## Abstract

Historically in the United States, the public sector has served as an equalizing institution through the expansion of job opportunities for minority workers. This study examines whether the public sector continues to serve as an equalizing institution in the aftermath of the Great Recession. Using data from the Current Population Survey, I investigate changes in public sector employment between 2003 and 2013. My results point to a post-recession double disadvantage for black public sector workers: they are concentrated in a shrinking sector of the economy, and they are more likely than white and Hispanic public sector workers to experience job loss. These two trends are a historical break for the public sector labor market. I find that race and ethnicity gaps in public sector employment cannot be explained by differences in education, occupation, or any of the other measurable factors that are typically associated with employment. Among unemployed workers who most recently worked for the public sector, black women are the least likely to transition into private sector employment.

## Introduction

The Great Recession of 2007–2009 was, according to most indicators, the most destabilizing recession in America since the Great Depression. After the mortgage securitization bubble burst in 2007, the Great Recession erased more than one-half of the stock market capitalization (Grusky et al. 2011). The financial collapse led to waves of job loss and unemployment. In March 2007, the national unemployment rate was 4.4 %. By October 2009, unemployment had increased almost 6 percentage points to 10.1 % (Bureau of Labor Statistics 2016).

After recessions, government employment usually expands (Elsby et al. 2011). After the Great Recession, however, government employment contracted. Aside from the two-week shutdown of the federal government in 2013, most of the government layoffs were made by state and local governments (Stevenson and Langan 2011).1 Severe cuts were made to police forces, fire departments, social service agencies, and school districts (Dewan and Rich 2012). Despite the billions of dollars allocated to preserve jobs through the American Recovery and Reinvestment Act of 2009, total government employment fell 3 % between December 2008 and December 2013. Similar to previous recessions, the Great Recession led to a decrease in sales, income, and property tax receipts (Gordon 2012). The drop in total tax revenue after the Great Recession, however, was especially deep and long-lasting. Compared with earlier recessions, workers stayed unemployed longer (Kroft et al. 2014). As a result of heightened demand for unemployment benefits and other government subsidies, budgets at both the federal and state level were stretched thin.

Political forces also contributed to the contraction of public sector employment. After the 2010 midterm election, the Tea Party and its supporters were vocal about their intent to cut public sector employee benefits and slash public spending (Skocpol and Williamson 2012). With tax revenues in free fall, political constraints against raising taxes, and a growing number of high-profile political attacks against public sector employees (including those led by governors in Wisconsin and Indiana), many states and municipalities resorted to drastic measures. In 2011, approximately 40 % of state and local governments reported layoffs (Center for State and Local Government Excellence 2012).

The effects of public sector decline will be uneven simply because women—particularly black women—are overrepresented in the public sector. In 2010, when state budget shortfalls reached their peak, 17 % of black women in the Current Population Survey (CPS) worked in the public sector, compared with 15 % of white women, 13 % of black men, and 12 % of white men (Oliff et al. 2012).2 Yet, evidence also suggests that inequality increased within the public sector. Among workers in the CPS who reported that their current or most recent job was in the public sector, there was a surge in the black-white unemployment gap between 2008 and 2011. Among workers in 2008 whose current or most recent job was in the public sector, 3 % of black public sector workers were unemployed compared with 2 % of white public sector workers. In 2011, those numbers grew to 8 % and 3 %, respectively.

What accounts for these new disparities among government workers? How does the public sector compare with the private sector in the post-recession era of public sector decline? This study considers three hypotheses. First, I test whether public sector employment inequality reflects compositional differences in education and occupation. Second, I test the hypothesis that employment stratification in the public sector is becoming increasingly similar to that in the private sector. Finally, I examine whether public sector whites, once unemployed, are more likely to find private sector employment. The dynamics of public sector stratification are important for at least two reasons. First, social scientists consider the public sector to be a major source of economic mobility for black workers (Parks 2011; Zipp 1994). Understanding public sector decline should inform debates about between-race and within-race inequality in the United States. Second, by clarifying the link between public sector decline and racial inequality, this analysis advances the literature on the social and economic consequences of the Great Recession (Grusky et al. 2011).

## Theoretical Framework

During the latter half of the twentieth century, the public sector provided an employment boom for groups that had been historically underrepresented in the labor market. Between 1961 and 1965, blacks gained 28 % of new positions in the federal government despite the fact that they made up only a little more than 10 % of the U.S. population (Krislov 1967). The share of female government workers rose by nearly 70 % between 1964 and 1974, and by another 28 % by 1981 (Abramovitz 2012; U.S. Department of Labor Women’s Bureau 1975, 1983). Since 1960, the proportion of blacks working for the government has surpassed the proportion of blacks in the population (Hellriegel and Short 1972; Pitts 2011).

