Changing Job Access of the Poor: Effects of Spatial and Socioeconomic Transformations in Chicago,1990

Abstract

Research has shown that the job access of the poor has been declining because of two major reasons: the spatial distribution of employment and housing, and socioeconomic restructuring. This paper aims to untangle the effects of the two factors on poor job seekers’ access to jobs in the Chicago metropolitan area from 1990 to 2010. Using census tracts as the unit of analysis, this research examines the effects of these two factors on the growth and distribution of poor job seekers and their matching jobs, as well as the consequential changes in job demand, supply and job access across the study area. Results show that spatial changes have increased job accessibility for the poor while socioeconomic transformation has adversely affected it.

1 Introduction

Since the 1940s, two kinds of transformation have significantly affected the employment prospects of the poor in the United States. The first is spatial redistribution of jobs and housing, and the second is social and economic restructuring in metropolitan areas.

John Kain (1968) proposed the spatial mismatch hypothesis (SMH) to explain how the spatial arrangements of jobs and housing, in addition to human capital and socioeconomic characteristics, lead to unequal labour market outcomes for the disadvantaged. Kain identified three spatial factors: disadvantaged groups had fewer residential choices and were constrained in the inner cities; jobs also suburbanised, especially those manufacturing jobs that were traditionally held by low-skilled and low-wage workers; and, inner-city residents have limited affordable and efficient transport options to reach suburban jobs. SMH suggests that the three factors combined result in low job access and unequal labour market outcomes of disadvantaged groups in the inner cities.

SMH was originally proposed to explain the employment difficulties of inner-city African Americans, but its three premises apply to the poor equally well (Arnott, 1998; Ihlanfeldt, 2006). Poor households face residential segregation (Abramson et al., 1995; Massey and Eggers, 1993) with their housing locations constrained to limited places such as inner cities (Kasarda, 1993). Relevant job opportunities for poor job seekers, specifically low-skilled and low-paid jobs, have suburbanised and dispersed from the inner cities. Poor job seekers have limited transport mobility. Therefore, the spatial transformation has increased spatial barriers between the residences of the poor and their matching job opportunities.

The second transformation is the social and economic restructuring that has affected the nature and size of the poor and their matching job opportunities. Economic restructuring has shifted jobs from manufacturing to service sectors (Muller, 2004; Sassen, 1990) and has reduced the number of better paying low-skilled jobs. Meanwhile, the poverty population in the United States increased from 25 million in 1970 to 46 million in 2010 (US census). A decreasing number of jobs and an increasing poverty population affect the poor’s job access.

This paper contributes to the literature by examining together the effects of the spatial and socioeconomic transformations on the job access of the poor. Existing literature has documented significant impacts of either the spatial or the socioeconomic transformation, but none has examined the two transformations together. Differentiating their effects is imperative for deciding effective policy measures to improve the employment prospects of the poor. Most current policies focus on remedies that address only one transformation. For example, the Job Access and Reverse Commute programme and the Moving to Opportunity programme aim to reduce spatial separation, whereas regional economic development programmes aim to increase the number of job opportunities. Individual programmes tend to have limited impacts (Clark, 2008), but this does not mean that the programmes are ineffective. A combination of policies and programmes that deal with both spatial and socioeconomic factors might be necessary. This research aims to examine to what extent the spatial and socioeconomic transformations affect the poor’s job access. The Chicago metropolitan area from 1990 to 2010 is used as a case study area.

The remainder of the paper is organised into four parts. The next section reviews the literature on spatial and socioeconomic transformations, as well as job accessibility. Section 3 describes the study area, data and methodology. Empirical results are presented in section 4, and the final section concludes with a discussion of the findings and policy implications.

2. Spatial and Socioeconomic Transformations, Job Accessibility

2.1 Spatial Transformation

Kain’s SMH is based on the post-World-War-II spatial transformation—the suburbanisation of the affluent majority and jobs. His research and the studies that immediately followed relied on the conventional monocentric model, which treats metropolitan areas as a dichotomy between the inner cities and the suburbs, and regarded spatial mismatch as an issue in the inner cities (for example, in Harrison, 1972; Kain, 1968; Kasarda and Ting, 1996; Price and Mills, 1985; Stoll, 1999b; Wilson, 1987).

Metropolitan areas, however, are no longer monocentric. In the past two decades, the urban spatial transformation has affected two of the three spatial premises of Kain’s SMH. First, jobs have suburbanised more extensively with greater shares of jobs located in the suburbs than in the central business districts (CBDs), but many suburban jobs have clustered in employment centres (Anas et al., 1998; Giuliano and Small, 1991). Access to suburban employment centres has become increasingly important. Secondly, the poor have increasingly more housing options and have begun to move to the suburbs (Berube and Frey, 2002; Jargowsky, 2003), although their suburbanisation is still characterised by segregation (Jargowsky, 1996; Schill and Wachter, 1995). Poverty concentrations have emerged in the suburbs (Kneebone and Garr, 2010).

Combining the two changes, one may argue that suburbanisation of both employment and the poor can reduce the spatial separation between housing and job opportunities, and thus lessen the spatial mismatch (Gordon et al., 1989b). This is not necessarily true. Employment concentration and housing segregation in the suburbs can still create great spatial barriers for the poor to reach jobs (Stoll and Covington, 2012).

2.2 Socioeconomic Transformation

Different from the spatial transformation, which affects the distribution and allocation of poor job seekers and their matching jobs within a metropolitan area, socioeconomic transformation affects the total numbers of job seekers and jobs, and hence the levels of job demand and supply.

Economic restructuring has shifted jobs from manufacturing to service sectors (Muller, 2004; Sassen, 1990). The shift has reduced the number of higher-paying manufacturing jobs traditionally held by low-skilled and low-income workers, thus exacerbating these workers’ disadvantages in the labour market (Kasarda, 1989). Social structure in US metropolitan areas has also changed. In addition to race/ethnicity, economic class has become an increasingly relevant indicator of population segments. Differences within the African American population have increased in terms of their socioeconomic status (Wilson, 1987; Fischer, 2003), while racial segregation has been reduced (Wilson, 1980; Massey, 2001). Hispanics from other countries have immigrated to gateway cities and increased their share in the total population. Many immigrants possess low skills and comprise an increasing share of the poor in metropolitan areas (Sullivan and Ziegert, 2008).

