Neighborhood Changes in Wayne and Oakland Counties, …



Neighborhood Changes in Wayne and Oakland Counties, Michigan

Glenn Wright

University of Texas, Austin

December 5, 2006

CRP 386 – Introduction to GIS

Executive Summary

Wayne County and Oakland County are two neighboring counties in southeastern Michigan. Wayne County contains the city of Detroit, often considered the embodiment of Rust Belt decline. Detroit is tied with Cleveland as the poorest large city in the nation and has the highest percentage African American population of any large city. Oakland County, in contrast, which contains several mostly white suburbs, is the 26th wealthiest county in the nation by per capita income. The economies of Wayne and Oakland Counties are intertwined: many of the most highly paid managers and employees of firms located in Detroit live in Oakland County.

The 2000 census revealed higher incomes, higher housing values, and less poverty in Detroit than did the 1990 census, reversing (at least temporarily) a trend of several decades of decline. Many observers were surprised, for there was no notable revival of industry in the area that would explain these changing fortunes.

Whether Detroit experienced a genuine revival in the 1990s is unknown. Incomes could have risen because wealthy residents were moving into the area or because salaries for existing residents were increasing, but incomes could also rise as a result of poor residents moving away. This paper investigates census demographic data for Wayne and Oakland County and in the City of Detroit between 1990 and 2000 in an attempt to gain a more detailed picture of the changes that took place.

Because the census data does not track individual families, it is impossible to know for certain how to interpret demographic changes. Nonetheless, the population of Detroit declined during the period of observation while the black population of nearby census tracts increased. Furthermore, median incomes in the center of Detroit increased while median incomes in surrounding areas decreased, which suggests that rising incomes in Detroit may have been the result of the poorest residents moving out, rather than increasing incomes among poor residents. Finally, some tracts in the downtown area experienced increasing white populations, which could be an early sign of gentrification.

Introduction

Over the past several decades, many cities in the Northeastern and Midwestern United States have suffered from a variety of interrelated problems. While whites have fled the cities for the suburbs, low-income blacks have become concentrated in the inner cities. At the same time, manufacturing industries - traditionally a gateway to the middle class for unskilled workers – have moved from cities to the suburbs and have also moved from the “Rust Belt” cities of the Midwest and Northeast to the “Sun Belt” cities of the South and Southwest or overseas. As a result of these changes, population and incomes have declined for decades in cities such as Cleveland, Pittsburgh, and above all, Detroit.

Detroit’s automobile industry exemplified these changes in industrial structure. While American automobile companies were facing increasingly fierce competition from foreign automobile companies, especially in Japan and Korea, the American companies were moving many of their factories out of Detroit – either to the nearby suburbs, to Sun Belt cities with fewer unions and lower labor costs, or to Mexico or other countries. In many cases, the populations of the Rust Belt cities have declined as people move to the Sun Belt in search of jobs. In the case of minorities, especially blacks, the migrations are frequently to the eastern areas of the South (Frey 2003, p. 1). Detroit’s growth during the first half of the twentieth century was due almost entirely to the automobile industry, and its decline in the second half of the twentieth century followed the fortunes of that industry as well.

This simple rise-and-fall story, however, masks a few complexities. First, while it is well known that the automobile industry has moved its factories to other states or oversees, fewer people realize that the industry has actually consolidated its management and research operations Detroit in the past several decades. Because many of the most highly paid professional and management workers in these industries live in the neighboring suburbs rather than in Detroit, Detroit gains few benefits for hosting these high tech industries (Wright, 2006.) Detroit may be able to recapture some of the benefits (tax dollars and local spending) of its industries if it can encourage wealthier workers to live in the city rather than in the suburbs. In many cities such “urban renewal” has been known as “gentrification” and has led to displacement of low-income residents from the renewed areas. But because Detroit has an extraordinarily high number of vacant buildings, these human costs may be considerably less than in other cities.

Second, the last ten years of the twentieth century featured an interruption in many of the trends that had plague Detroit for decades. The 2000 census revealed, to the surprise of many, higher incomes, higher housing values, and less poverty in Wayne County than did the 1990 census. These data, combined with recent reports of an influx of high-income residents into several Detroit neighborhoods, reveals that “urban renewal” may already be underway. Other cities, especially in the Midwest, experienced similar changes (Jargowsky 2003, p. 6.)