Both political and structural reasons for the overrepresentation of blacks and women in government jobs can be identified. In the decades following World War II, a series of executive orders and court decisions established equal opportunity employment procedures for government workers. In addition to building political pressure to enforce equal opportunity in the public sector, a structural component expanded public sector opportunities for blacks and women: demand for government labor expanded significantly during World War II. As a result, the overall number of public sector jobs increased sharply (Grandjean 1981; Hellriegel and Short 1972; Krislov 1967). With a large supply of jobs and a small supply of workers, government managers could change the composition of the public sector workforce without having to displace white men (Krislov 1967). The number of federal employees peaked again during the Vietnam War (Caplow et al. 2001). After the mid-1980s, the number of federal employees declined as the federal government increased the amount of work outsourced to contractors. The number of state and local jobs, however, continued to increase through the late 2000s (U.S. Census Bureau 2012).

The public sector has provided blacks and women with opportunities for good employment. Until the recent past, working for the government often meant having a pension, long-term job security, and regularly scheduled opportunities for upward mobility. Compared with the private sector, the public sector has offered black and female workers better pay, job stability, and more professional and managerial opportunities (Blank 1985, 1994; Carrington et al. 1996; Hollister 2011; Hout 1984; Pitts 2011; Smith 1977).3

However, between 2007 and 2001, after the Great Recession, state and local governments eliminated approximately 765,000 jobs (Cooper et al. 2012). Of those state and local government workers who lost their jobs, Cooper et al. found that women and African Americans represented approximately 70 % and 20 %, respectively, of those losses. Whether the public sector will continue to serve as an equalizing institution depends on the mechanism driving recent changes in public sector inequality. This study considers three hypotheses. First, the black-white gap in public sector employment might be partly accounted for by educational and occupational disparities among public sector employees. Faced with sudden pressure to downsize, public sector managers might try to protect high-skill workers who would be difficult to replace after tax revenues and personnel budgets start to rebound. Data from the CPS suggest that educational credentials reduce the likelihood of entering unemployment from both the private and the public sectors. Among public sector teachers (roughly one-third of the female public sector labor force), holding a master’s degree significantly decreases the odds of being laid off (Goldhaber and Theobald 2013). In the CPS, 45 % of white teachers hold a master’s degree or higher, compared with just 36 % of black teachers.

Independent of education, occupational disparities might partially explain public sector employment inequality. If black women, for example, are concentrated in the type of public sector jobs that tend to be eliminated or scaled back during a budget crisis—net of their individual levels of human capital—then their employment disadvantage may be linked to occupational sorting (Tomaskovic-Devey 1993). Since the early 1940s, the public sector has been a major source of clerical employment for black women (King 2003). Data from the CPS indicate that these positions may have been targeted during the recent downsizing of the public sector. Of the current and former government employees in the CPS between 2009 and 2013, those with an administrative or secretarial occupation are more likely to be unemployed.

My second hypothesis is that private and public sector levels of employment inequality are converging. Political accounts of black-white inequality point to specific policies and institutional interventions—rather than differences in education or occupation—as the source of contemporary race disparities. Sites and Parks (2011:62) argued that most of the twentieth century reduction in black-white employment and income inequality can be attributed to political forces, such as those that gave rise to civil rights policies in the 1960s and 1970s. Using data from the Panel Study of Income Dynamics (PSID) between 1985 and 2007, Wilson et al. (2013) described widening racial gaps in the incidence, determinants, and timing of downward mobility (including job loss) among public sector workers. Wilson et al. attributed their findings to the expansion of at-will hiring and the dismantling of bureaucratic employment protections beginning in the early 2000s. If private sector mechanisms of stratification are widely adopted in the public sector, then differences in public and private sector levels of employment inequality should be approaching zero.

Above and beyond compositional effects and political transformations, inequality in public sector employment may be exacerbated by race differences in the probability of finding private sector employment. Wilson et al. (2013) reported that in the decade prior to the Great Recession, white men were more likely than black men to leave the public sector for the private sector. My third hypothesis is that public sector whites will be insulated against public sector decline because they are more likely to find work in the private sector.