Socioeconomic conditions affect spatial distributions. In the 1990s, with a robust economy, many downtown areas revived with the affluent majority moving back (Sohmer and Lang, 2001). Meanwhile, inner-ring suburbs became increasingly vulnerable to socioeconomic decline, while the outer-ring suburbs continued to grow (Lee and Leigh, 2007). The economic recession of 2008–09 could have different effects. The recession started in mainly high-wage finance industries and then spread to low-wage industries, such as construction, manufacturing and service (Kuehn, 2011). Finance jobs tend to be located in the central business districts and other employment agglomerations, and low-wage jobs tend to be dispersed in the suburbs (Giuliano and Small, 1991). Thus the recession affected whole metropolitan areas. Examining the complicated interrelationships between the spatial and socioeconomic transformations can shed light on the mechanisms of and remedies for poor job seekers’ disadvantages in the labour market.

2.3 Job Accessibility

Accessibility is

determined by the spatial distribution of potential destinations, the ease of reaching each destination, and the magnitude, quality, and character of the activities found there (Handy and Niemeier, 1997, p. 1175).

Job accessibility reflects the relationships between job seekers and their matching job opportunities in terms of the locations and quantity.

Job accessibility is expected to be highly correlated with labour market outcomes. The spatial job search theory suggests that job accessibility denotes job searching costs, including transport and information costs. Declining searching costs, associated with enhancing job access, increase job searching intensities (Gobillon et al., 2007). More extensive spatial job searches lead to better employment and wage offers (Schwartz, 1976; Stoll, 1999a).

Measurements of job accessibility have evolved over time. This research adopts a simplified version of Shen’s (1998) relative gravity-based job accessibility measure, which considers job supply and demand at the same time. The job accessibility measure in this paper is place-based, ignoring personal and household characteristics and time–space constraints which are not directly related to the spatial and socioeconomic transformations.

3. Study Area, Data and Methodology

3.1 Study Area

The study area includes the six counties of Cook, DuPage, Kane, Lake, McHenry and Will in the Chicago metropolitan area. The geographical unit of analysis is the census tract and the study period is from 1990 to 2010. To maintain a consistent geography, all data are converted to the 2000 census tract geography. There are 1838 census tracts in 2000. Figure 1 shows the geography of the study area.

Figure 1.

The study area.

The Chicago metropolitan area has been significantly affected by spatial and socioeconomic transformations. The central city, the City of Chicago, had about 42 per cent of the total population and 45 per cent of the total jobs in 1990, but the percentages dropped to about 32 per cent for both population and jobs in 2010. Outside the central city, most employment centres are located in the other parts of Cook County, as well as in Lake and DuPage counties (McMillen and Lester, 2003). These areas form the inner-ring suburbs and started to attract employment concentrations after World War II, prior to the other suburban areas. The other three counties, McHenry, Kane and Will, are the outer-ring suburbs, which experienced fast growth after 1990.

The metropolitan area is a paradigmatic industrial metropolis and has been redefined in the deindustrialisation process. Between 1990 and 2010, the share of manufacturing jobs declined from 17 to 14 per cent, while the share of all service jobs increased from 33 to 61 per cent. Meanwhile, the metropolitan area experienced a demographic shift and the number of the poor in the study area grew from 808,400 in 1990 to slightly over 1 million in 2010.

3.2 Data

Census-tract-level demographic data come from the decennial census and American Community Survey (ACS). The tract-level data do not provide some essential information for this research, such as the 2010 one-year estimate of population in poverty or the share of poor workers in each industry. In these cases, individual-level 5 per cent Public Use Microdata Sample (PUMS) data collected from Integrated Public Use Microdata Series (Ruggles et al., 2010) are used and combined with the census tract data. PUMS data are only available for the Public Use Microdata Areas (PUMA), which are larger in size than the census tracts. The study area had 51 PUMAs in 1990 and 55 in 2000 and 2010.

Employment data come from Chicago Metropolitan Agency for Planning (CMAP), the region’s Metropolitan Planning Organization (MPO). The employment data are classified into six sectors (manufacturing, retail, service, TCUW (transport, communications, utilities and warehousing), government and other) and are based on quarter sections, which are much smaller than most census tracts.

Potential poor job seekers are defined as persons 18–64 years old in poverty. All working-age people, rather than just those in the labour force, are included because labour force participation is affected by the local labour market condition. In census tracts with low job access where job search is difficult and costly, potential labour force participants might seek substitutes for employment (Arnott, 1998).

The 1990 and 2000 census tract population of persons 18–64 years old in poverty is directly available from the decennial census. The census does not provide one-year estimates of population in poverty in 2010; instead, it provides 2006–10 estimates of persons 18–64 years old in poverty based on the ACS and 2010 one-year estimates of all persons 18–64 years old. The average margin of error of the five-year estimates is 33 per cent for the census tracts with more than 1000 poor job seekers, and the margin increases as the number of poor job seekers decreases. Nevertheless, the multiyear ACS estimates are still the most reliable demographic data source and are used widely in poverty studies (Kneebone et al., 2011; Lichter et al., 2012).

Another challenge is the temporal mismatch between the five-year ACS estimate of job seekers and the one-year estimates of employment. To adjust the five-year estimate, I assumed that all census tracts within a PUMA experienced the same growth or decline in the share of poor job seekers. The equations are

Rati o_{p} = \frac{PovJSShar e_{p, 2010}}{PovJSShar e_{p, 2006 - 10}}

(1)

\begin{matrix} PovJSShar e_{m, p, 2010} = PovJSShar e_{m, p, 2006 - 10} \\ * Rati o_{p} (m \in p) \end{matrix}

(2)

\begin{matrix} PovJ S_{m, p, 2010} = PovJSShar e_{m, p, 2010} \\ * J S_{m, p, 2010} (m \in p) \end{matrix}

(3)

where, $Rati o_{p}$ is the ratio of the share of poor job seekers based on the one-year estimates to the share based on the five-year estimates for PUMA p. $PovJSShar e_{p, 2010}$ is the share of poor job seekers in total job seekers 18–64 years old based on the 2010 one-year estimate for PUMA p; and $PovJSShar e_{p, 2006 - 10}$ is the share based on the five-year estimate. $PovJSShar e_{m, p, 2010}$ is the share of poor job seekers in all job seekers in census tract m that is located in p. $PovJSShar e_{m, p, 2006 - 10}$ is the five-year estimate share of poor job seekers in census tract m in p, based on the ACS data. $PovJSShar e_{m, p, 2010}$ is the number of poor job seekers in census tract m in PUMA p. $J S_{m, p, 2010}$ is the number of all job seekers in m based on the 2010 Census Summary File 1 (SF1) data.

