Young Black Men Labor Supply
By Gonzalo Dona and Rene Zamarripa
Fig. 1 This figure shows unemployment for men ages 25 to 54 at each decennial Census, by race. The red line for blacks and the blue line for whites.
Fig. 2 this figure shows the ratio between the two lines in Fig. 1. A value of 1 means identical unemployment rates. A value of 2 means blacks are twice more likely than whites to be unemployed.
Introduction
The figures above show a well established historical development: the racial unemployment gap. Regretfully, it was not until the Census of 1940 that the modern definition of unemployment was used. Furthermore, numbers for unemployment are not available in the 1900 and 1920 Census. Nevertheless, the data above for 1910 and 1930 is not directly comparable to the data for 1940 - 2000.
In this paper we do three things:
First, we show that for a certain population the definition for labor force participation did not change significantly throughout the twentieth century. This is important because labor force participation is available in every Census from 1900 to 2000, and it means the definition of inactivity for the group is comparable too.
Second, we establish that the racial gap did not exist prior to the 1930 Census. This is critical to consider potential sources that may have created the gap.
Finally, we evaluate which historical event was most likely to have created the racial divergence. From this exercise we are driven to conclude that the migration to the North and the Great Depression probably did not produced the gap. Instead, we pose that it was more likely created by the New Deal. Because of racism, moving North increased black men's chances to receive relief jobs, creating welfare migration. Furthermore, because welfare had continuity it made it harder for people to migrate in the future in search of employment opportunities. We believe in a world without social assistance many would have migrated again when employment opportunities appeared far from them.
Why is this old subject important?
If the New Deal help create the racial unemployment gap it means we have to be much more careful with social assistance than we know to be. Until now, there is some question as to whether some welfare programs may create perverse incentives, but this in many ways is larger than that. Additionally, knowing what caused the racial divergence may be useful to consider solutions to it.
2. Unemployment v. Inactivity
Researchers over the years have determined that the racial unemployment gap dates from the early 1930s (Gilman, 1995; Welch, 1990; Vedder and Gallaway. 1992; Sundstrom, 1992; and Fairlie and Sundstrom, 1999). However, because of the lack of data between 1900 and 1940 the reason for the gap has not been determined yet.
It is unquestionable that unemployment itself cannot be used. Not only did the definition change, but the data is not available for critical years. However, labor force participation is available in every Census. Labor force participation gives us inactivity (its complement), but raises two questions.
First, if the definition of labor force participation changed (as unemployment did), then we may not have a number that is consistent throughout the century. This concern is alleviated by Moen (1988) who showed that for a specific population there was practically no difference before and after 1940. Specifically, he found that at most the definition change would create a bias of just 2% for men 25 to 54 years old. Furthermore, the bias would come from on of four sources: seasonal workers, new workers, inmates, and retirees. The first and last were rare at the time for men 25 to 54 years old, inmates can be identified, leaving seasonal workers are the most likely source of bias. The older Census would have an excess of these seasonal workers because of the question used to determine labor force participation. However, this is not observed in the data for black men in the age range.
Second, labor force participation (and inactivity) will only be useful if the same racial gap is created by the event that created the unemployment racial gap. Otherwise, we would not be able to proxy unemployment using inactivity. However, the correlation between unemployment and inactivity in the data is 0.93 and the table below shows how closely the two series follow each other.
Vedder and Gallaway (1992) concluded that there was no significant gap between white and blacks prior to the early 1930s. The inactivity data shows the same, except for maybe a small advantage for black men in 1910. Overall, inactivity data between 1900 and 1930, the conclusions on the racial unemployment gap of previous researchers, and the two data series from the 1930 to the 2000 Census, lead us to conclude that inactivity is a very good proxy for unemployment.
