RACE, RESIDENCE, AND UNDEREMPLOYMENT: 50 YEARS IN COMPARATIVE PERSPECTIVE,

Transcription

1 RACE, RESIDENCE, AND UNDEREMPLOYMENT: 50 YEARS IN COMPARATIVE PERSPECTIVE, Tim Slack, Louisiana State University Brian C. Thiede, Penn State University Leif Jensen, Penn State University Submitted for the Rural Poverty: Fifty Years After The People Left Behind Conference, Washington, DC, March 21-22, 2018.

2 Introduction Fifty years after President Lyndon B. Johnson declared a War on Poverty in the United States, much attention has been paid to assessing the legacy of that initiative (e.g., Bailey and Danziger 2013). Less well-known than the broader War on Poverty was a 1967 report produced by the President s National Advisory Commission on Rural Poverty entitled The People Left Behind. The report documented the especially high and persistent poverty that afflicted much of rural America, a pressing problem that the Commission noted was overshadowed by the more visible economic troubles of the nation s inner-cities. Among the report s key findings was that unemployment and underemployment are major problems in rural America (1967: x) and that rural racial/ethnic minorities suffer even more than low income whites from unemployment and underemployment (1967: 5). Fifty years later, this paper returns to these themes. Drawing on data from the March Current Population Surveys (CPS) from 1964 to 2017, we examine inequality in underemployment between metropolitan (metro) and nonmetropolitan (nonmetro) areas of the United States, paying special attention to differences between white, black, and Hispanic workers. In short, our results show that while five decades have passed since the Commission issued its report, marked differences in underemployment persist by race and residence nearly two decades into the 21 st Century. Our results suggest further that the Commission s goal to achieve the elimination of underemployment and low income of rural people (1967: vi) remains an unmet aspiration and a continuing obstacle to rural economic well-being. 1

3 Background Twenty years ago, Snipp (1996) urged rural sociologists to recognize the unique contribution the field stood to make in the race and ethnic relations literature. He noted that racial segregation is as much a reality in the American countryside as it is in the cities (Snipp 1996: 126). Whether black populations in the Delta or Black Belt, Latino colonias, or American Indian reservations, all are identifiable by high concentrations of racial/minority residents. Moreover, Snipp (1996) argued, a major characteristic of rural areas with concentrated minority populations is that they also stand among the very poorest places in American society. For example, the Economic Research Service (Farrigan and Parker 2017) shows that in , the average nonmetro black and American Indian individual resided in a county where 22 percent of the population was poor, while the average nonmetro Hispanic individual lived in an area where the county-level poverty rate stood at 18 percent. The corresponding numbers were percent among their metro counterparts. The risk of living in a high poverty area is even more pronounced for poor nonmetro minorities: 67.6 percent of poor nonmetro blacks (vs. 20.0% of poor metro blacks), 60.5 percent of poor nonmetro American Indians (vs. 13.9% of poor metro American Indians), and 39.6 percent of poor nonmetro Hispanics (vs. 18.4% of poor metro Hispanics) lived in a county with a poverty rate over 60 percent during the same time period (Farrigan and Parker 2017). Finally, Snipp (1996: 127) points out: Perhaps most profound is that reservations, colonias, and rural African-American communities, unlike other communities, share the experience of living in close proximity to the historical remnants of institutions explicitly created to conquer, oppress and maintain their subordinate position in society. A list of these institutions is easy to construct: labor contractors, immigration authorities, slavery, Jim Crow, sharecropping, plantation agriculture, the Bureau of Indian Affairs, and tribal police, to name 2