Many other cities, however, that have experienced “urban renewal” have also seen its dark side: gentrification. Sociologist Ruth Glass first described the phenomenon of gentrification in 1964 to describe changes in working class London neighborhoods. Some working class neighborhoods became desirable places to live for high-income individuals, who bid up housing prices in the area, driving out the previous inhabitants. Though there is no official definition of “gentrification”, it is usually used to refer to physical improvements in low cost neighborhoods and the replacement of the prior low-income residents with new high-income residents. Though this process can bring new businesses and increased tax revenues to a city, it also displaces residents whose low incomes make them ill prepared to face unstable circumstances.

Detroit certainly experienced neighborhood changes between 1990 and 2000, but were those changes gentrification? If not, what were they? Geographic information systems (GIS) can be a useful tool for the study of gentrification and other issues of neighborhood changes. For example, a paper by Gina Clemmer (2000) uses GIS to combine data from the American Community Survey, the Home Mortgage Disclosure Act, and local business directories to analyze gentrification in Portland, Oregon. Several features of GIS make it useful for studying data of this kind. First, it allows simultaneous visualization of multiple layers of data, which makes it possible to recognize spatial patterns that might not otherwise be obvious. Second, it allows analysis of geographic units at several different scales, in this case counties, census tracts, and census block groups. Finally, GIS can produce output that is understandable even to those who do not have experience reading scholarly papers.

Research Questions

This paper will examine the changes in Detroit neighborhoods between 1990 and 2000 in an attempt to quantify “urban renewal.” A similar analysis will be performed for Oakland County, whose economy is highly interdependent with Detroit’s.

1. In many Detroit neighborhoods, average household income and housing values rose between 1990 and 2000, while poverty declined. Were these changes due to rising incomes of people already living in the neighborhoods, wealthier people moving in, poor people moving out, or some combination of the above?

2. If wealthier people are moving into some Detroit neighborhoods, what is the effect on poor people? Are rents rising? Do these neighborhoods have high percentages of vacant or owner-occupied housing units?

3. How do changes in Oakland County neighborhoods compare to those in Detroit?

Methodology

Neighborhood changes can be measured in a variety of ways, including race, income, housing values, and many other variables. The United States Census Bureau collects extensive data of this kind, some of it on the normal census form (SF1 - filled out by every householder) and some of it in the long form (SF3 - filled out by only one out of every six householders.) Other data sources have more detailed information on land use and building, but the primary focus of this paper was on demographic changes.

This study used data from the 1990 and 2000 United States Census. I downloaded census tract polygons for both years, roads, municipal boundaries, and SF1 demographics for 2000 ESRI’s geography network site, but SF3 demographics for 2000 and both sets of demographics for 1990 were not available at that site and I downloaded them from the Census Bureau’s website using their American FactFinder interface.

The following demographic variables were downloaded.

- Median Household Income

- Median Housing Value of Selected Units

- Median Contracted Rent

- Year Householder Moved Into Unit

- Owner Occupied Housing Units

- Vacant Housing Units

- Population

- Race

- Poverty

After I downloaded the demographic data, I joined the data tables to the census tract polygons for the appropriate year. Next, I compared the polygons from the two years visually to determine where tract boundaries had changed between 1990 and 2000. Because I did not have data at the block group level, I realized I would have to aggregate data from the two time periods always to the larger of the two tracts. For example, if two tracts from 1990 merged to form one tract in 2000, I aggregated the data from the two to form a unit comparable to the tract in 2000. If one tract in 1990 split into two tracts in 2000, I aggregated the data from the two to form a unit comparable to the tract in 1990. Thus, my final units of analysis (and display) were not identical to the set of census tracts from either 1990 or 2000, but in fact included several dozen “grouped” census tracts. I will hereafter refer to my units of analysis as “adjusted census tracts.” I decided to ignore minor changes to the shapes of tracts between the two time periods.

I created a table that showed which adjusted census tract each actual census tract in 1990 and 2000 was associated with. I then used the Dissolve tool on the adjusted census tract ID field of the 2000 shapefile to create a set of polygons that could contain demographic data from both years. Medians for aggregated tracts were averaged and other variables from aggregated tracts were added together to calculate variable values for the adjusted tracts. Finally, I joined the data table from the dissolved 1990 tracts to the shapefile for the dissolved 2000 tracts in order to create a single “adjusted tracts” shapefile that had values for every variable in both years.