3. Who was Responsible
Because of the timing, sometime during the 1930s, there are three large historical events that could have produced the racial gap: the migration North of black people, the Great Depression, and the New Deal. Additionally, it is possible that a host of individually small changes could be responsible for it if they affected the black more than the white population and happened during the decade. The latter suspect is hard to investigate. Fortunately, at least two large studies of black Americans took place at the perfect moment. Myrdal (1944) and Drake and Cayton (1945) were writing in the late 1930s and early 1940s. Neither of them records any significant change in the lives of black people in America, other than the three events already alluded at, that could explain the creation of a racial divergence.
3.1 The Great Migration
After the emancipation, blacks were able to move cross-country more freely. The data show that many took advantage of this opportunity before 1900. Furthermore, the Census show fewer and fewer blacks living in the South (as percentage of the total) decade after decade from 1900 to 1970, with the exception of the 1940 Census that shows nearly no difference to 1930. Could have their migration created the racial gap?
Conceptually, this is unlikely. Migration can be undone by migrating back, which would make such a shocking development very improbable. Nevertheless, we explore the possibility empirically. If migration caused the racial unemployment gap then something important must have taken place prior to its creation, most likely between 1930 and 1940. The event would leave an important change between the decennial Census of those two years. The figure below shows migration out of the South for black men 25-54 years old.
Nothing in the figure above suggest migration could have created the racial divergence. It is observed that between 1930 and 1940 migration North paused briefly, probably as a result of the Great Depression. The crisis was relatively more acute in the North, which would explain why fewer blacks moved there during the 1930s. More importantly, if migration is bad for black unemployment numbers, an acceleration would be necessary to predict the creation of the racial gap; instead, we observe the opposite.
3.2. The Great Depression
An arguably more historical event for the 1930s was the great economic downturn known as the Great Depression. The effects of this crisis were large and unprecedented, and centered on the right decade. The biggest conceptual objection against the depression is that it passed. Temporary shocks normally produce temporary consequences. Nevertheless, in terms of its uniqueness and timing, the Great Depression may have caused the racial gap.
To determine whether the Great Depression is the reason for the racial divergence, we need to establish an experiment. White men are not the best comparison to black men to establish this because it is a well established fact that the depression hit the latter much harder (see Sundstrom, 1992). However, blacks can be compared to other blacks, if we can find two groups that were differently affected by the Great Depression.
Difference-in-difference method: If we identify two groups for whom a particular event had differential effects (for example, one was not affected and the other was), but that are otherwise identical in determinants of labor force participation, we can identify the effect of that particular event by comparing the difference in labor force participation before and after the event. Then, the design allow us to conclude that there is a causal relation (because the event produced an answer, not the other way around).
Richardson and Troost (2009) provide these groups. The authors show that the Great Depression hit the north half of Mississippi harder than it did the southern half. According to them, the hands-off stance of the St. Louis federal reserve (8th region) led to a significantly higher number of banks collapsing in the north half of the state. Furthermore, they show that this translated into a deeper economic recession for the northern half of the state. The other part of the state was under the influence of the Atlanta fed (6th region), and fare better during the recession.
The figure below compares black men living in areas controlled by the Atlanta federal reserve (6th region) with black men living under the control of the St. Louis federal reserve (8th region), for Mississippi and Tennesse. Tennesse is also divided between the same two federal reserve regions. The effect is only significant thirty years too late, for the 1970 Census. It does not support the view that the recession created the racial divergence.
3.3. The New Deal
Finally, another large historical event that took place at the right time was the New Deal. In addition to being a large historical event, it has the advantage of continuity after the 1930 all the way to the present day. Its effects need not go away because welfare never went away, the key conceptual difficulty to understanding the Great Depression as the chief cause.
But how to test for the New Deal as the chief cause? According to Myrdal (1944) the best way is probably to look at relief. On the subject he wrote "public relief has become one of the major Negro occupations". 50% of Northern blacks received relief, and 25% of their Southern brethren. Myrdal notes that blacks receive relief with more probability in the North than in the South, and in the cities than in the country (statement which the 1940 Census confirms). Even though the money for relief came from the federal government, it was administered locally by the States; which is why it was possible for blacks to be discriminated against in the South.