Based on these equations, the total 2010 number of poor job seekers in the study area is 73,500 greater than the total of the 2006–10 five-year estimates. The method used here maintains the poverty distribution pattern at the census-tract level based on the multiple-year ACS data and addresses a common problem of the mismatch between the multiyear ACS estimates and one-year estimates from other data sources. The method creates errors, but they are expected to be small. Analysis of census data in 2000 and 2006–10 shows that the changes in the share of poor job seekers between the two time-periods in each PUMA are consistent with the changes in the census tracts with large numbers of poor job seekers; thus applying the ratios at the PUMA level to those census tracts does not create large errors. Differences in the changes in the share between the census tracts with small numbers of poor job seekers and their respective PUMAs are greater, but the magnitude of errors is small because of the small bases.

A second method was also used to evaluate the errors. Instead of using the share of poor job seekers, I calculated the ratios of the one-year numbers of poor job seekers to the five-year numbers by PUMA, and applied the ratios to the census-tract-level number of poor job seekers based on the five-year ACS estimates. The difference in the total number of poor job seekers between the two methods is marginal at 1,409, or 0.2 per cent of the total number of poor job seekers in 2010.

To identify poor job seekers’ matching job opportunities, the CMAP quarter-section-level employment data are aggregated to the census tracts. A census tract has all the employment of a quarter-section if the centroid of the latter is located within the census tract. This step incurs inaccuracy, particularly in small census tracts, but it does not affect the job accessibility scores because they are calculated based on the weighted sum of job opportunities (see equations (17) and (19)). Jobs, as long as they are in adjacent small census tracts, are weighted relatively equally.

To calculate the number of jobs suitable for poor job seekers, I assume that, within a PUMA, poor job seekers are similarly likely to be working or seeking jobs in a specific industry. For example in 2010, in the PUMA where the Chicago downtown is located, percentages of jobs held by labour force in poverty were 5.9 per cent in manufacturing, 29.7 in retail, 12.5 in service, 9.2 per cent in TCUW, 6.3 per cent in government and 2.1 per cent in other jobs. The percentages are applied to the number of jobs in the corresponding industry in all census tracts in the PUMA to get the number of jobs for poor job seekers. The equations are

Rat e_{p, y} = \frac{PovL F_{p, y}}{L F_{p, y}}

(4)

PovJo b_{m, p, y} = Jo b_{m, y} * Rat e_{p, y} (m \in p)

(5)

PovJo b_{m, p} = \sum_{y} PovJo b_{m, p, y}

(6)

where, $Rat e_{p, y}$ is the share of the industry y labour force that is in poverty within PUMA p; $PovL F_{p, y}$ is the number of the labour force in poverty in industry y in PUMA p; and $L F_{p, y}$ is the number of the labour force in industry y in PUMA p. $PovJo b_{m, p, y}$ is the number of jobs suitable for poor job seekers in industry y in census tract m in PUMA p. $Jo {b_{m,}}_{y}$ is the number of all jobs in industry y in census tract m. $PovJo b_{m, p}$ is the total number of jobs suitable for poor job seekers in census tract m that is in PUMA p. This method maintains the census-tract-level distribution of jobs by industry.

Some research uses other indicators of job opportunities, such as new job openings (Shen, 2001) and employment growth (Painter et al., 2007; Rogers, 1997). Those indicators are not as relevant as the total number of jobs for the potential poor job seekers, some of whom are not actively looking for jobs. Campbell and Rosenfeld (1985) found that “not searching is the most common method of search”. Moreover, the turnover rate (quits, lay-offs or discharges) of low-income jobs tends to be higher (Doeringer, 1968; Simpson, 1992); thus the total job supply for poor job seekers is more pertinent to this research.

3.2 Weighting Job Seekers and Their Jobs

To separate the effects of spatial and socioeconomic transformations, two weighting methods were used. The first method controls the total numbers of poor job seekers and jobs and highlights changes in spatial distributions.

Weighting Method to Explain the Spatial Transformation

PovJ S_{t, n} = TotPovJ S^{*} * % PovJ S_{t, n}

(7)

PovJo b_{t, n} = TotPovJo b^{*} * % PovJ o_{t, n}

(8)

\begin{matrix} PovJ S_{t, n} - PovJ S_{t - 1, n} = TotPovJ S^{*} \\ * % PovJ S_{t, n} - TotPovJ S^{*} * % PovJ S_{t - 1, n} \\ = TotPovJ S^{*} * (% PovJ S_{t, n} - % PovJ S_{t - 1, n}) \end{matrix}

(9)

\begin{matrix} PovJo b_{t, n} - PovJo b_{t - 1, n} = TotPovJo b^{*} \\ * % PovJ o_{t, n} - TotPovJo b^{*} * % PovJ o_{t - 1, n} \\ = TotPovJo b^{*} * (% PovJo b_{t, n} - % PovJo b_{t - 1, n}) \end{matrix}

(10)

where $PovJ S_{t, n}$ and $PovJo b_{t, n}$ denote poor job seekers and their matching jobs in census tract n at time t. $TotPovJ S^{*}$ and $TotPovJo b^{*}$ are the total numbers of poor job seekers and jobs that are held constant. This weighting method uses year 2000 total numbers as the control totals. $% PovJ S_{t, n}$ and $% PovJo b_{t, n}$ denote the shares of poor job seekers and their jobs that are in census tract n in the respective year t.

