2. Background
• suppose labour markets are competitive i.e.
each employer thinks they can hire as many
workers as they want at the going market wage,
B
• the wage paid will be equal to B - there is no
point in paying a higher wage and a lower wage
results in no workers being hired.
• the firm will choose the level of employment to
maximize profits which leads to MRPL=B:
R '( N ) = B
4. What happens to wages/employment if
there is a union?
• need to make some assumptions about:
– the preferences of the union
– the process of bargaining between employer
and union.
5. The preferences of unions
• large literature debating what the objective
function of a union is likely to be.
• But good reason to think unions like:
– higher wages (unions do set out to raise
wages for their members)
– higher employment (they normally fight
reductions in employment).
• A simple way of capturing these ideas is to
assume that the utility function has the
form:
U = N ( W − B)
γ
7. The Process of Bargaining
• Two main issues:
– the subject of bargaining
– whose preferences, employer or union, get most weight in
determining the outcome.
• On the subject of bargaining, there are two main models:
• the labour demand curve (or right-to-manage) model
– unions and employer negotiate the wage but employment is
determined unilaterally by the employer.
• the efficient bargain model
– unions and employer negotiate both the wage and employment.
8. The Labour Demand Curve Model
• Given wage, employment will be where
MRPL=w
• But what will be the wage?
• Consider extreme case where union has
all the bargaining power
• Will choose point on labour demand curve
that gives highest utility
10. What determines how high this
wage will be?
• One can show that the wage will be lower:
– the more sensitive is labour demand to the
wage (the higher the wage elasticity of labour
demand) – a high elasticity makes the trade-
off between wages and employment less
attractive.
– the more the union cares about employment
relative to the wage.
11. Between the extremes..
• What happens if we have some
distribution of bargaining power between
union and employer between the two
extremes we have discussed?
• It should not be hard to see that we will
have some outcome on the labour
demand curve between the two extremes
and that the wage will be lower:
– the more bargaining power the employer has
12. In the labour demand curve model..
• Unions raise wages
• Unions reduce employment
• Unions reduce profits
• In a model with capital they would also
reduce investment
• But perhaps this is not the final word
13. The Efficient Bargain Model
• The labor demand curve model is based
on a ‘realistic’ set of assumptions.
• The criticism is that the outcome has
unexploited gains from trade.
• To see this consider drawing iso-profit
curves on Figures. The have an inverted
u-shape and, more importantly, they are
horizontal at all points on the labour
demand curve.
14. Implication
• at the monopoly union outcome, the firm iso-profit
curves and union indifference curves are not tangential
• both parties could be made better-off by swapping a
reduction in the wage for some increase in employment
away from the labor demand curve.
• The efficient bargain model makes both wages and
employment the subject of bargaining
• leads to an outcome in which all gains from trade are
exploited and union indifference curves and firm iso-
profit curves are tangential.
• there are many such points differing in the distribution of
surplus between employer and union – the set of these
points is called the contract curve.
15. The Contract Curve
• Can show that contract curve has the form
• If γ=0 then same as labour demand curve
• If 1>γ>0 then downward-sloping
• If γ=1, then vertical
• If γ>1, upward-sloping
• Unions raise wages but may also raise
employment (depending on preferences)
16. Empirical Evidence: Wages
• very large literature on the effect of unions on
wages (the union wage mark-up).
• Typically this includes union status in an
earnings function together with other factors and
interprets the coefficient on the union variable as
the causal effect of union status on wages.
• Little experimental or quasi-experimental
evidence but people accept this literature as it
gives ‘sensible’ answers – a mark-up of 15+% in
the US, 10% in the UK (though zero in recent
years).
17. Empirical Evidence: Employment
• The literature on the impact of unions on
employment is tiny.
• In part this is because the equivalent of the
above wage regression – regress employment
on union status leads to a strong positive effect,
something people have been very reluctant to
believe.
• Perhaps for good reason, as unions may be
more likely to succeed in organizing workers in
large firms.
18. diNardo-Lee
• Contribution needs to be set against weaknesses of
other literature
• They exploit the fact that unions in the US have to win a
representation election to be allowed to negotiate.
• This leads to a RDD design in which one can compare
subsequent outcomes in firms where the union just won
with the outcome in firms where the union just lost.
• There is a good discussion in the paper of what one can
hope to estimate using this – the ‘impact of unionization’
is too simple. They summarize what one might expect
in:
19.
20. Conclusion
• union impact on the ‘bottom line’ is very
small on:
– business survival
– Employment
– Output
– Productivity
– wages
• i.e. unions seem to do very little. A
representative type of empirical finding is:
22. A puzzle
• Even the effect on wages is very small compared to
previous estimates.
• How can we interpret these results and reconcile them
with other findings:
– studies ‘young’ unions who have less impact – but wage effect
on cross-section is similar to other studies.
– can only identify effect at margin but not much evidence of other
effects.
• Its all a bit troubling – this study has the best quality data
but conclusions that are at odds with
– what many people think
– the fact that employers spend large sums of money trying to
avoid unions.
23. Conclusion
• Hard to get good research design to study
the effect of unions
• Best study we has fails to find any effects
at all
• But reluctant perhaps to believe that
• Research does not always give us clear
answers