## Data and Methods

I examine stratification in employment and unemployment using data from the merged outgoing rotation group of the Current Population Survey (CPS-MORG). The CPS, the source of the official U.S. monthly unemployment rate, is a monthly survey of approximately 60,000 households conducted by the U.S. Census Bureau for the Bureau of Labor Statistics (BLS). I use the 2003–2013 CPS MORG files (prior to 2003, there were major changes to the occupation scheme). I use the MORG files of the CPS instead of the CPS Annual Social and Economic Supplement (ASEC) because the MORG samples are larger and because the ASEC samples may be subject to seasonal or recall bias given that they are administered yearly rather than monthly (Akerlof and Yellen 1985; Horvath 1982; Morgenstern and Barrett 1974). I restrict the sample to working-age men and women between the ages of 16 and 64.4 To determine whether black-white employment disparities are the result of differential rates of switching between the public and the private sector, I use longitudinal IPUMS-CPS data linked across all months between 2003 and 2013 (Drew et al. 2014; King et al. 2010). I restrict the longitudinal analysis to black and white workers between the ages of 25 and 55 in order to limit transitions associated with school enrollment and retirement.

The CPS is a monthly survey, although new households are not interviewed each month. Households that enter the CPS are typically interviewed for four months, ignored for eight months, and then interviewed again for four more months. Households in months 4 and 8 are considered the “outgoing rotation groups” because they are about to leave the observation sample (temporarily or permanently). I drop CPS respondents in their eighth interview to avoid observing respondents twice in one sample. Because the CPS models include occupation as a control, I further restrict the sample to those from whom the CPS collects occupation information. The CPS collects occupation information from those who report one of the four following conditions the week prior to the survey: (1) employed, (2) laid off/unemployed/looking and ever worked, (3) retired and worked within last 12 months, or (4) disabled and worked within last 12 months or otherwise not in the labor force and worked within last 12 months.5 I use the BLS definition of unemployment: not currently working, have actively looked for work in the prior four weeks, and currently available for work.

The race/ethnicity categories I use are non-Hispanic white, non-Hispanic black, Hispanic, and non-Hispanic other race. A class of worker variable is an indicator of whether the respondent’s current job (or most recent job if the respondent is unemployed or out of the labor force) is in the private or public sector. Public sector workers can be further disaggregated by type of public sector employment (federal, state, or local).

I estimate the models separately by gender and sector (public versus private) based on the assumption that pathways to employment and opportunity structures vary by gender and by sector. In the full model, I control for education, occupation, age, age squared, marital status, parental status, and veteran status.6 Educational attainment, age, and being married are all associated with a lower risk of unemployment for both men and women (Farber 2005; Johnson and Mommaerts 2011). In the CPS, men living with children tend to have lower unemployment rates than men not living with children. For women in the CPS, living with children increases the probability of unemployment. I include the control for veteran status because veterans tend to have higher unemployment than nonveterans (U.S. Bureau of Labor Statistics 2012; Kleykamp 2013).7

The dependent variable in all the employment models represents three outcomes: employed, unemployed, and not in the labor force. I include workers who are not in the labor force for two reasons. First, I do not want to exclude discouraged workers. According to the BLS (2016), “not in the labor force” includes discouraged workers (those who are not working, available for work, have looked for a job during the past year but not during the past four weeks). Second, because the sample is restricted to individuals with nonmissing occupation information, all individuals not in the labor force identify as either public or private sector workers (even if they are not actively looking for work).

Because my dependent variable consists of multiple unordered nominal categories, I estimate the outcome probability for individual i using a multinomial logit model:
$lnPry=mXiPry=nXi=Xiβm−βn,$
where Xi is the matrix of explanatory variables, and the β coefficients correspond to outcomes m and n.8 I include state, metro/nonmetro, year, and month fixed effects to control for observed and unobserved geographic and temporal factors that give rise to differential rates of employment and unemployment.

CPS-MORG sample sizes and descriptive statistics for sector and employment are presented in Table 1, which shows the composition of the labor force by sector and the disparities in employment status by sector, sex, and race. Among those workers who report that their current or most recent job was in the public or the private sector, black men have the highest rates of unemployment and the lowest rates of employment. Black women are clearly overrepresented among government employees: roughly one in five work in the public sector (mostly in state and local government positions). Within each gender group, Hispanics have lower public sector representation rates than both whites and blacks.

Figure 1 shows the black-white unemployment gap among working-age individuals who report that their current or (if not working) most recent job is in the public sector.9 In the post-recession public sector, black women appear to be among the workers most affected by public sector layoffs. The race gap in female public sector unemployment rates increased dramatically from less than a percentage point in 2008 to a peak of 5.5 percentage points in 2011. Compared with their male counterparts, both black women and white women in the public sector experienced a steeper rise in post-recession unemployment.

The gaps in unemployment do not capture the full extent of the gaps in labor force participation. If an unemployed individual stops looking for work, then the BLS considers that worker to be out of the labor force (i.e., a discouraged worker) and no longer unemployed. As the economy improves and more jobs become available, individuals start looking for work again, temporarily increasing the ranks of the unemployed. The prime-age employment ratio captures trends in labor force participation without being affected by temporary changes in job search behaviors.

