Population Growth and Decline in City Neighborhoods

G. Thomas Kingsley, Kathryn L.S. Pettit

No. 1 in The "Neighborhood Change in Urban America" Series

The nonpartisan Urban Institute publishes studies, reports, and books on timely topics worthy of public consideration. The views expressed are those of the authors and should not be attributed to the Urban Institute, its trustees, or its funders.

---

It is now well known that populations in America's cities generally grew faster over the past decade than they did in the 1980s. But how was this growth distributed across neighborhoods in these cities? It matters considerably whether the poor inner-city communities that lost population so dramatically in the 1980s shared in the overall improvement or whether the higher growth rates were found only in areas that were better off to begin with. And what impact did racial composition have on neighborhood population change? This paper reviews the evidence from the U.S. Census for the central cities of the nation's 100 largest metropolitan areas (listed in appendix table A1).¹ Major findings are as follows:

• Almost all types of neighborhoods (census tracts) in these cities did better in the 1990s (median population growth of 2.7 percent) than the 1980s (median loss of 0.5 percent). Comparing the two decades, the share of all tracts that grew significantly (by 5 percent or more) increased from 36 percent to 44 percent, while the share with significant losses dropped from 34 percent to 27 percent.
• Improvement was not the same for all types of neighborhoods, nor was it enough to overturn past differences. For example, neighborhoods that had been predominantly (more than 60 percent) black in 1990 fared worst in the 1990s (median population decline of 7 percent), although this was a marked improvement over their 1980s median loss of 11 percent, whereas neighborhoods with no predominant race grew fastest (8 percent gain).
• Within all types of neighborhoods in both decades, higher poverty rates were closely associated with population loss. In the 1990s, high-poverty tracts (poverty rates of 30 percent or more in 1990) suffered a median loss of 5 percent, compared with a gain of 5 percent for tracts with poverty rates of less than 10 percent.
• Neighborhoods in sunbelt cities generally grew faster in the 1990s than those in rustbelt cities, but there were major differences within regions. In all regions, neighborhoods of all types in "Melting Pot" cities (in metropolitan areas with the largest minority populations, particularly nonblack minorities) typically did much better (median +7 percent) than those in "Largely White-Black" cities (-3 percent), and "Largely White" cities (-2 percent).²
• Despite the general trend, there was a non-trivial number of high poverty neighborhoods whose populations grew significantly (by 5 percent or more) in the 1990s. They are important because they demonstrate that it is possible to attract growth to some high-poverty environments. These neighborhoods made up 29 percent of all high-poverty tracts in the 100 cities overall, but even higher shares in the West and the Melting Pot cities of the other three regions.
• Racial change played a role, but it was not dominant. Hispanics increased their share of tract population significantly (by 20 percentage points or more) in only 5 percent of all tracts in the 100 cities but in 9 percent of the growing high-poverty tracts. Blacks experienced significant share increases in a smaller 4 percent of all tracts and in only 2 percent of the growing high-poverty tracts.

Purpose and Approach

The 100 largest metropolitan areas in the United States accounted for 60 percent of the country's population in 2000 but for much higher shares of its wealth and the sectors that will drive its economic future. One-third of the population in these areas (54 million out of 169 million) lived in their central cities—places that, few would deny, are still critically important to the nation's well-being.

In the early 1990s, the prospects for cities, particularly in the rustbelt states, seemed at a low point. Even after evidence of resurgence in some urban neighborhoods later in the decade, careful analysis of available data did not provide a basis for much optimism (Kasarda et al. 1997). However, results from the 2000 census have been analyzed (Glaeser and Shapiro 2001, Katz and Berube 2002, Simmons and Lang 2001), and while they do not indicate a dramatic turnaround, they reflect considerable improvement over those of the preceding decade.

Among the central cities of the 100 largest metropolitan areas, only 36 grew by 5 percent or more in the 1980s; in the 1990s, 49 did so. The number that suffered any decline shrank from 44 to 36. Altogether, the growth rate for these cities increased from 4.5 percent in the 1980s to 7.2 percent in the 1990s.

The implications of these results, of course, depend on how they vary by region and type of city (more will be said about that later), but the patterns of growth and decline within cities are also highly significant. A city's average growth rate could mask a serious deepening of disparities between neighborhoods or an alleviation of them. The most important measures of disparities rely on economic variables (including poverty) in the census long-form data files, which (as of this writing) have not been released for 2000. In the meantime, data on population change alone can provide useful insights.

