Summary
In 1960, 94% of American doctors and lawyers were white men: by 2010, the fraction was just over 60%. Similar changes have occurred throughout the US economy over the last 50 years, particularly among high-skilled occupations.
Since the innate talent for a profession among members of a group is unlikely to change over time, the change in the distribution of occupations since 1960 suggests that, historically, many innately talented people were not pursuing their comparative advantage. The resulting misallocation of talent potentially had important aggregate consequences.
The fact that the occupational distribution of women and black men has converged towards that of white men suggests that discrimination could have diminished since the 1960s. What remains to be assessed is the extent to which this decline in discrimination raised aggregate growth over the past five or six decades.
New analysis of the labor market measures the effect of discrimination on job preferences and allocation outcomes. By integrating inequalities (such as discrimination), it is able to clarify the level of inefficiency within historical labor markets in the United States.
For example, the results imply that labor market frictions for women lawyers can lower the wages of women in all occupations, not just in law. If the labor market discriminates against women lawyers, only women with the most legal talent choose to become lawyers, and this selection effect works to raise the earnings of women who choose law.
Furthermore, if women are under-represented in an occupation because they have lower preferences for working in it, their wages will be higher in that occupation. Intuitively, women require higher wages to choose an occupation that they dislike.
The research also finds no systematic relationship between the share of white women relative to white men in an occupation in 1980 and the wage gap in that occupation. For example, young white women were 64 times more likely to work as secretaries and only 25% as likely to work as lawyers in 1980. But the wage gap between young white women and young white men among secretaries was nearly identical to the wage gap among lawyers. It seems as though discrimination is found prominently in quantities rather than prices.
The study also estimates the extent to which the reduction in labor market distortions has increased GDP. The results are striking: over the entire period, it accounts for approximately 40% of the overall growth in GDP per person. This reflects better allocation of talent across occupations, better-targeted human capital investments, and rising labor force participation of women in response to falling discrimination.
Overall, the results suggest that the decline in occupational barriers facing women and black men has been a very important source of growth to the US economy. Much of the subsequent productivity gains came from drawing women and black men into high-skilled occupations. Whether women increased their propensity to become lawyers and doctors because of declining labor market barriers, or because they started liking those occupations more, have broadly similar implications for growth.
Main article
In 1960, 94% of American doctors and lawyers were white men: by 2010, the fraction was just 62%. Similar changes in other high-skilled occupations have occurred throughout the US economy over the last 60 years. Given that the innate talent for these professions is not likely to be any different across groups, the change in the occupational distribution suggests that a substantial pool of innately talented women and black men in 1960 were not pursuing their comparative advantage. This article reports evidence on the impact of convergence in the occupational distribution on aggregate productivity between 1960 and 2010: 20-40% of growth in aggregate market output per person can be explained by the improved allocation of talent.Over the last 60 years, there has been a remarkable convergence in the occupational distribution between white men, women, and black men in the United States. Groups that have been historically discriminated against in terms of elite professions have started to capture a greater share of employment in these high-skill sectors.
In 1960, 94% of doctors and lawyers were white men: by 2010, the fraction was just over 60%. Similar changes occurred throughout the economy over that period, particularly among high-skilled occupations.
To date, there has been little formal research into the effect of these demographic changes on aggregate economic performance. Since the innate talent for a profession among members of a group is unlikely to change over time, the change in the occupational distribution since 1960 suggests that a substantial pool of innately talented women and black men were not pursuing their comparative advantage 60 years ago. The resulting misallocation of talent could have important aggregate consequences.
By way of an example, it is useful to 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 overall in her class, the only job she could get then was as a legal secretary. Such discriminatory barriers might explain why white men dominated the legal profession at that time.
Another possibility is underinvestment in human capital. For example, Harvard, Princeton, and Yale systematically discriminated in admissions against women and black men and Jewish applicants until the late 1960s. In the 1950s, public schools for black children in the American South were underfunded relative to schools for white children. Furthermore, strong social norms against women and black men working in high-skilled occupations in the 1960s may have discouraged additional investment in human capital.
The fact that the occupational distribution of women and black men has converged towards that of white men suggests that discrimination may have diminished since then. What remains to be assessed is the extent to which this decline in discrimination improved the allocation of talent – and how much this shift has raised aggregate growth over the past five decades.
A theoretical model of workers
In a standard Roy (1951) model, workers are endowed with ‘idiosyncratic talent’ in each occupation. In our study, two additional forces are included within the model:
- First, it is assumed that workers are heterogeneous in either their talent or their idiosyncratic preferences over occupations.
- Second, forces that distort the allocation of workers across occupations are included within the model. These include discrimination in the labor market, barriers to human capital accumulation, and group-specific social norms.
Individuals invest in their human capital and choose an occupation in an initial ‘pre-period’, after which they work in the chosen market occupation or home sector for three working lifecycle periods (‘young’, ‘middle’, and ‘old’). It is also assumed that human capital investments and the choice of occupation are fixed after the pre-period.
The model assumes that a person’s skills are independent draws from a particular probability distribution. The independence assumption makes the analysis tractable, but it does have the strong implication that a person’s talent for a particular occupation (for example, law) is unrelated to her comparative advantage across other occupations.
Given an occupational choice, the occupational wage, and idiosyncratic ability in the occupation, the individual chooses consumption in each period (including the initial pre-period), with the aim of maximizing lifetime utility. Because individuals expect the same conditions in future periods as in the first period, expected lifetime income is proportional to income in the first period.
