Uneven land of opportunity: US regional employment futures

ABSTRACT This paper augments our understanding of the geography of work and opportunity in the United States, examining employment projections for a set of occupations deemed to have a particularly ‘bright outlook’ in the coming decade. Drawing labour and feminist geography insights into regional studies, this paper combines several sources of socioeconomic data to examine critically, first, the regional patterns of projected employment; and, second, how good these jobs really are. It finds that access to these jobs will be highly uneven across the United States; many of these jobs pay well below average, often not paying regional living wages; and patterned disadvantages are likely to hamper improved pay and conditions. It is argued that the mismatch between the optimistic rhetoric and the actual empirics shows that deep changes must be made for there to be a bright outlook for US employment futures. The paper concludes with suggestions on regional policy avenues that could improve this outlook.


INTRODUCTION
This paper augments our understanding of the geography of work and opportunity in the United States. It takes as a case the US Bureau of Labor Statistics (BLS) employment projections for a set of occupations deemed to have a particularly 'bright outlook' by O*NET, a leading source on occupational information supported by the US Department of Labor (DoL). The yellow sun logo marking such occupations throughout the O*NET website conveys a bright future for individuals willing to seize these opportunities, with similarly positive descriptions identifying 'promising career opportunities' (O*NET, 2017a). But what are the regional patterns in these projections and are these occupations actually 'good' jobs?
Using a range of publicly available data sets, this paper offers a critical assessment of the future of work and opportunity across the United States. It builds on research in regional studies and labour and feminist geographies. It illuminates problems of regional inequality and structured disadvantage identified in the literature by drawing on official labour force and projection data, providing future-oriented insights. In short, this paper shows the supposedly 'bright outlook' occupations are unevenly distributed across regions, often pay extremely low wages, and share characteristics associated with structured challenges to improved pay and working conditions. It argues that the challenges presented by many of these jobs are not amenable to individual choices, as the rhetoric of the 'Bright Outlook' suggests. It concludes with suggestions for potential policy intervention at the regional meso-scale of US states.

LITERATURE REVIEW
A decade after the Great Recession's start, one major economic indicator looks good again: US national unemployment has been less than 4% for most of 2018 (BLS, 2018). However, research in regional studies and related disciplines demonstrates that persistent regional differences and inequalities complicate national economic indicators (e.g., Wei, 2015). Additionally, employment does not guarantee adequate sustenance to workers, evidenced by research on the working poor (Gautié & Ponthieux, 2016), precarity (Strauss, 2018), and high costs of living (NLIHC, 2017). Feminist scholars and labour advocates also identify several structured disadvantages (Werner et al., 2017) that hamper pay and working conditions in particular types of work including: care work, which faces a relative wage penalty compared with work requiring similar skills and training (England, Budig, & Folbre, 2002); domestic work, which is excluded from national labour protections, only recently protected in a handful of states, and frequently performed under substandard conditions (NDWA, 2016;Theodore, Gutelius, & Burnham, 2013); and tipped minimum wage work, which is associated with poverty wages and elevated incidences of wage theft (ROC United, 2016). These insights from the extant literature demand that we examine more than national aggregates to assess occupational outlooks, including more consistently considering pay, working conditions and regional variation. This paper does so at the meso-scale regional level of US states. Metropolitan regions or commuting zones would capture labour market regions well (Scott, 1988;Walsh, 2000), but these economic regions rarely have corresponding cohesive and well-resourced governmental structures (Dreier, Mollenkopf, & Swanstrom, 2014). Instead, US states have wide-ranging authority in work-related matters such as determining minimum wages, corporate structures and working conditions. Implementation of national work regulations and social safety nets largely goes through states, allowing substantial administrative leeway. States also encompass rural, suburban and urban areas, enabling collection and distribution of resources needed to implement public programmes, which may be impractical at more local levels of government. This regional scale is therefore particularly useful for examining occupational outlooks.
This paper asks two main questions: How do projected job openings vary regionally at the scale of states? Can these Bright Outlook occupations reasonably be considered 'good' jobs? These questions will help shed a light on the rhetoric and reality of the US employment outlook.

