force to the resources devoted
by management to opposing unions, the resources devoted by unions to organizing,
and a vector of all other factors that might influence the outcome.
The second relation links the resources management devotes to opposing unionization
to economic factors likely to affect the profitability of such activity. One such
factor is the union wage premium, which is assumed to reduce profits and increase
management opposition. Another is the cost of opposing unionism in an NLRB
election, which will depend on the "technology" of battling unions at workplaces
(which has changed greatly in the past fifteen years due to the advent of labormanagement
consultants who specialize in training supervisors to pressure workers to
oppose unions and in running antiunion propaganda campaigns) and on the legal
penalty for committing unfair labor practices. In the private sector the legal penalty is
minuscule: the law requires that management reinstate workers unjustly fired for
union activity, pay the workers back pay less whatever income they earned in the
Freeman 81
Exhibit 7
Determinants of union organizing activity and management opposition
and their effect on unionization of new workers
The first relation
Determination of organizing success: PCTNEW = f(MAN, ORG, X)
where PCTNEW = number of workers organized/labor force;
MAN = resources devoted by management to opposing unions;
ORG = resources devoted by unions to organizing;
X = other factors that influence outcomes
The secondrelation
Management opposition: MAN = g(WAGEPREM, COST, ORG, Y)
where WAGEPREM = union wage premium;
COST = cost of opposing union in NLRB elections;
Y = other relevant factors, largely relating to product market
factors such as deregulation of industries, etc., which determine effect of unionism on profitability.
The third relation
Union and worker organizing effort: ORG = h(WAGEPREM, MAN, UDENS, Z)
where UDENS = union density at the beginning of the period;
Z = other relevant factors, largely relating to labor market.
period, and post a notice that the firm will not engage in such illegal activity again.
However, reinstating workers often occurs only after the representation election and
many choose not to return. Posting a notice about past illegal activity on the part of
the firm often has the effect of warning workers how far management is willing to go
to defeat unions rather than convincing them that management will forego such
tactics in the future. Finally, the management decision is also assumed to depend on
the resources the union devotes to organizing, and on other unspecified factors, largely
relating to product market conditions, that can be expected to make anticipated union
wage premiums more or less expensive in terms of lost profits.
82 Economic Perspectives
The third relationship in my analysis links union (and worker) organizing activity
to: the wage premium, presumed to raise the attractiveness of unions to workers and
thus increase organizing activity; the existing density of union workers, with a lower
density assumed to reduce organizing activity because the cost of organizing each
nonunion worker is higher to existing union members when there are proportionately
fewer unionists among whom to spread the cost; the amount of resources management
devotes to opposing unionism; and a catch-all vector of other relevant factors, largely
relating to labor market conditions.
One important feature of this model is that the union wage premium affects the
decisions of both the management and union (workers). In the management equation,
a higher premium induces more antiunion activity and thus reduces union organizing
success. In the union equation, a higher premium induces additional organizing
activity and thus raises union organizing success. Given opposing tendencies, is the
wage premium likely to be positively or negatively correlated with actual outcomes?
To the extent that unions extract rents from firms through monopoly wage increases,
the money loss to firms will exceed the transfer to workers (due to the "triangle"
inefficiency), giving management a potential incentive to spend more resources to
prevent unionization than unions/workers spend to organize (Freeman, 1986a). If
expenditures by the two sides have equal effects on outcomes, higher union wage
premium will reduce organizing success.
A second important feature of the model is the inclusion of management and
union activity in the equations determining the behavior of the other side. This
highlights the interactive nature of the organizing struggle—which can be viewed as a
two person (three person, if one distinguishes workers at an organizing site from the
union) game—and thus directs attention to the potential payoff from game-theoretic
analyses of union and management strategic behavior that goes beyond the scope of
this essay.
A third aspect of the analysis that deserves attention is the inclusion of union
density in the equation for union organizing activity, for it raises the possibility that a
fall in union density will produce a cumulative decline, as the increased cost of
organizing induces unions to lower organizing activity, while, conversely, rises in
union density will have the opposite cumulative effect.
How might one use the model embedded in these relationships to explain the
decline in private sector union density?
The most reasonable hypothesis, given the evidence in the previous section, is that
management opposition to unionism induced by changes in the cost of unionization
and the union wage premium reduced union organizing success, with cumulative
effects on union density and organizing effort. For this explanation to be valid, it is
necessary that management activity has an important impact on organizing success;
and that this activity responds to economic incentives.