Equations (9) and (10) indicate that differences in the weighted data among the three points in time (for example, $PovJ S_{t, n} - PovJ S_{t - 1, n}$ ) can show the changes in the shares of jobs and job seekers (for example, $% PovJ S_{t, n} - % PovJ S_{t - 1, n}$ ). The shares indicate spatial distributions and thus their changes suggest redistribution and relocation.

The second method controls the spatial distributions of jobs and job seekers, and highlights the changes in the total numbers associated with the socioeconomic transformations.

Weighting Method to Explain Socioeconomic Transformation

PovJ S_{t, n} = TotPovJ S_{t} * % PovJ {S_{n}}^{*}

(11)

PovJo b_{t, n} = TotPovJo b_{t} * % PovPovJo {b_{n}}^{*}

(12)

\begin{matrix} PovJ S_{t, n} - PovJ S_{t - 1, n} = TotPovJ S_{t} * % PovJ {S_{n}}^{*} \\ - TotPovJ S_{t - 1} * % PovJ {S_{n}}^{*} \\ = (TotPovJ S_{t} - TotPovJ S_{t - 1}) * % PovPovJ {S_{n}}^{*} \end{matrix}

(13)

\begin{matrix} PovJo b_{t, n} - PovJo b_{t - 1, n} = TotPovJo b_{t} \\ * % {PovPovJob}_{n}^{*} - TotPovJo b_{t - 1} \\ * % {PovPovJob}_{n}^{*} = (TotPovJo b_{t} - TotPovJo b_{t - 1}) \\ * % {PovPovJob}_{n}^{*} \end{matrix}

(14)

where, $TotPovJ S_{t}$ and $TotPovJo b_{t}$ are the total number of poor job seekers and jobs in time t. $% {PovJS}_{n}^{*}$ and $% {PovJob}_{n}^{*}$ denote the shares of job seekers and jobs in census tract n, and the shares in 2000 are held constant. All the other notations are the same as in the previous equations.

Equations (13) and (14) indicate that the differences across time are explained by the changes in the total numbers of job seekers and jobs, assuming that different parts of the region experienced similar rates of changes regardless of their spatial locations. As stated in the previous sections, changes in the total numbers are directly related to the socioeconomic transformation.

3.4 Job Accessibility

This research adopts Shen’s (1998) job accessibility model. The attractiveness of job opportunities declines with increasing travel distance or time. Such decline can be measured by a travel impedance function. The impedance factor is equal to 1 over the regional average commuting distance (Thomas and Huggett, 1980, pp. 161–162). Based on the origin–destination data in 2000 Census Transportation Planning Package (CTPP) Part III, the average commuting distance of the Chicago metropolitan area is 9.525 miles.¹ This results in an impedance factor of b = 1/9.525. Applying the friction factor in year 2000 to all places in a metropolitan area across different time-periods can affect the accuracy of job accessibility measures, but no other data sources allow me to calculate the commute distance in 1990 and 2010. Sensitive analysis based on different impedance factors is conducted, but the results, particularly changes in job accessibility over time, are only slightly different.²

Travel distance is defined as Euclidian distance among centroids of census tracts. Ong and Blumenburg (1998) used the same distance measure and found no bias in the estimated results. Intrazonal distance is calculated as a half of the square root of the census tract area.

The first step is to estimate demand for jobs that are located in census tract j from poor job seekers in census tract k

J D_{j} = \sum_{k} PovJ S_{k} f (C_{kj})

(15)

f (C_{kj}) = e^{- {bC}_{kj}}

(16)

where, $J D_{j}$ = job demand of census tract j; $PovJ S_{k}$ = the number of poor job seekers residing in tract k; $C_{kj}$ = Euclidian distance between k and j; and b = impedance factor = 0.105.

The next step is to estimate job supply from tract j for poor job seekers in census tract i

J S_{i} = \sum_{j} PovJo b_{j} f (C_{ij})

(17)

f (C_{ij}) = e^{- {bC}_{ij}}

(18)

where, $J S_{i}$ = job supply of census tract i for poor job seekers; $PovJo b_{j}$ = the number of suitable jobs located in tract j for poor job seekers; and $C_{ij}$ = Euclidian distance between i and j; other notations are the same as those in the previous equations.

Job accessibility of census tract i is found by summing up jobs weighted by job demand ( $J D_{j}$ ) and travel impedance $f (C_{ij})$ from i to j

A_{i} = \sum_{j} \frac{{PovJob}_{j} f (Cij)}{{JD}_{j}} = \sum_{j} \frac{{PovJob}_{j} f (Cij)}{\sum_{k} {PovJS}_{k} f (Ckj)} .

(19)

where, $A_{i}$ = job accessibility for poor job seekers residing in tract i.

Job demand as calculated in equation (15) indicates overall residential locations of poor job seekers and job supply as calculated in equation (17) indicates employment locations.

4. Results

This section provides description of the unweighted and weighted numbers of poor job seekers and their matching jobs, as well as job accessibility scores. This section also diagnoses how spatial and socioeconomic transformations affect changes in job accessibility.

4.1 Poor Job Seekers and Jobs

Table 1 shows the original and the weighted numbers of poor job seekers and their matching jobs (referred to as poor jobs), as well as their respective percentage changes by county from 1990 to 2010. The counties are ranked roughly by population and employment size. The ranking also indicates the relative location of each county in the study area, from the centre to the fringe.

Table 1.

