Chapter 9 Labor Market Discrimination by hcj


									Chapter 9 Labor Market
Why is there wage dispersion?
 Different jobs (compensating wage differentials)
 Discrimination
 Different characteristics of workers:
   Ability
   Schooling
   Choices about work and leisure
 Discrimination
What is discrimination?
 When workers earn different amounts even if they do the
  same job and have the same ability.
 Borjas: “where the costs and benefits of an economic
  exchange depend on the race and/or gender of the person
  involved in the exchange.”
Table 9-1: various measures of human capital and labor market outcomes in the US labor market.

                                                  White                         Black                     Hispanic

                                                  Male          Female          Male             Female   Male       Female

    % HS grad or more                             85.3          87.1            81.9             82.5     58.2       62.5

    % bachelor’s degree or more                   29.9          28.3            18.0             19.0     11.8       13.7

    LFPR                                          76.3          60.1            71.2             64.0     84.7       58.8

    Unemployment Rate                             3.7           3.6             7.9              6.7      4.6        5.5

    Annual Earnings (in $1,000)                   55.9          35.1            39.5             31.8     35.4       26.3

    Annual earnings (among workers employed       65.8          43.6            44.5             37.6     38.8       32.1
    full-time, year-round) (in $1,000)
What is interesting about table 9-1?
 Difference in earning are due in part to differences in labor
  supply: white man 59% more than white woman but if we
  compare only those who work full time the gap is 51%
 Differences in earnings are due in part to differences in
  educational attaiment: 15% of white men do not have a HS
  diploma, 20% of black men and > 40% of hispanic men.
 Look at college diplomas.
 If rate of return to schooling is between 7 and 9%, this could
  generate substantial wage differentials
Table 9-2: international differences in Female/Male Wage ratios

        Country                                        1979-1981   1994-1998

        Australia                                      0.800       0.868

        Canada                                         0.633       0.698

        Finland                                        0.734       0.799

        France                                         0.799       0.899

        Germany                                        0.717       0.755

        Ireland                                                    0.745

        Italy                                                      0.833

        Japan                                          0.587       0.636

        New Zealand                                    0.734       0.814

        Spain                                                      0.711

        Sweden                                         0.838       0.835

        United Kingdom                                 0.626       0.749

        United States                                  0.625       0.763
 So, the truth is, we can’t explain all of the differences in
  wages between men and women and across races.
 So, maybe there is discrimination in the labor market.
 There are several models of discrimination and ways to
  measure it. That’s what this chapter is about.
Models of discrimination
 The perfectly competitive market
 Gary Becker’s theory of discrimination
 Statistical discrimination
 Core and Periphery jobs
Neoclassical Model of Discrimination
 If a market is perfectly competitive, discrimination cannot
  exist for long.
 There is evidence of discrimination.
 Explore the EEOC webpage sometime.
 Hundreds of legal cases.
 Hundreds of stories:
   The movie “North Country”
   Wal Mart
Gary Becker’s Taste for Discrimination
 So, Becker wants to explain why we might see discrimination
  in a market, even if it is a competitive one.
 He comes up with the notion of a “taste for discrimination”
   Employers
   Employees
   Customers
Taste for Discrimination
 Suppose there are two types of workers in the labor market:
  white workers and black workers.
 A competitive employer faces constant prices for these
  Ww = wage rate for white workers
  Wb = wage rate for black workers
  If the employer is prejudiced against blacks, the employer gets a
     disutility from hiring black workers.
  Employer will ACT as though the cost of a black worker is:
     Wb(1+d) dollars.
  d is the “discrimination coefficient”
 Suppose that Wb = $10 per hour
 d = 0.5
 The employer will then act as if hiring a black worker costs
  $15 per hour, a 50% increase in the cost.
 The greater the prejudice, the greater the disutility from
  hiring blacks and the greater the discrimination coefficient.
 Some employers (maybe black-owned firms) might have a
  different type of prejudice: they PREFER to hire blacks
 Think Oprah
 This type of behavior is called nepotism
 It implies that an employer’s utility-adjusted cost of hiring a
  favored worker equals Wb(1-n) dollars
 They will act as if hiring a black worker is actually cheaper
  than it really is.
Employer Discrimination
 Assume black and white workers are perfect substitutes, so a
  firm’s production function looks like this:
        q = f(Ew + Eb)
The firm needs to decide which combination of inputs to hire.
  Ew is number of white workers hired and Eb is number of
  black workers hired. The firm’s output just depends on the
  total number of workers hired. Not their race.
Any differences that arise in the economic status of the two
  groups cannot be attributed to skill differencences.
Assume first: No Prejudice
 Both black and white workers have same VMP
 Ww = wage of white worker and Wb = wage of black
 Profit maximization? Hire the cheapest worker.
 If Wb < Ww, hire black workers until Wb = VMPe

 See figure 9-1.
Assume Prejudicial Firm
 Employer acts as if the black wage is not Wb but is Wb(1+d).