With respect to the effectiveness of management opposition, studies by a diverse
set of researchers ranging from management groups to the General Accounting Office
to the AFL-CIO to academics find that management activity reduces union success in
NLRB representation elections and organizing drives (see Exhibit 8). The sole
Contraction and Expansion of Unionism 83
Exhibit 8
Summary of studies of effects of management activity
on NLRB representation election results
exception (Getman, Goldberg, and Herman, 1976) has been the subject of considerable
controversy, with Dickens convincingly reversing their conclusion in a re-analysis
of the underlying data. Even absent this, however, the preponderance of the evidence
is that the extent of management opposition substantially determines the outcomesof organizing
campaigns, as posited. As management devotes considerable resources to opposing
unionization, and presumably acts rationally, this makes intuitive sense.
Because the studies in Exhibit 8 focus on representation elections/organizing
drives, rather than on the workers gained relative to the work force (PCTNEW) that
84 Economic Perspectives
Exhibit 9
Estimates of the effect of management unfair labor practices on percentage
of the nonagricultural workforce newly organized in the NLRB elections
enters the union density equation, and do not provide estimates of the extent to which
the downward trend in union success can be attributed to management opposition, I
have estimated the impact of the one indicator of management opposition for which
time series data exists—unfair labor practices—on workers won by unions as a share
of nonagricultural employment. My analysis uses pooled state (industry) time series
data, with time dummies included to control for omitted trend and cyclical factors
and state/industry dummies to control for omitted state/industry effects, and diverse
control variables. The results, summarized in Exhibit 9, show that unfair labor
practices reduce the number of those unionized relative to the work force and that the
trend in unfair labor practices can account for roughly half of the observed drop in
that outcome measure. To the extent that unfair practices substitute for lawful
management antiunion activity, this estimate, while sizeable, understates the full
impact of management's fight against unions. Alternatively, however, if the growth of
unfair practices is positively correlated with the growth of legal management opposition,
my estimates overstate the impact of illegal management activity on the decline
in union success, though they may accurately measure the impact of management
opposition, in total.
The role of changes in the relative cost of union deterrence in simulating the
management offensive has not been the subject of extensive research, making conclusions
here more speculative. Time series calculations show that some of the rise in
unfair practices is explicable by the rise in union wage premiums in the 1970s, but the
series is short and the measures are crude (Freeman, 1986a). Estimates of the relation
between decline of union density and union premiums at a one-digit industry level
indicate that where the premium has risen most, union declines have been greatest
Freeman 85
(Linneman and Wachter, 1986), which may reflect reduced organizing success or a
greater death rate for union plants in those sectors. As for other factors likely to raise
the cost of unionism to firms, it is highly plausible that such product market
developments as the growth of foreign competition and deregulation have made
existing union wages more expensive in terms of lost profits. As increased nonunion
competition and management policies against unionized firms in construction indicates,
however, new product market competition from any source, not simply trade or
deregulation, can significantly reduce union density. Finally, with respect to the
determination of union organizing efforts, the major work here—by Paula
Voos—suggests that one can treat organizing behavior as rational decision-making.
Is this analysis consistent with the increase in public sector density?
There are three basic facts to explain regarding the spread of collective bargaining
to the public sector: the geographic variation in unionization associated with
different labor laws; the spurt in unionization that followed passage of the laws; and
the different behavior of public and private sector managers under comparable laws.
If the benefits to management of operating with a union relative to the cost of
opposing unions are lower in the public than in the private sector, and those costs
have decreased over time, we could account for these facts in the same framework
used to explain certain developments in the private sector: in terms of the incentives
and options for management to oppose unions.
In fact, the incentives for management to oppose unions do appear to be lower in
the public sector. First, public sector workers constitute an especially active political
group able to punish or reward the politicians who are their employers at the ballot
box, even though they are only a small proportion of voters in most areas. Second, the
cost of illegal opposition is likely to be greater for public than for private officials, as
public officials who break laws are likely to face possible removal from office. Third,
unions can help public sector employers increase budgets through lobbying for
additional public expenditures, creating a greater jointness of interest than in the
private sector. Fourth, wage premiums tend to be smaller in the public sector (see
Lewis, forthcoming; Freeman, 1986). Put crudely, management opposition to unions
can gain profits in the private sector; in the public sector, it can cost votes. Given this,
and the fact that states passed laws favorable to unionization in the 1960s and 1970s
that increased the cost of management opposition in the public sector, the spurt of
unions there is consistent with the model.
Finally, is this analysis consistent with the differential experience of Canada?
