The National Minimum Wage and Wage Inequality: An Update

1
The National Minimum Wage and Wage Inequality An
Update

• Richard Dickens and Alan Manning

2
Why is an update needed?

• Some interesting things have been happening to
  wage inequality in the UK
• Used to it always getting worse but bottom-end
  wage inequality has been declining for some
  considerable time
• Tempting to put this down to the NMW
• But perhaps not all of it

3
Recent Trends in UK Wage Inequality
4
Where in distribution is change happening?
5
Change up to 25th percentile
6
What Could Be Going On Here?

• Impact of Minimum wage from 1999
• Other factors causing changes in wage inequality
• Strong labour market
• Rise in educational attainment
• Lovely vs. lousy jobs
• Immigration

7
Our Earlier Findings on Impact of NMW

• Dickens and Manning (JRSS, 2004)
• Used LFS
• Found effect of NMW and zero spill-over
• No effect on 10th percentile, only on 5th
• Dickens and Manning (EJ, 2004)
• Care homes data
• Minimum wage much more important (40 affected)
• Still no spill-over effect
• Reduction in wage inequality seems much larger
  than we would expect from these results
• Contrast with US
• Lee (QJE, 1998) implies quite large spill-overs

8
What might be happening?

• NMW has been increasing faster than median
  earnings
• NMW has spill-over effects over long period
  earlier studies were impact effect
• Measurement Error Problems
• Other factors at work

9
The Kaitz Index(NMW/Median Hourly Earnings)
10
But Measures of Spike Do Not Seem To Show Big
Increase
11
Relationship between Kaitz Index and Spike
12
Other estimates of spike

• ONS only seems to provide estimates of those paid
  below the NMW based on ASHE
• More or less in line with LFS
• LPC Report suggests 3.2 were beneficiaries of
  Oct 05 increase
• Cant seem to explain powerful reduction in wage
  inequality.

13
Measurement Error Problems

• Would expect the pattern we see a simple
  example
• Underlying wage dist log normal
• Perfect compliance with minimum wage approx 5
  affected
• But observed wage distribution true classical
  error

14
What Would We See?
15
Can Use Hourly Rate Measure

• LFS has collected hourly rate since March 1999
• Only available for 38 of people
• Have to impute for others
• ONS and Dickens/Manning describe methodologies
  for doing this

16
Propensity Score Re-weighting

• Assumption conditional on covariates, the
  distribution of hourly rate is independent of
  whether it is observed or not
• Estimate probit model for whether hourly rate
  observed
• A strong assumption probably leads to
  over-estimate of impact of NMW

17
Details

• Hourly rate only collected from March 1999 i.e.
  only one month prior to NMW
• Routing of question then different from
  subsequently
• We will compare 2001q1 with 2006q1 (2001q1 NMW
  low in real terms)

18
Comparison of Changes in Hourly Rate and Hourly
Pay Measures
19

• Changes reach too far up wage distribution to be
  direct effect of NMW
• Do not seem able to explain them by measurement
  problems
• What about spill-overs?

20
The Lee Model of Spill-Overs

• Latent log wage distribution w(F) assume
  normal
• Only direct effect of NMW
• Lee adds possibility of spill-over
• Spill-over parameter ß high value, big
  spill-over
• This model works well

21
Estimate Lee Model for 2001q1-2006q1
22
Goodness-of-fit for 2006q1
23
Implied Impact of NMW on Average Log Hourly Wages
24
Conclusions

• Fall in wage inequality can be explained using
  direct effect of NMW and modest spill-over effect
• Implies other factors not so important
• Can this really be true?

25
The Impact of Immigration