A better day in the neighborhood: The rise and decline of poverty concentration in the Twin Cities, 1970-2000

The percentage of individuals nationwide whose family income falls below the federal government's official poverty level varied only slightly between 1970 and 2000, according to the U.S. Census Bureau. Over the same period, the spatial concentration of poverty—the percentage of low-income individuals who live in low-income neighborhoods—was much more volatile. From 1970 to 1990, this and related measures of the concentration of poverty rose sharply across the nation—particularly in the Midwest, including Minneapolis-St. Paul.

Poverty's prolonged, pervasive tendency to concentrate, mostly in inner-city neighborhoods, led to gloom about the future, typified by poverty expert Paul Jargowsky's assessment that "Social conditions in high-poverty neighborhoods have deteriorated, fueling more abandonment in a cycle of decay that, with few exceptions, seems immune to policy interventions or private initiatives." 1/ Then, in the 1990s, poverty concentration unexpectedly fell, with especially steep declines in the Twin Cities and other Midwestern metropolitan areas.

The combination of a rapid decline in concentrated poverty with only a modest decline in the national poverty rate may seem to suggest that the deconcentration resulted mainly from the relocation of poor and nonpoor households. Indeed, relocating poor households out of high-poverty neighborhoods (HPN) was a goal of housing policy in the 1990s. In pursuit of that goal, many high-density public housing projects were torn down, including some in Minneapolis.

However, a closer look at the evidence suggests that a reduction in poverty, not just increased mixing of existing poor and nonpoor households, was an important cause of poverty deconcentration, at least in the Twin Cities and much of the Midwest. In the 1990s, these areas experienced strong growth in earned income and a sharp decline in the percentage of individuals in poverty. In the Midwest and Twin Cities, these positive economic developments benefited populations that are overrepresented among the poor and the concentrated poor. For example, poverty rates in Minneapolis-St. Paul fell sharply for most racial minorities and for children and adults living in single-parent households.

This article explains the concept of poverty concentration and summarizes its national and regional trends since 1970, including the 1990s reversal. Those facts then serve as background for a tentative discussion of some of the factors associated with poverty deconcentration in the Twin Cities and other regions.

Defining and measuring poverty concentration

Poverty concentration relates to whether low-income households are dispersed throughout the population or clustered together in areas where poverty is common. Some clustering is likely to occur simply because the cost of housing differs by neighborhood, and people with low incomes will gravitate to neighborhoods where housing costs less. Other factors, such as the advantages that low-income immigrants may find in living near each other, can give rise to neighborhoods with a high rate of poverty. To assess how all these factors affect the spatial intermingling of low-income and other households, researchers have developed precise definitions and measurements of poverty concentration.

Any definition begins with a definition of poverty itself. This article uses the federal government's official definition of poverty. Since the 1960s, the government has established annual poverty thresholds intended to represent the cost of basic needs for families of various sizes. In 1999, the year for which Census 2000 measures income, the poverty threshold was $13,290 for a typical family of three and $17,029 for a typical family of four. The government defines an individual as poor, or in poverty, if his or her household income is below the threshold for households of the same size. This definition of poverty has many faults, 2/ but it is widely used and has the advantage that the U.S. Census Bureau enumerates households in poverty for each census tract.

Measuring concentration also requires that we define the neighborhoods, or spatial areas, in which concentration will be measured. Census tracts, as defined by the U.S. Census Bureau, are the most convenient and widely used concept of neighborhood in the study of poverty concentration. According to the bureau, "census tracts in the United States ... generally have between 1,500 and 8,000 people, with an optimum size of 4,000 people. ... When first delineated, census tracts are designed to be relatively homogeneous with respect to population characteristics, economic status, and living conditions." 3/ I will use the terms "neighborhood" and "census tract" interchangeably. 4/

The poverty rate—the percentage of residents living in poverty—varies widely among neighborhoods. Field work and data analysis suggest that the character of a neighborhood changes significantly as the poverty rate rises above a threshold of about 30 to 40 percent. Above that level, outsiders are more likely to perceive the neighborhood as poor and unattractive. These neighborhoods often have a high proportion of vacant or rundown buildings and a dearth of neighborhood shops and businesses. Often, among their residents, rates of unemployment and single parenthood are above average, while rates of educational and vocational attainment are low. 5/ I will define an HPN as one in which 40 percent or more of the residents live in poverty. 6/

