THE DECLINE IN BLACK TEENAGE LABOR FORCE PARTICIPATION IN THE SOUTH, 1900-1970: THE ROLE OF SCHOOLING

Robert A. Margo T. Aldrich Finegan
Vanderbilt University

In 1950, approximately 61 percent of southern black teenagers, ages 16 to 19, were participating in the labor force. By 1970 their participation rate had declined to 34 percent, or by 27 percentage points in two decades. Perhaps the most widely accepted explanation of the post-1950 decline in black teen participation emphasizes decade-specific shifts in labor demand (John Cogan; Gavin Wright). Between 1950 and 1960 mechanization of agriculture in the South accelerated significantly. In particular, the introduction of the mechanical cotton picker disproportionately reduced the demand for the labor of black male teenagers, who were concentrated in cotton production. Black teens displaced from southern cotton farms did not find ready work in the South's non-farm economy. Their failure to find non- farm jobs, according to Cogan, was caused by the federal minimum wage, primarily through increases in its coverage in the 1960s. The combination of mechanization and the minimum wage drastically reduced job opportunities for black male teens in the South, resulting in the sharp decline in labor force participation between 1950 and 1970.

We argue that the quantitative importance of demand-side shocks in accounting for the post-1950 decline in black teen participation has been overstated. A significant portion of the post-1950 decline appears to have been a continuation of a downward trend whose origins can be dated to the early 1900s. In turn, the long-term downward trend in participation was the mirror image of a long-term upward trend in black school enrollment. Each successive generation of southern black males was more likely to be in school between the ages of 16 and 19 than out of school and in the labor force. Further, the labor force participation rate of black male teens simultaneously enrolled in school declined over time, especially before 1950. Because the factors that were responsible for the long-run increases in black schooling had not yet run their course by mid- century, a post-1950s decline in labor force participation of black male teens in the South would have happened anyway, even if mechanization in cotton agriculture had not occurred in the 1950s and had the level and coverage of the minimum wage not been increased in the 1960s.

Census data provide the basis for constructing estimates of pre-1950 labor force participation rates of black male teens in the South. At the turn of the century fully 86 percent of southern black male teens were participating in the labor force. Their participation rate fell by 10 percentage points between 1900 and 1920, and by an additional 15 percentage points between 1920 and 1950 -- or an average rate of decline of almost 5 points per decade between 1900 and 1950. Fully 48 percent of the decline in participation between 1900 and 1970 took place before 1950.

Cogan attributed the post-1950 decline in black teen participation to the negative effects of mechanization on labor demand in cotton agriculture and of the minimum wage on labor demand off the farm. Between 1950 and 1970 the number of man- hours required to produce a bale of cotton declined by 84 hours, principally due to the widespread adoption of the mechanical cotton picker. Because cotton production remained constant over the period, the decline in labor requirements translated into a one-for-one reduction in labor demand. Cogan claimed that virtually the entire decline in employment of black teens in the South between 1950 and 1970 could be explained by the decline in their employment in agriculture. Although some black teens found jobs in manufacturing, their employment growth in retail trade slowed considerably in the 1960s, precisely when coverage of the minimum wage was extended over this industry. Based on state- level regressions, Cogan concluded that technical change and the minimum wage were equally important in explaining the decline in participation between 1950 and 1970.

Whatever the merit of Cogan's analysis of post-1950 trends, census data demonstrate that participation rates of southern black male teens were already falling between 1900 and 1950. Indeed, had post-1950s participation rates continued to decrease at the average decadal rate experienced between 1900 and 1950 (4.96 percentage points), the predicted participation rate in 1970 would have been 51 percent. The difference between this predicted 1970 rate and the actual 1950 rate accounts for 37 percent (= 9.9/27) of the decline in participation between 1950 and 1970. If the predicted 1970 rate is calculated on the basis of the average decadal rate of decline experienced between 1930 and 1950, the percent explained increases to 49 percent. Clearly, the rate of decline in black teen participation accelerated after 1950, but a reduction in participation could have been predicted (and was, by John Durand) on the basis of pre-1950 trends.

The diffusion of the mechanical cotton picker cannot explain the decline in black teen participation prior to 1950, because less than 1 percent of southern cotton acreage was mechanically harvested in 1950. Might agricultural mechanization or the minimum wage account for any of the pre-1950 decline in participation rates? Consider first the decline in participation before 1930. The minimum wage cannot possibly explain the pre- 1930 decline in participation because minimum wage legislation was enacted in 1938. The principal type of mechanization occurring in American agriculture prior to World War Two was the diffusion of tractors. The diffusion of tractors was far slower in the South than elsewhere in the country. In 1930, only 4 percent of southern farms used tractors. Yet the participation rate of black teens fell by 12 percentage points between 1900 and 1920, before virtually any tractors had been introduced into southern agriculture.