Human capital barriers or labor market discrimination lower indirect utility for an individual from a particular group when choosing their occupation.
Occupational choice, worker quality, and wages
In our model the occupational choice problem reduces to picking the occupation that delivers the highest value of lifetime utility. Because heterogeneity is drawn from an extreme value distribution, the highest utility can also be characterized by an extreme value distribution – a result reminiscent of McFadden (1974) and Eaton and Kortum (2002).
For individuals with heterogeneous abilities, sorting affects the average quality of workers in an occupation for each group. For individuals with heterogeneous preferences, sorting has no effect on the average quality of workers in an occupation.
While sorting on productivities or sorting on preferences have similar effects on the number of individuals from a given group working in a given occupation, they have different effects on occupational wages. Average worker quality in an occupation is a weighted average of the quality of workers who sort on ability and those who sort on preferences.
When all individuals possess heterogeneous abilities, average quality is inversely related to the share of the group working in the occupation. This result captures the selection effect.
For example, the model predicts that if the labor market discriminated against women lawyers in 1960, only the most talented women lawyers would have chosen to work in this occupation. And if the barriers faced by women lawyers declined after 1960, less talented women lawyers would move into the legal profession and thus lower the average quality of women lawyers.
Conversely, in 1960, the average quality of white men lawyers would have been lower in the presence of labor market discrimination against women and black men. At the other extreme (all workers sort on preferences), this selection effect is absent.
These are the only two forces that generate differences in wages across occupations for the young when individuals sort completely on talent. Average earnings are no higher in occupations where a group faces less discrimination in the labor market, lower frictions in human capital attainment, a higher wage per efficiency unit, or where the group has more talent in the sector. The reason is that each of these factors leads lower quality workers to enter those jobs.
The composition effect would not be present if selection were driven by forces other than occupational ability. When all workers select based on idiosyncratic preferences, then selection affects the average utility of workers in an occupation but has no effect on average ability. It follows that, in this case, there is no quality offset due to selection.
The general point is that the wage gap is not a robust measure of the frictions (discrimination and barriers) faced by a group in a given occupation. The elasticity of the wage gap to occupational frictions depends on the share of individuals who sort on preferences instead of ability.
The data and sampling method
Our study uses data from the 1960, 1970, 1980, 1990, and 2000 decennial censuses and the 2010-2012 American Community Surveys. By restricting the sample to white men, white women, black men, and black women aged 25-54, the analysis is focused on individuals after they finish schooling and prior to retirement.
We create pseudo-panels by following synthetic cohorts from 1960 to 2010 over their lifecycle. We define three age periods within a cohort’s lifecycle: young (age 25-34), middle (age 35-44) and old (age 45-54).
We have information on eight cohorts for the time periods we study. We observe information at all three lifecycle points for four cohorts (the young in 1960, 1970, 1980 and 1990) and one or two lifecycle points for the remaining cohorts.
We define a person as either in the home sector or in the market sector based on their number of hours worked. We classify a person as being in the home sector if he or she is not currently employed or works fewer than ten hours per week. Individuals working more than 30 hours per week are classified as employed.
We measure earnings as the sum of labor, business, and farm income in the previous year. When calculating earnings, we restrict the sample to individuals who are currently working, who worked at least 48 weeks during the prior year, and who earned at least $1,000 (in 2010 dollars) in the previous year.
Results
The results from the framework we use suggest that there have been large reductions in occupational barriers concentrated in high-skilled occupations. We show that falling barriers explain roughly 40% of aggregate growth in market GDP per person and 20% of aggregate growth in market GDP per worker.
Figure 1 plots the standard deviation of the shares for women and black men relative to white men across market occupations since 1960. The sorting of women and black men has converged toward that of white men over time, suggesting that, overall, the allocation of talent has improved over the last five decades.
Figure 2 provides evidence on the selection margin, indicating that there is no systematic relationship between, on the one hand, the share of white women relative to white men in an occupation in 1980 and, on the other hand, the occupational wage gaps between young women and men in the same year.
For example, young white women were 64 times more likely to work as secretaries and one quarter as likely to work as lawyers in 1980. Yet the wage gap between young white women and young white men among secretaries was nearly identical to the gender wage gap among lawyers. Discrimination seems to show up prominently in quantities rather than in prices.
In terms of documenting the trend over time, Figure 3 plots the time series decomposition of growth in market GDP per person coming from changes in the effect of labor market discrimination. The top line in the figure shows market GDP per person in the data. The bottom line is the counterfactual path of market GDP if barriers were held fixed at their 1960 levels.
Over the entire period, falling distortions account for about 40% of the overall growth in GDP per person. This reflects better allocation of talent across occupations, better targeted human capital investments, and rising labor force participation of women in response to falling discrimination. Productivity growth would have been close to zero during the 1970s had it not been for the reduction in labor market barriers facing women and black men during that period.
Much of the productivity gains come from drawing women and black men into high-skilled occupations.
Conclusions
Our results suggest that the decline in occupational and human capital barriers to women and black men was a very important source of growth for the US economy, and that the slowing of progress may be one reason why growth has slowed down.
We realize that our model is only a ‘launching off point’ to address these important questions. We expect some of the assumptions we made for tractability to be relaxed as further research progresses.
This article summarizes ‘The Allocation of Talent and U.S. Economic Growth’ by Chang-Tai Hsieh, Erik Hurst, Charles I Jones, and Peter J Klenow, published in Econometrica in September 2019.
Chang-Tai Hsieh and Erik Hurst are at the University of Chicago. Charles I Jones and Peter J Klenow are at Stanford University.
