Statistical discrimination - PowerPoint PPT Presentation

1 / 11
About This Presentation
Title:

Statistical discrimination

Description:

Statistics Theories of Discrimination in Labor Markets.' Industrial and Labor Relations Review 30: 175-187. ... SELF-FULFILLING PROPHECIES ... – PowerPoint PPT presentation

Number of Views:79
Avg rating:3.0/5.0
Slides: 12
Provided by: petero8
Category:

less

Transcript and Presenter's Notes

Title: Statistical discrimination


1
Statistical discrimination
  • Phelps, E. S. 1972. The Statistical Theory of
    Racism and Sexism. American Economic Review 62
    659-661.
  •  
  • Aigner and Cain. 1977. Statistics Theories of
    Discrimination in Labor Markets. Industrial and
    Labor Relations Review 30 175-187.
  • Question How do firms behave if they have
    greater uncertainty about the skills of one group
    relative to others and they use group averages to
    predict skills of individuals in the group?

2
Notation
  • y an indicator of skill such as a test score
  • q true skill which the firm cannot observe
  • u error term N(0,su2)
  • q N(a,sq2)
  • Firm wants to predict q given y. Best linear
    predictor is of the form
  • E(qy) a by

3
Results
  • a (1-?)a lt a 0lt ?lt1
  • b ?
  • ? (sq2 )/(sq2 su2)
  • signal-to-noise ratio
  • If y perfectly predicts q, su2 0 b ? 1 a
    0

4
Assumption
  • Majority and Minority group have
  • Same average ability, a
  • Same distribution of true ability, q N(a,sq2)
  • Minority group has greater error variance, su2
  • Implies ?W gt ?N
  • (1- ?W )a lt (1- ?N )a

5
Statistical Discrimination Relationship between
test score (y) and expected skill (q)
45o
q
W
N
a
Slope ?W
Slope ?N
(1- ?N )a
(1- ?W )a
y
a
6
Results
  • N workers with y lt a are predicted to have higher
    q than W workers with the same test score
  • N workers with y gt a are predicted to have lower
    q than W workers with the same test score

7
Results
  • On average, N and W treated the same
  • But
  • high ability N will find it hard to compete for
    high q jobs
  • If firm gains additional information on y over
    time so that su2 gets smaller over time, ?N
    increases in size to be closer to ?W

8
Add risk aversion parallel shift down in
relationship
45o
q
W
WR
N
NR
a
y
a
9
Add risk aversion parallel shift down in
relationship
45o
q
WR
NR
a
y
a
Now at y a, N is paid less than W
10
Hiring errors
  • Type 1 False positive hire someone who is not
    qualified
  • Type 2 False negative do not hire someone who
    is qualified
  • Which error is more costly?

11
Self-fulfilling prophecies
  • If firm assumes N is higher risk, firm may
    eliminate N from the pool of top jobs
  • Eliminates incentive for N to acquire requisite
    skills or signals to compete for top jobs
Write a Comment
User Comments (0)
About PowerShow.com