HPNs are the basis for various measurements of the concentration and spatial arrangement of poverty. One measurement is the number of such neighborhoods, nationally or in a region or city. This number reflects the spatial "footprint" or extent of HPNs. A second measurement is the percentage of people, poor and nonpoor, living in such neighborhoods, which indicates the fraction of the total area population that is exposed to the negative effects often associated with HPNs. (For more information on these negative effects, see the sidebar.) A third measurement, the concentration of poverty,is defined as "the percentage of the poor in some city or region that resides in high-poverty neighborhoods ... [It] captures the percentage of the poor individuals who not only must cope with their own low incomes, but also with the economic and social effects of the poverty that surrounds them." 7/ Looking at these three measurements of the prevalence of HPNs provides a broad assessment of these neighborhoods' extent and significance.

1970-1990: Increasing concentration of poverty

From 1970 to 1990, all three indicators rose sharply. Paul Jargowsky analyzed these trends using a fixed set of 239 metropolitan areas for which data are available from 1970 on. 8/ His data, selectively reproduced in Table 1, show that the number of HPNs in these areas more than doubled from 1970 to 1990. Over the same period, the number of all people living in HPNs increased by over 90 percent, the number of poor persons living in HPNs essentially doubled and the concentration of poverty in these metropolitan areas—the percentage of the poor living in HPNs—jumped from 12.4 to 17.9 percent. Meanwhile, the total numbers of people, poor people, and neighborhoods in these metropolitan areas each grew much more slowly, by roughly a third, and the overall metropolitan poverty rate—the percentage of all people in those 239 metropolitan areas living in households with incomes below the poverty level—increased only moderately, from 10.9 to 11.8 percent.

Within major racial and ethnic groups in the metropolitan areas Jargowsky studied, exposure to HPNs increased significantly from 1970 to 1990 (see Table 1). The increases were not a direct reflection of poverty rates, for Table 1 shows that from 1970 to 1990, metropolitan poverty rates for all persons, African Americans and Hispanics were relatively stable.

The group Jargowsky labels "white" actually includes all non-Hispanic non-African American individuals in many of his tables, due to limitations in the data on race and ethnicity in the 1970 census. 9/ From here on, the label "whites/others" refers to this group, and "white" refers solely to non-Hispanic whites. For metropolitan whites/others, population growth from 1970 to 1990 was slow (18 percent), but HPN residency and poverty concentration rose rapidly from initially low levels. The number of metropolitan whites/others living in HPNs nearly doubled (up 90 percent), and their concentration of poverty—the percentage of poor individuals living in HPNs—more than doubled, to 6.3 percent.

For metropolitan African Americans, overall population growth was rapid (41 percent). The increases in HPN residency and poverty concentration were somewhat less steep than for metropolitan whites/others, but from a much higher initial level. The number of individuals living in HPNs grew rapidly, by 70 percent, but the percentage living in HPNs rose more moderately, from 14.4 to 17.4. African American poverty concentration rose from 26.1 percent in 1970 to 33.5 percent in 1990.

The number of metropolitan Hispanics grew rapidly, but there was relatively little change in their rates of HPN residency and poverty concentration. Their overall metropolitan population rose 148 percent between 1970 and 1990, contributing to a 171 percent increase in the number of Hispanics living in HPNs. But the percentage of all metropolitan Hispanics who lived in HPNs rose moderately, from 9.6 in 1970 to 10.5 in 1990, while the percentage of the Hispanic poor living in HPNs fell slightly to 22.1 in 1990.

The incidence of HPNs showed divergent regional patterns between 1970 and 1990. The number of HPNs in the metropolitan areas Jargowsky studied grew relatively slowly in the South and West, from 798 to 1,255, a 57 percent increase. In other parts of the country, such as New England, the Mid-Atlantic and the Midwest, the number of HPNs nearly tripled, from 379 to 1,471. The spread of HPNs was especially rapid in old industrial cities in Wisconsin, Illinois, Indiana, Michigan and Ohio. 10/

The Ninth Federal Reserve District's Twin Cities metropolitan area was also sharply affected, as Table 1 indicates. 11/ The number of HPNs in the Twin Cities more than quadrupled, from just 7 in 1970 to 33 in 1990. The total population of Twin Cities HPNs increased almost sevenfold. In 1990, about 57,000 of the 79,000 HPN residents were whites/others (up over sixfold from 1970) 12/ and more than 19,000 were African Americans (up almost ninefold from 1970). The Hispanic population of these HPNs rose rapidly from 1970 to 1990 but remained small.

