تغییرات داخل شهری در تقاضای نیروی کار و تنظیم بازارهای محلی

I offer new evidence on the adjustment of local labor markets to geographic shifts in labor demand within US metropolitan areas using a unique data set in which metropolitan subregions are geographically matched across the 1970–1980 and 1980–1990 decades. The evidence uncovered paints the following picture. Workers, especially those with less education, make incomplete adjustments within metropolitan areas in response to intra-metropolitan demand shifts. Although blacks may not make especially limited adjustments, they have disproportionately suffered deleterious effects from job movements because the demand shifts have tended to be away from their places of residence.

According to Kain’s [19] spatial mismatch hypothesis, the decentralization of employment in US cities has caused relative declines in the employment and earnings of inner-city blacks, whose ability to make complete residential adjustments is constrained. Yet, the extensive line of empirical studies testing this hypothesis has focused primarily on cross-sectional comparisons, leaving changes in labor market outcomes largely unexamined. In this paper, I provide new evidence on the relationship between intrametropolitan job movements and changes in labor market outcomes. Using Decennial Census of Population Data matched by geography over the 1970s and 1980s for a sub-set of metropolitan areas, I examine the effects of changing job locations on the employment and earnings of metropolitan area residents. In addition to estimating separate effects among blacks, I also measure the effects separately for several different demographic subgroups. Further, I implement an instrumental variables strategy using geographic variation in industry compositions to isolate the effects of intra-metropolitan demand shifts. Two different strands of literature provide relevant theoretical models. The spatial mismatch literature offers some depictions of how job movements can affect residentially constrained workers in an urban setting. In Brueckner and Zenou [8], blacks are at first residentially concentrated in the central city (due to a lower land consumption and a consequently steep bid–rent curve). When jobs exogenously decentralize, housing discrimination in the suburbs prevents blacks from following these job movements. The resulting mismatch causes central-city blacks to undergo diminished wages and, under an efficiency wages regime, increased unemployment. Arnott [2] establishes a general equilibrium framework in which discrimination forces blacks to live in the central city, from which commuting to the suburbs is costly. He suggests but does not fully explore comparative statics experiments in which exogenous factors such as changes in the technology of transportation cause exogenous increases in the demand for labor in the suburbs relative to the central city. Gobillon et al. [10] offer a review of several other recent papers describing how job decentralization may adversely affect labor market outcomes among residentially constrained workers. A second strand of the literature has modeled the effects of movements of labor demand between less proximate local labor markets, such as those native to different states or metropolitan areas. Topel [33] describes the spatial equilibrium occurring between locales and shows that geographic labor demand shocks affect equilibriumwages, especially among less mobile populations. Blanchard and Katz [4] model the dynamics of local employment, population, unemployment and wages. In their model, wages and unemployment rates respond to shocks in the short run but are unaffected in the long run. Bound and Holzer [6] also differentiate between the short- and long-runs, but emphasize that demand shocks will have different impacts on differently mobile groups of workers; they also explicitly describe a means by which nationwide shocks to different industries can generate such local labor demand shocks. The literature studying demand shifts between distant labor markets has provided various empirical estimates of the shifts’ effects on labor market outcomes of different workers [3,4,6,33]. But this literature has not explored the dramatic spatial variation in job growth within metropolitan areas. Meanwhile, empirical studies of spatial mismatch have most frequently related wage or employment levels at a single point in time to measures of proximity to jobs, job growth, or the city’s center (Ellwood [9], Ihlanfeldt and Sjoquist [16,17], Ihlanfeldt [15], Raphael [28],Weinberg [34,35]). These cross-sectional estimates of spatial mismatch may not very well approximate changes over time in job location and their effects. The employment and earnings levels among residents of the central-city or a particular neighborhood may reflect long-standing characteristics of the area or its population. In measuring the recent changes in labor market outcomes, levels estimates will confound such long-standing characteristics with the object of interest. More specifically, the literature on inter-urban wage differences (Roback [29,30], Greenwood et al. [13], Mueser and Graves [26]) highlights the importance of residential and productive “amenities” in determining spatial equilibria. It is likely that such amenities would have important effects on the geographic distribution of wages and employment within metropolitan areas as well. For example, the effects of natural features such as rivers may cause one city to be more centralized than another; special features of the housing stock built up many generations ago may also favor one neighborhood over another. Since these amenities will affect the sorting of workers across the metropolitan area, they may cause comparisons of wage and employment levels to be poor proxies for comparisons of wage and employment changes. Few studies have analyzed the effects of intra-metropolitan job movements by directly examining changes in labor market outcomes. Zax and Kain [36] perform a case study by following the employees of one large employer that relocated; they obtain some provocative results indicating racial differences in mobility. To measure labor market adjustment on a larger scale, some studies exploit aggregated data on locality-based employment rates (Kasarda [20]; Mouw [25]); these studies have also provided evidence that decreases in the proximity of jobs to residences adversely impact black workers’ labor market outcomes. Martin [24] uses aggregated data to document the evolution of differences between job locations and black residences and the extent to which black migration has ameliorated them. Perhaps due to limitations in the available data, no studies of intra-metropolitan job movements have examined measures of changes in labor market outcomes using large microdata sets. In this paper, I overcome these data limitations for a subset of US metropolitan areas by matching geographical entities across decades using US Census of Population microdata. Using these data, I estimate the effects of job location changes on employment and earnings changes while controlling for the population composition. My findings, in brief, are as follows. A relative increase in labor demand in a given metropolitan subregion results in increased nominal wages and employment probabilities among the subregion’s residents. However, these effects are smaller than would be measured by analyzing outcomes in levels. The effects are present among virtually every broad subpopulation, but they are significantly larger among less-educated workers. The evidence on racial differences, however, does not consistently show that blacks are less able to adjust to shifting job locations. The results show that supply of labor within metropolitan areas is not perfectly elastic across space, and that the adjustments of some groups of workers are especially constrained.

In this paper, I have exploited a unique data set matching county groups over decades to characterize the impacts of geographic shifts in labor demand within metropolitan areas. The results show that labor demand shifts over ten-year periods have significant impacts on the geographic distribution of both the employment rates and the wages of metropolitan residents. Additionally, these impacts on both wages and employment rates are greater among the relatively less educated. These results support a number of conclusions. They indicate that the supply of workers between subregions of metropolitan areas is imperfectly elastic, especially among lesseducated workers. They do not, however, consistently show that black workers have less elastic supply between subregions than whites, once other demographic differences are accounted for. Nevertheless, they confirm that job movements within metropolitan areas during the 1970s and 1980s contributed to the deterioration in relative wages and employment rates of blacks, because black residences at the outset were relatively concentrated in areas that had low job growth. The validity of these results rests on the extent to which the data have captured the relevant variation in job opportunities between metropolitan subregions and the sufficiency of the variables included to control for residential sorting. On the first score, this study has improved greatly on previous research by considering changes over time in the job opportunity and labor market outcome measures. These data much more likely capture demand shifts rather than permanent differences between areas in the abundance of jobs. On the second score, the issue of completeness of controls, this study has improved significantly on the few studies that have examined changes in labor market outcomes and opportunities.