What Unions Do:

Insights from Economic Theory

B R U C E E. K A U F M A N
Georgia State University, Atlanta, GA 30303

I. Introduction
At the end of What Do Unions Do?, Freeman and Medoff (F&M 1984, p. 247) con-
clude that "unionism on net probably raises social efficiency" and (p. 250) "recent
trends have brought the level of union density below the optimal level." For these
reasons, they advocate revision in the nation's labor laws in order to foster greater
unionization in the American economy.
These conclusions are controversial and, it seems fair to say, go against the main-
stream of opinion among economists. In making their case, F&M appeal to both the-
ory and empirical evidence. The empirical evidence is examined in considerable detail
in other articles in this symposium; my purpose in this article is to introduce the the-
ory side of the subject. Other authors will, however, also consider theory, albeit more
selectively. To proceed, I first review the theoretical literature on what F&M call the
"monopoly face" of unions. The emphasis is on basic models and their implications
for the effect of unions on key economic and social outcomes. The same is then done
for the voice/response face of unions. As much as possible, I present a neutral, straight-
forward account of the different theoretical perspectives and issues in debate, reserv-
ing until the conclusion a brief assessment and commentary.
My focus is on the modern, microeconomics-based theory of unions. Naturally,
this perspective emphasizes the economic function and effects of unions over the
noneconomic, particularly since the issue of efficiency occupies center stage in mod-
ern economics. An alternative, more historical and heterodox perspective is provided
by the author in another paper in this multi-part symposium.

II. The Monopoly Face of Unions

F&M claim that most economists view unions as akin to a labor market monopoly. The
first question that may be asked is whether this characterization is indeed accurate.
Among mainstream economists, and speaking at a broad level of generality, the
answer is probably "Yes." In the Theory of Wages, Hicks (1932, p. 137) states that
"monopolistic combination is common enough in all parts of the economic system;
very much the same motives which drive business men to form rings and cartels drive
their employees to form unions." A similar point of view is expressed by Friedman

JOURNAL OF LABOR RESEARCH

Volume XXV, Number 3 Summer 2004
352 JOURNAL OF LABOR RESEARCH

(1951, p. 206): "From this strictly economic point of view, labor unions and enter-
prise monopolies are conceptually similar if not identical phenomena and have simi-
lax effects." He then says (p. 239), "It seems to me like all other monopolies labor
monopolies are undesirable. This goes back to my belief in a competitive society."
More recently, Hirsch and Addison observe (1986, p. 21), "The monopolistic view of
unionism, firmly held by most economists, begins with the presumption that unions
raise wage rates above competitive levels in the union sector," leading to the predic-
tion that (p. 22), "Society suffers net welfare losses from unionism owing to the result-
ing inefficient factor mix and the misallocation of resources between the union and
nonunion sectors." In a similar vein Booth (1995, p. 51) remarks, "The standard view
of trade unions is that they are organizations whose purpose is to improve the mate-
rial welfare of members, principally by raising wages above the competitive wage level.
There is little dispute that unions are frequently able to push wages above the com-
petitive level - - what is called the 'monopoly' role of trade unions . . . . In a competi-
tive framework the trade union does introduce into the economy a variety of distortions
and inefficiencies."
When economists talk about the monopoly face of unions, they typically have a
standard, simplified model in mind as the baseline of comparison (Oswald, 1985; Far-
ber, 1986; Addison and Chilton, 1997), which I show in Figure 1, Panel A. Shown there
is a competitive labor market, with an equilibrium wage of W0 and employment of L 0
determined by the market labor demand and supply curves. Now introduce a union into
this industry and assume that it successfully organizes all the firms. The union gains
market power through its control of the supply of labor to the firms, either through
the strike threat or apprenticeship programs and union security provisions (e.g., a closed
shop), and uses this power to raise the wage above the competitive level. In this model,
it is assumed that the goal of the union is to maximize utility of the membership, and
utility is a function of the wage rate and level of employment. This assumption yields
union indifference curves, such as I I. Assuming the union has the bargaining power
to obtain its preferred outcome, the optimal wage is W 2 where the indifference curve
lj is tangent to the labor demand curve. The firm then chooses the employment level
L 29
The market power of the union provides its employed members a wage gain of
W 2 - W o. They presumably would report a higher level of satisfaction with their jobs.
Employment in the unionized industry in the short run falls to L 2, however, and indus-
try output contracts. In the long run, capital is variable, and the labor demand curve
becomes more elastic, leading to a further decline in employment. The costs of pro-
duction for unionized firms increase and, despite a rise in the market price of the prod-
uct and a possible long-run increase in productivity (as unionized firms substitute
capital for labor), the firms' profits shrink. If firms were just earning a competitive
return on capital before unionization, the higher costs and lower profits drive some
out of business. An incentive also opens up for nonunion firms to enter this industry,
since they can earn above-normal profit given the higher product price in the market
and their lower cost structure.
BRUCE E. KAUFMAN 353

Figure 1

The Union Effect on Wages and Employment

Wage
Ra[e
Wage
Rate
Wage
Rare

~
/ MCL S
s
w2 W2 .
L---
wl. ---~.~ F(W)
w1 _
~- U(W)
Wo %.
W4 w
~ . . . . . 0
.

iJi o
w3

Labor Time Labor

L2 L0 L3 T1 0 L4 L1

A B C

There are yet other implications. At the wage of W2, the market supply curve (S)
shows L 3 workers want union jobs, creating an excess supply of labor of L3-L 2, These
workers may queue for union jobs, remaining unemployed until a vacancy opens, or
migrate to the nonunion sector and search for jobs there. In the case of the latter, the
supply of labor in the nonunion sector increases, putting downward pressure on the
nonunion wage. If the nonunion wage is flexible, it declines to (say) W3, leading to an
expansion in nonunion employment and an expansion of output in that sector. Although
nonunion workers suffer reduced wages (W0-W3), nonunion firms earn higher prof-
its, temporarily at least. Assuming workers in the two industries are identical except
for union status, the introduction of unionism leads to greater income inequality among
production workers - - before unionization all were paid W0' after unionization one
group gets W2 while the other gets W3. Aggregate income inequality may decline, how-
ever, if owners of capital originally received much higher incomes than owners of labor
and the effect of unionization is to much reduce the former. [Wage inequality also
declines if unions standardize wages across firms. In the simple version of the com-
petitive model, however, all firms pay W0, making this a moot issue.]
Rent-Seeking and the Incidence of the Union Tax. These conclusions lead most
economists to take a dim view of unionism and collective bargaining, at least in its eco-
nomic function. Seen from this perspective, unions are rent-seeking institutions that
use monopoly power to transfer economic "surplus" (rents) from other economic agents
to union members. This point of view is well captured in Hicks' observation (1932, p.
137), "where it is possible for men to snatch gains, real or apparent, permanent or
temporary, from the abandonment of individual action, it would be surprising if they
did not sometimes attempt it." In Panel A, one measure of the amount of rent trans-
ferred (or "snatched") by the union is the difference in the wage bill before and after
unionization - - the difference between areas WoAL00 and WzBL20. It is in principle
354 JOURNAL OF LABOR RESEARCH

no different than the income transferred from oil-consuming to oil-producing coun-

ties by the OPEC cartel when it uses its market power to restrict oil production and
increase the price.
The union wage gain can be thought of as a union-imposed tax on employers,
nonunion workers, and consumers. The degree to which unions can transfer rents thus
raises issues quite similar to those addressed in the public finance literature on the inci-
dence of a tax (Bronfenbrenner, 1971; Bennett and DiLorenzo, 1984). In general, a
tax will raise more revenue the more inelastic (less price sensitive) is the demand for
the good being taxed and the less opportunity economic agents have to evade it. Many
aspects of union behavior are directly interpretable in this light.
One crucial ingredient for union success is an inelastic demand for labor. The
more inelastic is the labor demand curve, the smaller is the loss in employment for
any given union wage increase and the greater is the increase in the wage bill for union
labor. Indeed, if the labor demand curve were elastic a higher union wage would actu-
ally reduce the wage bill - - an unattractive result for union members.
Factors contributing to an inelastic labor demand curve are highlighted by the
Marshall-Hicks laws of derived demand (Rees, 1989). These laws state that labor
demand will be more inelastic: (1) the more inelastic is the product demand curve, (2)
the more difficult is substitution in production between union labor and capital or
no~union labor (i.e., the lower is the elasticity of substitution), (3) the smaller the share
of union labor cost to total cost (this condition is reversed if the elasticity of substitu-
tion is greater than the elasticity of the product demand curve), and (4) the less elas-
tic the supply of other competing factors of production.
It is usually argued that the likelihood of unionization and the potential gains from
organization increase as these laws individually and collectively are more fully met
(Ulman, 1955; Hirsch and Addison, 1986). These laws also influence who pays the
union tax. Consider condition 1, for example. The firm's demand for labor is derived
from the consumers' demand for the product and, hence, the less price sensitive is prod-
uct demand the less wage sensitive is labor demand. Thus, when substitution possi-
bilities for the final product are small, unionized firms are better able to pass to
consumers the cost of union wages through higher product prices, given the modest
decline in unit sales. One implication is that unions have greater ability to raise wages
when they organize all firms in the product market, since the market product demand
curve is more inelastic than the product demand curve for any one firm. A second is
that union wage gains are also potentially greater among imperfectly competitive firms
(e.g., oligopolies and monopolies) since their product demand curves are also more
inelastic than those of competitive firms. The same may also apply to public sector
organizations.
Condition 2 augments the ability of unions to raise wages and transfer rents by
making it more difficult for the firm to replace union workers with nonunion workers
in the short run and capital in the long run. The former can be accomplished by vari-
ous contract restrictions, such as union security provisions mandating a union (or
BRUCE E. KAUFMAN 355

closed) shop, and picketing and threats of violence that deter nonunion workers from
taking the jobs of strikers; the latter may be an inherent feature of the production tech-
nology or arise from union work rules that restrict the introduction of new technology
or mandate a minimum crew size or employment of a particular craft or occupation.
Condition 3 facilitates union wage gains because a rise in wages leads to only a small
increase in total production cost, thus having only a small negative effect on profits,
prices, and product demand. This condition ordinarily favors craft unions over indus-
trial unions, since the former typically represent only one trade or occupation in a plant
or company while the latter represent all production workers. Condition 4 also limits
the substitution of nonunion labor or capital for union workers, given that an increase
in demand for the substitute factor drives up its price and makes it no longer as eco-
nomically attractive to replace the union labor.
Unions' activities in the political process at the federal, state, and local levels are
also interpretable in this economic framework (Johnson, 1989; Delaney and
Schwochau, 1993). Union lobbying for import restrictions and defeat of free-trade leg-
islation, for example, are attempts to limit the substitution possibilities of consumers
in the product market and firms in the labor market, thus making the demand curve
for union labor more inelastic than otherwise. Likewise, restrictive immigration and
minimum wage laws reduce the competitive threat of lower cost nonunion labor. In the
public sector, unions use their political power to limit substitution possibilities in other
ways, such as leading opposition to privatization of public services.
Costs of Unionization. From a monopoly perspective, not only does union rent-
seeking appear to lack a compelling public interest rationale, it also imposes a variety
of economic costs. These costs were touched on earlier but deserve a more detailed
examination. Four major areas of cost are considered.
9 Allocative Inefficiency. A fundamental insight of microeconomic theory is that
competitively determined relative prices lead to an efficient allocation of resources
across firms and industries. Efficient resource allocation is a virtue because it is Pareto
optimal - - e.g., it is impossible to reallocate resources so as to make at least one eco-
nomic agent better off without simultaneously making another worse off. When union-
ism is introduced into an otherwise competitive economy, it distorts relative prices,
leads to a misallocation of resources, and thus creates inefficiency and a social wel-
fare loss.
For example, economy-wide allocative efficiency requires that units of an iden-
tical input be allocated in such a way that the marginal products are equalized across
firms and industries. Thus, if the cost of labor is the same in two industries but its
marginal product is higher in one, a net gain in output and social welfare could be
achieved by redistributing labor from the industry with a low marginal product to the
one with a high marginal product. The virtue of wage/employment determination in a
competitive market is that market forces bring about exactly this distribution of labor,
while a cost of collective bargaining is that it leads to "too little" employment in the
unionized sector and "too much" employment in the nonunion sector. The resulting
356 JOURNAL OF LABOR RESEARCH