Poor job seekers and jobs by county

	Unweighted				Weighting method to explain the spatial transformation				Weighting method to explain the socioeconomic transformation
	Poor job seekers		Poor jobs		Poor job seekers		Poor jobs		Poor job seekers		Poor jobs
	Number (1000s)	Percentage change	Number (1000s)	Percentage change	Number (1000s)	Percentage change	Number (1000s)	Percentage change	Number (1000s)	Percentage change	Number (1000s)	Percentage change
1990
Study Area	418		249		458		250		418		249
Cook	367		219		403		220		353		203
Lake	14		8		15		8		18		12
DuPage	12		11		14		11		17		18
Will	11		3		12		3		12		5
Kane	10		6		11		6		13		8
McHenry	3		2		4		2		5		2
2000
Study Area	458	9.6	250	0.6	458	0.0	250	0.0	458	9.6	250	0.6
Cook	387	5.3	204	–6.7	387	–3.9	204	–7.2	387	9.6	204	0.6
Lake	20	45.4	12	62.1	20	31.4	12	61.6	20	9.6	12	0.6
DuPage	19	52.2	18	55.3	19	38.9	18	54.4	19	9.6	18	0.6
Will	13	21.5	5	69.2	13	10.9	5	68.3	13	9.6	5	0.6
Kane	14	34.7	8	31.1	14	23.0	8	30.4	14	9.6	8	0.6
McHenry	5	67.1	2	59.7	5	52.5	2	58.9	5	9.6	2	0.6
2010
Study Area	613	33.7	259	3.5	458	0.0	250	0.0	613	33.7	259	3.5
Cook	474	22.6	185	–9.5	355	–8.3	179	–12.6	517	33.7	212	3.5
Lake	29	46.3	18	42.4	22	9.1	17	38.8	26	33.7	13	3.5
DuPage	40	112.5	30	67.4	30	59.0	29	61.7	25	33.7	18	3.5
Will	28	111.1	9	69.0	21	57.9	9	63.3	18	33.7	5	3.5
Kane	30	111.0	14	70.0	22	57.8	13	64.2	19	33.7	8	3.5
McHenry	12	118.1	4	75.9	9	63.1	4	69.9	7	33.7	3	3.5

The left panel of the table shows the unweighted numbers. Between 1990 and 2010, the number of poor job seekers constantly grew faster than poor jobs in the study area: 9.6 per cent versus 0.6 per cent in the 1990s and 33.7 per cent versus 3.5 per cent in the 2000s. Across the study area, Cook County still had the largest numbers of poor job seekers and jobs, but it experienced the slowest growth of poor job seekers and jobs. All the other counties had faster growth than the regional averages. The suburbanisation trend is obvious.

The middle and right panels display the weighted numbers of poor job seekers and jobs based on the two weighting methods that explain spatial and socioeconomic transformations respectively. Because both methods use data in year 2000 as the controls, the 2000 data remain the same as in the left panel. Based on the first weighting method that explains the spatial transformation (the middle panel), the weighted total numbers of poor job seekers and jobs of the study area in 1990 and 2010 are the same as the 2000 totals. Positive percentage changes signify increasing shares of poor job seekers or jobs due to spatial redistribution, and negative changes signify declining shares. All counties except for Cook County gained shares. Again, both poor job seekers and jobs suburbanised. It is noteworthy that poor job seekers suburbanised slower than their matching jobs, as all the suburban counties gained higher shares of poor jobs than job seekers while Cook County lost higher shares of poor jobs.

The right panel of the table shows the statistics of the second weighting method that explains the socioeconomic transformation. Because changes in the numbers of job seekers and jobs in each census tract are associated only with the changes in the total numbers, the growth rate of each county is the same as that of the whole study area. Poor job seekers grew faster than poor jobs due to economic restructuring and demographic shift. The unbalanced growth is expected to reduce job accessibility scores.

4.2 Job Accessibility of the Poor

Table 2 presents job accessibility scores and their changes by county. The left panel shows the average accessibility scores based on the unweighted data. Job accessibility in the whole study area declined, from 0.510 in 1990 to 0.501 in 2000 and to 0.423 in 2010. Most of the decline can be attributed to the decline of job access in Cook County. Nevertheless, Cook constantly has the highest job accessibility scores (0.459 in 2010), followed by DuPage (0.419 in 2010) and Lake (0.327 in 2010).

Table 2.

Job accessibility by weighting method, 1990–2010

	Unweighted		Weighting method to explain the spatial transformation		Weighting method to explain the socioeconomic transformation
	Access	Change	Access	Change	Access	Change
1990
Study area	0.510		0.467		0.546
Cook	0.594		0.545		0.622
DuPage	0.388		0.356		0.451
Lake	0.262		0.235		0.330
Will	0.191		0.175		0.237
Kane	0.264		0.242		0.311
McHenry	0.185		0.169		0.244
2000
Study area	0.501	−0.009	0.501	0.034	0.501	−0.045
Cook	0.570	−0.024	0.570	0.025	0.570	−0.052
DuPage	0.413	0.025	0.413	0.058	0.413	−0.038
Lake	0.307	0.045	0.307	0.072	0.307	−0.023
Will	0.217	0.026	0.217	0.042	0.217	−0.020
Kane	0.285	0.021	0.285	0.044	0.285	−0.026
McHenry	0.224	0.039	0.224	0.055	0.224	−0.021
2010
Study area	0.423	−0.078	0.547	0.046	0.388	−0.113
Cook	0.459	−0.111	0.593	0.023	0.442	−0.129
DuPage	0.419	0.006	0.542	0.129	0.320	−0.093
Lake	0.327	0.020	0.420	0.113	0.234	−0.073
Will	0.216	−0.001	0.279	0.062	0.168	−0.049
Kane	0.311	0.025	0.402	0.117	0.221	−0.064
McHenry	0.241	0.017	0.312	0.088	0.173	−0.050

Note: the average scores area weighted by the number of poor job seekers.

Figure 2 visually presents job accessibility scores of the poor in 1990, 2000 and 2010, based on the unweighted data. The highest accessibility is shown in the darkest shade and the lowest accessibility in the lightest shade. Job accessibility forms a peak in the Chicago downtown area and decreases with increasing distance from the peak. Poor job seekers in the inner city still have higher job access than those in the suburbs. Although the concentration of the poor in the inner city incurs intense job competition, concentrated job opportunities offset the competition and benefit the poor job seekers residing there. This finding is different from what the original SMH suggested, but consistent with recent findings (for example, Hess, 2005; Shen, 2001).

Figure 2.

Job accessibility of the poor, 1990–2010.

Socioeconomic transformation intertwines with spatial reorganisation of jobs and job seekers, and they jointly affect job accessibility. The next section separates the effects of the two factors.