This paper reviews data from the 2000 census short-form files for these 100 cities to determine what kinds of neighborhoods grew and declined in the 1990s and how those results compared with patterns of change in the 1980s.³ The analysis uses census tracts as its units of reference (we use the terms "tracts" and "neighborhoods" interchangeably throughout).

We start by looking at changes for all 14,041 tracts in these cities, characterizing them by their 1990 racial composition and poverty levels.⁴ We then review how population changes in the 100 cities themselves varied by region and type of city. This review provides a basis for understanding the next topic: regional and type variations in patterns of neighborhood growth and decline within the cities. We close by answering two questions of special interest. First, what share of all neighborhoods experienced significant racial change, and where were they located? Second, did any tracts with high poverty rates grow significantly? If so, how many and where were they located?

Data for this analysis came from the Neighborhood Change Database (NCDB), the only source of tract-level census data in which tract boundaries are defined consistently over time. We also hold metropolitan area and city boundaries constant; data for all geographic units and years are presented for boundaries as defined in 2000.⁵

For simplicity, we consistently divide the population for all years into four groups by ethnicity and race: "Hispanic" and three racial groups (white, black, and other), always defined to exclude Hispanics of those races.⁶ To understand the discussion of neighborhood change that follows, it is helpful to know in advance that these groups made very different contributions to city growth in the 1990s (figure 1; see also Berube 2001). In the 100 cities, whites remain the largest group, but their number declined by 10 percent (from 26.1 million to 23.5 million). The numbers of blacks increased, but only by a modest 8 percent (from 12.9 million to 14.0 million). The Hispanic and "other" groups made the largest contribution, together increasing by 43 percent (from 11.7 million to 16.7 million), in the process surpassing the black total.

Neighborhood Population Change—National Overview

Research on urban spatial patterns over the past decade has stressed the influence of both poverty levels and race on variations in neighborhood conditions (e.g., Jargowsky 1997). We find that both were also critically important to neighborhood population change in the 1980s and 1990s.

In the 1980s, there was a consistent relationship between poverty and population change: The higher the poverty rate, the higher the probability of population loss (figure 2). The 1990s exhibited the same pattern, but with modest improvement in every category (larger gains, smaller losses). The median decennial population change rate for all tracts in the 100 cities went from a small loss (-0.5 percent) in the 1980s to a modest gain (+2.7 percent) in the 1990s. The tracts with the lowest poverty rates as of 1990 (under 10 percent) had a much better record, with a median gain of +2 percent in the 1980s, rising substantially to +5 percent in the 1990s. At the other extreme, tracts with the highest poverty rates (30 percent or more) experienced serious losses in the 1990s (median of -5 percent), although less serious than in the 1980s (-8 percent).

There were also substantial differences by race/ethnicity. We look at the results for tracts in four categories: where the predominant race (more than 60 percent of the population in 1990) was white, black, or Hispanic, and where there was no predominant race. (Since there were only 109 tracts where the predominant race was "other" in 1990, we left them out of this analysis.)

Results were similar to those for the poverty groups. Although each category had a distinctly different experience, measures for the 1990s typically represented an improvement over those in the 1980s. For tracts that were predominantly white, for example, the median population change went from -0.5 percent to +3 percent. This was a larger gain than the average for the 100 cities nationally, although population gains in tracts that were predominantly white in 1990 are partially explained by in-migration of Hispanics and people of other races.

Predominantly black tracts, like high-poverty tracts, experienced serious losses in the 1990s (median of -7 percent); nonetheless, this was an improvement over their median loss of -11 percent in the 1980s. Tracts with no predominant race experienced above-average growth in both periods, with the median increasing modestly from +7 percent to +8 percent.

For tracts that were predominantly Hispanic in 1990, the pattern was different. They had by far the highest median growth rate in the 1980s (+10 percent), but their median dropped to +4 percent in the 1990s. The absolute number of Hispanics in these tracts was quite small in 1980 and much larger in 1990, which makes it not that surprising that the rate of growth dropped, even though the absolute growth continued to be high. Clearly, an important share of the impressive net increase in Hispanics in cities during the 1990s took place in tracts that were not predominantly Hispanic at the start of that decade.