METHODOLOGY AND CASE STUDY OVERVIEW
The O*NET ® programme, sponsored by the US DoL, provides detailed descriptions of a full range of occupations (O*NET, 2017a). While often used in research (e.g., Cooke & Kemeny, 2017), a major aim of O*NET is to provide information to job seekers planning their careers. To this end, one prominent feature is the highlighted 'Bright Outlook' occupations, first introduced in 2010 (O*NET, 2017a). In educational material aimed at workforce development professionals, the designation's utility is introduced in this way: 'it is important to know which occupations have the best outlook for people preparing for and searching for jobs' (Wall,n.d.,p. 1). This corresponds to the sponsoring agency's agenda. One of the guiding principles of the DoL's Employment and Training Administration is: 'We will strive to turn individuals into career entrepreneurs by equipping them with the information they need to develop the knowledge, skills and abilities sought after in the new economy' (US ETA, 2010). Thus, the Bright Outlook designation is part of broader efforts to guide individuals toward choices with an eye to the near future.
O*NET's designation includes occupations meeting at least one of two definitions: (1) rapid projected growth (at least 10% over the decade); and (2) large numbers of expected openings (at least 100,000 over the decade) (O*NET, 2017a). The growth rates and openings come from long-term employment projections produced by the BLS, the most recent being 2016-26 (BLS, 2017. There is wide variation among Bright Outlook occupations: from water resource specialists and neurologists, to teachers and forest fire inspectors, to cashiers and cooks. BLS projections demonstrate they will all feature prominently in the future of US work. By 2026, the hundreds of Bright Outlook occupations will account for roughly two-thirds of all jobs, with 2.5 times the job openings for Bright versus other occupations (O*NET, 2017b; BLS, 2017).
To answer how Bright Outlook occupation openings vary regionally and to assess whether it is reasonable to categorize these as 'good' jobs, the following section presents descriptive statistics for these occupations. State-specific occupational projections (Projections Central, n.d.) and American Community Survey (ACS) microdata (Ruggles, Genadek, Goeken, Grover, & Sobek, 2017) reveal the highly uneven regional geography of these occupations. Assessment of remuneration levels draws on the Occupational Employment Survey (OES), detailing median annualized earnings (BLS, 2017), and the ACS, which contains reported earnings and typical weeks and hours worked. The MIT Living Wage Calculator allows the comparison of typical earnings with state-specific costs of living (Glasmeier, 2017). Additional characteristics of key occupations are drawn from the literature.

EMPIRICS
To offer a more grounded sense of the Bright Outlook occupations, Table 1 presents occupations with the largest number of expected annual openings. These occupations are those most likely to be open to job seekers at any point. Openings for the top 25 occupations account for nearly 40% of all projected annual openings.
Dramatic regional variation of projected Bright Outlook job growth runs from 3% in Maine and West Virginia to more than 25% in Colorado, Nevada and Utah (Projections Central, n.d.) ( Figure 1). These patterns relate to, but are more extreme than, projected population growth: while Maine expects no population growth, Utah will grow by 16.5% (Weldon, 2016). This can translate to the uneven density of jobs: Idaho will have 170 Bright Outlook jobs per 1000 people; Washington, DC, 580 per 1000. Since jobs and openings differ, openings per capita will also vary substantially. Thus, the answer to the first key question is that the employment forecast will be much brighter in some regions than in others.
Turning to the second question, there are important differences in pay. Assuming stability in relative compensation across occupations, the average Bright Outlook worker in 2026 will have below-average wages (BLS, 2017). Currently, occupations with numerous projected openings largely have below-median annualized earnings (20 of the top 25 occupations; Table 1). Annualized compensation in OES data assumes a 40-hour work week year round. However, actual yearly pay for people currently in these occupations is frequently lower (Table 1, ACS columns). Parttime work and high turnover reflected in occupations with numerous openings but low projected growth (e.g., cashiers) suggest steady, year-round employment is elusive. For some, the difference is remarkable: cashiers (ranked third) and childcare workers (18th) report earnings of less than half the annualized earnings (Tables 1 and 2). Without substantial shifts in pay structure, scheduling and tenure, many of the jobs open every year will pay a pittance.  Even accounting for regional variation in the cost of living, Bright Outlook occupations frequently fail to pay 'minimum subsistence wages' as Glasmeier (2017) interprets her 'living wage' thresholds. Typical weekly earnings (from ACS median earnings, hours and weeks worked) reveal that only 23% of Bright Outlook occupations pay a two-adult living wage in all states, including only seven of the top 25. However, the average US household has 2.4 people, including half a child and with only one adult in the labour force (Ruggles et al., 2017). It is notable that only five IPUMS Bright Outlook occupations typically pay above a family living wage in all states: health service managers, software developers, civil engineers, physicians and nurses.
Meeting a less onerous benchmarkthe living wage for a single adultwould reduce Bright Outlook occupations by 34%. This translates to substantial reductions in the number of such jobs, again varying across regions: Nevada and Hawaii would maintain slightly over half their projected Bright Outlook jobs, but Utah and other states would see reductions of over 60% (Figure 2). This pervasive gap between actual and living wages means regions with numerous projected jobs are not systematically more affordable. Indeed, there is a positive (though not statistically significant) relationship between projected Bright Outlook growth and cost of living at the state level.
While compensation for many of these jobs is currently below average and largely below living wages, it is worth considering their prospects for wage growth. For several of these occupations immediate past trends show increased employer demand, but with minimal wage increases. Employment for personal care aides nearly tripled from 2006 to 2016, but mean wages rose only 55 cents per hour (OES data adjusted to 2016 US dollars). This is likely partially due to the number of people nominally qualified for this work, but also to the persistence of deeply gendered, racialized and immigration status-specific stratifications in US labour markets and educational opportunities. Many of the numerous and low-pay Bright Outlook occupations are dominated by women, particularly women of colour and those foreign born. These occupations ( Table 2) fit several of the categories of work facing documented disadvantages identified in the literature: care, domestic and tipped minimum wage work. Taken together, these results clearly demonstrate a highly uneven regional geography of employment opportunities across the United States. Further, even in regions with large numbers of projected job openings, many will be for occupations that currently pay very low wages and share structured barriers to improving pay and working conditions.

