Although public sector unionism is growing throughout the industri-
alized nations, and there is widespread interest in the question of how
such labour markets operate, little theoretical work has been done on
these problems. The paper develops a two sector general equilibrium
model in which a government and a public sector union together negotiate
an efficient contract. Public sector wages turn out to be sticky com-
pared to those in the private sector, and there is over—employment in
the publicly run half of the economy. The model's features are rather
different from those in the private contract literature. Extensions
of the model to include union—run insurance, endogenous membership,
missing capital markets and public goods are also considered.

This paper re-examines the connection between unions and wage
inequality, focussing on three questions: (1) How does the union wage
effect vary across the wage distribution? (2) What is the effect of
unionism on the overall variance of wages at the end of the 1980s?
(3) How much of the increase in the variance of wages over the 1970s
and 1980s can be attributed to changes in the level and distribution
of union coverage?
Cross-sectional union wage gap estimates vary over the wage
distribution, ranging from over 30 percent for lower wage workers to
-10 percent for higher wage workers. Using a longitudinal estimation
technique that accounts for misclassification errors in union status,
I find that this variation represents a combination of a truly larger
wage effect for lower-paid workers, and differential selection
biases.
The estimated effect of unions on the variance of wages in the
late 1980s is relatively modest. Nevertheless, changes in the level
and pattern of unionism -- particularly the decline of unions among
lower wage workers -- have been an important component of the growth
in wage inequality. Changes in unionization account for one-fifth of
the increase of the variance of adult male wages between 1973 and
1987.

Unlike existing models which rely heavily on assumptions regarding unions’
distributional preferences, we present a very simple model in which union
seniority-layoff rules and rising seniority-wage profiles result from
optimal price discrimination against the firm. Surprisingly, even when
cash transfers among union members are ruled out, unions’ optimal
seniority-wage profiles are likely to be completely unaffected by their
distributional preferences because of a kink in the utility-possibility
frontier. This suggests that the simple technology of price discrimination
may play a key role, hitherto unappreciated, in explaining union policies
that affect the relative wellbeing of different union members.