4.3 Effects of Spatial Transformation and Socioeconomic Transformation

In Table 2 the middle panel shows job accessibility scores based on the data from the first weighting method that explains the spatial distribution. The changes in the scores reflect the effects of urban spatial transformation. Positive changes in job accessibility scores in the whole study area (0.034 between 1990 and 2000, and 0.046 between 2000 and 2010) indicate that spatial reorganisation reduces the barriers between poor job seekers and their matching jobs, all else being equal. Among all counties, spatial transformation has the smallest positive effect in Cook County, which experienced a job accessibility increase of 0.025 in the 1990s and 0.023 in the 2000s, while job accessibility increased more in the other counties.

The right panel in Table 2 shows the results from weighting method 2, which controls the spatial distribution pattern. Changes in the numbers of job seekers and their matching jobs are mainly associated with changes in socioeconomic conditions. As the number of poor job seekers grew faster than their matching jobs due to the socioeconomic transformation, the whole study area experienced a decline in job accessibility: -0.045 in the 1990s and -0.113 in the 2000s. Cook County consistently had the largest decline, -0.052 and -0.129, because it has the largest numbers of poor job seekers and jobs, and proportional changes based on weighting method 2 result in the greatest reduction in the absolute numbers than all the other counties.

To understand changes in job accessibility of the poor, we need to dissect the two components of the job accessibility measure: job supply and job demand. Job supply is calculated based on equation (17) and job demand is calculated based on equation (15). The following analysis focuses on the decade between 2000 and 2010 to explain how spatial and socioeconomic transformations independently affect job supply and demand, and consequently job accessibility. The effects of both spatial and socioeconomic transformations in the 1990s are similar to those in the 2000s. The only notable difference is the magnitude of changes in job supply and demand based on the data from weighting method 2. Because of the consistency of the findings, the analysis of the changes in the 1990s is not included.

Figure 3 visually presents changes in job accessibility, job supply and job demand between 2000 and 2010 based on the two weighting methods. The top three graphs display changes based on the first weighting method that explains the spatial distribution, and the bottom three graphs are based on the second weighting method that explains the socioeconomic transformation. Positive changes are shown in solid grey, and the shading darkens as the change increases. Negative changes are shown in hatch.

Figure 3.

Changes in job accessibility, supply, and demand based on the two weighting methods, 2000–2010.

Figure 3(a1) shows that, due to spatial reorganisation, most census tracts experienced an increase in job accessibility. Out of the 1838 tracts, 1292 tracts where 68 per cent of poor job seekers resided in 2010 showed increases. The other 546 census tracts around downtown Chicago experienced job access decline. The inner-ring suburbs to the north-west of the central city experienced job access increases greater than 0.1 and these places housed 21 per cent of poor job seekers in 2010. The changes in job access due to spatial redistribution can be explained by changes in job demand and supply, shown in Figures 3(b1) and 3(c1). Suburbanisation of both job seekers and jobs is evident: jobs and job seekers shifted away from the inner city and their numbers grew in the suburbs. The spatial changes resulted in a decline of both job demand and supply in most parts of Cook County and an increase in all suburban counties. In the City of Chicago, the average change of supply was -28,389 and the average change of demand was -20,087. The relatively greater decline of job supply can largely explain why job accessibility decreased in the central city. Figures 3(b1) and 3(c1) also show that, in the suburban counties, both job supply and demand grew, but the magnitude of growth in supply was greater than growth in demand. On average, the change of supply was 5402 and the change of demand was 3364. Lake County showed the greatest difference, 5400 in supply change and 423 in demand change.

The whole study area exhibited a decline in job accessibility based on the data from the second weighting method. The increase of job demand outweighs that of job supply in the whole study area; thus the study area experienced declining job accessibility due to the social and economic transitions. The central city shows the greatest decline. Because it has the largest concentration of poor job seekers and their matching jobs, their proportional changes associated with the socioeconomic transformation result in the greatest changes in magnitude. The decline diminishes with an increasing distance from the central city.

In general, the suburbanisation of poor job seekers and their jobs increased their proximity and benefited the poor job seekers who reside in suburbs. On the other hand, job seekers grew faster than their matching jobs in the study area, and this reduced job accessibility. These two forces work together and contribute to the changes in job accessibility.

5. Conclusion and Discussion

Literature has shown that spatial and socioeconomic transformations affect the job access of the poor, but no research has tested both factors together. This research shows that, in the Chicago metropolitan area between 1990 and 2010, spatial transformation—suburbanisation of poor job seekers and their matching jobs—benefits the poor and improves their job accessibility, while socioeconomic restructuring—reduction in relevant jobs and increase in poverty—adversely affects poor job seekers.

The finding that spatial transformation has positive effects appears contradictory to the conventional SMH. Nevertheless, the effects can be reasonably explained by the similar relocation patterns of poor job seekers and their matching jobs, which reduced the spatial barriers between jobs and housing. The finding is consistent with the observation of Gordon et al. (1989a), who argued that suburbanisation of population can result in their proximity to jobs. The suburbanised poor households have started to catch up with the suburbanisation process and thus have improved their job accessibility. On the other hand, some premises of the SMH still hold true. Although the poor have increasingly more housing options, they still face affordability and other housing market constraints. Overall, the poor job seekers have suburbanised slower than their matching jobs.

Socioeconomic transformation presents great challenges for the poor by increasing the number of poor job seekers but relatively reducing the number of their job opportunities. It is consistent with Jargowsky’s (1997) and Gans’s (2010) claim that a good metropolitan economic climate is essential for the poor’s economic prospects. Without a strong metropolitan economic base, it is difficult to improve the local labour market environment.

The results help us to understand the multiple factors that affect job accessibility for the poor. Policies must address both spatial and socioeconomic factors to effectively improve poor job seekers’ employment prospects. The results also suggest shifting policy focus areas. Housing policies should expand the focus from the inner cities to parts of the suburbs where the suburbanised poor can benefit from job concentrations. Historically, poverty policies have neglected the suburbs, particularly the inner-ring suburbs (Puentes and Orfield, 2002). Increasing poverty there imposes pressure on the ageing housing, school systems, infrastructures and social services (Cooke, 2010). Suburbanised poor households need political and fiscal support similar to that provided to their inner-city counterparts.