welfare loss is shown in Panel A of Figure 1. [Assume the diagram represents the mar-
ginal product schedule for two separate but identical industries, one union and one
nonunion.] It has two components - - a deadweight loss in the unionized industry
denoted by the triangle ABC (the value of foregone output above its opportunity cost),
and a deadweight loss in the nonunion industry equal to the triangle AEF (the excess
of opportunity cost over the value of additional output). Empirical studies
provide varying estimates of the size of the union-induced welfare loss, generally in
the range of 0.1-0.3 percent of GDP but as high as 0.8 percent (Hirsch and Addison,
1986; Vedder and Galloway, 2002).
A second form of allocative inefficiency arises in the long run. Not only will the
higher union wage cause unionized firms to reduce employment and output but it will
also induce them to substitute capital for labor. Inefficiency caused by too little labor
in the unionized sector is thus exacerbated by too much capital (in relative terms) in
the long run. Perhaps, however, the amount of capital employed in the unionized indus-
try may decline in absolute terms if the scale effect (the decline in desired capital as
output falls in the unionized industry) dominates the substitution effect (the shift from
labor to capital in producing a given level of output). Whatever the case, allocative
inefficiency results.
9 Technical Inefficiency. An economy is technically efficient when the maximum
possible output is being produced from a given combination of inputs. Orthodox micro-
economics assumes that the forces of competition motivate employers to operate firms
with technical efficiency. Technical efficiency implies, in turn, cost minimization and
absence of organizational slack.
From a monopoly perspective, unions interfere with the efficient operation of
firms, thus contributing to technical inefficiency. Critics of unions cite a long list of
restrictive rules and practices that raise cost and retard productivity (Reynolds, 1984;
Northrup, 1997). Examples include featherbedding (make-work) requirements, narrow
job classifications that split up job tasks among a number of trades or occupations,
restrictions on management's ability to promote workers on the basis of merit or to dis-
cipline/terminate low-performing workers, and a "work to the contract" attitude toward
job performance. The motive of unions in bargaining for these things is partly to restrict
employers from shifting higher union wage costs back on to union workers through
work intensification ("speed ups") or replacement with lower cost nonunion workers;
another part is to gain new benefits for workers, such as improved job security. From
an economic perspective, however, these restrictive union practices are again a form
of rent-seeking that entail substantial costs in terms of inefficient use of scarce
resources. According to one knowledgeable observer (Rees, 1963), the social cost
unions create from technical inefficiency may well be two-to-three times as large as
those arising from allocative inefficiency.
9 Reduced Capital Investment and Higher Firm Deaths. It was noted above, in
discussing allocative inefficiency, that in response to a higher union wage firms may
increase or decrease the amount of capital used in production. The cost in this case
BRUCE E. KAUFMAN 357

comes from an inappropriate mix of labor and capital. But when additional consider-
ations are taken into account, theory suggests that it is quite likely that unionized firms
will decrease investment spending on long-lived capital projects and research and
development (R&D). The cost of unions in this case is not a static form of inefficiency
but a dynamic one and takes the form of reduced productivity growth and innovation.
It is also possible that unionism leads to earlier "deaths" (closures) of plants and firms
(Freeman and Kleiner, 1999).
The key point is that a higher union wage not only raises the relative price of labor,
but it also acts as a tax on capital (Hirsch, 1991). To the extent that firms cannot pass
on higher union wage costs to consumers, they must finance them from profits, thus
reducing the net return on the firm's invested capital. Faced with a lower return on
investment, union-ized firms are likely to reduce capital spending, or redirect capital
spending from unionized plants (or divisions) to nonunion plants - - possibly in another
country. The same reasoning applies to investment in R&D,
Since reduced capital spending and R&D means lower output and employment
in the future, unions have an incentive to pledge to firms in return for undertaking
new capital investment that they will restrain their wage demands and cooperate in
improving production efficiency. Firms fear, however, that once they invest in long-
lived capital a union will renege and use its bargaining power to extract not only prof-
its but also quasi-rents (revenue above average variable cost), knowing that in the short
run as long as the firm can at least cover variable costs it will continue in oper-ation
(Grout, 1984; Lawrence and Lawrence, 1985). Thus, rather than have future capital
investments held hostage or "milked" by a union, firms forego at least a portion of
investment spending, particularly on projects that are highly immobile or have very
long payback periods. Union milking of capital is also likely to lead to earlier shut-
down of organized firms and plants because they underinvest in maintenance and new
equipment or find in cyclical economic downturns that revenue no longer covers oper-
ating costs.
9 Inflationary Bias and Greater Cyclical Instability. The areas of union cost con-
sidered so far are microeconornic in nature; the next is macroeconomic. Two aspects
are identifiable.
The first is that the monopoly face of unions may impart an inflationary bias or
"price creep" to the economy (Mitchell, 1980; Hirsch and Addison, 1986). [The degree
of centralization in the bargaining structure is an important intervening variable, how-
ever (Naylor, 2003).] Unionization of an industry, such as assumed in Figure 1, leads
to a one-time rise in the price level, assuming the increase in wages (from W0 to W2)
is not offset by an increase in productivity. Other things equal, the union wage should
then remain stable at Wz and, for a given level of unionization, the price level is also
stable. But other things may not stay equal. One possibility is that the union demands
a higher wage in each subsequent bargaining round, thus moving firms gradually up
their labor demand curves and causing them to match higher wages with higher prices.
This result is not consistent with the simplest monopoly model of unions (per the dic-
358 JOURNAL OF LABOR RESEARCH

tum that "monopoly causes high prices but not rising prices"), but an alternative median
voter model of unions (discussed shortly) predicts this form of upward ratchet in prices.
A similar result can emerge from models of inter-union rivalry and wage imitation
(Ross, 1948; Budd, 1997).
A second possibility is the unions and other special interest groups lobby gov-
ernment officials to use expansionary monetary and fiscal policies to create jobs for
the L3-L2 people who are (potentially) unemployed at the union wage of W2. If these
policies are implemented, the rightward shift of the aggregate demand curve and
increase in national production and employment will likely lead to some increase in
the price level (moving along an upward sloping aggregate supply curve). Furthermore,
the expansion of employment in the unionized industry shifts out the labor demand
curve and, under reasonable assumptions, leads to a new and higher equilibrium union
wage (not shown in Panel A, Figure 1, but above W2).
The second form of macroeconomic cost is that unions may exacerbate the cycli-
cal instability of the macroeconomy or lengthen periods of boom and recession
(Mitchell, 1980; Neumark, 1993). Union wages are less flexible than market-deter-
mined wages because they are typically specified in multi-year contracts, During a
cyclical boom, nonunion wages rise under the pressure of increased labor demand, thus
helping maintain a balance between supply and demand. Union wage rates, however,
are more rigid and thus cannot rise as quickly. Not only does the rigidity of union wages
interfere with self-correcting processes in the economy (thus exacerbating instabil-
ity), policymakers can misread the stability in wages as a sign that there is still more
room for fiscal/monetary stimulation. The opposite chain of events happens in reces-
sions, potentially exacerbating downturns. Unions can also make it more difficult for
fiscal and monetary policy to end a boom or recession since multi-year contracts impart
considerable inertia into union wages and make them more impervious to aggregate
demand conditions.

III. Extensions and Qualifications

Although the monopoly model of unions sketched above provides the baseline frame-
work for analyzing the economic effects of unions, it is highly stylized and simpli-
fied. To provide a more complete picture, therefore, extensions and qualifications must
be introduced. Some of the most important are briefly described below; note, however,
that the voice/response function of unions is reserved for separate treatment. In some
cases these extensions and qualifications to the monopoly model strengthen the nega-
tive picture of unions, while in other cases they lead to a more positive assessment.
Endogenous Union Membership. The simplest version of the monopoly model
treats union formation and the level of union membership as exogenous variables.
Initially, a labor market has no union; then, without explaining why or how, the work-
ers are assumed to unionize and the economic effects are examined. A more realistic
model, however, makes union formation and union membership endogenous.
BRUCE E. KAUFMAN 359

One approach is to look at union membership (or a union card) as an asset that
workers purchase for price P per time period and then hold in order to gain the net ben-
efits that accrue from the services unions provide (Pencavel, 1971). The decision to
join a union and the level of union membership can then be modeled as the outcome
of a demand/supply process - - workers demand union services and unions supply the
services (Hirsch and Addison, 1986; Farber and Krueger, 1993). The price of union
services is the membership dues and fees workers pay to unions to finance their oper-
ations. Other things equal, the larger is workers' demand for unions or the greater the
supply of union services, the larger will be observed union membership.
Workers' demand for union services can be modeled as a function of a standard
set of variables: own price, price of related goods, net benefit of union services, income,
and tastes. Variation in own price (dues) causes movements along a union services
demand curve, changes in the other variables cause a shift in the curve. The latter are
most important in explaining the level and change over time in union membership.
Perhaps most significant in this regard is the net benefit that workers gain from
union membership. One such benefit is a higher union wage, and the usual assump-
tion is that the greater the union-nonunion wage differential the greater workers'
demand for union membership (Farber, 1983). This consideration also points out, how-
ever, that the relevant variable in the demand function is the n e t benefits of union mem-
bership, since higher wages also brings with it potential offsetting costs, such as threat
of job loss. Obviously, the more elastic the labor demand curve the greater the threat
of job loss and hence the less attractive is unionization.
Job loss, either threatened or real, can also arise from coercive actions undertaken
by employers to forestall union formation. In this respect the size of the union-nonunion
wage differential has a two-sided effect on the net benefits of union services and work-
ers' demand for unions (Freeman and Kleiner, 1990). On one hand, a larger wage dif-
ferential increases the benefits of union services and draws workers to unions; on the
other it causes employers to resist unions more strongly and undertake actions, such
as firing union supporters and closing plants, that reduce the net benefits of unioniza-
tion. The extent of legal protection and encouragement given to union organizing thus
has an important influence on the strength of union demand.
Other factors also influence net benefits. One consideration is demographic char-
acteristics of workers, such as race, gender, age, and education/skills. Workers who
suffer discrimination or low pay in labor markets, for example, may well gain more
from a union and potentially have a larger demand; while the wage-leveling policies
of unions may be unattractive to workers in the top tier of a firm's wage distribution,
such as those with advanced degrees or unique talents.
The next important variable in the union demand function is the price (or avail-
ability) of related goods. Most important in this regard are substitute goods for unions.
Three such unions substitutes are "good markets," "good management," and "good
laws" (Bennett and Kaufman, 2002). One reason the monopoly model leads to a rel-
atively negative view of unions is that most often it is built on the assumption that labor
360 JOURNAL OF LABOR RESEARCH

