Upgrade to remove ads
Only $35.99/year

Economic Applications - Discrimination in the labour market

Terms in this set (14)

What is taste based discrimination?
People discriminate as they do not like a certain group.
This is essentially what most non-economists think of when they
hear discrimination: One group does not like another group and
behaves accordingly.
Consequences differ depending on whether employers, co-workers or
customers are prejudiced.
What is the effect of employers having taste-based discrimination?
If employers do not like to interact with minority group, they behave as
if workers from that group were more expensive.
An employer who is prejudiced against foreigners would
behave as if the wage for foreign workers was WAGES(F) + D where d is a
measure of the strength of that employer's prejudice.
Employers will only hire foreign workers if
they are sufficiently cheaper, i.e WAGES(F) + D < WAGES(H).
Why might employer discrimination not necessarily lead to wage
differences between majority and minority group?
Minority workers
could work for non-prejudiced employers, while prejudiced employers
only hire majority workers.
Example: Say there are 100 workers and 100 employers each hiring
one worker. 10 of the workers are foreigners and 10 of the employers
are prejudiced against foreigners.In this example, foreign workers would have 90 non-prejudiced
employers to choose from.
We might not actually observe any
wage difference.
However, when might we see wage differences because of the employer?
Say we have 50 foreign workers and 60/100 prejudiced employers. In
this case, some foreign workers will end up working for prejudiced
employers and we would expect to see a wage difference.
How do we work out how discriminative employers are?
Let's say the 100 employers differ in their value of d, such that
d1 < d2 < ... < d99 < d100, i.e., we order them by the strength of
their prejudice.
The most prejudiced employer will go for a native as w
F + d100 will
be very large.
The second-most prejudiced employers will also go for a native as
w
F + d99 will also be very large.
On the other end, the least prejudiced employer will likely hire a
foreigner as w
F + d1 will be small.
In other words you will likely get sorting in the sense that natives are
selected by the most prejudiced employers and foreigners by the least
prejudiced employers.
What matters now for observed wage differences is the value of d of
the employers who end up hiring foreigners.
If all these employers are non-prejudiced, i.e., their d = 0, we will
not observe any wage difference.
If these employers are prejudiced, we should observe some wage
differences.
What is the effect of co-workers having taste-based discrimination?
Prejudiced workers will demand higher wages if they have to work
alongside members of the minority group.
They demand a wage equal to w + d if working alongside members
of the minority group and w if not.
The prediction is segregated workplaces: Employers will only
hire majority or minority workers (depending on who is cheaper
relative to productivity), but never both.
What is the effect of customers having taste-based discrimination?
Prejudiced customers don't like to interact with
members of minority group when purchasing goods or services. ⇒
Behave as if price is p + d instead of p.
Prejudiced native might still go to foreign barber if
significantly cheaper.
Prejudiced customers effectively makes is more expensive for firms to
hire minority workers into roles with customer contact.
The prediction is minority workers more likely to be found in back
offices and occupations without customer contact.
What is statistical discrimination?
Group membership reveals something about
productivity-relevant characteristics that cannot be observed on an
individual level.
Here we assume that no one has any
prejudice.
Firms just want to hire the most productive workers, but are unable
to observe all relevant characteristics.
Some of these unobservable characteristics will be correlated with
some observed group characteristic that firms use as a proxy.
What's an example of statistical discrimination?
Firms might be reluctant to hire women as they fear that they might
become pregnant and leave the company.
A specific individual's desire to have children and leave the job is
unobserved.
If the firm thinks that women are more likely to leave a job after
having children, they might be less willing to hire women in an age
in which they are likely to become pregnant.
Note that this is prediction is a bit different from taste
discrimination: Discrimination against women disappears when they
reach an age where the firm's child fear stops to matter
How can statistical discrimination be overcome?
by evidence that a
specific individual does not share a group's characteristics:
If firms are less willing to hire foreign workers because they fear
language difficulties, proving that someone speaks the respective
language perfectly will attenuate this form of discrimination.
Or if universities are less willing to take on students from worse schools
because they fear that these are of lower ability (even with the same
marks), some additional proof of academic aptitude will attenuate
these concerns.
Not every difference between two groups is due to
discrimination, what is an alternative explanation?
Preferences.
Easiest to
see with non-labour market discrimination - the fact that foreigners
are on average less likely to its of fish & chips could be because
chippy owners discriminate against them or it could be because they
don't like fish & chips.
ANOTHER EXAMPLE
Say a society believes the "proper place" for a woman is with her family at
home. If women are raised in that belief, they might end up with
preferences that make them prefer the life of a housewife over a
career. They might also invest in less education and end up being
less productive workers.Let's say we observe that women in that society have worse labour
market outcomes than men: Is the reason differences in productivity,
differences in preferences or is this some form of pre-labour market
discrimination?
How would we calculate wage differences between men and women, due to differences in productivity?
One way to do this is to run regressions of the form
wi = α + τ ∗ Femalei + βXi + ei
Where female is 1 for woman and Xi contains productivity information. τ measures the earnings difference between men and women after
controlling for differences in Xi.
We can see how τ changes when
we include additional elements in Xi
What are the problems with regression based?
We can always come up with more elements to include in X.
We can always think of some elements that shouldn't be in there,
e.g., differences in education can be caused by expected
discrimination in the labour market that makes it less worthwhile for
a discriminated group to obtain it.
Reasonably useful to figure out how much some of the elements in
X matter quantitatively, but generally not particularly convincing to
claim that all remaining differences are due to discrimination.
How is the validity of the test of sending two people (e.g., one man, one woman) with identical CVs and
trained to behave identically into job talks?
Hard to consider as evidence as not particularly reliable.
Also, for taste-based discrimination market outcome will depend on
the preferences of the marginal employer. ⇒ Just finding some
discriminating employers might not actually matter at all for market
outcomes if they aren't the ones hiring foreigners.
Sets found in the same folder