Once the final shapefile was created, I experimented with many five-level choropleth visualization schemes to see which ones revealed the most about demographic changes in the area. In many cases, I add to calculate additional fields in order to normalize variables the way I wanted, because ArcGIS will only allow normalization of one variable by another. In order to display absolute change in percent black population, for example, I had to calculate percent black in 2000 and then subtract from it percent black in 1990.

After much experimentation, I chose the set of choropleths that seemed to best provide the best evidence addressing my research questions. I decided that some variables were of interest chiefly within Detroit’s municipal boundaries, so I created a secondary shapefile by clipping the adjusted census tracts file by Detroit’s boundaries. Finally, I adjusted the symbology of these choropleths to provide the best information possible and visually analyzed them, looking for spatial patterns in demographic change.

Findings

Maps A1 through A9 show information about Wayne and Oakland Counties. Maps B1 through B7 show Detroit only, in greater detail.

Map A1 shows municipal boundaries from the year 2000 in Wayne and Oakland Countries. Detroit is by far the largest area, but there are several dozen other municipalities of varying sizes in the area. Three areas merit special mention. Highland Park and Hamtramck are municipalities within Detroit that have historically chosen to remain independent from the city government. Gross Pointe and surrounding municipalities (Gross Pointe Woods, Shores, Park, and Farms; also Harper Woods), though located between Detroit and Lake Huron, are also independent of Detroit, and these suburbs, unlike the Highland Park and Hamtramck, are very wealthy. Finally, the city of Pontiac in Oakland County is urbanized and resembles Detroit demographically and economically more than it resembles the rest of Oakland County. Map A1 can be compared with other maps in the report to determine which political boundaries correspond to measured phenomena.

Map B1 shows neighborhood boundaries in Detroit. This map is associated mostly with maps B2 through B7, but the areas on this map can also be used to describe some trends from maps A2 through A9.

Maps A2, A4, A6, A8, and B5 show values for demographic variables in 2000, with low values represented by light shades of blue and high values represented by dark shades of blue. Maps A3, 5, A7, A9, B2-B4, B6, and B7 show changes in demographic variables between 1990 and 2000, with large negative changes represented by dark red, small negative changes by pink, small positive changes by pale green, and large positive changes by dark green. The changes are measured in different ways on different maps, so it is important to check the analysis section or the appendix to resolve ambiguities. The maps of absolute change in white and black population (B3 and B4, respectively) are shown with the same scale to allow easy comparison.

Several of the Detroit maps show areas of rapidly decreasing population or rapidly increasing rent using a diagonal line fill pattern overlayed on the choropleth. This provides yet another way to compare tracts.

On every choropleth map in this section, the middle three categories cover equal intervals, and whenever possible those values were chosen so that they divide the largest parts of the sample distribution into roughly equal sizes. The highest and lowest categories, however, extend to the highest and lowest points in the distribution, regardless of the size of the interval. In cases where data was missing or a normalization variable was zero, the tract is not shown.

[pic]

[pic]

[pic]

[pic]

[pic]

[pic]

[pic]

[pic]

[pic]

[pic]

[pic]

[pic]

[pic]

[pic]

[pic]

[pic]

Analysis

Maps A2 through A9 describe demographic and economic trends in Wayne and Oakland Counties for 1990 and 2000, providing context for changes in Detroit during that time period.

Map A2 shows population density (persons per square mile) in 2000. Detroit is the most densely populated area, followed by the city of Pontiac and the suburbs immediately surrounding Detroit. The downtown area of Detroit is less densely populated, probably because that area is partially given over to commercial activity and public buildings.

Map A3 shows percent change in population between 1990 and 2000. The area’s population loss continued: most of the tracts in the area experienced population declines of between 0 and 50 percent. Interestingly, while all but a few tracts shrank in the downtown area, the tracts within Detroit near the northeastern and southwestern edges of the city experienced population growth. A few other areas in western Wayne County and Southwestern Oakland County experienced significant growth during the time period.

Map A4 shows blacks as a percentage of total population. Clearly the area is highly segregated – most of the tracts fall in the extreme categories, either less than 20% black or more than 80% black. Detroit is overwhelmingly black, except for the southern tip and the enclosed area of Hamtramck, which is mostly non-black (Hamtramck hosts a large Polish American community.) The cities of Pontiac and Inkster are also largely black, as are two inner suburbs to the north of Detroit, Southfield and Oak Park. The rest of the suburban and rural areas are overwhelmingly white.