 Now firm hires the worker with the lowest utility-adjusted
 Decision rule:

       Hire only black workers if Wb(1+d) < Ww
       Hire only white workers if Wb(1+d) > Ww

  Big Implication: Segregated Workforce even under competitive
Consequences of Employer
 See Figure 9-2.
   “white firms” pay more and hire less
   “black firms” who are still discriminatory, hire fewer black
    workers depending on size of d
Labor Market Equilibrium
 See figure 9-4
 Initially assume Wb > Ww: no firm will hire black workers
 Wb falls.
 Note: even where Wb = Ww demand is zero. Why?
 R is a “threshold” where some firms have a low enough d to
  begin to hire black workers.
 Note that the equilibrium black-white wage ratio
  (Wb/Ww)* occurs BELOW the point where the black
  white wage ratio = 1. Employer discrimination generates a
  gap between equally skilled black and white workers.
Employee Discrimination
 Suppose that whites dislike working alongside blacks and blacks
  are indifferent about the race of their coworkers.
 These white workers will act as if their wage (Ww) is only Ww(1-
  d) where d is the white worker’s discrimination coefficient.
 Example: Ww = $15/hour. In a discriminating worker’s view, an
  integrated firm’s wage is less than $15/hour so that firm will have
  to pay more to get white workers to work there.
 But it does not pay to do this if you are a color-blind firm… Thus
  this model implies a segregated workforce as well.
 Note however, that a race wage gap is NOT generated by this
 Why? Explain.
Customer Discrimination
 If customers have a taste for discrimination, their purchasing
  decisions are not based on the actual price of a good, p, but
  on the utility-adjusted price or p(1+d), where d is the
  discrimination coefficient.
 If whites dislike purchasing from black sellers, customer
  discrimination reduces the demand for goods and services
  sold by minorities.
 Can you think of a case in which you have “customer
Customer Discrimination and the wage
 Adverse effect on black wages when the firm cannot easily
  hide its black workers from public view.
 A firm employing a black worker in a sales position will have
  to lower the price of the product so as to compensate white
  buyers for their disutility.
 The wage of black workers will then fall because black
  workers have to compensate the employer for the loss in
Empirical Study
 Recent survey tried to get at how the interaction between
  the customers’ racial background and the extent of the
  contact between workers and customers alters the hiring
  decisions of firms.
 Table 9.3: 58% of newly hired workers are black in contact
  firms where most customers are black.
 9% of newly hired workers are black in contact firms where
  most customers are white.
 Difference between the two suggests that customer
  discrimination reduces the fraction of blacks among newly
  hired workers by 49.0 percentage points.
Need to compare to a control
 Important to note that the black employment gap between these
  two types of firms may be attributable to other factors. What if
  contact firms with mainly black customers are in black parts of a
  city… they’d be more likely to hire black workers.
 Firms in survey where workers don’t have much contact can serve
  as a control group.
 Fraction of newly hired workers who are black falls from 46.6
  percent to 12.2 percent as the customer base shifts from black to
  white. … this is a 34.4 percentage point difference.
 Because we can’t blame this on customers, this is what we’d
  expect to happen even if we didn’t have customer discrimination.
Diff in Diff Example
                     Over ½ of Firm’s   Over 75% of        Difference
                     Customers are      Firm’s Customers
                     Black              are White
Type of Firm
Contact between      58.0%              9.0%               49.0%
customers and
No contact between   46.6%              12.2%              34.4%
customers and
Difference-in-       --                 --                 14.6
Model 3: Statistical Discrimination
 Racial and gender differences can arise in perfectly
  competitive situations even in the absence of prejudice when
  membership in a particular group carries information about a
  person’s skills and productivity.
 Job searches are expensive, employers use all info available
  to them. Statistical averages are part of that.
 Example: statistical averages about women and maternity
  leaves etc.
 Illegal to apply this information unless it is a BFOQ
Experimental Evidence on
 Hiring audits:
   Experiments are cleverly designed to induce employers to
    reveal preferences about hiring women and minorities.
   Sent out about 5,000 fake resumes in response to about 1,300
    job ads that actually appeared in Boston and Chicago newspapers.
   Included black sounding and white sounding names
   Holding skills in the resume constant, the applicants with white
    -sounding names got about 1 callback for every 10 resumes sent
    whereas the black-sounding names got 1 callback for every 15
    resumes sent.
Measuring Discrimination
 ∆W = Wm – Wf
  where W indicates the average