DISCUSSION
O*NET's marking of 'promising career opportunities' is intended as a signal that supports more informed choices by job seekers and advisors wading through the data. This logic of supporting individual choices is not unique to O*NET. Recent recommendations to the DoL for bolstering occupation and labour market information double down on it. These expert-generated recommendations to the Secretary of Labor were introduced this way: 'Through its recommendations, the Council aims to support Congress's intent and your efforts to ensure that individual students, workers, educators, and employers have the information each needs to make good decisions in the realms of education and employment' (WIAC, 2018, p. i). Such data improvements would be invaluable for regional research on work and opportunity. However, the highly uneven regional geography of the projected openings and the sheer number of lowpaying Bright Outlook jobs imply broader challenges that most individuals cannot 'choose' their way out of. This gap between the rhetoric and reality of the employment outlook might be viewed as a technical problem to be fixed with adjustments to the Bright Outlook criteria. However, inspired by the feminist analytical approach of 'mapping for difference', this gap itself can be a source of information, prompting different questions (Pavlovskaya & Bier, 2012). Fixing the defining criteria will not solve the underlying problems with these jobs. If these Bright Outlook occupations reflect the work that will be needed in the futureas the employment projections suggestthen should they not provide decent and secure livelihoods?
As of now, too many of the supposedly Bright Outlook jobs are not good ones. They provide only precarious livelihoods with persistent and patterned disadvantages in wages and working conditions. Furthermore, the regional gaps between where opportunities will be and what they will pay raises important issues related to differential mobility. Cross-state moves imply a loss of in-person social networks that often act as a safety net. Moving for a position as a general manager (with a median salary near six figures) may be logical, but moving for the more numerous openings in food preparation or sales might make less sense as a 'survival strategy' (Gilbert, 1998). These socially structured and regionally informed problems, combined with well-documented labour market failures (for a recent example, see Azar, Marinescu, & Steinbaum, 2017), show the very real limits of individual choice around these jobs, no matter how clear the information about them is. Instead, these jobs must be made more robustly 'bright'.

CONCLUSIONS
In considering the future of work and opportunity across the United States, the wealth of highquality, publicly accessible data from the government and entities such as O*NET is welcome. However, this paper has shown the current Bright Outlook criteria include many occupations that do not reasonably fit the rhetoric of 'promising careers'. Instead, these are regionally uneven and often low-paying occupations with structured challenges to improvement. This contradiction between the rhetoric and reality highlights problems that are unsolvable purely by informed individual choices. They necessitate deeper changes. What could regional policy do to brighten occupational futures? US states have obvious policy levers: 1 increasing minimum wages; improving overtime, paid leave and unionization rules; and expanding coverage of such policies (e.g., Appelbaum, 2010;Engeman, 2015;Rhee & Zabin, 2009). Further, states can marshal resources for enforcement of these rules, as well as anti-wage theft and anti-discrimination laws. They can enable cooperative corporate structures, where workers share ownership and decision-making. Funding and incentivizing improvements to metropolitan public transit systems can help states mitigate uneven access to jobs. Bolstering equitable access to quality public education can underwrite broader access to high-paying occupations. Funding and incentivizing affordability in housing (Hsieh & Moretti, 2015), particularly rentals (NLIHC, 2017), as well as in childcare, eldercare and healthcare would narrow the gap between living wages and actually existing wages. Models for these policies and others, including social insurance programmes and coordinated long-term care for the elderly, are readily available (Poo, 2015). A significant commitment to a mixture of these regional policies, appropriate to the varied regional economic dynamics, could improve many of the Bright Outlook occupations by augmenting pay, expanding legal protections, supplementing public services and addressing rising costs of living.
How could further regional studies research contribute to this project? First, close assessment of countries with similar employment structures but different policies could identify new policy options and suggest necessary adjustments for the US context (e.g., Carré & Tilly, 2017). Second, additional research on the interaction of declining internal migration and uneven regional labour markets would shed light on the potential for regional efforts to improve pay and conditions in low-wage but numerous jobs. Third, more research on the impact of technology change on these occupational futures would help illuminate what jobs will still require people to do them, further clarifying where interventions will be needed to ensure these are decent jobs.
As with nearly all progressive policy change, and because governments are not neutral agents (Bergene, Endresen, & Knutsen, 2016), regional policy aimed at brightening occupational futures will require active public pressure (e.g., Rhee & Zabin, 2009). Without such pressure, the critical assessment presented here strongly suggests a future of work in the United States that is a constellation of unevenly bright stars with vast dark spaces in between.

NOTE
1 There are also important policy levers at the national scale, but the geographical variation in these projections suggests substantial room for regionally tailored intervention. Furthermore, gerrymandering (drawing geographical boundaries of political representation to favour one party, allowing it to win a higher percentage of electoral seats than votes cast for its candidates) has entrenched the more corporate lobby-friendly, anti-labour, anti-regulation party at the federal level for the next few election cycles. Immediate progress seems likelier at the state scale.