The Great Discrimination: Borders as a Labor Market Barrier

1
The Great DiscriminationBorders as a Labor
Market Barrier

• Lant Pritchett
• (paper with Michael Clemens and Claudio
  Montenegro)
• April 21, 2008
• Yale University

2
Outline

• Motivation
• Estimates of wages of observationally equivalent
  workers across the US border for 42 countries
• Data
• Regressions
• Results
• From observationally equivalent to equal
  productivity?five attempts
• Implications/extensions

3
Preview of coming attractions

• Median wage ratio 3.41 (Bolivia)
• Five techniques suggest the bias of self
  selection 1.2
• Border distortion far and way the largest wage
  discrimination ever (on par with slavery)
• Compare its removal to trade liberalization and
  all other anti-poverty efforts

4


5
Wages ratios are enormously larger than those
that set in motion the first globalizations
migration
Today larger, but (relative) flows smaller
Prior to 1920
Source Pritchett 2006
6
What to make of the existing ratios

• How much of the wage ratio is
• differences in productivity in the individual
  (e.g. education) or individual-market match (e.g.
  language)
• workers of equal productivity (if in the same
  place) earn different amounts?
• Very few papers on this topic, Altonji and
  Rosenzweig (2008) using actual migrants before
  and after

7
Estimates of wage ratios of equivalent workers

• Assume that earnings and basic demographics
  sufficiently comparable in different countries
  formal sector wage surveys
• Convert earnings to PPP dollars
• Pool data across the available HH level data sets
  and US Census

8
(No Transcript)
9
All kinds of issues

• Gross versus net income
• Wage earners only (so self-selection in the
  foreign labor market)
• Wage periodicity (weekly, monthly)
• Lets come back to all of this later

10
Comparing our data to National Accounts
Clear trouble at low end
11
Bilateral regressions Wages of naturals on
either side of the border

• Left-hand side ln(wage)
• Right-hand side
• Schooling, Age (and its square), Gender, Rural
• Dummies Foreign resident, Late-arrival
  US-resident Foreign-born, Early-arrival
  US-resident foreign born
• Interactions of schooling and dummies
• Different specifications

12
(No Transcript)
13
Wage ratios of observationally equivalent workers

• Same country of birth (e.g. Bolivians)
• Age 35, Schooling 9, Male, Urban
• Compare non-migrants with those characteristics
  to other categories
• US born
• Bolivian born early arrivers (perhaps US
  educated)
• Bolivian born late arrivers (likely Bolivian
  educated).

14
(No Transcript)
15
(No Transcript)
16
Functional form makes little difference
comparing Mincer to Categories specification,
Naturals to late arrivers
17
Wage ratios of observationally equivalent low
skill workers

• Range from 1.3 (Costa Rica) to 10 (Haiti,
  Nigeria) with median 3.4 (Bolivia3.43)
• In absolute terms, Bolivia early arrivers
  2163/month (12.3/hr), late arrivers 1579/month
  (8.97/hr), Bolivians in Bolivia 460/month
  (2.61/hr)
• Income difference is 13,425 per year (at the
  median), 5,600 for Mexico, 17,000 for Haiti

18
How much of observed differences are eliminated?
About 50 percent
19
What is the relation between observationally
equivalent and equal productivity?

• Simulation
• Pooled regressions, adjusted for education
• Visa lottery natural experiment
• Open borders (e.g. Puerto Rico)
• Macroeconomic comparison

20
1st Model bias of observed wage ratiofor
marginal migrant
?Ratio of home to foreign variance.84 rhoCorre
lation of home and foreign unobserved assumed.5,
hm-hbar.33 at those
21
Comparison of residualsfrom wage regressions
Vietnam
22
2nd Used pooled regressions to predict average
wages of observationally equivalent in home and
foreign (allowing for country specific schooling)

• Doesnt involve movers at allso should
  understate the marginal mover if there is
  positive selection.
• In fact, these are larger than bilateral
  estimatesmedian of 3.75 vs. 3.18 (for S9)
• But one has to correct for the quality of
  schooling as S in Bolivia is not S in USA

23
Using scores to adjustfor years equivalent
24
Using estimated Mincer of naturals in USA to
estimate S evaporation
25
3rd Experimental estimate

• Movers from Tonga to New Zealand chosen from
  applicants based on a lottery
• OLS wage ratio 6.14
• Experimental wage ratio 4.91.
• Bias 6.12/4.911.25

26
4th US and Puerto Rico

• Full labor mobility
• Ratio 1.5
• Upper bound on self-selection bias
• If there are any natural barriers (language,
  information, family, etc.) then upper bound on
  selection gets lower

27
5th Comparison to macro data using growth
accounting
28
Wage ratios from labor marketbarriers of various
types
29
The Great Discrimination
Slavery in Virginia, 1840-1860, ratio rental to
subsistence 3.8
30
The total present value of access to a lifetime
of micro-credit is the wage difference of 11
weeks work of the same worker in USA versus in
Bangladesh
Source Gains estimated from Pitt and Khandker
for micro-credit (annual gain of 14 percent of
per capita HH consumption, CMP (forthcoming) for
wages
31
Debt Relief? How about more aid? Fairer/freer
trade? Total gains versus a small increase in
labor mobility
32
Conclusion

• Wage differentials are huge
• Wage ratios of observationally equivalent
  low-skill workers are about half the total
• This appears to mostly reflect barriers to
  mobility, with modest self-selection
• This makes allowing a person across a border the
  single largest impact developmental intervention
  available