In the case of cotton production before the diffusion of mechanical cotton pickers, tractors did not reduce labor requirements in farm tasks that were intensive in the use of labor. Further, the diffusion of tractors in cotton production was comparatively slow, yet it was this sector of southern agriculture that was relatively intensive in the use of black teen labor. Nevertheless, it is possible that tractorization or the minimum wage might account for some of the decline in participation rates between 1930 and 1950. The proportion of southern farms using tractors rose to 26 percent by mid-century. Recent research demonstrates that tractorization was associated with a decrease in sharecropping and other forms of tenancy and with a redistribution of cotton production to the Southwest and California. Because tenant labor was frequently supplied in family groups, the decline in tenancy may have reduced, pari passu, the demand for the labor of black male teens on the farm, as might have the spatial reallocation of cotton production. Estimates of coverage rates of the minimum wage indicate that about 18 percent of employed southern black teens in 1950 were in covered employment. The establishment of the federal minimum wage in 1938 may have reduced job opportunities for black teens even before 1950, and hence made labor force participation less likely.

To evaluate the pre-1950 impact of tractorization and the minimum wage, we estimated a pooled time-series cross-section fixed-effects regression using state-level data for 1930 and 1950. The dependent variable is the participation rate of black male teens; the independent variables are the proportion of farms using tractors, the coverage rate of the minimum wage, and dummy variables for year (1950) and for states (the state dummies are the fixed effects). The effects of tractorization and the minimum wage are negative as hypothesized, but are economically and statistically insignificant: together, tractorization and the minimum wage can explain about 10 percent of the 1930-50 decline in participation. Put another way, the likelihood a black male teen would participate in the labor force fell between 1930 and 1950 throughout the South. Variations across states in the rate of decline were essentially unrelated to the diffusion of tractors or the extent of coverage of the minimum wage.

Our analysis might be criticized, however, because of alterations in labor force statistics over the period. In 1940 the definition of the labor force changed from the "gainful worker" concept to the "labor force week" concept. As Durand and others have pointed out, the pre-1940 participation rates are not, strictly speaking, comparable with the post-1940 estimates. Two responses to this criticism can be made. First, changes in the definition of the labor force have little bearing on the finding of a pre-1940 decline in participation, because the gainful worker definition was used in 1900 and 1930. Second, Durand calculated adjustment ratios in an attempt to make the pre- and post-1940 participation rates comparable. The force of Durand's adjustments was to reduce the size of the labor force prior to 1940. Had the 1940 census question been asked in 1900 and 1930, participation rates in both years would have been lower.

Unfortunately, data limitations prevented Durand (and us) from estimating decade-specific adjustment ratios for southern black male teens, but he did estimate an adjustment ratio for all males ages 14 to 19 in 1930: 0.9756. Multiplying our estimate of the 1930 participation rate for southern black male teens by Durand's adjustment ratio produces an adjusted participation rate of 72.1 percent, compared with the measured participation rate of 73.9 percent. Although there is reason to believe that the true adjustment ratio for southern black male teens was higher than 0.9756, even if we assume an adjustment ratio equal to the lowest estimated by Durand for any group of males (0.8953, ages 75 and over) there still was a decline in black teen participation between 1930 and 1950. In sum, it is doubtful that adjusting for definitional changes would reverse the decline in labor force participation by southern black male teens before 1950.

If the minimum wage, agricultural mechanization, or biases in census data cannot explain the majority of the pre-1950 decline in black teen participation, which factors can? Some clues are by examining participation rates by farm-nonfarm and school enrollment status, along with the percent on farms and percent enrolled in school. At the turn of the century over 90 percent of southern black male teens not enrolled in school were participating in the labor force. Participation rates of teens enrolled in school were lower than participation rates of those not in school, but were still relatively high by comparison with the rates experienced after World War Two. Despite a lower participation rate among in-school non-farm teens, the overall participation rate did not differ between the farm and non-farm population because school enrollment was more common among farm teens.