Among individuals who reported that their current or most recent job is in the public sector, the female black-white employment gap showed a large increase (Fig. 2). Between 2009 and 2011, the employment rates for black women in the public sector declined steeply. The recovery for women also lagged. By 2013, employment rates for black men had rebounded to the pre-recession range. Employment rates for black women in the public sector bottomed out in 2011. In 2013, prime-age employment rates for black female public sector were still 4.6 percentage points lower than the 2008 peak.

As the public sector declines, so too will the effect of the public sector on employment inequality. However, had the public sector not contracted, black and white female employment would be measurably higher. Figure 3 shows the effect of public sector decline on overall employment rates. Holding public sector employment at the 2003 levels increases the 2013 employment rate for white and black women by 3.3 and 4.3 percentage points, respectively. The increase is smaller for men: 1.5 percentage points for white men and 2.2 percentage points for black men.

I next consider potential explanations for the aforementioned trends. I examine whether growing public sector employment inequality is a function of compositional differences in education and occupation, and I test whether employment patterns in the public sector are becoming increasingly similar to employment patterns in the private sector.

## Results

### Compositional Differences in Education and Occupation

In Tables 2 and 3, I present the results of multinomial regression models that test whether education or occupation can account for differences in the probability of unemployment among workers with a current job or (if not working) a recent job in the public sector.10

Among the public sector sample, blacks have higher odds of unemployment (vs. full-time employment) than whites or Hispanics. The second model takes into account differences in educational attainment. Including education only slightly reduces the differences between groups, suggesting that unemployment disparities by race are not attributable to differences in educational attainment. The third column of Tables 2 and 3 includes education and the 22-category occupation variable (see footnote 6 for a list of the occupations). Again, I find minor changes in the race effects. It is clear from Tables 2 and 3 that race disparities persist even after compositional differences in educational attainment and occupation are controlled for.

Of the men in the public sector sample, Hispanic men have unemployment probabilities that are statistically indistinguishable from white men. The pattern is similar for Hispanic women, until 2011. After the public sector started to contract, Hispanic women were significantly more likely than white women to be unemployed.

The coefficients for the control variables are generally consistent with prior research. Age is negatively associated with unemployment (at a decreasing rate). I conducted additional analyses (results available upon request) restricting the sample to workers ages 55–64 to look for race or ethnic disparities among public sector workers who are approaching retirement (and in some cases, pensions). After including all controls (the full model in Tables 2 and 3), the only statistically significant disparity for this restricted sample was between black and white women after the Great Recession: Hispanics and whites were statistically indistinguishable. Among women ages 55–64 in the public sector sample, black women had significantly higher probabilities of being unemployed compared to white women. That said, the age slope is consistently negative for all subgroups, before and after the Great Recession. In other words, the protective effect of seniority is evident in all models.

As expected, the odds of unemployment decrease with each education level. Being married reduces the odds of unemployment, although the effect is much stronger for men. Among women, veterans have significantly higher unemployment but only after the onset of the Great Recession. Being a parent with children at home increases the odds of unemployment for women. The parent effects may reflect a household specialization model, in which women’s household obligations take away time that could be spent looking for a job.

At first glance, the direction of the effects of working for state and local governments (vs. the federal government) appear to be inconsistent with media accounts of public sector layoffs being concentrated at the state and local level. However, less than one-fifth of public sector workers are employed by the federal government. Unemployment may be higher among individuals recently employed by the federal government simply because jobs with the federal government are relatively scarce.11 Given that most public sector employment is at the state or local level and that the federal government is located in a metro area (Washington, DC) with a high concentration of black workers, I also present a specification that omits all federal workers. Tables S3 and S4 in Online Resource 1 show results only for state and local public sector workers. Overall patterns of stratification are similar across all the tables, with some exceptions. Among public sector workers, roughly 25 % of men work for the federal government compared with just 15 % of women. As a result, the coefficients change more for men after federal workers are omitted. The odds of unemployment for black men increases slightly after I exclude federal workers.

To get a more complete picture of how race gaps in public sector employment have changed over time, I also analyze models with a race × year interaction (no year fixed effects). Figures 4 and 5 show the predicted probabilities of unemployment and employment from the models with the interaction. I generate the predicted probabilities by holding the control variables at their mean, thereby creating a hypothetical situation in which blacks and whites have the same distribution across the covariates (including education and occupation). Unemployment probabilities increased for all public sector workers as a result of the Great Recession; black workers, however, experienced a much larger increase than white workers. Considering the extent of the financial shock to the public sector, white workers appear to have been well protected.

Unemployment rates reflect only the population of active job-seekers. The trends in Fig. 4 could understate or overstate labor force participation, depending on whether race differences exist in the propensity to stop looking for a job.

