The geography of nonstandard employment across U.S. metropolitan areas

ABSTRACT

The U.S. nonstandard workforce remained at around 10% of the total employed population for the past decades, although the subnational levels reveal variation. Insufficient scholarly attention has been devoted to understanding its spatial distribution and associated causes. This paper addresses this gap by analyzing the contextual factors that help explain the geographic unevenness of the nonstandard workforce across U.S. metropolitan areas from 1997 to 2017. We find evidence that the urban context matters, but unevenly across arrangements and time. Three out of four of the nonstandard arrangements studied are more prevalent in metropolitan areas, while on-call workers are typically rural. Independent contractors are more concentrated in cities with higher fissuring, contrary to temporary and contracted out workers. Higher unemployment rates seem to push workers toward on-call arrangements, and inequality to temporary jobs. While the city effects change substantially over time, individual determinants are consistent.

KEYWORDS:

• Alternative work
• nonstandard work
• metropolitan areas
• workforce development

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. See, Kalleberg (2000) for a detailed discussion on the various concepts and operationalizations, and Liu and Kolenda (2012) for an overview of estimates prior to CWS.

2. The CPS is a national survey, sponsored by the Bureau of Census and the Bureau of Labor and Statistics, with roots in the 1930s. The CPS is carried every month and comprise the primary source of information of labor statistics for the U.S. population. There are several supplements to CPS, covering specific topics. The CWS is one of such supplements, introduced in 1995 to gather specific data on contingent and alternative (or nonstandard) work arrangements. It is the first national survey designed to specifically capture these workers, and was carried in February of the years 1995, 1997, 1999, 2001, and 2005, and in May of 2017.

3. McLaughlin and Coleman-Jensen (2008) define nonstandard workers as part-time workers, workers with varying hours and contingent workers, based on the description of Kalleberg (2000). Our work follows the definition of the CWS, based on the work Polivka (1996) and Polivka and Nardone (1989), who made a clear distinction of nonstandard and contingent workers as overlapping but different groups. In CWS, contingent work refers to jobs in which there is no explicit or implicit contract for long-term employment. Differently, alternative or nonstandard work correspond to the definitions already presented. A practical implication that follows from McLaughlin and Coleman-Jensen’s approach is that their object of study is mainly nonmetropolitan or rural, whereas ours in mainly urban.

4. This has been shown, for example, across industries (Boschma et al., 2013; Neffke et al., 2011), occupations (Farinha et al., 2019; Shutters et al., 2016), and technologies (Kogler et al., 2013).

5. This process has been described by several authors, but the fissuring as a concept was developed by Weil (2014).

6. We compared the metropolitan characteristics obtained from 2016 CPS with equivalent calculations from the American Community Survey (ACS), a more conventional source for local level analysis. Means tests did not reveal significant differences for most variables, except for unemployment rates and share of highly educated. The former is known to be higher in ACS than in CPS due to differences in sample design (Kromer & Howard, 2011). The latter may also originate from differences in the survey questionnaires, which provide the respondents with slightly different choices for educational attainment.

7. This is true for all years, except for 2005, as a change in metropolitan area definitions in the prior year made the 2004 data incompatible.

8. The industrial diversity index is calculated for each metropolitan area as: $\mathit{D}=(1-\sum _{\mathit{i}=1}^n{(\mathit{pi})}^2))\ast 100$, where pi is the share of industry i. This formula is equivalent to (1-HHI)*100, in which HHI stands for the Herfindahl-Hirschman Index (Hirschman, 1964). D ranges from 0 to 100, with higher numbers meaning higher industrial diversity. Studies that have adopted a similar strategy to capture industrial diversity include Chen (2020) and Ženka et al. (2015).

9. The weekly earnings ratio is a measure of inequality calculated as the ratio between the wages at the 90th and the 10th percentile in each metropolitan area and year. Higher ratios indicate higher inequality. When defining this variable, we trimmed the bottom 1.5 percentile of the distribution in every year, corresponding to weekly wages below the following thresholds: 49.50 nominal dollars in 1996, 52.00 in 1998, 60.00 in 2000, 64.00 in 2005, and 85.00 in 2016.

10. CPS is structured in rotation samples, and each month sample is split into eight subsamples or rotation groups. Households in each rotation group are interviewed consecutively for four months, leave the sample for eight months, and are interviewed for a second time in the following four months, when they are dropped from the sampe. CPS gathers detailed information on weekly earnings and union status of workers who are in months-in-sample 4 and 8. Therefore, these variables are only available for a subset of the sample in each month.

11. The SUBS data series is available from 1997 and, therefore, for that year it was not possible to use the lagged information from 1996. It is reasonable to assume, however, that average firm size would not change substantially from one year to the next. We assume then that average firm size between 1996 and 1997, as well as racial/ethnic diversity index between 2004 and 2005, remained relatively stable.

12. “One set of estimates that can be produced from CPS microdata files should be treated with caution. These are estimates for individual metropolitan areas. Although estimates for the larger areas such as New York, Los Angeles, and so forth, should be fairly accurate and valid for a multitude of uses, estimates for the smaller metropolitan areas (those with populations under 500,000) should be used with caution because of the relatively large sampling variability associated with these estimates.” (https://cps.ipums.org/cps-action/variables/METAREA#description_section)

13. We identified 123 large metropolitan areas in the years 1997–2001, 85 in the year 2005, and 104 in the year 2017, of which 74 were present in all years. Restricting the analysis to the 74 subset prevents us from confounding time effects with changes in the areas of study across samples. However, we note that the area consistency is imperfect and does not comprise a panel, as the metropolitan areas definitions change over time. Throughout the twenty-year period of our analysis, three distinct classifications were in place. In 1997, 1999, and 2001, CWS adopts the Office of Management and Budget (OMB) definition of June 30, 1993; in 2005, the definition employed was OMB’s June 30, 2003, and in it was OMB’s February 28, 2013. Therefore, we assume the area of study across years to be approximately the same.

14. The dummy for metropolitan area equals 1 if the individual lived in a metropolitan area, whether identified or not, and zero if in a nonmetropolitan area. Individuals with missing information for this variable (14,018 or 4.66% of the total) were excluded from the sample.

Additional information

Notes on contributors

Luísa Nazareno

Luísa Nazareno is a PhD candidate of Public Policy at Georgia State University, Andrew Young School of Policy Studies. Her research focuses on alternative work arrangements, worker well-being, inequalities, and the implications of technology development to jobs and economic development. Her work has been published in Technology in Society, The Russell Sage Foundation Journal of the Social Sciences, and Social Indicators Research.

Cathy Yang Liu

Cathy Yang Liu is a Professor and Chair in the Department of Public Management and Policy at the Andrew Young School of Policy Studies, Georgia State University. She conducts research and publishes widely in the areas of community and economic development, urban labor market and inequality, migration and entrepreneurship, as well as international urban development. Her edited book “Immigrant Entrepreneurship in Cities: Global Perspectives” was published by Springer in 2021.