Regional economic conditions have great impacts on job accessibility. It is noteworthy that a booming regional economy does not necessarily benefit the poor, mainly because of skill mismatch. In the 1990s, the economy of the study area was strong, but jobs suitable for poor job seekers grew only 0.6 per cent, while overall job growth was 12.8 per cent. Most of the job growth benefitted non-poor job seekers; very little benefited the poor. Therefore, policies should emphasise workforce development to help poor job seekers acquire the skills and knowledge needed for the growing employment sectors. Such workforce development programmes should include some combination of education and training, direct ties to employers or industries, and support and services (such as child care and transport) (Holzer, 2008).

I suggest some directions for future research. This research adopts a new approach to dissect two macro transformations that affect place-based job accessibility. It does not delve into how individual personal and household characteristics such as automobile ownership and time–space constraints affect job access. Future research needs to examine the macro and micro factors together.

Results of this research are circumstantial. The Chicago metropolitan area is selected because, like many other metropolitan areas, the region is significantly affected by both spatial and socioeconomic transformations. Unlike some metropolitan areas, the study area still has a strong central city and thus the weighting method, to explain socioeconomic transformation, results in greater decrease in job access in the central city. It might not be true in areas that are dominated by a polycentric structure, such as Los Angeles. Moreover, other regions, particularly the ‘sun belt’, do not suffer as much from the reliance on traditional manufacturing industries and thus are not affected to the same extent by economic restructuring. Comparative studies with other metropolitan areas would be beneficial.

Footnotes

Acknowledgements

The author is grateful to Claire Bozic of the Chicago Metropolitan Agency for Planning (CMAP) for providing data, to Stephen Hannon for research assistance, and to the three anonymous referees for their comments to improve the paper.

Funding

Research funding was provided by University of Wisconsin-Milwaukee Research Committee Awards.

Notes

References

Abramson

A. J.

Tobin

M. S.

VanderGoot

M. R.

(1995) The changing geography of metropolitan opportunity: the segregation of the poor in US metropolitan areas, 1970 to 1990, Housing Policy Debate, 6(1), pp. 45–72.

Anas

Arnott

Small

K. A.

(1998) Urban spatial structure, Journal of Economic Literature, 36(3), pp. 1426–1464.

Arnott

(1998) Economic theory and the spatial mismatch hypothesis, Urban Studies, 35(7), pp. 1171–1185.

Berube

Frey

W. H.

(2002) A Decade of Mixed Blessings: Urban and Suburban Poverty in Census 2000. Washington, DC: Brookings Institution Press.

Campbell

K. E.

Rosenfeld

R. A.

(1985) Job search and job mobility: sex and race differences, Research in the Sociology of Work, 3, pp. 147–175.

Clark

W. A. V.

(2008) Reexamining the moving to opportunity study and its contribution to changing the distribution of poverty and ethnic concentration, Demography, 45(3), pp. 515–535.

Cooke

T. J.

(2010) Residential mobility of the poor and the growth of poverty in inner-ring suburbs, Urban Geography, 31(2), pp. 179–193.

Doeringer

(1968) Ghetto labor markets: problems and programs. Paper presented at the Conference on Transportation and Poverty.

Fischer

(2003) The relative importance of income and race in determining residential outcomes in U.S. urban areas, 1970–2000, Urban Affairs Review, 38(5), pp. 669–696.

10.

Gans

H. J.

(2010) Concentrated poverty: a critical analysis, Challenge, 53(3), pp. 82–96.

11.

Giuliano

Small

K. A.

(1991) Subcenters in the Los Angeles region, Regional Science and Urban Economics, 21(2), pp. 163–182.

12.

Gobillon

Selod

Zenou

(2007) The mechanisms of spatial mismatch, Urban Studies, 44(12), pp. 2401–2428.

13.

Gordon

Kumar

Richardson

H. W.

(1989a) Congestion, changing metropolitan structure, and city size in the United States, International Regional Science Review, 12(1), pp. 45–56.

14.

Gordon

Kumar

Richardson

H. W.

(1989b) The spatial mismatch hypothesis: some new evidence, Urban Studies, 26(3), pp. 315–326.

15.

Handy

S. L.

Niemeier

D. A.

(1997) Measuring accessibility: an exploration of issues and alternatives, Environment and Planning A, 29, pp. 1175–1194.

16.

Harrison

(1972) The intrametropolitan distribution of minority economic welfare, Journal of Regional Science, 12(1), pp. 23–43.

17.

Hess

D. B.

(2005) Access to employment for adults in poverty in the Buffalo-Niagara region, Urban Studies, 42(7), pp. 1177–1200.

18.

Holzer

H. J.

(2008) Workforce development as an antipoverty strategy: what do we know? What should we do? Discussion Paper No. 3776, IZA, Bonn.

19.

Ihlanfeldt

K. R.

(2006) A primer on spatial mismatch within urban labor markets, in: Arnott

McMillen

(Eds) A Companion to Urban Economics, pp. 404–417. Malden, MA: Blackwell Publishing.

20.

Jargowsky

P. A.

(1996) Take the money and run: economic segregation in US metropolitan areas, American Sociological Review, 61(6), pp. 984–998.

21.

Jargowsky

P. A.

(1997) Poverty and Place. New York: Russell Sage Foundation.

22.

Jargowsky

P. A.

(2003) Stunning Progress, Hidden Problems. Washington, DC: Brookings Institution Press.

23.

Kain

J. F.

(1968) Housing segregation, negro employment, and metropolitan decentralization, Quarterly Journal of Economics, 82, pp. 175–197.

24.

Kasarda

J. D.

(1989) Urban industrial transition and the underclass, The Annals of the American Academy of Political and Social Science, 501, pp. 26–47.

25.

Kasarda

J. D.

(1993) Inner-city concentrated poverty and neighborhood distress: 1970 to 1990, Housing Policy Debate, 4(3), pp. 253–302.

26.

Kasarda

J. D.

Ting

K. F.

(1996) Joblessness and poverty in America’s central cities: causes and policy prescriptions, Housing Policy Debate, 7, pp. 387–419.