Figure 5 shows predicted probabilities of employment among those who reported that their current or most recent job was in the public sector. The trends in Fig. 5 are consistent with those shown in Fig. 4. Among public sector workers, black workers experienced a much larger drop in employment, even after education, occupation, and other observable factors associated with employment are controlled for. The unemployment and employment probabilities together suggest that the public sector lost black workers—particularly black women—at a disproportionate rate following the Great Recession.

### Public Sector Versus the Private Sector

To what extent are the trends in public sector inequality consistent with theories about public/private sector convergence? Figure 6 shows the gaps between public and private sector predicted probabilities of employment (with all covariates held at their means). With a few exceptions, the controls for the private sector model are the same as the controls in the public sector model. All models used to generate the predicted probabilities include race × year interactions. Instead of controlling for type of government employment (federal, state, or local), the private sector model includes dummy variables for nonprofit employment, incorporated self-employment, and nonincorporated self-employment (all of which reduce the odds of unemployment). Model coefficients are presented in the tables in Online Resource 1. Although black workers in the public sector are at a disadvantage relative to white workers, private sector blacks have a much larger disadvantage relative to public sector blacks, even after the public sector started downsizing. Black men have the largest gap between public and private sector probabilities of employment.

### Differences in Public/Private Transition Rates

I next consider whether white workers leaving the public sector are more likely than their black and Hispanic counterparts to find private sector employment using the matched CPS monthly files. Figure 7 shows transition rates among unemployed workers who were most recently employed in the public sector. The four transition outcomes are (1) continued unemployment, (2) private sector employment, (3) public sector employment, and (4) labor force exit.

Among unemployed workers who were most recently employed in the public sector, black men are the least likely of all the race-sex groups to exit unemployment. White women are the most likely to find public sector employment within a month. Black women are the least likely to transition into private sector employment. Table S5 in Online Resource 1 shows the full matrix of transition probabilities for blacks, whites, and Hispanics. Compared with other employed public sector workers, white men are the least likely to enter unemployment. Compared with other unemployed workers in the public sector sample, black and Hispanic women are the most likely to exit the labor force.

## Discussion

This study examines the extent to which the public sector serves as an equalizing employment institution after the Great Recession. Blacks are overrepresented in a shrinking sector of the economy, and they are more likely than other public sector workers to experience unemployment. These two trends are a historical break for the public sector labor market.

I test three hypotheses related to these trends. I first consider explanations based on compositional differences in education and occupation. Model results show that even after controlling for education, occupation, and a host of other measurable factors associated with labor force attachment, significant racial and ethnic gaps exist in public sector employment probabilities, especially after the Great Recession (Tables 2 and 3, Figs. 46).

My second hypothesis is that public and private sector patterns of employment stratification converged after the Great Recession. Compared with the private sector, employment disparities are relatively narrow in the public sector (Fig. 6, and Tables S1 and S2 in Online Resource 1), even after the Great Recession, when nearly one-half of all state and local governments reported layoffs. Regardless of the state of the economy, racial and ethnic employment differences are attenuated in the public sector.

My third hypothesis is that public sector whites, after becoming unemployed, are more likely to find private sector employment. The linked IPUMS-CPS monthly files present a grim outlook for minority public sector workers. Compared with similarly situated whites and Hispanics, unemployed black workers from the public sector are the least likely to transition into private sector employment. Among black and Hispanic women from the public sector who are unemployed for at least one month, roughly one in five will make a complete exit from the labor force. White men in the public sector have advantageous employment outcomes partly because they have low rates of unemployment entry, low rates of labor force exit, and high rates of transitioning out of unemployment into both public and private sector opportunities.

This analysis is largely descriptive. More research and more data are needed to identify specific causes of the recent increase in public sector inequality. In some states, public sector inequality might be the result of anti–Affirmative Action ballot initiatives. In other states, the source may be legislative responses to budget deficits.12 Wisconsin, for example, implemented sharp and immediate funding cuts for municipalities in 2011. Black public sector employment started to rebound after 2011 in most states, but in Wisconsin, black public sector employment continued to plummet into 2012. Perhaps black workers are more likely to be laid off when the layoffs are triggered by a sudden and significant reduction in funding. When presented with more layoff decisions, managers have more opportunities to discriminate.

The protective effect of working in the public sector decreased substantially for minority workers after the Great Recession, whereas white workers were relatively insulated. The preceding analyses suggest that without a course correction, further efforts to dismantle the public sector will most likely have a negative effect on the workers who have historically gained the most from public sector employment.