markets are competitive. This assumption is double-edged: it suggests, first, that the
terms and conditions of employment are already well-adjusted (or "optimal" in an effi-
ciency sense) in light of relative prices and factor supplies/productivities and, second,
that workers who have a legitimate dissatisfaction or grievance have a readily avail-
able and low (zero) cost means to solve the problem - - quit and find a job elsewhere.
Competitive markets are thus a good substitute for union services, except in the area
of rent-seeking. Where labor markets are less than perfectly competitive, well-adjusted
terms and conditions of employment, and corrections to any departures therefrom,
can also potentially be obtained from two other union substitutes - - professional man-
agement and a well-run corporate human resource function and various laws and reg-
ulations, such as protective labor legislation, social insurance programs, and income
redistribution programs that favor workers. Thus, the more readily available and lower
cost are union substitutes the smaller the demand for unions, other things equal.
Two other demand-shift variables are income and tastes. Whether union mem-
bership is a normal or inferior good is not well established either theoretically or empir-
ically, so not much can be said. The taste variable reflects individual preferences for
union membership, as well as the effect of broader social and cultural factors. Demand
for unions, for example, would most likely be stronger in a community with a strong
working-class culture but weaker where people have a strong ethos of individualism.
Variations in demand curves for union services thus explains, in part, the extent
of trade unionism. The supply side of union services also deserves brief mention, how-
ever, for demand by itself will not yield collective bargaining - - unions have to be
willing and able to provide this service at a reasonable cost. Thus, the supply curve of
union services will lie further to the right, other things equal, the larger are bargain-
ing units (providing economies of scale in servicing members), the greater the pro-
portion of dues-paying members in a bargaining unit (e.g., such as union or closed
shop, thus eliminating free riders), and the greater the legal protection given to pick-
eting, strikes, and other forms of union pressure (Farber, 1984; Taras and Ponak, 2002).
In most markets economists presume that price adjusts to equilibrate demand
and supply. In the market for union services, however, it is likely that disequilibrium
prevails. The price of union services - - monthly d u e s - - is not flexible and thus can-
not clear the market. Equally important, union services are in part a collective good.
That is, an individual worker may have a demand for unionism but cannot gain union
representation unless a majority of workmates also have a demand for unions. Fur-
thermore, in countries with a Wagner Act model of union recognition, the majority of
workers must be willing to express this desire through a formal organizing campaign
and election, running the risk of employer retaliation. Many workers, therefore, may
have a frustrated demand for unionism (Farber and Krueger, 1993; Freeman and
Rodgers, 1999).
The Union's Objective Function. A second aspect of the monopoly model that
requires extension and qualification concerns the specification of the union's objective
function. This issue is relevant because the objective function specifies the union's
BRUCE E. KAUFMAN 361

goals in collective bargaining - - a critical assumption in any effort to predict what

unions do.
Analytically, profit maximization is a relatively straightforward objective for a
firm. This assumption follows from two facts: first, making a profit is the unambigu-
ous goal of firms in a capitalist economy and, second, everyone who owns part of a
firm - - whether a sole proprietor, a partner, or individual shareholder - - gains when
profit increases, thus making each person's goal consistent with the organization's goal.
A union's objective is far more difficult to specify (Pencavel, 1991; Kaufman,
2002). A union does not own the labor it sells but, rather, sets a common price that indi-
vidual workers and the firm agree to honor. In this respect a union is more like a car-
tel of independent producers who band together to achieve a monopoly-like outcome.
Furthermore, while firms are interested in profit, workers are interested in the full range
of terms and conditions of employment, including not only wages but also benefits,
job security, working conditions, and style of management. Collapsing all of these
disparate items into a meaningful dollars and cents maximand is very difficult, so the
only recourse is to either focus on one or two variables at the expense of the others or
assume the union's goal is the general (and unobservable) one of maximizing "util-
ity," where utility is a function of all the terms and conditions of employment. Finally,
unlike a firm's shareholders, individual union members do not usually share equally
in the "profits" (rents) gained from collective bargaining - - as the union raises the
wage some members gain more income but others lose their jobs and are worse o f f - -
and thus it is generally impossible to derive an aggregate union objective function from
the diverse and partially conflicting objective functions of the individual members.
No ideal solution has yet been found to these problems. In practice, the theoret-
ical literature goes in two directions (Oswald, 1985; Flanagan, 1993). One is to use
the general utility function approach, such as depicted in Panel A and finesse the col-
lective choice (aggregation) problem through simplifying assumptions (e.g., that lay-
off is by random draw). The second is to develop an explicit collective-choice
mechanism that explains how the diverse preferences of the individual union members
are aggregated and reconciled to yield a well-defined bargaining goal for the union.
Insider-outsider, property rights, and median voter models are among this genre, with
the latter most widely used (Martin, 1980; Bennett and Johnson, 1980; Hirsch and Addi-
son, 1986; Kaufman, 2002).
The median voter model assumes that the union's bargaining goals are chosen
by the union leader(s). To keep the model tractable, it is often assumed that the only
variable on the bargaining table is wages. Once the wage rate is chosen, the firm uses
the labor demand curve to determine the level of employment. The union leader's
goal is to gain re-election, and to do this he or she must win in negotiations a wage
rate favored by a majority of the members. The smallest majority possible is 50 per-
cent plus one of the membership. To make the outcome interesting, it is further assumed
that the union members differ in their probability of layoff, say due to seniority pro-
visions (last in, first out). The result gives rise to a well-ordered distribution of union
362 JOURNAL OF LABOR RESEARCH

members ranked by their most preferred wage rate - - the least senior member prefers
the competitive wage since at any higher wage he or she loses the job, the median (mid-
dle) member in the seniority queue prefers the wage be raised just up to the point on
the demand curve where he or she is threatened with layoff, and the most senior mem-
ber prefers a yet higher wage on the demand curve until he or she is also threatened
with layoff. The most preferred wage demands of the other union members can be
determined in similar fashion. Given this, it can be shown that the union leader will
adopt the median member's preferred wage as the union's bargaining goal. The rea-
son is that any lower wage will be opposed by the median member and everyone of
higher seniority, thus leading to the union leader's electoral defeat, while the same hap-
pens (but for opposite reasons) if the leader chooses a higher wage.
The median voter model has several important implications for understanding
union behavior. First, it introduces into the monopoly model the important piece of
realism that unions are, at least within broad parameters, democratic organizations.
Whether "democratic" rent-seeking is any less objectionable on economic efficiency
grounds is uncertain, but from a social/ethical perspective the democratic nature of
unions undoubtedly frees them from some of the negative stigma imparted by the
monopoly/cartel image.
A second implication is that unions not only redistribute income from consumers,
firms, and nonunion workers to union members but also redistribute income among
the union members themselves (Kaufman and Martinez-Vasquez, 1988; Devinatz,
2002). As the union raises wages and bargains for various benefits and work rules,
some union members gain more than others; indeed, some members may be worse
off, e.g., if employment declines and they lose their jobs. Union bargaining goals, there-
fore, always in part reflect internal political dynamics within the organization. "Make
work" featherbedding rules, for example, are not likely to be a major bargaining
demand of a union with expanding membership, while they may well be a major con-
cern for a union whose members face substantial displacement by new technology.
Likewise, a union is likely to reject a company's request for a wage cut unless the threat
to jobs is so large that a majority of the members view a "Yes" vote as in their self-
interest.
An interesting application of this insight concerns the effect of unions on firms'
structure of wages and benefits. Widely noted effects of unionism are wage standard-
ization and wage-leveling (Card, 2001). They can partly be rationalized as unions'
attempt to eliminate competition based on labor cost differentials among workers and
firms, or as an attempt to promote worker solidarity and internal equity. But Freeman
(1980a) has argued that these policies are also likely to arise for median-voter rea-
sons. A majority coalition in a union always has an incentive to use collective bar-
gaining to restructure wages in their favor, in effect taxing the minority. He suggests
that most likely the majority coalition will form at the lower end of the wage distri-
bution since a larger amount of income can be redistributed. The result will be a more
compressed wage structure (entry-level rates increased and maximum pay rates
reduced) and a flatter age-wage profile.
BRUCE E. KAUFMAN 363

Freeman (1981) also points out another predicted effect of unionism is to increase
the share of benefits in total labor compensation and tilt the benefits package toward
pensions and health insurance and away from bonuses. Nonunion firms structure their
compensation package to attract/retain the "marginal" worker who is more likely to
be younger and unmarried or married without children. If these firms are unionized,
however, the union will bargain for a compensation package that reflects the prefer-
ence of the median member - - a person likely to be older and with a spouse and chil-
dren. Freeman reasons that the marginal worker will want most compensation in the
form of wages, while the median worker will have a greater demand for benefits. Like-
wise, the median worker will have a greater demand for certain benefits, such as pen-
sions and health insurance, reflecting his or her greater age.
A third implication of the median voter (not considered by F&M) concerns the
dynamic path of the union-nonunion wage differential. Assuming for simplicity that
the union can dictate the wage outcome, in the first bargaining around the union raises
the wage from the competitive level to the median voter's preferred wage (where he
or she is next to be laid-off), forcing the firm up its labor demand curve and cutting
employment by one-half. In the next contract period, there is a new median voter and
the wage is again forced up, moving the firm further up the demand curve and again
eliminating the bottom half of the seniority distribution. This upward ratchet effect
(or quest for "more") continues until the union bargains itself out of existence (Burda,
1990; Kaufman, 2002).
This scenario is naive, partly because the median voter is modeled as a myopic
one-period maximizer and partly because the union is assumed to have sufficient bar-
gaining power to attain the median voter's preferred wage in each negotiation. Build-
ing in more realistic assumptions, however, does not alter the fundamental conclusion
- other things equal, a democratic union with seniority-based layoffs will gradually
-

nibble its way up a firm's labor demand curve, leading to declining employment, a
widening union-nonunion wage gap, and a bias toward cost-push wage inflation. These
predictions led to an early debate (e.g., Shister, 1943) whether it is possible to recon-
cile union democracy and "responsible" union wage policy (a question still unresolved).
Likewise, this model suggests one reason why unions are strong supporters of full-
employment macroeconomic policies; such policies shift the labor demand curve to
the right and remove the threat of unemployment, thus allowing the upward union wage
ratchet to continue.
A final implication of the median voter model concerns the union effect on cap-
ital investment and R&D expenditures. Hirsch (1991) argues, for example, that the
decision-making horizon of most union members - - particularly the senior members
who may control the union's wage policy - - often may be shorter than the economic
life of the firm's long-lived, specialized capital equipment. This fact, combined with
the inability of union members to sell or bequeath their property rights in union jobs
to others (Martin, 1980), causes unions to act "rationally myopic" and raise wages to
the point where firms do not earn a competitive rate of return on capital. The result is
a slow process of dis-investment and industrial and union decline.
364 JOURNAL OF LABOR RESEARCH