Generally, within each of these racial/ethnic groups, population growth was inversely correlated with 1990 poverty rates. More specifically:

• Predominant race = white (6,894 tracts). Consistent patterns in both decades: The higher the poverty rate, the weaker the growth, but growth in every category was better in the 1990s than in the 1980s. In tracts with poverty rates below 10 percent, the median growth rose from +0.9 percent to +5 percent. In tracts with poverty rates above 40 percent, the median rate of decline dropped from -5 percent to -3 percent.
• Predominant race = black (2,983 tracts). Similarly consistent relationships but lower on the chart: declines in all categories in the 1980s, generally less severe in the 1990s, and turning positive in the lowest poverty category. The median rates of decline in the highest poverty category (above 40 percent) are extreme: -18 percent in the 1980s, dropping only to -14 percent in the 1990s.
• Predominant race = Hispanic (1,058 tracts). Substantial rates of growth in all categories in both decades, although, as with other groups, growth diminishes as poverty rates increase. There is an important internal difference, however. For tracts in the lowest poverty category, the 1990s median growth rate (+23 percent) was much higher than that for the 1980s (+15 percent), but in all other categories (those with higher poverty rates), 1990s growth was noticeably slower than 1980s growth.
• No predominant race (2,957 tracts). Median growth rates decline regularly as poverty goes up, and there is not much difference between experiences in the two decades.

City-Level Population Change by City Type and Region

Our next objective was to find out whether and how these national findings vary across the nation. Before we looked into how neighborhoods changed, however, we characterized the variation in population growth and decline among the 100 cities themselves.

To be sure, a large part of the story is regional. The sunbelt continued to outperform the rustbelt, but in all regions populations of these cities grew during the 1990s, and in all regions but one they performed better than they had in the preceding decade (see table 1 and discussion in Glaeser and Shapiro 2001). The lowest 1990s decennial growth rate was recorded by the cities in the Midwest: only +0.1 percent, but still a marked improvement over their -5 percent during the 1980s. Rates for cities in the Northeast picked up from +0.3 percent to +4 percent and, in the South, from +5 percent to +9 percent. The highest 1990s growth rate was the +14 percent achieved in the West. This rate is down from the extraordinary +18 percent this region's cities recorded in the 1980s, but in these circumstances that sort of drop should probably be seen more as a relief than a problem, in policy terms.

TABLE 1. Population Level and Change, 1980-2000 (Central Cities of 100 Largest Metropolitan Areas)
	No. of metro. areas	2000 population (million)		Population change Central City (%)		Percent of total population 2000 (%)		Population change 1990-00 (%)
		Metro.	Central City	1980-90	1990-00	Black	Hispanic & other	Black	Hispanic & other
Northeast
Melting Pot	4	13	9	3	9	27	39	10	33
Largely White	15	16	3	-2	-4	27	20	11	47
Largely White-Black	1	5	2	-6	-4	43	14	6	52
Total	20	34	13	0	4	29	32	9	36
Midwest
Melting Pot	2	11	4	-6	4	33	29	4	42
Largely White	13	13	4	0	2	27	8	15	87
Largely White-Black	5	11	2	-11	-7	61	9	3	55
Total	20	35	10	-5	0	37	15	7	52
South
Melting Pot	10	23	7	9	14	22	44	4	46
Largely White	8	8	2	0	7	21	12	17	70
Largely White-Black	19	23	7	3	5	46	7	12	117
Total	37	55	16	5	9	31	25	10	53
West
Melting Pot	18	37	13	21	14	9	51	5	39
Largely White	5	8	2	1	12	9	24	14	63
Total	23	45	15	18	14	9	47	6	40
Total U.S.
Melting Pot	34	84	32	9	11	19	44	6	39
Largely White	41	45	11	0	3	23	15	14	62
Largely White-Black	25	40	11	-2	0	49	8	9	81
Total	100	169	54	5	7	26	31	8	43
Source: Neighborhood Change Database, 1980, 1990, 2000. Note: Excludes tracts with less than 200 population in both 1990 and 2000 and tracts with zero population in 1990.