## Acknowledgments

This research was supported by NICHD Training Grant 5T32HD007543 to the Center for Studies in Demography and Ecology at the University of Washington. Thanks to Jake Rosenfeld, Stew Tolnay, Kyle Crowder, and anonymous reviewers for insights on prior versions of this article.

## Notes

1

The majority of public sector workers are employed by local government; less than one-fifth are employed by the federal government.

2

In this article, I treat blacks, whites, and Hispanics as separate categories: “black” refers to non-Hispanic black, and “white” refers to non-Hispanic white. Within each gender group, Hispanics have lower public sector representation rates than both whites and blacks (see Table 1).

3

Not all public sector jobs are good jobs. In both the private and the public sectors, black workers have a lower median wage than whites, and women have a lower median wage than men (results of weekly earnings models available upon request). However, black-white and male-female income inequality is significantly lower in the public sector, even after public-private differences in occupation and education are controlled for (Gornick and Jacobs 1998; Grodsky and Pager 2001). Using data from the Panel Study of Income Dynamics (1976–1998), Heywood and Parent (2012) reported that although the raw black-white wage gap in the public sector is positive and significant, the gap disappears after observable characteristics (e.g., education, experience, region, occupation) are controlled for. Heywood and Parent attributed the absence of a positive wage gap between similarly situated white and black workers in the public sector to explicit pay scales and rules determining compensation.

4

Members of the military who reside in military barracks are excluded from the CPS. Because the CPS is designed to measure unemployment in the civilian labor force, members of the armed forces are not part of the universe for many employment-related questions. For these reasons, members of the armed forces are not included in this analysis.

5

Those who are not working, available for work, have looked for a job during the past year but not during the past four weeks are considered by the BLS to be discouraged workers. Approximately 77 % of the 3,885 discouraged workers in the CPS sample have missing occupation information and are therefore dropped from this analysis. Given that discouraged workers are disproportionately male and black, the CPS results most likely underestimate employment disadvantages among men and among blacks.

6

I use the CPS “two-digit” detail occupation recode. The categories are management, business, and financial operations; computer and mathematical science; architecture and engineering; life, physical, and social science occupations; legal occupations; education, training, and library occupations; arts, design, entertainment, sports, and media occupations; healthcare practitioner and technical occupations; healthcare support occupations; protective service occupations; food prep and serving occupations; building and grounds cleaning and maintenance; personal care and service; sales; office and administrative support; farming, fishing, and forestry; construction and extraction; installation, maintenance, and repair; production; and transportation and material moving. I use this occupation scheme because it identifies occupation groups that were disproportionately affected by the recent recession (e.g., administrative support and construction). With several hundred categories, the more detailed occupation scheme would yield cell counts that are too small to quantify race differences within sectors.

7

For public sector workers, veteran status should theoretically reduce the odds of unemployment because preference for veterans is commonly used in in the civil service hiring process (Ban 1995; Lewis 2013). I did not find that veteran status reduced the odds of unemployment in any of my models (even those restricted to the public sector).

8

The independence of irrelevant alternatives assumption of multinomial logit requires that an individual’s probability of being in one outcome category relative to another outcome category should not change if a third (irrelevant) category is added to or dropped from the analysis (e.g., there is a chance that an individual’s probability of voting for a Democrat versus a Republican will change if a third-party candidate is added to the ballot). Under the independence of irrelevant alternatives (IIA) assumption, no systematic change should be evident in the coefficients if one of the outcomes is excluded from the model. I performed a Hausman test for a violation of IIA, comparing the results from the full model and a model that excludes those who are not in the labor force. According to the results of the Hausman test (available upon request), I find no evidence that the IIA assumption is violated in this analysis.

9

I use the tile package in R to produce the figures in this analysis (Adolph 2016).

10

In models that combine public and private sector workers, I find statistically significant interactions between sector (public vs. private) and other predictors of employment, including education, age, and occupation (results available upon request). For this reason, I model employment separately for public and private sector workers. Online Resource 1 includes models predicting the odds of not being in the labor force, as well as results for the private sector. The coefficients in the tables in Online Resource 1 show that the race gaps in unemployment cannot be attributed to differential likelihoods of being out of the labor force.

11

Who is the typical federal, state, or local government employee? Approximately one-quarter of state employees and one-third of local government employees are teachers. Nearly one-third of federal employees work in office or administrative support occupations; one-half of federal administrative support workers are postal workers. Among all public sector employees, 50 % work for local governments, 30 % work for state governments, and nearly 20 % work for the federal government.

12

The CPS does not have enough minority public sector workers in every state to allow a state-level analysis. The large samples in the American Community Survey (ACS) are better suited for an analysis of state-level variation in public sector employment, but the ACS does not measure the timing of employment as precisely as the CPS. (ACS responses can relate to any weekly period throughout the year, whereas CPS responses refer to a particular week.) Figure S1 in Online Resource 1 shows state and local government employment rates in the ACS for those states that had at least 100 black public sector workers (men and women) in 2008.