4 only a few. These institutions were first established in rural areas, and they have survived longer in rural areas than anywhere else. The People Left Behind Report (1967) did not ignore this reality at the middle of the 20 th Century, and made clear that a pronounced dimension of this issue was high and persistent underemployment among rural minorities. Research by Slack and Jensen (2002) picked up the historical record from 1968 to 1998 and showed that over the course of the latter half of the 20 th Century nonmetro minorities remained significantly more likely to be underemployed than their metro (or central city) counterparts even when controlling for a full range of worker characteristics. The aim of this paper is to examine underemployment by race/ethnicity and metro/nonmetro residence over the longer arc of history spanning 1964 to In doing so, we seek to bring focus to the experience of employment hardship among rural minorities in the 50 years following The People Left Behind (1967). Analytic Strategy Our analysis traces underemployment among whites, blacks, and Hispanics in metro and nonmetro areas from 1964 to We use descriptive statistics to identify important group differences and track trends over time. We then use logistic regression analysis to model the likelihood that a worker is underemployed controlling for an array of additional worker characteristics. Data and Methods To meet our objectives, we draw on data from the March Current Population Surveys (CPS) from 1964 to The CPS, sponsored by the U.S. Census Bureau and the U.S. Bureau 3

5 of Labor Statistics (BLS), is a monthly survey that serves as the primary source of labor force statistics for the population of the United States. The March survey, known as the Annual Social and Economic Supplement, provides a wealth of detailed information related to the social and economic characteristics of households, families, and individuals as of the interview date (U.S. Census Bureau 2017). Using concatenated (linked) data files, we examine employment hardship among all individuals aged 18 to 64 years those in the prime of their working years living in sampled households. The stratified cluster sampling design of the CPS requires the use of weights to produce reliable population estimates. In the analysis that follows, we use the CPS person weights divided by their means to yield weighted cases sizes that are approximately equal to the CPS sample size. Further, we limit our analysis to individuals in their fifth through eighth month in the survey to avoid problems with repeated observations owing the rotating sampling design of the CPS. We measure underemployment according to the Labor Utilization Framework (LUF) developed by Clogg and Sullivan (Clogg 1977; Clogg and Sullivan 1983; Sullivan 1978). This measure, designed specifically for use with the CPS, defines the operational states of underemployment as: Discouraged workers: individuals who would like to be employed but are currently not working and did not look for work in the past four weeks due to discouragement with their job prospects (official measures do not count these workers as in the labor force because they are neither employed nor looking for work). Unemployed workers: consistent with the official definition, individuals who are not employed but (a) have looked for work during the previous four weeks, or (b) are currently on layoff but expect to be called back to work. Low-hour workers (or involuntary part-time): consistent with the official definition of those who are working part-time for economic reasons (i.e., those employed less than 35 hours per week only because of slack work or being unable to find full-time employment). 4

6 Low-income workers: includes workers employed full-time (i.e., 35 or more hours per week) for at least 48 weeks in the previous year whose earnings were less than 125 percent of the individual poverty threshold. All other workers are defined as adequately employed, while those who are not employed and do not indicate a desire to be so are defined as not in the labor force. The latter are excluded from our analysis. Our other key measures include residence, which we define using the metropolitan (metro) and nonmetropolitan (nonmetro) dichotomy, and race/ethnicity, which we define as non- Hispanic white (white), non-hispanic black (black), Hispanic of any race (Hispanic), and non- Hispanic other race (other). The CPS did not begin reporting Hispanic ethnicity until 1971, so in periods prior to that time we cannot distinguish Hispanics from other groups. Most Hispanics in the early years of our analysis likely defined themselves as white. Results Table 1 shows the percentage of the U.S. labor force that was underemployed, both overall and by type, in 3-year periods from 1964 to Over this 50-year span, underemployment averaged 17.4 percent and ranged from a low of 14.1 percent in , following the great economic expansion related to the dotcom boom of the late 1990s, to a high of 22.6 percent in , following the deep recession at the beginning of that decade. Also apparent here is the high underemployment (22.1%) in in the wake of the Great Recession of 2007 to Figure 1 illustrates these time trends, showing a clear lagging countercyclical relationship between underemployment and the state of the macro economy. [Table 1 about here] 5