Union-Firm Bargaining. The next extension to introduce is the bargaining process

between the union and firm. To this point it has been assumed that the union has suf-
ficient power to dictate a "take it or leave it" wage outcome to the firm. In reality, wages
and the other terms and conditions of employment in unionized firms are determined
through a sequential-step bargaining process with the threat and occasional use of a
strike.
The bargaining process is depicted in Panel B of Figure 1. This model originates
with Hicks (1932) and has subsequently been extended and developed in a number of
directions (Cross, 1969; Kaufman, 1992; Gallagher and Gramm, 1997). The vertical
axis measures the wage rate, the horizontal axis measures bargaining time (e.g., days).
Total time available to reach an agreement without a strike is Tl; disagreement past
this point necessarily entails a strike of some duration. Without strike costs and other
costs of disagreement, the firm and union have no motivation to compromise and reach
an agreement. Thus, the firm's desired wage rate is the competitive wage W0 in Panel
A, while the union's preferred wage is W2. When costs of disagreement are introduced,
however, both sides have to calculate the benefits and costs of intransigence. In gen-
eral, holding out for W0 and W2 now carries potentially large costs for the firm and
union in the form of a strike and other sanctions, so immediately in the first round of
bargaining they are induced to "put some money on the table" in order to move toward
agreement. Thus, in Panel B the firm's initial wage offer is modestly higher than W0
and the union's demand is modestly lower than W2.
The two lines that originate from the vertical axis are the union's "resistance
curve" U(W) and the firm's "concession curve" F(W) which depict the wage rates (pre-
sumed to be the only subject of bargaining) that each side offers/demands in each round
of the bargaining process. The two lines gradually converge as each bargainer makes
compromises - - motivated in large part by the looming strike deadline. As shown in
Panel B, the two bargainers reach an agreement at the strike deadline T 1. The wage rate
is Wj, determined by the intersection point of the resistance and concession curves. In
a world of perfect foresight, this bargaining process could be dispensed with and the
two sides could immediately move to Wt (saving themselves the costs of bargaining).
Bargaining models based on Nash-type game theory essentially make this theoretical
leap (Addison and Chilton, 1997; Manzini, 1998). When the bargainers have bounded
rationality and imperfect/asymmetric information, however, they have to grope their
way to an equilibrium, as pictured in Panel B. In some cases the parties start out with
differences in their positions that are too large for them to compromise in the allotted
bargaining time; in other cases one or both sides are inexperienced, maladroit, or oppor-
tunistic bargainers, and the compromise process breaks down; and yet in other cases
strong emotions or commitment to principle make one party unwilling to consider com-
promise. Whatever the cause, a strike occurs and lasts until an agreement is reached
or one of the parties exits the relationship.
Introducing bargaining into the monopoly model provides several new insights.
First, it reveals that union wage policy is shaped by two fundamental constraints - -
BRUCE E. KAUFMAN 365

the labor demand curve and strike costs (real and expected), rather than just the for-
mer as in the simplest monopoly model.
Second, it is evident that the union wage gain with bargaining is smaller than
predicted by the simple monopoly model. In Panel A, the wage gain is Wz-W0; in Panel
B it is WI-Wo. Bargaining thus serves to moderate the size of the union monopoly
wage gain, keeping the misallocation of resources and rent transfer smaller than oth-
erwise predicted. Moreover, when the right to strike is limited or banned, as in many
public sector situations, the union wage gain is smaller yet. Bargaining's positive con-
tribution to efficiency is counterbalanced, however, by an additional form of resource
cost and inefficiency. The restraint bargaining provides on union rent-seeking is not a
free good. To "produce" collective bargaining requires scarce time and factor inputs
of union, firm, and government negotiators and mediators. More importantly, when
coercive weapons are used, such as strikes, boycotts, and slowdowns, production is
lost and resources remain idle. Although strikes in recent years have occurred in only
a small proportion of union negotiations and, in general, resulted in very modest social
cost, in the 1940s-1960s when union density and power was much higher so too were
the incidence and cost of strikes (Gunderson and Melino, 1987; Kaufman, 1992).
Finally, an intangible but real cost of collective bargaining is that it heightens or gives
outlet to an adversarial, sometimes bitter "we against them" relationship between the
workers and the firm.
Third, consideration of the bargaining process reveals yet another reason why
the efficiency loss from unionism may be overstated by the monopoly model. A large
literature on "efficient contracts" has developed showing that a union and firm have
an incentive to replace the traditional noncooperative "on the demand curve" bargain-
ing strategy with a cooperative "off the demand curve" approach, given that the latter
yields wage/employment outcomes that are potentially more desirable for both sides
(Booth, 1995; Addison and Chilton, 1997). Without going into details, the coopera-
tive outcome entails moving from the monopoly (noncooperative) outcome of W2/L2
in Panel A to a point off the labor demand curve, such as depicted by Point G, where
the wage is lower but the level of employment is higher. Point G is closer to the orig-
inal competitive outcome of point A and thus less efficiency-distorting. Indeed, under
certain conditions, Point G may lie directly above Point A (a vertical contract curve in
the efficient bargain model), and the level of employment and output in the union and
nonunion sectors remains the same as in the competitive outcome (Abowd, 1989). Since
the wage rate at Point G is higher, however, unionism still redistributes rents to the
workers.
Note also that the efficient bargain model leads to a more sanguine view of union
work rules. An efficient contract moves the firm off its labor demand curve. As this
literature demonstrates, once the contract is signed the employer has an incentive to
renege on the negotiated employment level and reduce jobs, since by moving left-
ward from Point G back to the labor demand curve it can increase profits. To prevent
(albeit imperfectly) such employer opportunism, a union can negotiate work rules, such
as minimum crew sizes (Johnson, 1990). In this interpretation, union work rules are a
366 JOURNAL OF LABOR RESEARCH

contract-protection device that helps police a win-win agreement, rather than a pro-
ductivity-sapping interference in production. Given this optimistic conclusion, a cau-
tionary note is required. Despite superficial appeal, economists have pointed out that
the efficient contract model paradigm has serious conceptual and practical difficulties
that raise significant questions about its validity and relevance. Conceptually, for exam-
ple, it is not apparent why the senior majority of union members would vote for a wage
cut when the increase in jobs goes to unemployed low-seniority workers in the layoff
pool (Kaufman, 2002). Empirically, researchers have found little evidence from col-
lective bargaining negotiations and contracts that unions have successfully obtained
employment commitments from firms that represent off-the-demand curve settlements
(Oswald, 1993).
Labor and Product Market Structure. The final extension is to introduce non-
competitive product and labor market structures into the model. Both are important,
the former because it provides a long-lasting source of rents for unions to transfer and
the latter because union power gains social and economic legitimacy to protect the
underdog and offset employer exploitation of labor (Chaison and Bigelow, 2001).
Consider first imperfect competition in product markets. If product markets are
perfectly competitive, the union's ability to raise wages and other labor conditions
above market levels is very limited. Ease of entry and exit into the industry means
that firms in the long run only earn a competitive return on capital, leaving zero sur-
plus lbr unions to capture. If unions do push up labor costs in organized firms, in the
end the result is self-defeating since these firms go out of business - - both because
they do not earn a competitive return on capital and because lower cost nonunion rivals
underprice them and capture their customers.
This conclusion is modified only modestly by further qualifications. First, even
in a competitive product market unions can raise wages and other labor standards to
the extent they also contribute to greater productivity, since the latter offsets the for-
mer and leaves the firm's cost structure the same as nonunion rivals. Second, the "long
run" may be ten or 20 years in the future when capital is very bulky and long-lived,
so even if unions can only raise labor costs above competitive levels in the short run
(by capturing quasi-rents or temporary excess profits) this may be attractive to work-
ers. Third, if some firms have lower cost structures than others, say due to non-repro-
ducible advantages such as superior management or ownership of a strategic natural
resource, they are able to earn rents (above-normal profits) even in the long run which
are available for union capture. And, finally, if a union can organize the entire prod-
uct market, including all new entrants, it can then take labor cost out of competition
and push up wage rates, moving up an inelastic industry labor demand curve with
only modest employment loss. Parenthetically, for this reason employers in very com-
petitive industries have in some cases in the past welcomed industry-wide unioniza-
tion as a way to stabilize prices during depressions or as a cartelizing device to raise
prices and joint surplus (Vittoz, 1987; Maloney et al., 1979).
When product markets are imperfectly competitive, prospects for successful union
rent-seeking increase significantly, particularly among regulated firms and large oli-
BRUCE E. KAUFMAN 367

gopolists with above-normal profits. [A union first has to organize these firms, how-
ever, which is frequently a substantial challenge.] In imperfect competition, individ-
ual firms have market power and can set the price on their product demand curves to
maximize profit. Above-normal profit is not guaranteed for an imperfectly competi-
tive firm, but it is more likely in both the short and long run due to factors such as dif-
ferentiated products, one or only a few sellers, barriers to entry, collusive price
coordination, and government regulatory barriers. Imperfect competition thus helps
create additional rents for unions to capture, leading to the prediction (other things
equal) that the union wage effect should be greater in this sector (Rees, 1989). This
prediction is also supported by the fact (already noted) that the labor demand curve of
an imperfectly competitive firm is more inelastic than for a competitive firm, thus
diminishing the union's worry of employment loss.
Threat of capital dis-investment and plant shutdown is also reduced to the degree
the union transfers the monopoly portion of the finn's rents while leaving untouched
the normal rate of profit. To the extent that imperfect competition allows firms to earn
large (monopoly) profits, union rent transfer appears less objectionable (or even
benign/meritorious) since it redistributes to labor some of the excessive gains or
"cream" going to capital. Indeed, redistribution of monopoly rents to labor may increase
efficiency in at least two ways. Rent sharing, for example, can increase (or maintain)
employee morale and productivity, given that workers feel on equity grounds they
deserve a "fair share" of their employer's profits (the "ability to pay" argument) and
will reduce work effort and productivity if it is not forthcoming (Blanchflower et al.,
1986; Levine, 1993). A second possibility is that rent-sharing, by transferring income
from capital owners with a lower propensity to spend to workers with a higher propen-
sity, strengthens aggregate demand in the macro economy and maintenance of full
employment (Argyrous et al., 2003).
Next consider labor market structure. As in product markets, imperfect competi-
tion in labor markets gives companies a measure of control over price (and other terms
and conditions of employment) which they can use to increase profit. Imperfect com-
petition in labor markets arises from a variety of factors, many of which in generic
form are similar to product markets: differentiated workers and jobs, one or only a
few buyers of labor, barriers to worker mobility, and imperfect/asymmetric informa-
tion. The key difference is that while companies use market power in product markets
to raise price above the competitive level, they use market power in labor markets to
lower price (the wage) below the competitive level. In the former case, consumers are
"exploited" whereas in the latter case companies exploit workers, where "exploitation"
in labor markets is conventionally defined to mean a wage below labor's contribution
to production (marginal revenue product). The classic rationale for labor unions, in
turn, is to use collective bargaining to protect the worker underdog and eliminate
exploitation of labor by raising wages.
These ideas are illustrated in Figure l. The exemplar of imperfect competition in
product markets is monopoly and the analog in labor markets is monopsony (one buyer
of labor). Panel C illustrates a monopsonistic labor market. The firm faces an upward
368 JOURNAL OF LABOR RESEARCH