Map A5 shows the absolute change in percent black from 1990 and 2000 – in other words, percent black in 2000 minus percent black in 1990. The most striking change over the time period is that the northeastern and western areas of Detroit became proportionally much more black over the time period, while some downtown areas became less so. Most other tracts in the two counties increased the black share of their population only slightly, by between 0 and 10 percent, though percent black declined in several rural tracts in the northwest of Oakland County.

Map A6 shows median household income. Detroit and Pontiac had much lower median incomes than the suburban areas, and the suburbs in Wayne County had somewhat lower median incomes than those in Oakland County. Most of the tracts closest to the central business district in Detroit had lower incomes than those near the edges of the city.

Map A7 shows percent change in median household income between 1990 and 2000. The changes are measured in real dollars – adjusted by the change in Consumer Price Index (a measure of inflation) by a factor of 1.278 between 1990 and 2000. Incomes increased rapidly in the center city and in rural areas while increasing slowly or decreasing in the suburbs and near the edges of Detroit.

Map A8 shows median values for “selected housing units” in 2000. Median housing values in Detroit were lower, in general, than those in Oakland County or the rest of Wayne County. The greater wealth of Oakland County suburbs as opposed to Wayne County suburbs is more visible in this map than in Map A6.

Map A9 shows percent change in median housing values from 1990 to 2000, also corrected for changes in the Consumer Price Index (measures of housing inflation are available, but the Consumer Price Index was used as it is a measure of inflation closer to what laymen think of as “inflation.”) Housing values outside Detroit for the most part declined or stayed steady, while the story was more mixed in Detroit: some tracts had median values go up, whereas others (perhaps slightly fewer) had them go down.

The analyses of the area confirm media accounts of census findings. The area continues to lose population, and housing values and incomes rose rapidly in Detroit than in other areas of the two counties. Many of the wealthiest suburbs actually experienced declines in median income and housing values. These maps also reveal a phenomenon not mentioned in the media: the apparent movement of blacks from the center of the city to the edges and to several inner suburbs, and the faster increases in incomes and housing values in the center city.

These findings reveal at least the possibility that “gentrification” is taking place – it is possible that the blacks moving to the edges of the city had low incomes and had previously been residents of the center city, displaced by rising rents and replaced by former suburbanites. The one major difference between this case and classic cases of gentrification is that the overall population of Detroit is still declining, and the downtown is becoming less densely populated rather than more. Gentrification is more often associated with growing areas, especially on the West Coast.

The close-up maps of Detroit, B2-B7, allow a more nuanced look at changes in the city.

Map B2 shows percentage change in median contracted rent from 1990 to 2000. Interestingly, rent is falling in the business district and rising in the west tip of the city. In the rest of the city, rents are sporadically going up or down, with no obvious spatial pattern (a map not included in this report shows that the sporadic variation is typical of the areas outside Detroit as well.) Interestingly, rent is increasing in many of the tracts that lost population between 1990 and 2000, whereas ordinarily one would expect that rents in these tracts would decrease.

Maps B3 and B4 show the absolute change in the number of white and black persons, respectively, from 1990 to 2000. More so than income, race can gives clues about who is moving where, though because the census does not publish data on individuals we cannot know for certain where people are moving from and to. These maps seem to reveal that whites are leaving the edges of the city in great numbers. The white population is actually not increasing in many of the tracts in the center of the city, either – some tracts had more whites in 2000 than in 1990 while others had fewer in 2000 than in 1990. Blacks, on the other hand, seem to be uniformly leaving the center city and moving to the edges. Most of the tracts in which rent increased by more than 25% experienced a decline in black population, but these tracts showed no particular tendency to increase white population.

Maps B5 and B6 look at, respectively, percent of housing units vacant and absolute change in percent vacant (percent vacant units in 2000 minus percent vacant units in 1990.) Vacancy is an important issue in predicting the impact of neighborhood changes in Detroit, because in an area with many vacant buildings, new residents in a neighborhood need not necessarily displace existing residents. Map B5 shows that vacancy was higher in the central and southern areas of the city than near the northern, eastern, and western edges (a map not included in the report shows that vacancy rates outside Detroit are dramatically lower than rates in Detroit.) Map B7 shows that vacancy rates increased in the central city except along the waterfront and decreased in the western part of the city. Many of the tracts in which rent was increasing quickly have large numbers of vacant buildings, which is surprising because the owners of vacant units should have an incentive to lower their rents in order to attract tenants.