27.

Kneebone

Garr

(2010) The Suburbanization of Poverty: Trends in Metropolitan America, 2000 to 2008. Washington, DC: The Brookings Institution Press.

28.

Kneebone

Nadeau

Berube

(2011) The re-emergence of concentrated poverty. Center on Metropolitan Policy, The Brookings Institution, Washington, DC.

29.

Kuehn

(2011) Metropolitan job growth patterns in the great recession. Working Paper No. 1, The Urban Institute, Washington, DC.

30.

Lee

Leigh

N. G.

(2007) Intrametropolitan spatial differentiation and decline of inner-ring suburbs: a comparison of four US metropolitan areas, Journal of Planning Education and Research, 27(2), pp. 146–164.

31.

Lichter

D. T.

Parisi

Taquino

M. C.

(2012) The geography of exclusion: race, segregation, and concentrated poverty, Social Problems, 59(3), pp. 364–388.

32.

Massey

D. S.

(2001) Residential segregation and neighborhood conditions in U.S. metropolitan areas, in: Smelser

N. J.

Wilson

W. J.

Mitchell

(Eds) America Becoming: Racial Trends and Their Consequences, vol. 1, pp. 391–434. Washington, DC: National Academy Press.

33.

Massey

D. S.

Eggers

M. L.

(1993) The spatial concentration of affluence and poverty during the 1970s, Urban Affairs Quarterly, 29(2), pp. 299–315.

34.

McMillen

D. P.

Lester

T. W.

(2003) Evolving subcenters: employment and population densities in Chicago, 1970–2020, Journal of Housing Economics, 12(1), pp. 60–81.

35.

Muller

P. O.

(2004) Transportation and urban form: stages in the spatial evolution of the American metropolis, in: Hanson

Giuliano

(Eds) The Geography of Urban Transportation, pp. 59–85. New York: Guilford Press.

36.

Ong

P. M.

Blumenberg

(1998) Job access, commute and travel burden among welfare recipients, Urban Studies, 35(1), pp. 77–93.

37.

Painter

Liu

C. Y.

Zhuang

(2007) Immigrants and the spatial mismatch hypothesis: employment outcomes among immigrant youth in Los Angeles, Urban Studies, 44(13), pp. 2627–2649.

38.

Price

Mills

(1985) Race and residence in earnings determination, Journal of Urban Economics, 17(1), pp. 1–18.

39.

Puentes

Orfield

(2002) Valuing America’s first suburbs: a policy agenda for older suburbs in the Midwest. Center on Urban and Metropolitan Policy, The Brookings Institution, Washington DC.

40.

Rogers

C. L.

(1997) Job search and unemployment duration: implications for the spatial mismatch hypothesis, Journal of Urban Economics, 42(1), pp. 109–132.

41.

Ruggles

Alexander

J. T.

Genadek

Goeken

. (2010) Integrated public use microdata series: version 5.0 [machine-readable database]. University of Minnesota, Minneapolis.

42.

Sassen

(1990) Economic restructuring and the American city, Annual Review of Sociology, 16(1), pp. 465–490.

43.

Schill

M. H.

Wachter

S. M.

(1995) Housing market constraints and spatial stratification by income and race, Housing Policy Debate, 6(1), pp. 141–167.

44.

Schwartz

(1976) Migration, age, and education, The Journal of Political Economy, 84(4), pp. 701–719.

45.

Shen

(1998) Location characteristics of inner-city neighborhood and employment accessibility of low-wage workers, Environment and Planning B, 25, pp. 345–365.

46.

Shen

(2001) A spatial analysis of job openings and access in a US metropolitan area, Journal of the American Planning Association, 67(1), pp. 53–68.

47.

Simpson

(1992) Urban Structure and the Labour Market: Worker Mobility, Commuting, and Underemployment in Cities. New York: Oxford University Press.

48.

Sohmer

R. R.

Lang

R. E.

(2001) Downtown rebound. Center on Urban and Metropolitan Policy, The Brookings Institution, Washington, DC.

49.

Stoll

M. A.

(1999a) Spatial job search, spatial mismatch, and the employment and wages of racial and ethnic groups in Los Angeles, Journal of Urban Economics, 46(1), pp. 129–155.

50.

Stoll

M. A.

(1999b) Spatial mismatch, discrimination, and male youth employment in the Washington, DC area: implications for residential mobility policies, Journal of Policy Analysis and Management, 18(1), pp. 77–98.

51.

Stoll

M. A.

Covington

(2012) Explaining racial/ethnic gaps in spatial mismatch in the US: the primacy of racial segregation, Urban Studies, 49(11), pp. 2501–2521.

52.

Sullivan

D. H.

Ziegert

A. L.

(2008) Hispanic immigrant poverty: does ethnic origin matter?, Population Research and Policy Review, 27(6), pp. 667–687.

53.

Thomas

R. W.

Huggett

R. J.

(1980) Modelling in Geography: A Mathematical Approach. Lanham, MD: Rowman & Littlefield.

54.

Wilson

W. J.

(1980) The Declining Significance of Race: Blacks and Changing American Institutions. Chicago, IL: Universoty of Chicago Press.

55.

Wilson

W. J.

(1987) The Truly Disadvantaged: The Inner City, the Underclass, and Public Policy. Chicago, IL: University of Chicago Press.

Changing Job Access of the Poor: Effects of Spatial and Socioeconomic Transformations in Chicago,1990–2010

Abstract

1 Introduction

2. Spatial and Socioeconomic Transformations, Job Accessibility

2.1 Spatial Transformation

2.2 Socioeconomic Transformation

2.3 Job Accessibility

3. Study Area, Data and Methodology

3.1 Study Area

3.2 Data

3.2 Weighting Job Seekers and Their Jobs

Weighting Method to Explain the Spatial Transformation

Weighting Method to Explain Socioeconomic Transformation

3.4 Job Accessibility

4. Results

4.1 Poor Job Seekers and Jobs

4.2 Job Accessibility of the Poor

4.3 Effects of Spatial Transformation and Socioeconomic Transformation

5. Conclusion and Discussion

Footnotes

Acknowledgements

Funding

Notes

References