sloping supply curve of labor (S), rather than a horizontal labor supply curve as for an
individual competitive firm, because the monopsonist is the only labor buyer in the
local market and, hence, can lower the wage and not lose all of its labor supply to com-
petitors. Conversely, to acquire more labor it has to offer a higher wage, which by
assumption it also has to pay to both new and old hires. These considerations give
rise to a marginal cost of labor schedule (MCL) showing the increase in labor cost for
every new employee. The equilibrium, profit-maximizing employment level is L 4
(where marginal revenue product, given by the MRP schedule, equals MCL) and the
wage is W4 (given by the supply curve at L4). Instead of the competitive wage of Wo,
workers receive the lower wage W4. The amount of monopsonistic exploitation is the
difference between point H and J.
The monopsony model, strictly defined, applies to a single buyer of labor. The
classic form is a "one company town" where the only major employer is (say) a coal
mine or textile mill; another close equivalent is a community with only one buyer of
a specialized kind of labor, such as firemen. Monopsony-like conditions can arise in
any employment situation, however, where the firm faces an upward sloping supply
curve of labor. One example is oligopsony where there are only a few buyers of labor,
such as the market for nurses in a community with only one hospital and a small num-
ber of doctors. Another is inframarginal workers who have some element of immo-
bility or "job lock" due to seniority, firm-specific employee benefits (e.g., health
insurance), or employer-specific training. Monopsony-like conditions may also arise
when employers face a rising marginal cost of recruiting new employees due to worker
heterogeneity and imperfect job market information (Card and Krueger, 1995; Man-
ning, 2003). In each case the firm can pay less than competitive wages (or benefits or
other conditions) and not lose most of its work force to competitors.
Several important implications regarding unionism flow from this model. First,
it is evident why in a monopsony situation workers have a demand for union repre-
sentation. The firm has market power and provides below-competitive wages and other
conditions of employment. Collective bargaining in this context is thus a form of pro-
tection and countervailing power - - it gives workers as a group additional bargaining
power to confront the employer on a "level playing field" and raise wages closer to
what would prevail in a truly free market. This idea is illustrated in Figure 1, where
the wage rate W 1 determined by the bargaining process in Panel B is closer to the
competitive wage W0 in Panel C than is the monopsony wage of W4. Thus, while col-
lective bargaining is "monopoly-creating" in Panel A (a competitive labor market), it
is "monopsony-reducing" in Panel C (Reynolds and Taft, 1956).
A second implication is that by creating a level playing field in a monopsony labor
market collective bargaining can (but not necessarily will) also expand employment
and improve economic efficiency - - just the opposite results from the case of a com-
petitive labor market. As a nonunion firm, the monopsonist hires L 4 workers at the
wage of W4; as a unionized firm it is forced to pay the higher wage of W~ and, equat-
ing marginal revenue product and marginal cost of labor (no longer the MCL sched-
ule but a horizontal line at W l to the supply curve S, after which it jumps to MCL),
BRUCE E. KAUFMAN 369

hires Lj employees - - an increase of L1-L 4. This expansion of employment, in turn,

contributes to a more efficient allocation of resources - - output is increased closer to
what would prevail in a competitive situation and the marginal value of extra labor
(given by the MRP schedule) is brought closer to the opportunity cost of labor (given
by the labor supply curve S). [Observe, however, that all of these conclusions depend
on the assumption that the union's bargaining power is not so great, or repeatedly
applied, that it raises the wage above Point H.]
A third noteworthy implication is that collective bargaining in a monopsony labor
market may still drive plants/firms out of business in the short or long run. Just because
a monopsonist pays a below-competitive wage does not mean it always makes an
above-normal profit, or any profit at all (e.g., an old textile mill in a one-company town
that is barely covering operating costs). Collective bargaining, by raising labor costs,
may thus hasten the exit of capital. But, broadly viewed, this result is also beneficial
for allocative efficiency since the resources can flow to higher valued uses.
Finally, the monopsony case also reveals what appears to be a compelling case
for public protection and encouragement of trade unionism and collective bargaining.
With collective bargaining three "goods" are attained: exploitation of labor is dimin-
ished or eliminated, economic efficiency is improved, and workers are given an in-
dependent voice and democratic rights in a workplace and community otherwise
dominated by the employer.
Given this positive assessment, why don't the majority of labor economists sup-
port unionism and collective bargaining.'? The answer: most do not think monopsony
is prevalent or even a serious problem in labor markets (Reynolds, 1984; Boal and Ran-
som, 1997; Addison and Hirsch, 1997). Most economists believe labor markets have
become more competitive over time with increased geographic mobility, improved job
market information, and a larger number of employers in local areas. The "classic"
one-company-town form of monopsony is therefore widely seen as an historical curio-
sum, while other forms of quasi-monospony (oligopsony, monopsonistic competition)
or "dynamic" monopsony (arising from flows of employee quits and new hires) are
variously regarded as spotty in occurrence, of modest significance for wage determi-
nation, or of a short run, transitory nature (Manning, 2003). Another view is that
employers who have a degree of monopsony power nonetheless often refrain from
using it for exploitative purposes lest it damage morale and productivity or firm rep-
utation. Yet another is that sources of labor immobility, such as specific on-the-job
training, not only lock workers into jobs but also lock employers into workers (because
firms invest in training workers and lose the investment if workers quit) and thus cre-
ate a more balanced situation of bilateral monopoly. And, finally, the theoretical case
that unions are needed to offset deleterious labor market imperfections has faded badly
among economists in the face of new theoretical developments that argue these imper-
fections are either an economizing response to scarcity (e.g., job search as an adapta-
tion to imperfect information) or do not pose a serious obstruction to competitive forces
(e.g., the theory of contestable markets).
370 JOURNAL OF LABOR RESEARCH

Whatever the case, the economics literature seldom emphasizes or even men-
tions monopsony in labor markets. Perhaps the most telling evidence is that even Free-
man and Medoff, who desire to build a positive case for unionism, omit altogether an
appeal to monopsony conditions and accept from the start that unionism (on net) dis-
torts a competitive wage structure (also see Naylor, 2003). [As described below, F&M
motivate the voice/response face of unions with appeal to various forms of market
imperfection, but these (apparently) are not a source of monopsony power for employ-
ers.] The contrast in this matter is stark relative to the view of unionists and an earlier
generation of economists (see the other Kauffman paper in this symposium), The AFL-
CIO's web page (<www.aflcio.org>), for example, declares in the first sentence of
the section "How and Why People Join Unions" that "People who work for a living
know about the inequality of power between employers and employees," and shortly
thereafter repeats this theme, stating "Workers in unions counterbalance the unchecked
power of employers."
In ending this discussion, it is important to note that employers may gain a meas-
ure of control over wages and labor conditions for other reasons besides monopsony.
As an example, economists have recently developed an "efficiency wage" model of
unemployment (Akerlof and Yellen, 1986). The basic idea - - to some degree an
updated version of Marx's "reserve army of the unemployed" - - is that firm's delib-
erately pay above market-clearing wages so that a positive amount of involuntary unem-
ployment exists in labor markets, giving employers an additional method to motivate
employee work effort (lest they be fired). The mainstream literature, however, gener-
ally regards this form of employer control from a relatively positive/benign perspec-
tive as a legitimate method employers use to curb employee shirking (caused by
principal-agent problems) and, certainly, few contemporary labor economists - - includ-
ing Freeman and Medoff and the others just cited - - advocate unions as a means to
curb it.

IV. The Collective Voice~Institutional Response Face

As Freeman and Medoff suggest, the monopoly model of unions - - when combined
with a theory of competitive labor markets - - leads to a largely negative ver-dict on
the effects of labor unions. In years past, a traditional response to this verdict was to
challenge the competitive markets assumption and argue, per the monopsony discus-
sion above, that union power is needed to offset employer power in imperfect labor
markets. For reasons related to intellectual and real-world trends, this justification for
unions is waning in both generality and influence, certainly in the academic commu-
nity but probably also among the public. As this rationale for unions has declined in
influence, another has arisen. This new view of unions shifts attention from imper-
fections in the structure of markets to imperfections in employment contracts and inter-
nal firm governance (Williamson et al., 1975; Barker, 1997). Although there are several
diverse strands of theory, some of which are direct extensions of traditional micro-
economics and others that come from organizational economics and "new institutional
BRUCE E. KAUFMAN 371

economics," all have a common root in transaction costs and incomplete contracts. For
expositional simplicity, I refer to this new line of theory as "contract and transaction
cost economics" (CTCE), although in reality it comes under different labels. The voice
face of unions developed by F&M in What Do Unions Do? is interpretable as an early
application of CTCE.
Transaction Cost and Incomplete Contracts. One reason neoclassical economics
leads to a negative verdict on unions is because the core theory starts with a theory of
perfectly rational decision makers and perfectly competitive markets, so even when
these parts are amended and qualified it is difficult to avoid the conclusion that a labor
union is (largely) an interference in an otherwise well-functioning market system and
thus a source of inefficiency and welfare loss.
A CTCE model of unions reaches different and more positive implications because
it replaces these two fundamental neoclassical assumptions. Instead of modeling the
economic agent as perfectly rational (the "rational actor" model), CTCE theorists use
a model of bounded rationality. Formally developed by Simon (1982), bounded ration-
ality recognizes that human decision making is impacted by what Commons' (1934)
called "stupidity, ignorance and passion" and what in today's literature are called lim-
ited cognition, imperfect information, and emotional affect (Kaufman, 1999). In this
model, human behavior and decision making are purposeful and largely reasoned ("pro-
cedurally rational") but also prone in a substantively important way to suboptimiza-
tion, inertness or "slack," systematic biases and errors, occasional resort to violence
and aggression, and the influence of interpersonal emotional states arising from feel-
ings such as love, hate, anger, and jealousy.
The second amendment is to note that all institutional forms for organizing and
carrying out exchanges have imperfections, thus creating a larger potential for organ-
izations such as unions to have positive efficiency effects. The nature and impact of
the imperfections in CTCE are different, however, than in standard neoclassical eco-
nomics (Williamson, 1985; Dow, 1997). CTCE theorists note that in an interdepend-
ent economy people must engage in exchange for production and consumption to
proceed. What is exchanged, however, are not physical goods and services but prop-
erty rights to scarce things, including the services of workers. Every exchange, there-
fore, involves a transfer of property rights and this transfer is governed by an implicit
or explicit contract between the two parties. The central research issue in CTCE is to
discover how economic agents organize and transact the contracting process and the
efficiency implications thereof. With respect to labor, for example, various options
potentially exist; firms may use full-time employees, independent contractors, tempo-
rary workers or (law permitting) slaves, and will presumably choose the contractual
option that minimizes cost. The kind of cost that they seek to minimize, in turn, is trans-
action cost - - the real resources used to exchange property rights or, alternatively, to
construct, enforce, and administer contracts. [The concept of a transaction comes from
Commons (1934); the concept of transaction cost and its use for comparative institu-
tional analysis comes from Coase (1937) and Williamson (1985).]
372 JOURNAL OF LABOR RESEARCH