The percentage of people living in HPNs also increased rapidly in the Twin Cities between 1970 and 1990 (see Table 1). For all whites/others, this percentage doubled, from 1 to 2, and for poor whites/others, the percentage nearly quadrupled, from 4 to 15. For African Americans overall, the percentage living in HPNs rose even faster and from a higher initial level than for all whites/others, from 7 to 22. For the African American poor, the rate of change was less dramatic, but the level was again high. Thirty-three percent lived in HPNs in 1990, up from 21 percent in 1970. Twin Cities Hispanics experienced sharp increases in the extent of concentrated poverty, albeit from low/moderate levels in 1970. By 1990, therefore, measures of the extent of concentrated poverty were up sharply in the Twin Cities. The odds of living in an HPN were highest for Twin Cities African Americans, especially those in poverty.

Figure 1 shows the location of the Twin Cities' HPNs in 1990. They fell into four main clusters: North Minneapolis, South Minneapolis, downtown Minneapolis/University of Minnesota, and inner-city St. Paul. Inclusion of the University of Minnesota neighborhoods east and northeast of downtown Minneapolis reflects a large population of off-campus students and some public housing units. The North Minneapolis cluster lies northwest of downtown and centers around an HPN that had a high concentration of public housing units in 1990.

1990-2000: An unexpected reversal

In the 1990s, the national trend toward an increasing concentration of poverty unexpectedly and decisively reversed. Similar reversals occurred in many Midwestern metropolitan areas, including Minneapolis-St. Paul (see Table 2).

During a decade in which the number of Americans living in poverty rose by over 2 million and the percentage of U.S. households living in poverty declined only moderately, from 13.1 to 12.4, Jargowsky highlights the following features of the decline in poverty concentration nationally.

• The number of HPNs declined by more than a fourth.
• The number of people living in HPNs declined by 24 percent, or over 2.4 million.
• The number of poor people living in HPNs declined by 27 percent, from 4.8 million to 3.5 million.
• The concentration of poverty—the percentage of the poor living in HPNs—"declined among all racial and ethnic groups, especially African Americans" 13/ (see Table 2).

Table 2 documents a dramatic decline in the extent of concentrated poverty in the Midwest in the 1990s. After experiencing the steepest increase in HPNs in the 1970s and 1980s, the Midwest's HPN numbers plunged in the 1990s, resulting in a 46 percent decline in the region's number of HPNs, number of HPN residents and concentration of poverty. Midwestern cities ranked among the national leaders in decreasing the number of people living in HPNs. Detroit experienced the largest decrease in the country, followed by Chicago and, in fifth place, Milwaukee-Waukesha. The poverty concentration also decreased significantly in the South (by 41 percent,) but fell much less in the Northeast and not at all in the West.

The nationwide turnaround in poverty concentration in the 1990s was largely unexpected. Although Jargowsky stresses that poverty concentrations fluctuate in individual cities over time, largely in response to regional economic conditions, 14/ the overall trend toward increasing concentration had been clear and accelerating. The extent of the decline in poverty concentration plainly surprised him. At a Brookings Institution forum in the spring of 2003, he commented that his first reaction to some of the Census 2000 data was disbelief. 15/ According to poverty researchers G. Thomas Kingsley and Kathryn L.S. Pettit, "Against the overwhelmingly negative mindset that long dominated America's thinking about cities, [the] story is astonishing ... No writer of a decade ago even hinted at so dramatic a reversal in the concentration of poverty by the end of the century." 16/

Although the pessimistic mindset regarding poverty concentration trends affected expectations in the Twin Cities, too, 17/ the area ended up playing a leading role in the story of deconcentration (see Table 2). From 1990 to 2000, the number of HPNs in the Twin Cities fell from 33 to 15, and the number of HPN residents fell by 32,000, the 14th-largest decline in numbers among all metropolitan areas. The overall percentage of poor people living in HPNs in the Twin Cities fell from 17.3 in 1990 to 8.6 in 2000. Among the 20 largest metropolitan areas in the U.S., this 8.7 percentage-point drop was the fifth-largest decline in poverty concentration, behind Detroit, Chicago, Baltimore and Houston.

