i. assume initially all firms discriminate (d>0) so at equal wages, when the non-white to white wage ratio is 1, or when Wnw>Ww, no non-white worker will be hired
ii. are the ratio falls, then increasingly prejudices firms are willing to hire non-whites
iii. so employer discrimination generates a wage gap between equally skills workers
iv. if the firm is nepotistic, their demand curve shifts upwards, so non-white employment remains the same due to an inelastic supply, but their wage increases
i. customers act as if the price of a product sold by a different race is utility adjusted : p(1+d)
ii. firms facing customer discrimination with a mixed workforce may strategically place their non-white workers in the production department
iii. if firms are looking to fill production positions, equally skilled whites and non-whites will earn the same wage
iv. if non-whites cannot be hidden, firms have to lower p to compensate the white buyer for their disutility so wages fall to compensate a loss in profits
shows how racial and gender differentials can occur in the absence of overt prejudice, but where membership of a particular group is perceived as carrying information about skills or productivity i.e. males are more likely to be hired due to females being more likely to quit/go on maternity leave due to pregnancy
explain statistical discrimination when whites have a higher average score
i. two parallel upward sloping curves depicting the relationship between wages and test scores
ii. if non-white workers on average score lower than white workers, a white worker who scored T* earns more than a non-white worker with the same score
explain statistical discrimination when the test is a better predictor for white workers
i. white curve is steeper with the non-white curve intercepting it, depicting the relationship between wages and test scores
ii. if the test is a better predictor of productivity for white workers, high scoring whites earn more than high scoring non-whites and low scoring whites earn less than low scoring non-whites
ß: wage returns to schooling. if employers value both schooling equally then the betas would be equal
alpha : would be equal if employers value the skills of males and females without any schooling equally
raw mean difference: split into difference due to discrimination, reflecting differences in prices which can arise if the employer values male schooling more (betas) or if they pay males more for any level of schooling (alphas)
the second component is the difference due to characteristics, i.e. schooling. if male and female schooling was identical then it would tend to 0
in practice, S is replaced by a some vector, which tries to capture all observable differences in skill etc that might influenced productivity
any important factor missing from this vector goes to the discrimination coefficient
so if one group has poorer skills and this is observed by the employer, it will be characterised as discrimination in a regression when in fact it is not.
pre-market factors, such as poorer access to schooling may be important as they lead to wage differentials which will not be picked up by this approach as discrimination.
Oaxaca decomposition ignores that similarly aged men and women have different labour market histories - discontinuity in women's LM participation may explain the gap
human capital will deteriorate for women while they are on leave, so they are better off entering occupations in which their skills do not deteriorate