7 [Figure 1 about here] Table 1 also shows the prevalence of underemployment by type from 1964 to In most years, underemployment by low income is the most pronounced form of underemployment, though in periods following downturns in the economy unemployment can be seen to rise to assume that position, followed by low hours and discouragement, respectively. Over the 50 years examined here, low income and unemployment have averaged 6.2 percent, while low hour work has averaged 3.8 percent and discouragement 1.3 percent. Tables 2 and 3 present the percentage of the labor force underemployed, both overall and by type, in 3-year periods from 1964 to 2017 in nonmetro and metro areas, respectively. Over the last 50 years, nonmetro underemployment has averaged 20.6 percent or 1-in-5 nonmetro workers compared to a lower rate of 16.4 percent in metro areas. Figure 2 illustrates these trends over time as well as the difference between nonmetro-metro underemployment rates. Two points stand out. First, underemployment among the nonmetro labor force has exceeded that of the metro labor force at every time point over the last 50 years. Clearly, the chronic rural underemployment noted in The People Left Behind report has remained a pressing issue over the half century following its publication. However, a second takeaway is that the nonmetro-metro gap has been shrinking over the last 20 years or so, falling to within a percentage point in , before ticking upward again. [Tables 2 and 3 about here] [Figure 2 about here] Tables 2 and 3 also demonstrate nonmetro and metro area differences in underemployment by type from 1964 to In nonmetro areas, low income work stands out as the most prevalent form of underemployment, averaging 8.1 percent over the last 50 years. In 6

8 contrast, unemployment has been the most prevalent form of underemployment on average (6.0%) in metro areas over the same time frame. In both contexts, low hour work is the third most common type of underemployment, followed by discouragement. Importantly, when averaged across the last half century, low income work (8.1% vs. 5.5%), unemployment (6.6% vs. 6.0%), and low hour work (4.5% vs. 3.5%) are all more pronounced in nonmetro compared to metro areas, while the prevalence of discouraged workers is similar in both contexts (1.4% vs. 1.3%). Tables 4 and 5 show similar data for white workers in nonmetro and metro areas, respectively. Underemployment among white workers parallels the trends and differences outlined above. As illustrated in Figure 3, nonmetro white underemployment has consistently exceeded that of metro white workers across the last half century. Nonmetro white underemployment has averaged 18.5 percent over this time period compared to 13.6 percent among white workers in metro areas. The underemployment gap between white workers in nonmetro and metro areas has averaged 4.7 percent over the last 50 years, though again there is evidence of convergence over the last two decades. [Tables 4 and 5 about here] [Figure 3 about here] As was the case overall, between 1964 and 2017 white underemployment by low income (7.5% vs. 4.7%), unemployment (5.9% vs. 4.9%), and low hours (4.0% vs. 3.0%) have all been higher on average in nonmetro compared to metro areas over this time period, while the prevalence of discouragement has been relatively similar (1.1% vs. 1.0%). Tables 6 and 7 present the percentage of black workers who are underemployed, both overall and by type, in 3-year periods from 1964 to 2017 in nonmetro and metro areas, 7

9 respectively. A few key points stand out. First, underemployment among black workers in both nonmetro and metro contexts is consistently much higher than is true among their white counterparts, demonstrating the lasting power of racial stratification processes in the U.S. Second, nonmetro black workers are clearly disadvantaged relative to their metro counterparts. Over the last 50 years nonmetro black underemployment has averaged 36.4 percent more than 1-in-3 black workers in nonmetro areas and stood as high as half of the nonmetro black labor force in As illustrated in Figure 4, the gap between black workers in nonmetro and metro areas has averaged approximately 10 percentage points. Taken together, this is a truly staggering level of employment hardship. Underemployment among metro black workers, while lower compared to nonmetro blacks, is nonetheless high. Over the last 50 years nonmetro black underemployment has averaged 26.5 percent more than 1-in-4 black workers in metro areas with a high of 35.7 percent in [Tables 6 and 7 about here] [Figure 4 about here] Tables 6 and 7 also demonstrate differences in underemployment by type between nonmetro and metro black workers between 1964 and Among black workers in nonmetro areas underemployment by low income stands as the most prevalent form of underemployment on average (12.6%), while unemployment (11.2%) is the most prevalent form of employment hardship among their metro counterparts over this time period. What is also notable in comparing these tables is the persistent disadvantage on average of nonmetro black workers compared to their metro counterparts across every type of underemployment: low income (12.6% vs. 7.3%), unemployment (12.2% vs. 11.2%), low hours (8.1% vs. 4.9%) and discouragement (3.4% vs. 3.0%). 8