This definition is not great because it compares apples to
There are many factors, other than discrimination that generate
  wage differentials between groups. Number of hours worked
  in the workplace, for example is one simple measure.
Other factors in the gender wage gap?
Better measure
 We would like to adjust the “raw” wage differential given above for
   differences in skills (and even choices to the extent that we can measure
   them) between men and women.
 Do this by estimating regressions that relate the earnings of men or
   women to a wide array of socioeconomic and skill characteristics:
          Male earnings function: Wm = αm + βmSm
          Female earnings function: Wf = αf + βfSf
α tells us how much an employer values men/women with 0 years
   of schooling
β tells us by how much a man’s or woman’s wage increases if
   he/she gets one more year of schooling
If employers value the education acquired by men and women
   equally, the coefficients should be equal
Wage differential
∆W = αm + βmSm - αf - βfSf
Oaxaca Decomposition
 What we would like to do is be able to tell how much of the
  differences in wages is due to differences in skills between
  men and women and the portion of the “wage gap” that is
  due to discrimination.
 We can “decompose” the above equation into two parts
  (using a little algebra and a little trick –add and subtract the
  term βm x Sf):
        ∆W = (αm – αf) + (βm – βf)Sf + βm(Sm – Sf)
What does that mean???
 ∆W = (αm – αf) + (βm – βf)Sf + βm(Sm – Sf)

      (αm – αf) + (βm – βf)Sf : portion due to discrimination
               this term will be positive if either employers value a
        man’s schooling more than they value a woman’s schooling
        (βm – βf), or if employers just pay men more than women for
        any level of schooling (αm – αf)

      βm(Sm – Sf): portion due to differences in skills
              this term is zero of men and women have the same
       average schooling.

      See figure 9-6
Validity of Oaxaca
 Depends largely on whether we have controlled for ALL of
  the dimensions in which the skills of the two groups differ…
 Wage equations tend to get to be like the “kitchen sink” and
  even then we have unknowns…
 What CAN we control for?
   Schooling, years of labor market experience, sex, race, region,
    union status, occupation, industry, hours worked…? Number
    of children? Age of first birth? Marital status.
  WHAT CAN’T we control for?
    Effort. Motivation. Quality of schooling. Major?
Section 9-11: Determinants of the
Female-Male Wage Ratio
                               Controls for Differences Controls for Differences
                               in Education, Age, Sex   in Education, Age, Sex,
                               and Region of Residence Region of Residence,
                                                        and Occupation and
Raw log wage differential      -0.286                   -0.286
 Due to difference in skills   -0.008                   -0.076
 Due to discrimination         -0.279                   -0.211
Table 9-6
 Women earned about 28.6 percent less than men in 1995
 Difference in education, age, and region of residence generate
  only a trivial wage gap between men and women, about 0.8
 Even after adjusting for occupation and industry, differences in
  observable socioeconomic characteristics between the two groups
  generate only a 7.6 percent wage gap.
 What is this Oaxaca decomposition missing? Labor market
  experience. Women tend to have kids and drop out of the labor
  market. By the late 1980s, the typical woman worked only 71
  percent of her potential years whereas men worked 93 percent.
Labor market attachment
 If drop out of labor market, lower payoff to human capital
 Lower gain in wages due to experience
 Human capital depreciation
 Study of University of Michigan law school classes of 1973 and
 Fifteen years after graduation, male attorneys earned $141,000
  annually as compared to $86,000 for female attorneys.
 2/3 of this wage gap is due to differences in work histories.
 If a female attorney worked part-time for 3 years her earnings
  were reduced by 17% over her lifetime.
Which came first? And what should we
 Low wages due to less LM attachment or
 Less LM attachment due to low wages?

 Are family friendly policies the answer? Are they politically
Last Discrimination Model: Bergman’s
Occupational Crowding
 Occupational Segregation
   Blue tribe v. Red tribe and the hunt for berries
Economics of Specialization
 Becker – household economics.
 Specialize in what you’re good at, then trade.
 If women earn less, they should work at home while men
  work for a wage in the market.
Conclusion and the Extra Hour
 One of my papers
 One of John’s papers
 One of our coauthored papers

 Choose one, read it, discuss it.

To top