Regions do not tell the whole story, however. As suggested by table 1, forces other than sunshine are working to create major variations within regions. An important part of this variation is captured by the framework developed by demographer William Frey (2001). He divides metropolitan areas into three basic categories: Melting Pot metros are those in which whites account for no more than 69 percent of the 2000 population and minorities other than blacks account for more than 18 percent. Largely White-Black metros are metropolitan areas in which blacks account for at least 16 percent of the population. All other metropolitan areas are classified as Largely White.⁷

Central cities in the Melting Pot metros account for a very large share of the growth in these 100 cities, with a 1990s growth rate of +11 percent nationally (compared with only +3 percent for those in Largely White metros and a negligible +0.3 for those in Largely White-Black metros).

In the Northeast, the Melting Pot cities include New York City and three cities in northern New Jersey. Together they grew by +9 percent in the 1990s (up from +3 percent in the 1980s), while cities in the other two categories declined in both decades. The 15 central cities in Largely White metros make up the only group on table 1 that actually did worse in the 1990s (-4 percent) than it had in the 1980s (-2 percent), but this group is mixed. Two of its cities (Boston, with +3 percent growth in the 1990s, and Providence, with +8 percent) were close to meeting the criteria for the Melting Pot category and behaved as if they had. Almost all the rest in the Largely White group lost population in the 1990s, three at a rate of more than -10 percent (Syracuse, Buffalo, and Hartford). Philadelphia is the only Largely White-Black metro in the Northeast. Its population dropped by -4 percent in the 1990s; nonetheless, a better performance than its -6 percent in the preceding decade.

There were only two Melting Pot metros in the Midwest (Chicago and Minneapolis), but their central cities also dominated city growth in their region(+4 percent in the 1990s, an impressive turnaround from their -6 percent in the 1980s). Cities in the 13 Largely White metros in this region had varied growth rates, but their populations together grew by only +2 percent in the 1990s. Again, it was the cities in the Largely White-Black metros that suffered most (decline of -7 percent, although that marked a clear improvement over their average loss of -11 percent in the 1980s). Two cities in this category (Gary and St. Louis) lost more than 10 percent of their populations in the 1990s.

In the South, the 10 Melting Pot metros are Washington, D.C., plus 3 in Florida and 6 in Texas. Together, their central cities grew by +9 percent in the 1980s, increasing to +14 percent in the 1990s. There are eight Largely White metros in the South, with central city growth of +0.3 percent in the 1980s, moving up to +7 percent in the 1990s. Unlike the experience in the northern regions, the cities in the Largely White-Black metros in the South have grown substantially since 1980 (+3 percent in the 1980s, +5 percent in the 1990s), if less rapidly than the other categories in this region.

Melting Pot metros make up a larger share of all metropolitan areas from the top 100 in the West than any other region (18 of 23). They include all of those in California, Nevada, Arizona, and New Mexico. The growth rate of their central cities in the 1990s was the highest across all categories in table 1 (+14 percent), although it represented a drop from an even higher +21 percent in the preceding decade. The five Largely White metros in this region include those in the Pacific Northwest plus Denver and Salt Lake City. Growth in their central cities, in contrast, accelerated over these two decades, moving up from +1 percent to +12 percent (major growth increases in Denver, Salt Lake City, Seattle, and Portland). There are no Largely White-Black metros in the top 100 in the West.

The last four columns of table 1 deal with racial/ethnic composition and change in these categories. The contrasts with respect to composition are sharp, as would be expected because of the way these categories are defined.

Rates of change in the 1990s also exhibit dramatic differences. Blacks increased in every category (average +8 percent), but in all regions they made more headway in the Largely White metros (+14 percent on average) than the other categories. The increases were notably higher in the Northeast and South regions (+9 and +10 percent) than in the West and Midwest (+6 and +7 percent).

The "Hispanic and other" category experienced the most pronounced change, increasing by 43 percent on average (more than five times the average growth rate for blacks). Their growth rates were highest by far in every region and category; substantially higher in the Largely White and Largely White-Black categories than in the Melting Pot group. The Melting Pot cities, however, had much larger Hispanic and Other populations to begin with and thus accounted for by far the largest absolute change: an increase of 4.0 million, compared with only 643,000 for the Largely White cities and 396,000 for the Largely White-Black cities.