Therefore, if the New Deal had a hand in creating the racial unemployment gap, we should expect migration for the purpose of receiving relief to places were relief was more likely. It may be possible to identify this effect by comparing migrants during the 1920s to migrants during the 1930s.
For Census data released more than seventy years ago there are projects that connect individuals from one year to the next. Therefore, we can create a sample with individuals observed in 1920, 1930, and 1940. However, the linking of individual uses data that not always gets a unique match such as race, sex, name, and address. Furthermore, its failure to link is not completely random. For instance, migrants are harder to link than non-migrants because the address will never be the same for the former group. Furthermore, if migrants have a tendency to change their name (because they are running from someone for example) the matching will not succeed. In such a scenario, the matching rate can be different not only by migrant v. non-migrant, but also by races, and ages, and sex (indeed, the matching rates are worse for women because they change their name when they marry).
We reduce the opportunity for bias by focusing exclusively on blacks and exclusively on migrants. The only difference between them will be whether they migrated during the 1920s or during the 1930s. The bias can still be a problem if we miss a group of migrants that is particularly important for our analysis, but then our results will likely yield no significant differences between the early and late migrants.
If the New Deal created a different type of migrant, we should find important statistical differences between the 1920s and the 1930s migrants. The tables below appears to confirm our suspicions. Unemployment, joblessness and relief were significantly more likely in 1940 for the later migrants, those that may have migrated for relief instead of employment opportunities (the more traditional reason for migration, see Hawkins 1965).
Conclusion
We think the analysis in this paper tells a novel story about the racial unemployment gap. The New Deal contributed to turn the temporary shock of the Great Depression into a sustained effect by motivating people to migrate and depend on welfare in the long run. Thereafter, when opportunities appeared away from families dependent on welfare requiring them to migrate in order to take advantage of them, this social assistance would become a barrier for them.
We believe our conclusions can make sense of the theory of spatial mismatch of black workers (see Kain, 1992). Hawkins (1973) shows that they have migrated before to improve their employment opportunities, welfare could be the reason why they chose to deal with the problem of job opportunities differently during the second half of the twentieth century.
References
Drake, S. C., & Cayton, H. R. (1945). Black metropolis. Harcourt, Brace.
Fairlie, R. W., & Sundstrom, W. A. (1997). The racial unemployment gap in long-run perspective. The American Economic Review, 87(2), 306-310.
Gilman, H. J. (1965). Economic discrimination and unemployment. The American Economic Review, 55(5), 1077-1096.
Hawkins, Homer C, “Trends in Black Migration from 1863 to 1960,” Phylon (1960-), 1973, 34 (2), 140–152.
Kain, J. F. (1992). The spatial mismatch hypothesis: three decades later. Housing policy debate, 3(2), 371-460.
Moen, J. R. (1988). From Gainful Employment to Labor Force: Definitions and a New Estimate of Work Rates of American Males, 1860 to 1980. Historical Methods: A Journal of Quantitative and Interdisciplinary History, 21(4), 149-159.
Myrdal, Gunnar et al., \An American dilemma; the Negro problem and modern democracy.(2 vols.).," 1944.
Richardson, G., & Troost, W. (2009). Monetary intervention mitigated banking panics during the great depression: quasi-experimental evidence from a federal reserve district border, 1929–1933. Journal of Political Economy, 117(6), 1031-1073.
Sundstrom, W. A. (1992). Last hired, first fired? Unemployment and urban black workers during the Great Depression. The Journal of Economic History, 52(2), 415-429.
Vedder, R. K., & Gallaway, L. (1992). Racial differences in unemployment in the United States, 1890–1990. The Journal of Economic History, 52(3), 696-702.
Welch, F. (1990). The employment of black men. Journal of labor Economics, 8(1, Part 2), S26-S74.