While private sector managers are likely to have similar profit incentives to oppose
unions in Canada as in the United States (Anderson and Gunderson (1982) show there
is no evidence that union wage effects are smaller in Canada than in the United
States), the Canadian system for organizing workers differs from the American system
by giving management less option to express opposition. Canadian labor boards rely
largely on "card checks" in which unions sign up workers to determine representation
at a work place rather than on adversarial elections and impose harsher penalties on
managements that break the laws (Weiler, 1983; Meltz, 1985), with the result that
Canada has not experienced anything like the massive outburst of management unfair
86 Economic Perspectives
labor practices that characterizes U.S. labor relations. The difference in institutional
procedures highlights the fact that the decline in private sector unionism in the U.S.
required two factors: increased incentive for management to oppose unions, and the
opportunity to turn that incentive into action.
Conclusion
As the reader will undoubtedly have observed, not all the pieces for a complete
explanation of the decline in private sector union density and rise in public sector
density in the United States are in place. But the available evidence seems consistent
with an explanation of private sector decline that stresses increased management
opposition to union organization, motivated in part by profit-seeking behavior, and
augmented by trade union responses; and an explanation of growth in the public
sector that stresses reduced management opposition due to passage of comprehensive
collective bargaining laws and vote-seeking behavior.
As for the likely impact of the change in unionization on the performance of the
economy, two points are worth attention. The first is that the increasingly public
sector locus of American unionism is likely to produce different union effects and
modes of operation than have been found for the traditionally private sector union
movement. Public sector unions have, for example, smaller wage effects and strike
rates than private sector unions and appear to increase rather than reduce employment
in union activities (see Freeman and Ichniowski, forthcoming). Second, without
making any judgment of whether a less unionized private sector will perform better or
not, one thing should be perfectly clear: analysts who have attributed national
economic problems ranging from unemployment to wage inflation to low productivity
to unions will have to find a new culprit to blame: unless there is a remarkable
renaissance in unionism, critics won't have unions to kick around any more.
References
American Federation of Labor—Congress of
Industrial Organizations, Department of Field
Services, AFL-CIO Organizing Survey, April, 1984,
Washington, D.C.
Anderson, John, and Morley Gunderson,
Union Management Relations in Canada, Addison,
Wesley, 1982.
Aspin, Leslie, A Study of Reinstatement Under the
National Labor Relations Act, Ph.D. dissertation,
MIT, 1966.
Brown, Charles and James Medoff, "The
Employer Size Wage Effect," mimeo, Harvard
University, 1986.
Catler, Susan, "Labor Union Representation
Elections: What Determines Who Wins?" B.A.
thesis, Harvard University, 1978.
Dickens, William, "The Effect of Company
Campaigns on Certification Elections,"Industrial
and LaborRelations Review, 1983, 36, pp. 560–75.
Dickens, William, and Jon Leonard,
Contraction and Expansion of Unionism 87
"Accounting for the Decline in Union Membership,
1950–1980," Industrial andLabor Relations Review,
April 1985, 38(3), pp. 323–334.
Drotning, John, "NLRB Remedies for Election
Misconduct: An Analysis of Election Outcomes
and their Determinants," Journal of Business,
April 1967, 40(2), pp. 137–48.
Foulkes, Fred, Personnel Policies in Large Nonunion
Companies.Englewood Cliffs, N.J.: PrenticeHall,
1980.
Freeman, Richard, "The Effect of the Union
Wage Differential on Management Opposition
and Union Organizing Success," American Economic
Review, May 1986a, pp. 92–96.
Freeman, Richard, "Unionism Comes to the
Public Sector," Journal of Economic Literature, Vol.
XXIV, March 1986b, pp. 41–86.
Freeman, Richard, "Why Are Unions Fairing
Poorly in NLRB Representation Elections?" in
Thomas A. Kichan, ed., Challenges and Choices
Facing American Labor. Cambridge, MA: MIT,
1985, pp. 45–64.
Freeman, Richard, and B. Ichniowski, "When
Public Sector Workers Unionise: Summary of
Findings." In Freeman, R.B. and C. Ichniowski,
eds., When Public Sector WorkersUnionise. NBER,
University of Chicago Press, forthcoming.
Freeman, Richard, and James Medoff, What
Do Unions Do? New York: Basic Books, 1984.
Freeman, Richard, and Rob Valletta, "The
Effects of Public Sector Labor Laws on Unionisation,
Wages, and Employment." In Freeman, R.B.
and C. Ichniowski, eds., WhenPublic Sector Workers
Unionise. NBER, University of Chicago Press,
forthcoming.
Freeman, Richard, B. Ichniowski, and J. Zax,
"Measuring Collective Organization in the Public
Sector." In Freeman, R.B. and C. Ichniowski,
eds., When Public Sector WorkersUnionise. NBER,
University of Chicago Press, forthcoming.