As Tables 2 and 3 indicate, poverty concentration in the Twin Cities also fell sharply among groups traditionally overrepresented in HPNs. Within the African American population, the concentration of poverty fell by more than 20 percentage points. This was the second-largest decline in African American poverty concentration among the 20 largest cities, behind only Detroit. The concentration of poverty within the Twin Cities Hispanic population dropped 12.3 percentage points, also the second-largest such decline (behind only Detroit) among the 20 largest cities. From 1990 to 2000, poverty concentration fell from 40.1 to 12.7 percent among the Twin Cities American Indian community and from 50.3 to 18.8 percent among the Asian American community. From the perspective of family composition, poverty concentration among members of single-parent households also plunged, from 16.7 percent in 1990 to 7.2 percent in 2000.

Figure 2 shows the location of the Twin Cities' HPNs in 2000. Most areas show a clear net decline in HPN numbers in the 1990s. There is little change in downtown Minneapolis and the adjacent University of Minnesota area, which includes many low-income students. However, in South Minneapolis (defined here as the area south of Interstate 94, which runs east-west through the Twin Cities), the number of HPNs dropped from 10 to 2. In St. Paul, three neighborhoods were HPNs in both 1990 and 2000, but the extreme poverty rates (above 60 percent) that prevailed in two of them in 1990 had been eliminated by 2000. Six other St. Paul neighborhoods saw their poverty rates drop below 40 percent in the 1990s, removing them from the HPN category.

In North Minneapolis, there are four HPNs in 2000, compared to seven in 1990. This is probably related to both the changes in some census tract boundaries and the demolition of numerous public housing units in the area in the 1990s. The data also show a decline from the extreme levels of poverty concentration that formerly prevailed in North Minneapolis. In 1990, the poverty rate exceeded 65 percent on two north side census tracts, with a peak of 76.5 percent in a tract of 2,700 residents. In 2000, by contrast, no north side census tract had a poverty rate greater than 50 percent.

Causes of the reversal

During recent decades, the spatial concentration of poverty in the U.S. has fluctuated much more dramatically than the national poverty rate. In the 1970s and 1980s, the poverty rate changed little, and the concentration of poverty soared. In the 1990s, the poverty rate again changed little, and the concentration of poverty plunged. This might seem to suggest that the concentration of poverty changes mainly through the relocation of people, and not through changes in the number or percentage of individuals who are poor. Nonetheless, much of the 1990s evidence, at least in the Twin Cities and Midwest, is consistent with Jargowsky's earlier conclusion that broad economic factors leading to income growth and poverty reduction are very important in explaining changes in poverty concentration. 18/

Relocation probably played a role in deconcentrating poverty. The cross-neighborhood relocation of individual poor and nonpoor households, mostly but not always voluntarily, contributed to the deconcentration of poverty. For example, throughout the 1990s, low-income housing programs were modified specifically to decrease the spatial concentration of the poor. 19/ The federal government's Hope VI program provided funding to tear down many high-density public housing projects across the country and help former residents move to new neighborhoods. For low-income individuals not living in public housing, assistance shifted toward vouchers that either allowed a wider choice of rental locations than before or even required recipients to rent outside HPNs.

These changes in housing policy helped deconcentrate poverty to some degree. The demolition of high-density public housing projects and their replacement with lower-density, mixed-income housing dispersed low-income individuals and changed the poverty level in some neighborhoods from extreme to high or moderate. For example, in the Minneapolis-St. Paul area, public housing units in an extreme poverty neighborhood were eliminated, vouchers helped some low-income residents seek housing outside of HPNs and tax breaks subsidized the construction of affordable housing units in non-HPNs.

However, there are reasons to doubt that poverty deconcentration in the 1990s can be explained mainly by changes in housing policy. Housing experts observed that the limited political support for funding and implementing the changes, combined with many people's limited willingness to move very far from their familiar neighborhoods, would also limit the aggregate impact of the policy changes on concentrated poverty. 20/ In addition, tight rental markets in the late 1990s limited voucher recipients' ability to find units in many higher-income neighborhoods. Furthermore, housing policies change across the nation, but poverty deconcentrated much more in some regions than others. And within the Twin Cities, the numbers of HPNs declined as much or more in areas without large-scale demolition of public housing as they did in North Minneapolis, where many public housing units were torn down.