## References

Abramovitz, M. (
2012
).
The feminization of austerity
.
New Labor Forum
,
21
,
32
41
. 10.4179/NLF.211.0000006.
2016
).
tile
Akerlof, G. A., & Yellen, J. L. (
1985
).
Unemployment through the filter of memory
.
Quarterly Journal of Economics
,
100
,
747
773
. 10.2307/1884377.
Ban, C. (
1995
).
How do public managers manage?
.
San Francisco, CA
:
Jossey-Bass Publshers
.
Blank, R. M. (
1985
).
An analysis of workers’ choice between employment in the public and private sectors
.
Industrial and Labor Relations Review
,
38
,
211
224
. 10.1177/001979398503800204.
Blank, R. M. (
1994
).
Public sector growth and labor market flexibility: The United States versus the United Kingdom
. In R. M. Blank (Ed.),
Social protection versus economic flexibility: Is there a trade-off?
(pp.
223
264
).
Cambridge, MA
:
National Bureau of Economic Research
.
Bureau of Labor Statistics
. (
2012
)
Employment situation of veterans—2011
.
Washington, DC
:
U.S. Department of Labor
.
Bureau of Labor Statistics
. (
2016
).
Employment statistics
Caplow, T., Hicks, L., & Wattenberg, B. J. (
2001
).
The first measured century: An illustrated guide to trends in America, 1900–2000
.
Washington, DC
:
AEI Press
.
Carrington, W. J., McCue, K., & Pierce, B. (
1996
).
Black/white wage convergence: The role of public sector wages and employment
.
Industrial and Labor Relations Review
,
49
,
456
471
. 10.1177/001979399604900305.
Center for State and Local Government Excellence
. (
2012
).
State and local government workforce: 2012 trends
.
Washington, DC
:
Center for State and Local Government Excellence
Cooper, D., Gable, M., & Austin, A. (
2012
).
The public-sector jobs crisis
(EPI Briefing Paper No. 339).
Washington, DC
:
Economic Policy Institute
.
Dewan, S., & Rich, M. (
2012
,
June
20
).
Public workers face new rash of layoffs, hurting recovery
.
The New York Times
Drew, J. A., Flood, S., & Warren, J. R. (
2014
).
Making full use of the longitudinal design of the Current Population Survey: Methods for linking records across 16 months
.
Journal of Economic and Social Measurement
,
39
,
121
144
.
Elsby, M. W. L., Hobijn, B., Sahin, A., & Valletta, R. G. (
2011
).
The labor market in the Great Recession—An update to September 2011
.
Brookings Papers on Economic Activity
, Fall,
353
371
.
Farber, H. S. (
2005
).
What do we know about job loss in the United States? Evidence from Displaced Worker Survey, 1984–2004
.
Economic Perspectives
,
2Q/2005
,
1
28
.
Goldhaber, D., & Theobald, R. (
2013
).
Managing the teacher workforce in austere times: The implications of teacher layoffs
.
Education Finance and Policy
,
8
,
494
527
. 10.1162/EDFP_a_00111.
Gordon, T. (
2012
).
State and local budgets and the Great Recession
.
Stanford, CA
:
Stanford Center on Poverty and Inequality
.
Gornick, J. C., & Jacobs, J. A. (
1998
).
Gender, the welfare state, and public employment: A comparative study of seven industrialized countries
.
American Sociological Review
,
63
,
688
710
. 10.2307/2657334.
Grandjean, B. D. (
1981
).
History and career in a bureaucratic labor market
.
American Journal of Sociology
,
86
,
1057
1092
. 10.1086/227354.
Grodsky, E., & Pager, D. (
2001
).
The structure of disadvantage: Individual and occupational determinants of the black-white wage gap
.
American Sociological Review
,
66
,
542
567
. 10.2307/3088922.
Grusky, D. B., Western, B., & Wimer, C. (
2011
).
The Great Recession
.
New York, NY
:
Russell Sage Foundation
.
Hellriegel, D., & Short, L. (
1972
).
Equal employment opportunity in the federal government: A comparative analysis
.
,
32
,
851
858
. 10.2307/974645.
Heywood, J. S., & Parent, D. (
2012
).
Performance pay and the white-black wage gap
.
Journal of Labor Economics
,
30
,
249
290
. 10.1086/663355.
Hollister, M. (
2011
).
Employment stability in the U.S. labor market: Rhetoric versus reality
.
Annual Review of Sociology
,
37
,
305
324
. 10.1146/annurev-soc-081309-150042.
Horvath, F. E. (
1982
).
Forgotten unemployment: Recall bias in retrospective data