10 Tables 8 and 9 present the percentage of Hispanic workers underemployed, both overall and by type, in 3-year periods from 1964 to 2017 in nonmetro and metro areas, respectively. The data show a persistent Hispanic disadvantage relative to white workers in both residential settings, again illustrating the lasting power of racial/ethnic stratification processes in the U.S. Greater employment hardship among nonmetro Hispanics relative to their metro counterparts is also apparent over the period as a whole. Over the last half century, nonmetro Hispanic underemployment averaged 29.2 percent compared to 25.0 percent among Hispanic workers in metro areas. Moreover, nonmetro Hispanic underemployment has been higher than among metro Hispanics on average in terms of low income (10.9% vs. 8.9%), unemployment (9.4% vs. 8.7%), and low hours (7.3% vs. 5.9%). However, as illustrated in Figure 5, a unique feature of the Hispanic employment experience is that the historical nonmetro/metro disadvantage reversed itself following the Great Recession of After 2009, metro Hispanics have been subject to higher underemployment than is true among Hispanic workers in nonmetro areas. The greatest driver in this reversal of residential fortunes appears to be a substantial increase in metro Hispanic underemployment by low hours, and to a lesser degree unemployment and discouragement, relative to Hispanic workers in nonmetro areas in the wake of the Great Recession. [Tables 8 and 9 about here] [Figure 5 about here] Table 10 shows results from logistic regression models predicting underemployment (yes=1) among all workers in the U.S. as well as those in nonmetro areas and metro areas, specifically. In addition to the variables presented in Table 10, all models also control for age, age-squared, sex, marital status, educational attainment, industry of employment, region of the 9

11 country, and 3-year period effects. For the sake of parsimony, we focus our discussion here on the effects of race/ethnicity and metro/nonmetro residence. The model for the U.S. as a whole confirms the disadvantages by race and residence uncovered by our descriptive analysis. Over the last 50 years, the odds of being underemployed is 37 percent higher for nonmetro workers compared to their metro counterparts. In terms of race/ethnicity, the odds of being underemployed is 75 percent higher among blacks and 48 percent higher among Hispanics compared to whites. That these inequalities persist in the presence of the full range of controls underscores the independent effects of race and residence as axes of inequality in the U.S. labor market over the last half century. The residence-specific models further demonstrate the persistent employment hardship faced by non-white workers in both nonmetro and metro areas. What is especially notable, however, is the difference in the odds of underemployment for black and Hispanic workers compared to whites between nonmetro and metro contexts. Blacks are clearly subject to greater disadvantage in nonmetro areas, facing nearly twice the odds of being underemployed compared to their white counterparts. Hispanics, in contrast, experience greater disadvantage in metro areas, their odds of being underemployed being 48 percent higher than that of comparable white workers. Discussion In 1967, as part of the nation s broader War on Poverty, the President s National Advisory Commission on Rural Poverty produced a report entitled The People Left Behind. The Commission documented the especially high and persistent poverty that afflicted much of rural America, an issue it argued was overshadowed at mid-century by the more visible economic 10

12 troubles of the nation s inner-cities. One of the report s key findings was that unemployment and underemployment are major problems in rural America (1967: x) and that rural racial/ethnic minorities suffer even more than low income whites from unemployment and underemployment (1967: 5). Our analysis suggests the Commissions basic observations in this regard have persisted in the 50 years since it issued its report. More specifically, nonmetro workers have continued to face relatively greater employment hardship compared to their metro counterparts, both overall and by type of underemployment, and nonmetro racial/ethnic minorities have experienced the greatest employment disadvantages. This finding holds especially true for nonmetro black workers, among whom underemployment has remained persistently high in comparison to both white workers and their metro black counterparts. Hispanic workers, in contrast, have witnessed a reversal of residential economic fortunes over the last 50 years, with the problem of underemployment coming to be more pronounced in metro compared to nonmetro contexts. On balance, our analysis tells a story of both persistence and change. It also makes clear that the Commission s goal of achieving the elimination of underemployment and low income of rural people (1967: vi) remains very much an unmet challenge. 11