In 1950, the participation rate of farm teens not enrolled in school was essentially the same as in 1900. The participation rate of farm teens enrolled in school, however, declined by 20 percentage points and the proportion enrolled increased by 21 percentage points, between 1900 and 1950. Among non-farm teens, the percent enrolled in school increased by 34 percentage points and the participation rate of in-school black teens declined by 22 percentage points between 1900 and 1950 but, unlike the farm population, the participation rate of nonfarm teens not enrolled in school decreased by 12 percentage points. As a consequence of these changes, the overall participation rate of nonfarm teens fell 33 percentage points between 1900 and 1950, almost three times the decline experienced by farm teens. In short, the overall participation rate for nonfarm teens fell more than the rate for farm teens because of a larger rise in nonfarm school enrollment and a larger fall in participation among nonfarm teens not in school.

School enrollment rates in both sectors continued to climb between 1950 and 1970, at a quicker pace than before 1950. Among farm teens, participation rates of those enrolled in school fell by 34 percentage points, but the participation rate of non-farm teens enrolled in school remained stable at slightly over 20 percent. In contrast to the pre-1950 period, participation rates of farm teens not in school declined sharply, by 23 percentage points between 1950 and 1970. Participation rates of nonfarm teens not in school continued to fall as they had before 1950, with the majority of decline concentrated in the 1950s. Thus, reversing the patterns for the pre-1950 era, from 1950 to 1970 the overall participation rate for farm teens fell more than twice as much as the rate for nonfarm teens.

This analysis suggests that increasing school enrollments and reductions in labor force activity among those enrolled in school played a key role in generating the pre-1950 decline in black teen participation. The hypothesis is confirmed by counterfactual participation rates for 1900 and 1950 under various assumptions about school enrollment rates and participation conditional on enrollment status. Had enrollment rates (farm and nonfarm) in 1900 been equal to their 1950 values, the overall participation rate would have been 77 percent. Had enrollment rates and participation rates of in-school black teens in 1900 equalled their 1950 values, the overall participation rate would have been 68.2 percent. The difference between the actual 1900 participation rate and the latter counterfactual rate accounts for nearly three-quarters of the decline in participation between 1900 and 1950. By contrast changes in participation among those not enrolled were less important, explaining about 18 percent of the decrease in the overall participation rate before 1950.

Increasing school enrollments and decreasing participation rates of those enrolled in school continued to be important after 1950, but the relative significance of changes in participation rates of those not in school was greater after 1950 than before. Because nonfarm participation rates were lower than farm participation rates, the shift of population out of agriculture between 1950 and 1970 (35 percentage points) also helped to lower the overall participation rate.

The sharp decline after 1950 in participation rates of farm teens not in school is prima facie evidence that mechanization reduced the agricultural demand for teen labor. The shift of black labor out of agriculture, the increase in school enrollments, and the decrease in participation among those in school might also have been a response to technical change, except that downward trends in these variables began before the mechanization of cotton agriculture. What needs to be explained are the relatively low rate of school enrollment and high participation rate of black teens in school prior to 1950.

We hypothesize that changes before 1950 in school enrollment rates and in participation rates of in-school black teens were driven by increases in educational attainment of successive generations of southern black children. Historically black parents in the South were poor and heavily concentrated in rural areas. Because black parents were poor, because the returns to schooling were perceived to be low in agriculture, and because the labor of teenagers was valued on the family farm or in the nonfarm labor market, black children were not expected to remain in school at older ages. In 1910, for example, the peak age of black school enrollment in the South was age 11 (71 percent) and enrollment rates fell off sharply at age 14. The upshot was that southern blacks educated in the late nineteenth and early twentieth centuries typically completed very few years of schooling before reaching adulthood -- an average of about 5 years for the birth cohorts 1886 to 1905.

That completed years of schooling were so low can account for the low enrollment rate of black teens ca. 1900 but it cannot explain why the labor force participation of in-school black teens was so high in the early 20th century. A child whose education stopped at the sixth grade could, in theory, finish his schooling by age 12 or 13, and then enter the labor force. In fact, however, a significant fraction of southern black children suffered from high rates of age-in-grade retardation, and simply did not complete a given number of years of schooling in the customary amount of time. According to the 1940 census, for example, 47 percent of twelve year-old southern black males enrolled in school were attending the fourth grade or lower. Grade retardation in the pre-teen years had many causes, including adult poverty, and employment of young children in agriculture. But there is little doubt that a major factor retarding black educational achievement was the poor quality of segregated public schools attended by southern black children, especially the shorter school terms characteristic of the black schools.

As long as the demand for completed years of schooling was low, age-in-grade retardation at an early age was not a binding constraint. Completion of a given number of grades could be spread over several years and, if necessary, combined with labor force participation at later ages. Once the desired years of schooling reached modern norms, however, enrollment rates at later ages increased and participation of in-school black teens fell. A black teen struggling to finish his schooling with the sixth grade could work and still expect to achieve his goal before leaving home. But a black teen whose parents wished him to complete high school would, under the best of circumstances, normally graduate at age 18. He could ill afford to fall several grades behind, at an early or a later age, and his parents could best accomodate by keeping him out of the labor force during his teenage years. Over time, more black parents sought to have their children complete high school, and improvements in the quality of black schools lessened the extent of early age-in- grade retardation.