.
Monthly Labor Review
,
150
,
40
43
.
Hout, M. (
1984
).
Occupational mobility of black men: 1962 to 1973
.
American Sociological Review
,
49
,
308
322
. 10.2307/2095276.
Johnson, R. W., & Mommaerts, C. (
2011
).
Age differences in job loss, job search, and reemployment
(Program on Retirement Discussion Paper 11–01).
Washington, DC
:
Urban Institute
.
King, M., Ruggles, S., Alexander, J. T., Flood, S., Genadek, K., Schroeder, M. B., . . . Vick, R. (
2010
).
Integrated Public Use Microdata Series, Current Population Survey: Version 3.0
Minneapolis, MN
:
Minnesota Population Center [producer and distributor]
.
King, M. C. (
2003
).
Black women’s breakthrough into clerical work: An occupational tipping model
. In E. Mutari, & D. M. Figartl (Eds.),
Women and the economy: A reader
(pp.
221
233
).
Armonk, NY
:
M.E. Sharpe
.
Kleykamp, M. (
2013
).
Unemployment, earnings, and enrollment among post 9/11 veterans
.
Social Science Research
,
42
,
836
851
. 10.1016/j.ssresearch.2012.12.017.
Krislov, S. (
1967
).
The Negro in federal employment: The quest for equal opportunity
.
Minneapolis
:
University of Minnesota Press
.
Kroft, K., Lange, F., Notowidigdo, M. J., & Katz, L. F. (
2014
).
Long-term unemployment and the Great Recession: The role of composition, duration dependence, and non-participation
(NBER Working Paper No. 20273).
Cambridge, MA
:
National Bureau of Economic Research
.
Lewis, G. B. (
2013
).
The impact of veterans’ preference on the composition and quality of the federal civil service
.
Journal of Public Administration Research and Theory
,
23
,
247
265
. 10.1093/jopart/mus029.
Morgenstern, R. D., & Barrett, N. S. (
1974
).
The retrospective bias in unemployment reporting by sex, race, and age
.
Journal of the American Statistical Association
,
69
,
355
357
. 10.1080/01621459.1974.10482953.
Oliff, P., Mai, C., & Palacios, V. (
2012
).
States continue to feel recession’s impact
(Report).
Washington, DC
:
Center on Budget and Policy Priorities
.
Parks, V. (
2011
).
Revisiting shibboleths of race and urban economy: Black employment in manufacturing and the public sector compared, Chicago 1950–2000
.
International Journal of Urban and Regional Research
,
35
,
110
129
. 10.1111/j.1468-2427.2010.00942.x.
Pitts, S. (
2011
).
Research brief: Black workers and the public sector
.
Berkeley
:
University of California, Berkeley Center for Labor Research and Education
.
Sites, W., & Parks, V. (
2011
).
What do we really know about racial inequality? Labor markets, politics, and the historical basis of black economic fortunes
.
Politics & Society
,
39
,
40
73
. 10.1177/0032329210394998.
Skocpol, T., & Williamson, V. (
2012
).
The Tea Party and the remaking of Republican conservatism
.
New York, NY
:
Oxford University Press
.
Smith, S. (
1977
).
Equal pay in the public sector: Fact or fantasy
.
Princeton, NJ
:
Princeton University
.
Stevenson, B., & Langan, A. (
2011
).
Comment and discussion [on “The labor market in the Great Recession—An update to September 2011”]
.
Brookings Papers on Economic Activity
, Fall,
372
384
.
1993
).
Gender and racial inequality at work: The sources and consequences of job segregation
.
Ithaca, NY
:
Cornell University Press
.
U.S. Census Bureau
. (
2012
).
Annual survey of public employment and payroll summary report: 2010
(Report G10-ASPEP).
Washington, DC
:
U.S. Department of Commerce
.
U.S. Department of Labor Women’s Bureau
. (
1975
).
1975 handbook on women workers
(Bulletin 297).
Washington, DC
:
U.S. Government Printing Office
.
U.S. Department of Labor Women’s Bureau
. (
1983
).
Time of change: 1983 handbook on women workers
(Bulletin 298).
Washington, DC
:
U.S. Government Printing Office
.
Wilson, G., Roscigno, V. J., & Huffman, M. L. (
2013
).
Public sector transformation, racial inequality and downward occupational mobility
.
Social Forces
,
91
,
975
1006
. 10.1093/sf/sos178.
Zipp, J. F. (
1994
).
Government employment and black-white earnings inequality, 1980–1990
.
Social Problems
,
41
,
363
382
. 10.2307/3096968.