It is widely agreed (e.g., Reder, 1999) that neoclassical microeconomics omits

transaction cost, since its core assumptions allow buyers and sellers to construct com-
plete and fully contingent contracts at zero cost. Since exchange carries no cost, effi-
ciency is always maximized through market exchange - - to the point that, as Coase
first recognized, there is no rationale for multi-person firms and the economy dissolves
into a system of "perfect decentralization" where all production is done by sole pro-
prietorships (Demsetz, 199 I; Kaufman, 2003). No employment relationship exists and
the labor market disappears! With nonzero transaction costs, however, it may be more
efficient to shift from obtaining labor through spot transactions with other firms (oper-
ating in the guise of independent contractors) to hiring employees and entering into
long-term contracts. That is, positive transaction cost arises from factors such as
bounded rationality, imperfect information, asset specificity, incompletely specified
or nonpartitionable property rights, and imperfectly enforced contract law. Because
of nonzero transaction cost, every new contract thus entails additional expense, pro-
viding a motive for the firm to hire longer term employees over short-term independ-
ent contractors and to minimize churn through "hiring and firing." Thus, consistent
with the argument of F&M, in a positive-transaction-cost world the "exit" option of the
free market is not necessarily the most efficient way to re-contract for labor services.
Equally important, nonzero transaction cost means that all employment contracts
are incomplete - - i.e., the employer and employee cannot possibly fully specify in writ-
ing all terms and conditions of employment, such as the precise tasks and work effort
expected of the employee, the risk of injury or level of noise, and the management style
of supervisors. Thus, with incomplete contracts both the employer and employee are
exposed to a host of potentially costly contract breaches arising from opportunism,
malfeasance, adverse selection, moral hazard, and principal-agent problems (Miller,
1991; Dow, 1997). As examples, an employee may promise to be a diligent worker
but then loaf at every opportunity, while the employer may promise a promotion in
six months but later renege. Since positive transaction costs make the exit option expen-
sive, the employer and employee need an alternative method to prevent and resolve
these problems lest they cause hard feelings and conflict that sap cooperation and effi-
ciency. Following F&M, one solution is to instead construct a "voice" option in the
firm, such as a labor union, so employer and employees can communicate, talk out dif-
ferences, and construct a regime of enterprise rules and regulations that are fair and
"win-win" for all parties.
In the presence of bounded rationality and positive transaction cost, a labor union
can increase efficiency in a number of ways. The case is far from ironclad, however,
and a union may also decrease efficiency or be an inferior solution to transaction cost
problems relative to a different type of voice institution.
P u b l i c G o o d s in the Workplace. As earlier described, a central part of F&M's
justification for the voice function of unions rests on the idea that many workplace con-
ditions have characteristics of a public good. A public good presents a contracting prob-
lem and source of market failure because for technological or cost reasons establishing
a well-defined individual property right is impossible or prohibitively expensive. As
BRUCE E. KAUFMAN 373

a result, the good or service is "public": provision to one person means provision to
many or all (as with national defense). The problem for market exchange is that indi-
vidual economic agents have an incentive to understate their true demand for the good,
hoping others will pay the full cost and they can free ride and consume it at no (or lit-
tle) cost. Due to this type of transaction cost problem, the good will be under-pro-
vided relative to the socially optimal level. One solution is to have the good provided
by a collective organization, such as a government, where each person (voter) is
assessed a prorated share of the cost. Unable to free ride, the voters have an incentive
to declare their true demand, allowing government to provide a more efficient level
of the good than the private market.
Many workplace conditions may have a public goods quality. Increased safety,
improved ergonomics, reduced heat and noise, flexible work schedules, and an end to
management harassment or insensitivity are nonrivalrous in consumption since enjoy-
ment by one worker does not reduce the amount available for others. Individual work-
ers thus have an incentive to free ride, particularly since confronting the employer may
have significant costs. Although many nonunion firms say they have an open-door pol-
icy, and many managers express a desire for feedback from subordinates, workers worry
that speaking up will brand them as troublemakers or people who are not team play-
ers, resulting in a bad image with the boss, lower performance evaluations and pay
increases, and perhaps discriminatory treatment or even termination (Lewin, 2004). A
union can thus enhance efficiency because it replaces ineffective individual voice with
a stronger collective voice, leading to an increase in the supply of workplace public
goods closer to the social optimum (Flanagan, 1983).
Higher Productivity.Collective voice provided by a labor union can also promote
higher productivity, thus contributing to a second form of efficiency gain. This posi-
tive effect can arise through several channels (Kuhn, 1985).
One channel is by reducing turnover costs (Freeman, 1980b). The competitive
market model assumes worker mobility and the accompanying hiring and quitting are
costless. With bounded rationality and imperfect information, however, firms must
invest considerable resources in recruiting and screening prospective employees, while
workers invest similar resources in job search and interviewing. Similarly, once an
employment relationship commences, in many situations both the employer and
employee incur match-specific costs, such as investments in firm-specific job skills
and training, that are lost if the worker quits or is terminated. If these forms of trans-
action cost are sizable, using labor mobility to sort workers into good job matches
and provide signals to firms about workers' preferences may be inefficient. A labor
union, on the other hand, may promote greater productivity if it reduces employee churn
by substituting direct communication with the employer for indirect communication
through quitting. An important form of direct communication, for example, is a for-
mal grievance system where management and union representatives meet to resolve
work performance, discipline, and discharge disputes (Lewin and Boroff, 1996). With-
out such a system, the employee separation rate may be higher as workers either quit
or are terminated.
374 JOURNAL OF LABOR RESEARCH

Bounded rationality and the existence of organizational slack (sometimes called

"X-inefficiency") is a second route through which unions can improve productivity.
As long as the firm's profits are above some satisfactory level, managers may "loaf'
or pursue other goals at the expense of profits, such as plush offices or "empire build-
ing" through numerous mergers and acquisitions. Unionization or a union wage
increase, therefore, may reduce the level of organizational slack and force managers
to tighten operations, reduce unnecessary expenses, and focus more single-mindedly
on profit (sometimes called a "shock effect"). All of these actions will increase pro-
ductivity and yield a gain in efficiency (Kaufman, 1999; Altman, 2001).
The incomplete nature of employment contracts creates a third channel for unions
to stimulate higher productivity. The employment contract commits a worker to pro-
vide a certain number of hours of labor, but within some range the employee chooses
how much work effort (or "labor power") to provide. If employees provide the mini-
mum required amount of effort, productivity will be low; if they provide their maxi-
mum work effort, productivity will be high. One variable that strongly influences work
effort is whether employees feel fairly treated by the company, If the company gives
employees a "gift" of above-market wages, generous benefits, and employment secu-
rity, they will reciprocate with a gift of hard work, loyalty, and attention to customer
satisfaction; if on the other hand workers feel unfairly treated, they react by punish-
ing the company through greater absenteeism, loafing on the job, and a bad attitude
toward customers (Akerlof, 1990). A union may thus contribute to higher productiv-
ity by creating conditions that motivate greater employee work effort, either by elim-
inating practices and conditions that seem unfair or establishing terms and conditions
that are superior to other firms.
Another dimension of voice that unions may facilitate is employee involvement
in the production process. The traditional management model is "command and con-
trol" where managers do the "brain work" and give the orders and employees do the
"back work" and follow the orders. Companies using a new "high performance"
employment model have found, however, that productivity is higher when workers
are empowered and involved, giving them more responsibility, say, and influence in
how the work is organized and performed (Levine, 1995). Employee involvement may
not develop beyond a superficial level in nonunion firms, however, because managers
resist ceding control and fear that empowering workers will lead to a demand for a
sharing of the gains (rents). A union, however, can exert greater pressure on compa-
nies to expand the breadth and depth of employee involvement, thus promoting greater
productivity. Paradoxically, firms will nonetheless often prefer to forego the produc-
tivity gains of unionism because their profits are higher when they do not have to share
the gains with workers (Freeman and Lazear, 1995).
A final way in which union voice can promote higher productivity is by creating
"credible commitments" between companies and workers. Companies try to motivate
employees to work harder and smarter with an overt or implicit promise that employ-
ees will share in the greater profits in the form of higher wages, better benefits, and
other rewards. Employees fear, however, that once they live up to their end of the deal
BRUCE E. KAUFMAN 375

companies will renege and keep all the profits. So, afraid of employers' double-deal-
ing, employees hold back and the potential gain to productivity is lost (Miller, 1991;
Kaufman and Levine, 2000). Unionism, however, can solve this type of prisoner's
dilemma by getting the employers to agree in a written contract how the gains from
cooperation will be shared.
Agency Services. A third way unions can reduce transaction cost and increase effi-
ciency is by providing agency services to workers and firms. The principle involved
is well illustrated in sports and entertainment labor markets where professional ath-
letes and rock stars hire agents to represent them. These agents improve the "quality"
of the employment contract between the employer and employee; in effect, agents serve
as intermediaries and brokers (Barker, 1997). As the employees' agent, a union can
provide three kinds of efficiency-enhancing services (Faith and Reid, 1987). The first
is to supply relevant labor market information (wages at competing firms, alternative
pension plans) and aid in contract negotiation with the employer. A second benefit is
monitoring and evaluating the employer's contract performance (governance services).
A third benefit is communication (voice services) of the workers' preferences regard-
ing the structure of compensation and personnel practices. While rock stars and pro-
fessional athletes typically hire individual agents, workers have less money and
specialized needs and thus find agent sharing to be more cost effective. A labor union
is one such arrangement, a professional employees association is another. Although not
directly related to efficiency, it is important to note that a fourth service provided by
a collective agent is to represent workers' interests in the political and regulatory
process (Delaney and Schwochau, 1993).
Firms can also find it advantageous to hire or create an agent to act as an inter-
mediary with workers (Kaufman and Levine, 2000). Rather than the employer having
to meet one-on-one with each employee to negotiate and administer the employment
contract, this job can be assigned to the agent, thus reaping economies of scale. The
agent may also be granted some independence by the employer, thus increasing the
trust of workers that the agent will perform as an "honest broker" in mediating con-
flicts of interests. A labor union is one such agent; others include a personnel/human
resource department and employee representation plan ("company union").
Qualifications. All of the arguments of this section provide theoretical rationales
for why union voice may improve economic efficiency. Four significant qualifica-
tions to this otherwise positive account have to be noted, however.
First, on a conceptual level F&M's theory has a potential lacuna. They portray the
monopoly and voice/response functions as largely separable - - the former is the effi-
ciency distorting exercise of "muscle," the latter is the efficiency-enhancing effect of
communication or "voice." In reality, however, unions often are only able to induce
firms to provide more collective goods or honor workers rights when voice is backed
up by muscle. It is not self-evident, therefore, how the muscle function of unions with
respect to wages and benefits can be significantly weakened or redirected, per F&M's
proposal (p. 248), yet the voice function strengthened.
376 JOURNAL OF LABOR RESEARCH