Map B7 shows the same data on median income change that map A7 does, but on a smaller scale and with additional information about population changes overlayed. Many but not all of the inner city tracts with rising incomes also saw declines in population of more than 10%. This suggests that some, but not all, of the measured increases in income between 1990 and 2000 were due to emigration of low-income residents.

Conclusions and Further Research

None of the maps in this report, considered individually, tells a full story about neighborhood changes in Detroit over the time period in question. Even taken together, they do not present firm conclusions. The great limitation of the census data is that it does not track individuals between 1990 and 2000, so we can only make educated guesses about what people are doing based on aggregate patterns. Nonetheless, some of these patterns are clear.

The area that seems most likely to have experienced gentrification is the area near the center of the city, surrounding but not including the business district. In these areas, the number of black residents decreased, incomes, rents, and property values went up, and in some of these tracts, the number of white residents increased. The numbers of blacks moving out of the area, however, greatly exceed the number of whites moving in, and the number of vacant housing units increased rapidly during the time period. In other words, this looks like a continuation of Detroit’s decades-long trend of residents leaving to seek better opportunities elsewhere rather than a new response to the “physical improvement” that gentrification entails. In addition, the increase in rent in this area is not dramatically larger than the increase in the rest of the city: in all areas of the city, including the ones that saw increases in black population, the great majority of tracts experienced rent increases between 0 and 25 percent. It seems unlikely that a change of this size could trigger migrations on the scale that the maps show. Finally, a comparison of individual tracts within the area across maps shows that the tracts with the highest increase in rents were not the ones with the highest increase in white population or the highest decrease in black population. This could be evidence that three phenomena affecting the city center do not have the direct causal relationship with each other that would suggest gentrification.

Nonetheless, the possibility of future gentrification seems obvious: if the center city tracts continue to attract new middle-income residents and increase rents, low-income residents may be pushed out. The high vacancy rates in the area, however, suggest a solution: if the City enacts policies (such as land banking) that encourage renovation of abandoned housing, then housing supply will greatly exceed demand and competition among renters may keep rents stable. This would not only protect the low-income residents in the area, but it would also encourage additional middle-income residents to move in (admittedly, if this process continued for long enough, rents would eventually start rising.)

More generally, these analyses caution against accepting an overly optimistic interpretation of the 2000 census findings. At the same time incomes were increasing in Detroit, very large numbers of people were moving out of the city and a (smaller) new migration of whites to the center city was taking place. Thus, the new income figures probably reflect not alleviation of poverty but rather dispersion of poverty – a process that may well be desirable, but only in less direct ways. On the other hand, there were many tracts in which population increased that also experienced increases in median income.

A next logical step in understanding these issues would be research of a more finely grained or qualitative variety. An attempt could be made to locate families or individuals who are representative of those who moved out of the center city in the 1990s and find out what reasons led them to move. The tracts that showed possible signs of “revitalization” could also be examined to determine whether those trends were only random deviations or whether they continued between 2000 and 2006.

Another avenue for further research is investigation of data that reflects physical improvements to neighborhoods (new buildings and the like). In particular, an analysis could focus on changes in areas where people have expressed concern about gentrification. Concerns about gentrification in two areas, the Cass Corridor and Brush Park (which together make up the “Lower Woodward” neighborhood on map B1), have become common since 2000, though the maps in this report show no evidence of gentrification in those tracts between 1990 and 2000.

Appendix

Census tract polygons for 1990 and 2000, roads, municipalities, and SF1 demographic data for 2000 were obtained from the Census Tiger/2000 section of ESRI’s Geography Network website (). SF3 demographic data for 2000 and STF1 and STF3 data for 1990 were downloaded from the American FactFinder section of the United States Census Bureau website ()

Name: Census 2000 TIGER/Line data

Provider: U.S. Bureau of the Census

Coverage: Wayne and Oakland Counties, Michigan

Coordinate System: Geographic coordinates NAD83

Units: Decimal Degrees

Delivery: Shapefile

Detroit neighborhood boundaries (not an important part of this report) were downloaded from the GIS Layers section of the City of Detroit Department of Planning and Development website ()

Name: Neighborhood Clusters

Provider: City of Detroit Department of Planning and Development

Coverage: Detroit, Michigan

Coordinate System: US State Plane Coordinate System, Michigan 2113, Southern Zone, NAD83

Units: US Survey Feet

Delivery: Shapefile

The correct projections were defined for these files, and then all shapefiles were reprojected into the the US State Plane Coordinate System, Michigan 2113, Southern Zone, NAD83 coordinate system.