The residual group (not shown on the table) is the white population. It declined in every category but one—the Largely White metros of the West, where it increased modestly. In every region, loss rates for whites were highest in the Largely White-Black metros, particularly in the Midwest (-30 percent). However, these averages mask the fact that their numbers did increase in 17 cities: 9 in the South (including Atlanta, Austin, Charlotte, and Jacksonville) and 8 in the West (including Denver, Phoenix, San Francisco, and Seattle).

Neighborhood Change in Different Regions and Types of Cities

How did our overall story about neighborhood growth and decline hold up in different regions and types of cities? Fairly well in general, but there were some exceptions to each of those findings (table 2).

• Neighborhoods that were predominantly black in 1990 fared worse than others virtually everywhere in the 1990s. This meant major losses for these tracts in all city types in the Midwest (median of -11 percent), South (-8 percent), and Largely White and Largely White-Black cities of the Northeast (-12 and -10 percent, respectively). However, blacks did register gains in the Melting Pot cities of the Northeast and in the West as a whole (although it is important to point out that there were very few predominantly black tracts in the West).
  
  Given the comparative overall growth experiences, it is not surprising that other types of neighborhoods generally grew fastest in Melting Pot cities. For neighborhoods that were predominantly white and those with no predominant race, patterns were similar. Their highest growth rates by far were in Melting Pot cities of the Northeast and South, with somewhat lower growth in Melting Pot cities of the other two regions and the Largely White cities in the West. The contrasts were sometimes striking within regions. The median growth rates for predominantly white tracts in the Northeast Melting Pot cities, for example, was +11 percent, compared with a median loss of -3 percent for those in the Largely White cities of that region. The pattern for predominantly Hispanic tracts was not that different, but the number of such tracts in the non-Melting Pot cities of most regions was too small to convey meaningful differences.
• Neighborhoods with high poverty rates in 1990 did worse than average in almost all racial categories and locations. The only notable exception that involved a sizeable number of tracts was predominantly white tracts in the West. In the West as a whole, the median growth rate for white high-poverty tracts (those with poverty rates of 30 percent or more) was +17 percent, compared with +6 percent for all white tracts. Large cities in the West typically have more undeveloped land in their boundaries than cities in other regions, and high-poverty white tracts are more likely to be found in the fringe portions of those cities where such land is prevalent, so this finding is not too surprising.
• Neighborhoods of almost all types and in almost all locations performed better with respect to population change in the past decade than they did in the 1980s (data not shown in table 1), but there were two major exceptions. One relates to the troubled Largely White cities in the Northeast. All types of neighborhoods in these cities suffered serious losses in both decades, but for all groups except predominantly white tracts, loss rates were notably higher in the 1990s. The second exception relates to neighborhoods of all types in the Melting Pot metros of the West. As with Hispanic tracts nationwide, this is a case where extraordinarily high growth rates in the 1980s simply settled down to more sustainable levels in the next decade.

TABLE 2. Change in Tract Population by Predominant Race and Poverty Level (Central Cities of 100 Largest Metropolitan Areas)
		Median population change of tracts, 1990-2000, by predominant race and poverty rate
	Total No. of Tracts	White		Black		Hispanic		No predominant race
		Total	Poverty > 30%	Total	Poverty > 30%	Total	Poverty > 30%	Total	Poverty > 30%
Northeast
Melting Pot	2,337	11	6	5	1	10	7	13	6
Largely White	839	-3	-6	-12	-16	-16	-16	-4	-9
Largely White-Black	361	0	sn	-10	-13	sn	sn	-2	-6
Total	3,537	5	-3	-2	-7	7	4	8	1
Midwest
Melting Pot	1,039	6	9	-7	-11	4	2	8	7
Largely White	1,264	-2	-9	-14	-16	sn	sn	-6	-8
Largely White-Black	889	-1	-2	-12	-17	sn	sn	-3	-12
Total	3,192	-1	-6	-11	-16	2	0	1	-4
South
Melting Pot	1,629	11	5	-8	-8	2	-1	15	14
Largely White	664	4	-1	-8	-10	sn	sn	2	3
Largely White-Black	1,730	4	-2	-8	-11	sn	sn	0	-3
Total	4,023	5	-1	-8	-10	2	-1	8	6
West
Melting Pot	2,792	7	12	6	7	4	1	9	7
Largely White	497	6	20	16	sn	20	sn	15	11
Total	3,289	6	17	6	9	5	2	10	8
Total U.S.
Melting Pot	7,797	8	9	-2	-6	4	2	11	8
Largely White	3,264	0	-5	-13	-14	-3	-10	-1	-5
Largely White-Black	2,980	2	-2	-10	-14	-15	-15	-1	-6
Total	14,041	4	-2	-7	-12	4	2	8	3
Source: Neighborhood Change Database, 1980, 1990, 2000. Note: Excludes tracts with less than 200 population in both 1990 and 2000 and tracts with zero population in 1990; sn = small number of tracts (10 or fewer); predominant race = race or ethnic group accounting for 60 percent or more of tract population; predominant race and poverty rates as of 1990.