A second qualification is that on an empirical level the putatively positive effect

of union voice may in fact hide a more negative monopoly effect. Unions, for exam-
ple, may lower the employee quit rate but if this is done by making it difficult for firms
to fire "dead wood," the result may be lower productivity. Likewise, unions may
achieve economies of scale in negotiating collective contracts with employers, but if
the collective bargaining process creates or heightens adversarial relations and feelings
of distrust the result may be less effective teamwork and employee involvement.
A third qualification is that many of the gains in efficiency from union voice accrue
in the short run, often in the form of a "one-time" improvement. A union is organized
and a new governance regime is established, including a written contract, grievance
system, and a plant safety committee. For the first several bargaining rounds, a sig-
nificant proportion of the union's bargaining power may be expended to win these
improvements, thus keeping the monopoly wage effect to a modest level. In the long
run, however, further gains in the voice area are likely to be subject to substantial dimin-
ishing returns, causing the union to shift a growing proportion of its collective bar-
gaining demands to improvements in the economic package. Over time, therefore,
one may expect that the negative monopoly face will gradually displace the positive
voice face. [A similar conclusion is a reasonable hypothesis with respect to the monop-
sony-reducing effect of unions.]
Finally, even if unionism contributes to increased efficiency in all the ways enu-
merated above, the possibility remains that some other kind of workplace institution
may perform better. Examples include a works council, employee representation plan,
legally binding employee handbook, or labor court. Before one can conclude that
national labor policy should encourage more unionism, a comparative institutional
analysis should be done to determine that unions are superior to other forms of col-
lective representation and work force governance. Ideally, such analysis should also
be performed to determine the optimal form of union and configuration of labor law.

V. Conclusion
I have attempted to distill a quite sizable body of literature written by modern econo-
mists on the theory of labor unions, with particular emphasis given to the "two faces"
view of unions developed by Freeman and Medoff in What Do Unions Do? Many
important ideas and contributions have necessarily been omitted, but hopefully the
main contours and implications are captured.
From this review, it seems clear that F&M have correctly called attention to the
fact that unions in theory and practice have both a positive and negative face. They
are also correct, I think, to suggest that most economists believe, as a generalization,
that the negative side of unions outweighs the positive side, at least with respect to
resource allocation and efficiency. When other noneconomic outcomes are introduced,
such as income and wealth inequality, social justice, and workplace democracy, the
verdict of economists may become more favorable (see Rees, 1989, for example). But
BRUCE E. KAUFMAN 377

this conclusion is highly tentative since modern-day economists typically refrain from
making pronouncements on what are perceived to be normative issues.
The most controversial aspect of What Do Unions Do? is F&M's claim that even
on efficiency grounds unionism is most likely a net plus or, at worst, a very small
negative. The theoretical models and implications reviewed in this chapter do not pro-
vide definitive evidence on this matter, but they do reduce our range of ignorance and
inform our priors. In this regard, several points are worth noting.
First, it appears that unions in their economic function have not two faces but
three. There is the negative monopoly-creating face and the positive voice face, but
F&M and most modern labor economists omit a third face - - the positive monop-
sony-reducing face. The term "monopsony-reducing" endeavors to capture the idea
that unions are necessary to offset employer power in the labor market and firm, thus
protecting workers from managerial abuse and substandard wages and working con-
ditions. Employer power, however, has broader origins than classic monopsony, for it
originates in a variety of restrictions on competition and mobility - - including job lock
from seniority provisions and health care plans, discrimination and segmented labor
markets, and lack of jobs due to involuntary unemployment. A fair generalization is
that the most compelling social rationale for unions among citizens and policymakers
is the traditional one of "protecting the underdog" and "leveling the playing field,"
which appears to speak to the monopsony-reducing role. It is thus paradoxical that
F&M, who desire to construct a positive rationale for unions, omit this aspect of union-
ism altogether. Instead, they follow most modern labor economists and assume that
labor markets are largely competitive, at least in wage determination. This view may
be entirely correct as an empirical matter; my point is that it is too fundamental to "what
unions do" to so easily dismiss or assume it away.
When one looks at the theoretical literature on unions, other union "faces" also
emerge that seem important but under-emphasized by F&M. One such distinction, for
example, is the "short run" and "long run" faces of unions. Much of the theory and
analysis in What Do Unions Do? is static, cross sectional, and short run. On one hand
this emphasis is to a degree unavoidable, given modeling complexities and data avail-
ability. But economic theory also suggests that a preponderate focus on the short run
is likely to impart a positive bias to the economic scorecard on unionism. One reason
is that more of the efficiency benefits from unions are realized in the short run (from
the voice and monopsony-reducing effects); a second is that more of the efficiency
costs from the monopoly face of unions appear in the long run (due to factor misallo-
cation and lower productivity growth). As a concrete example, F&M use the median-
voter model of unions to derive a positive union effect on the short-run provision of
collective goods in the workplace but fail to consider the negative long-run implica-
tion of this model for employment, capital investment, and cost-push inflation. Another
example is the primary (but not exclusive) focus on the determinants of the cross-sec-
tional differences in the level of productivity across firms, rather than on the dynamic
path of productivity growth between union and nonunion firms. Also largely neglected
378 JOURNAL OF LABOR RESEARCH

by F & M are the macroeconomic costs o f unions, such as a tendency for cost-creep
inflation and higher u n e m p l o y m e n t in order to restrain union bargaining demands.
Given these shortcomings, it is also evident that F & M in W h a t D o Unions D o ?
have made significant and enduring contributions to the theory and understanding o f
unions. Certainly their most important contribution is their emphasis on and elabora-
tion o f the "voice/response" face of unions. As demonstrated in m y other article in
this symposium, some o f these ideas were noted by an earlier generation o f econo-
mists, but F & M gave these ideas a systematic treatment and theoretical foundation
grounded in modem economic theory. No subsequent study o f unions can discuss union
voice without referencing F & M . Two other noteworthy contributions also deserve high-
light. Prior to F&M, most of economic theorizing focused on the wage effects of unions.
More than any prior study, their book forced economists to broaden their models to
include the gamut o f union effects, including not only wages but also benefits, pro-
ductivity, quits, and a variety of other outcomes. And, finally, F & M also made a sig-
nificant mark on the theoretical literature with their application of the median-voter
model to the union effect on the structure of wages, benefits, and personnel practices.
I estimate that twenty years after the publication o f W h a t D o Unions D o ? the large
majority of economists remain convinced that economic theory points to a net nega-
tive impact of unions on resource allocation and economic efficiency. In this respect
F & M have failed to move professional opinion. Without question, however, they have
forced economists to carefully reconsider the case for and against unionism and have,
possibly, moved the verdict in a modestly more favorable direction. In the grand scheme
of things, these are powerful accomplishments for one book.

REFERENCES

Addison, John and Barry Hirsch. "The Economic Effects of Employment Regulation: What Are the Lim-
its?" In Bruce Kaufman, ed. Government Regulation of the Employment Relationship. Madison, Wisc.:
Industrial Relations Research Association, 1997, pp. 125-78.
Addison, John and John Chilton. "Models of Union Behavior." In David Lewin, Daniel Mitchell, and Mah-
mood Zaidi, eds. The Human Resource Management Handbook, Part If. Greenwich, Conn.: JAI Press,
1997, pp. 157-96.
Akerlof, George. "The Fair-Wage-Effort Hypothesis and Unemployment." Quarterly Journal of Econom-
ics 105 (May 1990): 255-84.
_ _ and Janet Yellen. Efficiency Wage Models of the Labor Market. New York: Cambridge University
Press, 1986.
Altman, Morris. Worker Satisfaction and Economic Performance. Armonk, N.Y.: M.E. Sharpe, 2000.
Argyrous, George, Mathew Forstater, and Gary Mongiovi. Growth, Distribution, and Effective Demand.
Armonk, N.Y.: M.E. Sharpe, 2003.
Barker, George. An Economic Analysis of Trade Unions and the Common Law, Brookfield, Vt.: Avebury,
1997.
Bennett, James T. and Bruce E. Kaufman. "Conclusion: the Future of Private Sector Unionismin the U.S.
--Assessment and Forecast." In James Bennett and Bruce Kaufman, eds. The Future of Private Sec-
tor Unionism in the United States. Armonk, N.Y.: M.E. Sharpe, 2002, pp. 359-86.
B R U C E E. K A U F M A N 379

Bennett, James T. and M. Johnson. "The Impact of Right to Work Laws on the Economic Behavior of Local
Unions - - A Property Rights Perspective." Journal of Labor Research 1 (Spring 1980): 1-28.
Bennett, James and Thomas DiLorenzo. "Unions, Politics, and Protectionism." Journal of Labor Research
5 (Summer 1984): 309-14.
Blanchflower, David, Andrew Oswald, and Peter Sanfey. "Wages, Profits, and Rent-Sharing." Quarterly
Journal of Economics 111 (February 1996): 227-5 l.
Boal, William and Michael Ransom. "Monopsony in the Labor Market." Journal of Economic Literature
35 (March 1997): 86-112.
Booth, Alison. The Economics of Trade Unions. New York: Cambridge University Press, 1995,
Bronfenbrenner, Martin. Income Distribution Theory. Chicago: Aldine, 1971.
Budd, John. "Institutional and Market Determinants of Wage Spillovers: Evidence from UAW Pattern Bar-
gaining." Industrial Relations 36 (January 1997): 97-116.
Burda, Michael. "Membership, Seniority and Wage Setting in Democratic Labour Unions." Economiea 57
(November 1990): 455-66.
Card, David. "The Effect of Unions on Wage Inequality in the U.S. Labor Market." Industrial and Labor
Relations Review 54 (January 2001 ): 296-315.
__ and Alan Krueger. Myth and Measurement: The New Economics of the Minimum Wage. Princeton:
Princeton University Press, 1995.
Chaison, Gary and Barbara Bigelow. Unions and Legitimacy. Ithaca, N.Y.: Cornell University Press, 2001.
Coase, Ronald. "The Nature of the Firm." Economica 4 (November 1937): 386-405.
Commons, John. Institutional Economics: Its Place in Political Economy. New York: MacMillan, 1934.
Cross, John. The Economics of Bargaining. New York: Basic Books, 1969.
Delaney, John and Susan Schwochau. "Employee Representation Through the Political Process." In Bruce
Kaufman and Morris Kleiner, eds. Employee Representation: Alternatives and Future Direc-
tions. Madison, Wisc.: Industrial Relations Research Association, 1993, pp. 265-304.
Demsetz, Harold. "The Theory of the Firm Revisited." In O. Williamson and S. Winter, eds. The Nature of
the Firm. New York: Oxford University Press, pp. 159-78.
Devinatz, Victor. "The Fears of Resource Standardization and the Creation of an Adversarial Workplace Cli-
mate: The Struggle to Organize a Faculty Union at Illinois State University." In David Lewin and Bruce
Kaufman, eds. Advances in Industrial and Labor Relations, Vol. 11. New York: Elsevier Science, 2002,
pp. 145-80.
Dow, Gregory. "The New Institutional Economics and Employment Regulation." In Bruce Kaufman, ed.
Government Regulation of the Employment Relationship. Madison, Wise.: Industrial Relations Research
Association, 1997, pp. 57-90.
Faith, Roger and Joseph Reid. "An Agency Theory of Unionism." Journal of Economic Behavior and Orga-
nizationon 8 (No. 1, 1987): 39-60.
Farber, Henry. "'The Determination of the Union Status of Workers." Econometrica 51 (September 1983):
1417-38.
. "Right-to-Work Laws and the Extent of Unionization." Journal of Labor Economics 2 (July 1984): 319-52.
. "The Analysis of Union Behavior." In Orley Ashenfelter and Richard Layard, eds. Handbook of
Labor Economics, Vol. 1. New York: North-Holland, 1986, pp. 1039-89.
_ _ and Alan Krueger. "Union Membership in the United States: The Decline Continues." In Bruce Kauf-
man and Morris Kleiner, eds. Employee Representation: Alternatives and Future Directions. Madi-
son, Wisc.: Industrial Relations Research Association, 1993, pp. 105-34.
380 JOURNAL OF LABOR RESEARCH