Assembling data for this analysis required several significant steps. The first was to join demographic information from the Census SF1 and SF3 (for 1990 STF1 and STF3) data to tracts. This was accomplished using the Join procedure.

The most challenging step was to create the “adjusted census tracts” that allowed for comparison between 1990 and 2000. Because block group level data were not easily available, some of the conversions between 1990 and 2000 tracts were inexact. Nonetheless, in most cases the transformation between the two years was very simple: either two tracts in 1990 became one tract in 2000, or vice versa. In a few cases, two tracts in 1990 mapped onto two differently oriented tracts in 2000. In any case, I created an Excel table that instructed on how to link census tract numbers from 1990 and 2000 to “adjusted census tracts”, polygons drawn from the more general of the two layers.

I then Joined that table to both county shapefiles and used the Dissolve procedure. I chose the adjusted census tract ID field as the Dissolve Field and created Statistic Fields for each of the demographic variables. Every variable that represented a count was aggregated using the SUM option, and every variable that represented a median was aggregated using the MEAN option. This is not ideal; it would have been better to weight the means. For example, if two census tracts are aggregated, one with 100 people and a median income of $20,000 and the other with 50 people and median income of $26,000, the larger tract should receive twice the weight of the smaller, such that the calculated mean of median income is $22,000. ArcGIS does not support this technique and my attempts to calculate it in other ways were met with repeated errors. Instead, I used the unweighted mean (which in this example would be $23,000.) In most cases, the tracts that were aggregated were of similar sizes, but in a few cases a small tract was aggregated with a large one. In those few cases (the island just south of eastern Detroit is a notable example) the values for median income, median rent, and median value are untrustworthy.

Next, the attribute tables from the 1990 Dissolved shapefiles were joined to the 2000 Dissolved shapefiles in order to create a single shapefile that had data for both years. The polygons from the two counties were then placed in the same shapefile using the Append procedure. The Detroit municipal boundary shapefile was created by opening the attribute table for the municipalities shapefile, selecting Detroit, and exporting that single record to another shapefile. The major roads shapefile was created by selecting from the downloaded roads shapefile by road type such that only major roads and highways were selected, then exporting those to a separate shapefile. The major roads file for the Detroit maps was created by Clipping the larger major roads file by the Detroit municipal boundary.

Numerous extra fields were calculated in order to display information on the choropleth maps. Most notably, normalizing by area required a formula found at :

Dim dlbArea as double

Dim pArea as IArea

Set pArea = [shape]

dblArea = pArea.area

In the maps showing change in income, rent, and value, fields were created by dividing the variable for 2000 by 1.278, the ratio of the Consumer Price Index in 2000 to that in 1990. Maps showing percent changes in numeric variables required no additional calculations, but maps showing absolute change in percentile values required some algebraic manipulation to make sure ArcGIS never divided by zero and generated an overflow error. An equation of the form (A/B – C/D), which could, for example, represent (black population 2000 / total population 2000 – black population 1990 / total population 1990), can be rewritten as (A*D – C*B) / (B*D), which allows ArcGIS to use its normalization algorithms (which render polygons with zero denominators invisible) rather than its Calculate Field algorithms, which crash when division by zero is attempted.

Bibliography

Gina Clemmer. Quantitative and Spatial Analysis Techniques for Analyzing Gentrification Patterns. Independent Research Project at Portland State University.

William H. Frey. Diversity Spreads Out: Metropolitan Shifts in Hispanic, Asian, and Black Populations Since 2000. March 2006. The Brookings Institution.

Paul A. Jargowsky. Stunning Progress, Hidden Problems: The Dramatic Decline of Concentrated Poverty in the 1990s. May 2003. The Brookings Institution.

Glenn Wright. Economic Profile of Detroit, Michigan and Wayne County. November 8, 2006. Finale report for PA393L – Economics of Urban and Regional Policy, professor Robert Wilson.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download