To determine the explanatory power of this hypothesis, we estimate the following two regressions:

(1) LFPRij = a0i + a1j + a2ELEMij + e1ij

(2) ENRij = b0i + b1j + b2ELEMij + e2ij

where:

LFPR: labor force participation rate of in-school teens ENR: percent enrolled ELEM: percent of enrolled black teens, ages 16-19, attending the elementary grades e : random error term

The subscript i refers to the urban, rural nonfarm, and rural farm sectors, and the subscript j refers to states. The dummy variables for sectors and for states control for a variety of factors that may have influenced participation and school enrollment but which cannot be measured directly at the sectoral level. The equations are estimated using weighted least squares; the weights are the population counts (black male teens, ages 16- 19). Data are drawn from the state volumes of the 1950 census.

The regression results confirm our hypotheses: the coefficient of ELEM is negative in the enrollment regression and positive in the labor force participation regression. Every ten percentage point decrease in the proportion of teens enrolled in the elementary grades increases the enrollment rate (holding the sector and state constant) by 2.8 percentage points and decreases the labor force participation rate of in-school teens by 6.6 percentage points.

We use the regression coefficients to calculate counterfactual enrollment and participation rates based on different assumptions about the value of ELEM. In 1950, the average value of ELEM for southern black male teens was 0.376. Had ELEM in 1950 been equal to its value in 1970 (8.5 percent) the estimated participation rate of southern black male teens in 1950 would have been 46.1 percent. The difference between the actual and counterfactual 1950 participation rates accounts for 59 percent of the decline in black teen participation between 1950 and 1970.

In 1940, the value of ELEM for southern black male teens was 51.2 percent. Comparable figures for the pre-1940 period are not available, but it is possible to estimate the value of ELEM from information on the age structure of enrollments and the number of pupils enrolled in high school. Using such data we estimate that ELEM was 58.9 percent in 1930. Thus the percent of black teens enrolled in the high school grades had been increasing for at least two decades prior to 1950, producing a decline in labor force participation that predated the mechanization of cotton agriculture and the minimum wage and which could have been expected to continue after 1950.

Racial and regional differences in black teen participation can also be attributed to differences in the value of ELEM. The value of ELEM for southern white male teens was 11.8 percent in 1950. Had there been no racial difference in ELEM in the South in 1950, we estimate that 48 percent of black male teens would have been participating in the labor force, 7 percentage points less than the actual participation rate of southern white male teens in 1950. The fact that black teens were less likely to participate in the labor market than white teens, holding ELEM constant, suggests that racial discrimination may have played a role in limiting labor market opportunities for southern blacks before 1950.

In 1950 the average value of ELEM for nonsouthern black male teens was 15.7 percent. Had the value of ELEM for southern black teens equalled its value outside the region, we predict a participation rate of exactly 50 percent. Since the participation rate of nonsouthern black teens in 1950 was 40.4 percent (compared with 60.8 percent in the South), the regional difference in ELEM explains 52.9 percent [=10.8/20.4] of the regional difference in participation rates.

The decline in labor force participation of southern black male teenagers after 1950 has frequently been attributed to the effects of agricultural mechanization and the minimum wage. The post-1950 decline in participation, however, originated before the diffusion of the mechanical cotton picker and the Fair Labor Standards Act of 1938. A key factor behind the long-term decline was the long-term increase in educational attainment of successive cohorts of southern black children. Once large numbers of black male teens desired to graduate from high school, low rates of school enrollment and high rates of in-school labor force participation were no longer feasible.

Although a significant part of the post-1950 decline in participation was the continuation of a long-term trend, census data also indicate that the decline in participation accelerated after 1950, and that this accelerated decline was reflected primarily in reductions in labor force activity of black teens who had left school. It may be that the post-1950 decline in participation among black teen dropouts was largely due to agricultural mechanization and the minimum wage. Black teen dropouts in the 1950s and 1960s, however, were increasingly undereducated compared with the average young worker. Recent work on the structure of earnings in the United States suggests that the overall demand for educated workers has risen more or less steadily in the post-World War Two period. The corresponding reduction in the demand for less-educated labor, including black teen dropouts, could have produced a decline in their participation, independent of specific effects of agricultural mechanization or the minimum wage.