Employers and Wages

1
Employers and Wages
2
The Puzzle

• Wage paid to a given type of labour should be
  independent of employer characteristics
• But wages seem correlated with employer
  characteristics
• Size employer size-wage effect
• Productivity/profitability e.g. Blanchflower,
  Oswald, Sanfey, QJE 96
• Industry e.g. Krueger-Summers, Ecta 88

3
Explanations

• Perfectly competitive model right but something
  wrong with the evidence
• Unobserved Worker Quality
• Compensating differentials
• Rent-sharing
• Monopsony

4
Unobserved Worker Quality

• Basic idea is that it is hard to directly observe
  worker quality so will see apparent effect of
  employer characteristics if correlated (omitted
  variables bias)
• True that higher levels of education in bigger,
  more profitable firms so might think same is true
  of unobserved worker quality

5
Ways to test this

• Use panel data does a given individual changing
  jobs get a change in wage that is correlated with
  change in employer characteristics (Gibbons-Katz
  QJE 1992)
• Effect is weaker but is still there so
  correlation of wages with employer
  characteristics seems a true phenomenon.
• Abowd Kramarz (2000) use matched
  (employer-employee, LEED) data

6
AK decompose firm/worker effects
7
Compensating Wage Differentials

• Have assumed workers only care about wages
• But non-pecuniary aspects very important
• Really utility not wages that should be equalized
  in a perfectly competitive market.

8
Simple Model of Compensating Differentials

• Utility is given by uw-e
• e is effort
• Perfect competition equalizes u so that wages for
  individual i can be written as
• wiuiei
• If jobs require more effort (so have higher
  productivity) than must pay higher wages)
• But can be applied to other working conditions

9
Evidence on Compensating Differentials

• Simple regressions of wages on working conditions
  often give wrong sign
• Most common explanation is that good working
  conditions are a normal good so tend to be
  negatively correlated with ui
• Looking at separations gives more sensible
  estimates

10
An example Night-WorkingUK LFS
1 2 3
Coeff on Night -0.079 (0.013) -0.050 (0.010) 0.029 (0.015)
Other Controls No Yes Yes
C-S/Panel C-S C-s Fixed effects
11
Link to Employer Characteristics

• Employer characteristics correlated with non-wage
  characteristics of work
• But if higher wages are not associated with any
  difference in utility then no reason for any
  difference in quits
• But quits are lower in high wage firms so
  suggests workers are better-off in these firms

12
Rent-Sharing

• Basic idea is that some firms get rents and
  workers manage to get a share of them
• Might be through unions but also other possible
  mechanisms
• There are problems with simple regression of
  wages on productivity as productivityAPL which
  is correlated with MPLw under perfect competition

13

• Better studies e.g. Abowd-Lemieux, QJE 1993 use
  instruments they use exchange rate movements
• Many studies seem to find that rent-sharing as
  strong or stronger in non-union as compared to
  union sector
• This seems a bit odd

14
Monopsony explanation

• Defining idea of monopsony is that labour supply
  curve to firm is upward-sloping have to pay
  higher wages to get more workers
• Can obviously explain the employer size wage
  effect
• But more productive employers will also want to
  choose to pay higher wages

15
A Simple Picture
16
Conclusions

• Does seem to be the case that better employers
  pay same worker higher wages
• Workers do seem to be better-off in these
  employers