Getman, Jules, Stephen Goldberg, and Jeanne
Herman, Union Representation Elections: Law and
Reality.New York: Russell Sage Foundation, 1976.
Goldfield, M., The Declineof Organized Labor,
University of Chicago Press, 1987.
Ichniowski, C., "Public Sector Union Growth
and Bargaining Laws." In Freeman, R.B. and C.
Ichniowski, eds., When Public Sector Workers Unionise.NBER,
University of Chicago Press, forth-
coming.
Kochan, Thomas, "How American Workers
View Labor Unions," MonthlyLaborReview, 1979
102(4), pp. 23–31.
Kochan, Thomas, Robert McKersie, and John
Chalykoff, "The Effects of Corporate Strategy
and Workplace Innovations on Union Representation,"
Industrialand LaborRelations Review, July
1986, 39(4),pp. 487–501.
Kramer, Leo, Labor's Paradox–theAmerican
Federation of State, Council,andMunicipal Employees,
1962.
Lawlor, John, "The Influence of Management
Consultants on the Outcome of Union Certification
Elections," Industrial and Labor Relations Review,
1984, 38, pp. 38–51.
Linneman, Peter and M. Wachter, "Rising
Union Premiums and the Declining Boundaries
Among Noncompeting Groups," American Economic
Review, May 1986, 76, 103–108.
Lipset, Seymour M., "North American Labor
Movements: A Comparative Perspective" in
Lipset, ed., Unions in Transition. San Francisco:
Institute for Contemporary Studies, 1986.
Lewis, H. Gregg, "Union/Nonunion Wage
Gaps in the Public Sector." In Freeman, R.B. and
C. Ichniowski, eds., When Public Sector Workers
Unionise. NBER, University of Chicago Press,
forthcoming.
Meltz, Noah, "Labor Movements in Canada
and the United States: Are They Really That
Different?" in Thomas A. Kochan, ed., Challenges
andChoices Facing American (Cambridge, MA; MIT,
1985).
Millward, Neil and Mark Stevens, British
Workplace Industrial Relations 1980–84. England:
Gower Publishing Co. Ltd., 1986.
National Labor Relations Board, Annual Report
(Washington, DC; USGPO).
Neumann, George, and Ellen Rissman,
"Where Have All the Union Members Gone?"
Journal of Labor Economics, April 1984, 2(2), pp.
175–192.
Prosten, Richard, "The Longest Season Union
Organizing in the Last Decade," Proceedings of
the 31st Annual Meeting of the Industrial Relations
Research Association, Madison, Wisconsin,
1978, pp. 240–49.
Roomkin, Myron, and Richard Block, "Case
Processing Time and Outcome of Elections: Some
Empirical Evidence," University of IllinoisLaw Review,
1981, V1, pp. 75–97.
Saltzman, G.M., "Bargaining Laws and Cause
and Consequences of the Growth of Teacher's
Unions," Industrial andLabor Relations Review,1985,
38, 335–57.
Saltzman, G.M., "The Effects of Recent
Bargaining Laws in Ohio and Illinois," In Freeman,
R.B. and C. Ichniowski, eds., When Public
Sector Workers Unionise. NBER, University of
Chicago Press, forthcoming.
Seeber, R., and W. Cooke, "The Decline in
Union Success in NLRB Representation
Elections," Industrial Relations, Winter 1983, 22(1).
Troy, Leo, and Neil Sheflin, Union Sourcebook.
88 Economic Perspectives
New Jersey: Industrial Relations Data and Information
Services, 1985.
U.S. Department of Labor, Bureau of Labor
Statistics, Directory of National Unions and Employee
Associations, various editions.
U.S. General Accounting Office, Concerns Regarding
Impact of Employee Charges Against Employers
for Unfair Labor Practices. Washington, D.C., GAOHRD
82–80, June 21, 1982.
Voos, Paula, "Union Organizing: Costs and
Benefits," Industrial and Labor Relations Review, July
1983, 576–91.
Voos, Paula, "Trends in Union Organizing
Expenditures, 1953–1977," Industrial and Labor Relations
Review,October 1984, 38, pp. 52–63.
Weiler, P., "Promises to Keep: Securing
Workers' Rights to Self-Organisation under the
NLRA," Harvard Law Review, June 1983, 96(8),
pp. 1769–1827.
Weiler, P., "Striking A New Balance: Freedom
of Contract and Prospects for Union Representation,"
Harvard Law Review, 1985, 98 pp. 351–420.