Flanagan, Robert. "Workplace Public Goods and Union Organization." Industrial Relations 22 (Spring 1983):
224-37.
_ . "Can Political Models Predict Union Behavior?" In R.J. Flanagan, K.O. Moene, and M. Waller-
stein, eds. Trade Union Behavior, Pay-Bargaining, and Economic Performance. Oxford: Oxford Uni-
versity Press, 1993, pp. 5-62.
Freeman, Richard. "Individual Mobility and Union Voice in the Labor Market." American Economic Review
66 (May 1976): 361-8.
. "Unionism and the Dispersion of Wages." Industrial and Labor Relations Review 34 (October
1980a): 3-23.
. "The Effect of Unionism on Worker Attachment to Firms." Journal of Labor Research 1 (Spring
1980b): 229-62.
. "The Effect of Unionism on Fringe Benefits." Industrial and Labor Relations Review 34 (July 1981 ):
489-509.
Freeman, Richard and James Medoff. What Do Unions Do? N.Y.: Basic Books, 1984.
Freeman, Richard and Morris Kleiner. "Employer Behavior in the Face of Union Organizing Drives." Indus-
trial and Labor Relations Review 43 (April 1990): 351-65.
. "Do Unions Make Enterprises Insolvent?" Industrial and Labor Relations Review 52 (July 1999):
510-27.
Freeman, Richard and Edward Lazear. "An Economic Analysis of Works Councils." In Wolfgang Streeck
and Joel Rogers, eds. Works Councils. Chicago: University of Chicago Press, 1995, pp. 27-52.
Freeman, Richard and Joel Rodgers. What Do Workers Want? Ithaca, N.Y.: ILR Press, 1999.
Friedman, Milton. "Some Comments on the Significance of Labor Unions for Economic Policy." In David
McCord Wright, ed. The Impact of the Union. New York: Harcourt, Brace, 1951, pp. 204-34.
Gallagher, Daniel and Cynthia Gramm. "Collective Bargaining and Strike Activity." In David Lewin, Daniel
Mitchell, and Mahmood Zaidi, eds. The Human Resource Management Handbook, Part II. Green-
wich, Conn.: JA1 Press, 1997, pp. 65-94.
Grout, Paul. "Investment and Wages in the Absence of Binding Contracts: A Nash Bargaining Approach."
Econometrica 52 (March 1984): 449-60.
Gunderson, Morley and A. Melino. "Estimating Strike Effects in a General Model of Prices and Quantities."
Journal of Labor Economics 5 (January 1987): 1-19.
Hicks, John. The Theory of Wages. New York: MacMillan, 1932.
Hirsch, Barry. Labor Unions and the Performance of Firms. Kalamazoo, Mich.: W.E. Upjohn Institute for
Employment Research, 1991.
_ _ and John Addison. The Economic Analysis of Unions. Boston: Allen and Unwin, 1986.
Hirschman, Albert. Exit, Voice, and Loyalty. Cambridge: Harvard University Press, 1970.
Johnson, George. "Organized Labor's Political Agenda." In Wei-Chiao Huang, ed. Organized Labor at the
Crossroads. Kalamazoo, Mich.: W.E. Upjohn Institute for Employment Research, 1989, pp. 63-90 9
9"Work-Rules, Featherbedding, and Pareto-Optimal Union-Management Bargaining." Journal of
Labor Economics 8 (January 1990): $237-59.
Kaufman, Bruce. "Research on Strike Models and Outcomes in the 1980s: Accomplishments and Short-
comings." In David Lewin, Olivia Mitchell, and Peter Sherer, eds. Research Frontiers in Industrial Rela-
tions and Human Resources. Madison, Wisc.: Industrial Relations Research Association, 1992, pp.
77-130.
B R U C E E. K A U F M A N 381

. "Emotional Arousal as a Source of Bounded Rationality." Journal of Economic Behavior and Orga-
nizationon 38 (February 1999): 135--44.
. "Models of Union Wage Determination: What Have We Learned since Dunlop and Ross?" Indus-
trial Relations 41 (January 2002): 110-58.
. "The Organization of Economic Activity: Insights from the Institutional Theory of John R. Com-
mons." Journal of Economic Behavior and Organization 52 (September 2003): 71-96.
_ _ and Jorge Martinez-Vasquez. "Voting for Wage Concessions: The Case of the 1982 GM-UAW Nego-
tiations." Industrial and Labor Relations Review 41 (January 1988): 183-94.
Kaufman, Bruce and David Levine. "An Economic Analysis of Employee Representation." In Bruce Kauf-
man and Daphne Taras, eds. Nonunion Employee Representation: History, Contemporary Practice, and
Policy. Armonk, N.Y.: M.E. Sharpe, 2000, pp. 149-75.
Kuhn, Peter. "Union Productivity Effects and Economic Efficiency." Journal of Labor Research 6 (Sum-
mer 1985): 229-48.
Lawrence, Colin and Robert Lawrence. "Manufacturing Wage Dispersion: An End-Game Interpretation."
Brookings Papers on Economic Activity (No. 1, 1985): 47-106.
Levine, David. "Fairness, Markets, and Ability to Pay." American Economic Review 83 (December 1993):
1 2 4 1 - 5 9 .

. Reinventing the Workplace. Washington, D.C.: Brookings Institution, 1995.

Lewin, David. "Dispute Resolution in Nonunion Organizations: Key Empirical Findings." In Samuel Estre-
icher, ed. Alternative Di,~pute Resolution in the Employment Arena. New York: Kluwer, 2004.
_ _ and Karen Boroff. "The Role of Loyalty in Exit and Voice: A Conceptual and Empirical Analysis."
In David Lewin, Bruce Kaufman, and Donna Sockell, eds. Advances in Industrial and Labor Relations,
Vol. 7. Greenwich, Conn.: JAI Press, 1996, pp. 69-96.
Maloney, Michael, Robert McCormick, and Robert Tollison. "'Achieving Cartel Profits through Unioniza-
tion." Southern Economic Journal 46 (October 1979): 628-34.
Manning, Alan. Monopsony in Motion: Imperfect Competititon in Labor Markets. Princeton: Princeton Uni-
versity Press, 2003.
Manzini, Paola. "Game Theoretic Models of Wage Bargaining." Journal of Economic Surveys 12 (Febru-
ary 1998): 1-41.
Miller, Gary. Managerial Dilemmas. New York: Cambridge University Press, 1991.
Mitchell, Daniel. Unions, Wages, and Inflation. Washington, D.C.: Brookings Institution, 1980.
Naylor, Robin. "Economic Models of Union Behavior." In John Addison and Claus Schnabel, eds. The Inter-
national Handbook of Trade Unions. Northampton, Mass.: Edward Elgar, 2003.
Neumark, David. "Declining Union Strength and Labor Cost Inflation in the 1980s." Industrial Relations
32 (Spring 1993): 204-22.
Northrup, Herbert. "Construction Union Programs to Regain Jobs: Background and Overview." Journal of
Labor Research 18 (Winter 1997): 1-16.
Oswald, Andrew. "The Economic Theory of Trade Unions: An Introductory Survey." Scandinavian Jour-
nal of Economics 87 (No. 2, 1985): 167-93.
. "Efficient Contracts Are on the Labour Demand Curve: Theory and Facts." Labour Economics 1
(1993): 85-113.
Pencavel, John. "The Demand for Union Services: An Exercise." Industrial and Labor Relations Review
24 (January 1971): 180-90.
382 JOURNAL OF LABOR RESEARCH

. Labor Markets and Trade Unionism. Cambridge: Blackwell, 1991.

Reder, Melvin. Economics: The Culture of a Controversial Science. Chicago: University of Chicago Press, 1999.
Rees, Albert. "The Effects of Unions on Resource Allocation." Journxd of Law and Economics 6 (October
1963): 69-789
9 The Economics of Trade Unions, 3rd ed. Chicago: University of Chicago Press, 1989.
Reynolds, Lloyd and Cynthia Taft. The Evolution of Wage Structure9 New Haven, Conn.: Yale University
Press, 1956.
Reynolds, Morgan. Power and Privilege." Labor Unions in America. New York: Universe Books, 1984.
Ross, Arthur. Trade Union Wage Policy. Berkeley: University of California Press, 1948.
Schwartz-Miller, Ann and Wayne Talley. "Technology and Labor Relations: Railroads and Ports." Journal
of Labor Research 23 (Fall 2002): 513-34.
Shister, Joseph9 "The Theory of Union Wage Rigidity." Quarterly Journal of Economics 57 (August 1943):
522-42.
Simon, Herbert. Models of Bounded Rationality, Vol. 2. Cambridge: MIT Press, 1982.
Taras, Daphne and Allen Ponak. "Mandatory Agency Shop Laws As an Explanation of Canada-US Union
Density Divergence." In James Bennett and Bruce Kaufman, eds. The Future of Private Sector Union-
ism in the United States. Armonk, N.Y.: M.E. Sharpe, 2002, pp. 260-91.
Ulman, Lloyd. "Marshall and Friedman on Union Strength." Review of Economics and Statistics 37 (Novem-
ber 1955): 384-401.
Vedder, Richard and Lowell Galloway. "The Economic Effects of Labor Unions Revisited." Journal c#fLabor
Research 23 (Winter 2(/02): 4 I-5(l.
Vittoz, Victor. New Deal lxlbor Policy and the American Industrial Economy. Chapel Hill: University of
North Carolina Press, 1987.
Williamson, Oliver. The Economic Institutions of Capitalism. New York: Free Press, 1985.
__, Michael Wachter, and J. Harris. "Understanding the Employment Relation: The Analysis of Idio-
syncratic Exchange." Bell Journal of Economics 6 (Spring 1975): 250-78.