If housing policy does not account for the magnitude and spatial pattern of poverty deconcentration, could other relocation factors be at work? Jargowsky found that the factors behind racial and socioeconomic segregation in a metropolitan area affect its concentration of poverty. 21/ Thus, changes in these factors—such as exclusionary zoning in affluent suburbs or the relative availability and quality of public services outside inner cities—could lead to increased mixing of low-income and more affluent households and thus a deconcentration of poverty without a reduction in the poverty rate. No comprehensive analysis of how these factors changed during the 1990s seems to be available yet. The relevant anecdotal evidence is mixed at best 22/ and appears to provide no more than limited support for a decrease in segregation factors as a significant explanation of the deconcentration of poverty.

Relocation within metropolitan areas is not the only possible answer to the question of how the national concentration of poverty could decrease while the national poverty rate remained stable. Depending on where they settled, the many low-income immigrants who arrived in the 1990s might be part of a different explanation. For example, the large number of Hispanic immigrants, the rising proportion of Hispanics among the poor in the U.S. (17 percent in 1990 versus 23 percent in 2000), and the growing presence of Hispanics in areas with relatively few HPNs in 1990 (e.g., the West, rural small towns) could help explain the situation. By this logic, a stable poverty rate could result if a large influx of poor immigrants into non-HPNs offset a significant decline in poverty among residents of current or former HPNs. However, the relationship between immigration and poverty concentration in the 1990s has not yet been fully investigated.

Broader forces leading to income growth and poverty reduction were probably important, at least in the Twin Cities and Midwest. Jargowsky's extensive data analysis and literature review on the causes of concentrated poverty stress metropolitan area forces, especially an area's "overall level of income and the inequality in its income distribution." 23/ Jargowsky also stressed that these factors differed significantly among metropolitan areas. This points toward a consideration of regional economic forces.

From a regional perspective, changes in the concentration of poverty and the poverty rate in the 1990s seem to fall in line with each other. In the Midwest and South, where concentrated poverty declined the most in the 1990s, income per household grew faster in the 1990s than in the West and, especially, the Northeast (see Table 2). Partly as a result, poverty rates in the Midwest and South fell by almost 2 percentage points in the 1990s but rose in the West and Northeast. In the Midwest, the number of people living in poverty declined by 611,000, or almost 9 percent. Thus, although housing policy changes affected the whole nation, poverty concentration fell sharply only in the regions where incomes grew fastest and poverty fell (Table 2).

The Twin Cities followed the Midwestern pattern in the 1990s. Median household income in the Twin Cities grew by almost 18 percent between the 1990 census and Census 2000, above average within the Midwest region. Poverty did not just move around in the Twin Cities; it declined, and more rapidly than in the nation as a whole. The number and percentage of poor people living in the core cities (Minneapolis and St. Paul) fell, and the percentage of metropolitan area residents living in poverty dropped from 8.1 to 6.7. Poverty rates among area minority and single-parent households remain above average, but fell sharply between 1990 and 2000 in most cases, as shown in Table 3. (The main exception was in the Twin Cities' rapidly growing Hispanic population, where the poverty rate decreased only slightly.) These declining poverty rates help explain why the number of HPNs in the Twin Cities area fell from 33 in 1990 to 15 in 2000. That, in turn, helps explain why the number of HPN residents fell by 40 percent and why poverty concentration was cut in half.

Looking ahead

From 1970 to 1990, the increasing concentration of low-income households in HPNs was often seen as an unstoppable source of growing social burdens for poor families and inner cities. Midwestern cities experienced some of the highest levels of and rapid increases in poverty concentration, and the Twin Cities were not immune. Then, in the 1990s, the concentration of poverty unexpectedly declined in much of the Midwest and nation. A decade of income growth that reached low-income households is probably responsible for much of the reversal in those regions, with housing policies aimed at deconcentration playing a supporting role.