13 References Bailey, Martha J. and Sheldon H. Danziger Legacies of the War on Poverty. New York: Russell Sage Foundation. Clogg, Clifford C Measuring Underemployment: Demographic Indicators for the United States. New York, NY: Academic Press. Clogg, Clifford C. and Teresa A. Sullivan Labor Force Composition and Underemployment Trends, Social Indicators Research 12: Farrigan, Tracey and Timothy Parker The Concentration of Poverty is a Growing Rural Problem. Available at: december/concentration-of-poverty. Paternoster, Raymond, Robert Brame, Paul Mazerolle, and Alex Piquero Using the Correct Statistical Test for the Equality of Regression Coefficients. Criminology 36: President s National Advisory Committee on Rural Poverty The People Left Behind. Washington, D.C.: U.S. Government Printing Office. Slack, Tim and Leif Jensen Race, Ethnicity, and Underemployment in Nonmetropolitan America: A 30-Year Profile. Rural Sociology 67: Snipp, C. Matthew Understanding Race and Ethnicity in Rural America. Rural Sociology 61: Sullivan, Teresa A Marginal Workers, Marginal Jobs: Underutilization of the U.S. Work Force. Austin, TX: University of Texas Press. U.S. Census Bureau Annual Social and Economic Supplement (ASEC) of the Current Population Survey (CPS). Retrieved from: 12

14 Table 1. U.S. underemployment, Year Underemployed Low income Low hours Unemployed Discouraged % 7.5% 2.9% 4.4% 2.2%

15 Table 2. Nonmetro underemployment, Year Underemployed Low income Low hours Unemployed Discouraged % 11.4% 3.9% 4.6% 2.3%

16 Table 3. Metro underemployment, Year Underemployed Low income Low hours Unemployed Discouraged % 5.5% 2.4% 4.4% 2.2%

17 Table 4. Nonmetro white underemployment, Year Underemployed Low income Low hours Unemployed Discouraged % 9.8% 3.1% 4.2% 2.2%

18 Table 5. Metro white underemployment, Year Underemployed Low income Low hours Unemployed Discouraged % 4.8% 2.0% 3.9% 2.0%

19 Table 6. Nonmetro black underemployment, Year Underemployed Low income Low hours Unemployed Discouraged % 27.2% 11.6% 7.9% 4.1%

20 Table 7. Metro black underemployment, Year Underemployed Low income Low hours Unemployed Discouraged % 10.9% 5.2% 8.6% 3.7%

21 Table 8. Nonmetro Hispanic underemployment, Year Underemployed Low income Low hours Unemployed Discouraged % 5.1% 6.9% 6.9% 4.5%

22 Table 9. Metro Hispanic underemployment, Year Underemployed Low income Low hours Unemployed Discouraged % 5.6% 3.7% 7.6% 2.6%

23 Table 10. Logistic regression models predicting underemployment (yes=1), United States Nonmetro Metro Coef. S.E. OR Coef. S.E. OR Coef. S.E. OR Residence Metro (ref). Nonmetro 0.317*** Not identified 0.230*** Race/ethnicity White (ref.) Black 0.558*** *** *** Hispanic 0.392*** *** *** Other 0.262*** *** *** Notes: Ref.=reference group. Coef.=unstandardized coefficient. S.E.=robust standard error. OR=odds ratio. Models control for age, age-squared, sex, marital status, educational attainment, industry, region, and 3-year period effects. *p<.05; **p<.01; ***p<.001. Nonmetro and metro coefficients are significantly different at p<.001, using the test for equality of regression coefficients outlined in Paternoster et al (1998).

24 Figure 1. Percent underemployed

25 Nonmetro Metro Difference Figure 2. Percent underemployed in metro and nonmetro areas,

26 Nonmetro Metro Difference Figure 3. Percent white workers underemployed in metro and nonmetro areas,

27 Nonmetro Metro Difference Figure 4. Percent black workers underemployed in metro and nonmetro areas,

28 Nonmetro Metro Difference Figure 5. Percent Hispanic workers underemployed in metro and nonmetro areas,