un'analisi molto puntuale sull'uso dei "talenti" neri e donne nella crescita USA IntroductionOver the last 50 years, there has been a remarkable convergence in the occupational
distribution between white men, women, and blacks. For example, in 1960, 94 percent
of doctors and lawyers were white men. By 2008, the fraction was just 62 percent.
Similar changes occurred throughout the economy during the last fifty years, particularly
among highly-skilled occupations.1
A large literature attempts to explain why white men differ in their occupational
distribution relative to women and blacks and why those differences have been changing
over time. Yet no formal study has assessed the effect of these changes on aggregate
productivity. Given that innate talent for many professions is unlikely to differ across
groups, the occupational distribution in 1960 suggests that a substantial pool of innately
talented blacks and women were not pursuing their comparative advantage. The
resulting misallocation of talent could potentially have important effects on aggregate
productivity.2
This paper measures the aggregate productivity effects of the potential misallocation of
talent among women and blacks from 1960 to 2008. To do this, we examine
the differences in labor market outcomes between race and gender groups through
the prism of a Roy (1951) model of occupational choice. While it is not our goal to
causally identify any specific friction that explains differences in occupational sorting,
our model is broad enough to encompass many of the common explanations highlighted
in the literature. Specifically, we begin by assuming that every person is born
with a range of talents across all possible occupations and chooses the occupation
with the highest return. Differences in the occupational distribution between men and
women can be driven by differences in the distribution of talent between groups.
Rendall (2010), for example, shows that brawn-intensive occupations (such as
construction) in the U.S. are dominated by men, and that changes in the returns to brawn- vs.
brain-intensive occupations can explain changes in the occupational distribution of
women vs. men since 1960; see also Pitt, Rosenzweig and Hassan (2012). Related,
Goldin and Katz (2002) and Bertrand, Goldin and Katz (2010) provide evidence that
innovations in contraception and increased labor market flexibility for
women had important effects on the occupational choices of women.However, we
also allow other forces to play a role in explaining differences in occupational
sorting. Consider the world that Supreme Court Justice Sandra Day O’Connor
faced when she graduated from Stanford Law School in 1952. Despite being ranked
third in her class, the only private sector job she could get immediately after graduating
was as a legal secretary (Biskupic, 2006). Such barriers might explain why white
men dominated the legal profession at that time. And the fact that private law firms
are now more open to hiring talented female lawyers might explain why the share of
women in the legal profession has increased dramatically over the last fifty years.
Similarly, the Civil Rights movement of the 1960s was surely important for the changing
occupational distribution of blacks.3
To capture these forces, we make several changes to the canonical Roy framework.
First, we allow for the possibility that each group faces different occupational frictions
in the labor market. We model these frictions as a group/occupation-specific “tax”
on earnings that drives a wedge between a group’s marginal product in an occupation
and their take home pay. One interpretation of these “taxes” is that they represent
preference-based discrimination as in Becker (1957). For example, one reason why
private law firms would not hire Justice O’Connor is that the law firms’ partners (or their
customers) viewed the otherwise identical legal services provided by female lawyers as
somehow less valuable.4
Second, we allow for frictions in the acquisition of human capital. We model these
frictions as a group-specific tax for each occupation on the inputs into human capital
production. These human capital frictions could represent the fact that some groups
were restricted from elite higher education institutions, that black public schools are
underfunded relative to white public schools, that there are differences in prenatal or
early life health investments across groups, or that social forces steered certain groups
towards certain occupations.Finally, we allow for changes in the returns to skill across
occupations. If thesechanges are common to all groups, then they will not affect the
occupations of menand women differently. However, some technological changes may
be group-specific.The innovations related to contraception mentioned earlier are a prime
example. Skill biased technical change may have changed the occupational distribution
of womenrelative to men if women are relatively more endowed with skill.
In our augmented Roy model, all three forces — barriers to occupational choice,
relative ability across occupations, and relative returns to occupational skills —
will affect the occupational distribution. To make progress analytically, we follow McFadden
(1974) and Eaton and Kortum (2002) and assume that talent obeys an extreme value
distribution. This assumption gives us two key results. First, we get a closed-form
expression relating the share of a group in an occupation to the frictions faced by the
group in the occupation relative to frictions faced by the group in all occupations. Talent is misallocated only when the frictions differ across occupations. Second, the misallocation of talent due to the dispersion of occupational frictions lowers the averagewage of the group in all occupations. Larger barriers in an occupation lead to a selection effect in which only the most talented choose that occupation, and these two
forces exactly net out in the model. As a result, frictions specific to an occupation show
up in quantities rather than in wages in that occupation. These two results allow us
to back out the occupation-specific frictions for each group from data on occupational
shares and average wages. Using data from the decadal U.S. Censuses and the
American Community Surveys, we find that the dispersion of the occupational frictions faced
by women and blacks, and thus the misallocation of talent, decreased substantially over
the last fifty years.We close the model by introducing the demand for skills in each occupation,
wherethe price of skills is determined by the supply and demand for skills in each occupation.
In our general equilibrium Roy model, the observed changes in occupational choicerelative
wages, and aggregate productivity can be explained by a combination of our
three forces (occupational barriers, talent distribution, occupation-specific technical
change). We then use this framework to isolate the effect of improved allocation of
talent among women and blacks from 1960 to 2008.
We freely admit this calculation makes no allowance for model misspecification and
thus should be viewed as only an illustration of the potential magnitude of the effect of
declining occupational barriers. In addition, without further information, we cannot
disentangle the effect of labor market discrimination from that of frictions in the human capital
market. However, while only illustrative, this calculation captures forces
that a simple back-of-the-envelope calculation (based on changing wage gaps alone)
does not. First, our calculation isolates the potential effect of changes in occupational
barriers, whereas the observed changes in the wage gap are also driven by innovations
in sector-specific productivities, skill requirements, and demographics. Second, our
calculation hones in on the effect of talent misallocation which, in the model, is only
driven by the dispersion in occupational barriers. Third, our calculation incorporates
the impact of changes in occupational barriers on white men, whereas the wage gap
calculation does not. Our results imply that changes in occupational barriers facing blacks
and womencan potentially explain 15 to 20 percent of aggregate growth in output per
worker between 1960 and 2008. These estimates are 40 percent larger than what we
find witha simple back-of-the-envelope calculation. Furthermore, essentially all of the gain is
driven by the movement of women into high-skilled occupations. We infer that changes
in occupational barriers may have raised real wages by roughly 40% for white women,
60% for black women, and 45% for black men, but lowered them by about 5% for white
men. Again, that wages of white men may have suffered is one important reason why
a simple back-of-the-envelope calculation could underestimate the importance of
declining labor market frictions in explaining aggregate productivity growth.The paper
proceeds as follows. Section 2 lays out the basic model of occupationalchoice. In
Section 3, we provide micro evidence for one of the key predictions of oursorting
model. We then use our framework to measure the frictions in occupational choice
between blacks and women versus white men in Section 4. In Section 5, we
explore the macroeconomic consequences of the changes in occupational frictionsacross
groups. We offer some closing thoughts in the final section.
Nessun commento:
Posta un commento