Although the statistics on Twin Cities poverty and poverty concentration in the 1990s were surprisingly positive, many challenges remain. Poverty rates remain high in some Twin Cities neighborhoods and probably increased somewhat during the recent recession and ensuing period of slow employment growth. Poverty concentration may have risen since 2000 as well, and research on whether its decline in the 1990s yielded concomitant reductions in social ills is still at an early stage. Looking ahead, low-income individuals continue to face impediments to housing mobility, such as restrictive zoning that limits the supply of low-cost housing in many jurisdictions, public transit systems with limited service outside inner cities and the reluctance of some landlords to accept Section 8 rental vouchers. Additional reductions in poverty and enhancements in residential choice for low-income households will depend on robust economic growth and effective income, housing and land-use policies and policy implementation. These, in turn, can benefit from further inquiries into the causes and consequences of poverty, in the tradition of William Julius Wilson's The Truly Disadvantaged.

Research assistance for this article was provided by Jovana Trkulja and Michael Grover. The poverty-rate maps were prepared by Laura Smith.

1/ Paul Jargowsky, Poverty and Place: Ghettos, Barrios, and the American City, Russell Sage Foundation, 1997, p. 3.

2/ Jargowsky, 1997, p. 21–22; Patricia Ruggles, Drawing the Line: Alternative Poverty Measures and Their Implications for Public Policy, Urban Institute Press, 1990.

3/ U.S. Census 2000, Appendix A.

4/ I use data from Jargowsky, Stunning Progress, Hidden Problems: The Dramatic Decline of Concentrated Poverty in the 1990s, The Living Cities Census Series, The Brookings Institution, May 2003. Jargowsky uses contemporaneous tracts for each census, rather than a fixed set of tracts. Although researchers G. Thomas Kingsley and Kathryn L.S. Pettit advocate and use the alternative of fixed tracts, plus other modifications of Jargowsky’s methods, they reach essentially the same conclusions about concentration of poverty in the U.S. from 1970 to 2000 in “Concentrated Poverty: A Change in Course,” Neighborhood Change in Urban America, No. 2, The Urban Institute, May 2003. This suggests that the facts summarized here may not be too sensitive to details of the definitions and measurements used.

5/ Jargowsky, 1997, p. 11.

6/ The 40 percent definition is taken from Jargowsky, 1997. However, Kingsley and Pettit find results similar to those presented here using a 30 percent threshold.

7/ Jargowsky, 2003, p. 3.

8/ Jargowsky’s analysis appears in chapter 2 of his 1997 book. He leaves out areas that were not officially declared metropolitan areas until after 1970. The omitted metropolitan areas are relatively small (totaling less than 10 percent of the metropolitan population in 1990) and “excluding them has little impact on the aggregate national figures” (p. 33).

9/ Jargowsky, 1997, p. 31–32.

10/ Ibid., p. 44–45.

11/ The rising concentration of poverty in the Twin Cities was highlighted by the Metropolitan Council’s Trouble at the Core study, authored by Mike Munson (1992).

12/ Jargowsky’s method of estimating the whites/others population—subtracting the African American and Hispanic population from the total population to maintain comparability with limited 1970 data—should be kept in mind, given the rapid growth in the Asian American—and, especially, the Hmong—population in the Twin Cities between 1970 and 1990. The Hmong population had a high rate of poverty during this period and was concentrated in low-income, inner-city neighborhoods.

13/ Jargowsky, 2003, p. 1.

14/ Jargowsky, 1997.

15/ Jargowsky appeared at a May 19, 2003, forum titled “Stunning Progress, Hidden Problems: Declines in Concentrated Poverty in the 1990s,” sponsored by the Brookings Institution and Living Cities: The National Community Development Initiative.

16/ Kingsley and Pettit, p. 10.

17/ See the cautionary words from the president of the Minneapolis NAACP, quoted in Edward G. Goetz, Clearing the Way: Deconcentrating the Poor in Urban America, The Urban Institute Press, 2003, p. 1.

18/ Jargowsky, 1997, p. 145–183.

19/ Goetz, Chapter 3.

20/ Goetz, p. 10.

21/ Jargowsky, 1997, Chapter 6.

22/ Goetz.

23/ Jargowsky, 1997, p. 145.

Table 1: Population and Poverty Measures for Selected Metropolitan Areas, 1970 and 1990