Growth, Decline, and Racial Change

Researchers have long been interested in racial succession in urban neighborhoods, and there has been considerable anecdotal evidence of its occurrence in the 1990s; for example, concerning Hispanics moving into formerly black ghettos and blacks replacing whites in inner-ring suburbs. There has been very little analysis, however, of the degree to which succession actually occurs.

Our data show that it does indeed occur, but dramatic shifts in proportions are not common, at least not within the span of one decade. For a large majority of tracts in the 100 cities, racial/ethnic composition did not change markedly. Between 1980 and 1990, black shares of total population changed by more than 10 percentage points (up or down) in only 16 percent of all tracts. The comparable proportions were 18 percent for Hispanics and only 5 percent for the "other race" category, but a notably higher 37 percent for whites.

For whites, almost all the tracts with share changes in excess of 10 percentage points experienced losses. The white share declined by 10 to 20 points in 21 percent of the tracts, by 20 to 40 points in 12 percent, and by 40 or more points in 2 percent.

For the other groups, changes in excess of 10 percent were almost always increases in share, but here too, few were dramatic. Only 2 percent of all tracts saw Hispanic share increases of 40 percentage points or more; the percentage was the same for blacks, and there were no tracts where "other" races had share increases that large. We decided to define a share increase of 20 percentage points or more as "significant" and to look at the numbers for those changes more closely: 4 percent of all tracts (558) experienced share increases that large for blacks, and 5 percent (731 tracts) did so for Hispanics.

There were some notable differences in the types of neighborhoods in which black and Hispanic shares grew significantly. First, about half of the tracts where the black share increased significantly had grown substantially (by 5 percent or more) in the 1990s, whereas the other half were either stable or declining. The comparable tracts for Hispanics, however, were almost all (91 percent) in the substantial growth category.

Other differences show up in the spatial pattern of these tracts by region and city-type (table 3). For blacks, the highest concentrations are in the cities of the Largely White-Black metros. They account for a high 13 percent of all tracts in such cities in the Midwest and 9 percent of all tracts in cities of that type in both the Northeast and South. We have seen that tracts that were predominantly black in those cities in 1990 suffered significant population losses in the 1990s. Many African Americans from those tracts no doubt moved out to the suburbs, but these data on share increases suggest that blacks spread out within their central cities as well. This theory is corroborated by the second column in table 3, showing that, in all areas, significant increases in black share were much less prevalent in high-poverty tracts (those most likely to be losing population) than they were in general. Significant black share increases were found in only 2 percent of high-poverty tracts, compared with 4 percent of all tracts.

TABLE 3. Population and Racial/Ethnic Change in High-Poverty Neighborhoods (Central Cities of 100 Largest Metropolitan Areas)
	Tracts with black share increase of 20 percentage points or more, 1990s			Tracts with Hispanic share increase of 20 percentage points or more, 1990s			High-poverty tracts with population growth > 5%, 1990s
	Percent of all tracts	Percent of high-poverty tracts	Percent of high-poverty growth tracts	Percent of all tracts	Percent of high-poverty tracts	Percent of high-poverty growth tracts	No. of tracts	Percent of high-poverty tracts
Northeast
Melting Pot	3	1	2	3	1	1	247	49
Largely White	4	2	0	5	5	17	35	16
Largely White-Black	9	1	0	2	1	0	10	12
Total	4	1	2	3	2	3	292	36
Midwest
Melting Pot	3	2	6	8	3	8	105	32
Largely White	6	3	3	2	3	17	35	11
Largely White-Black	13	5	10	4	4	24	41	9
Total	7	4	6	5	3	13	181	17
South
Melting Pot	1	0	0	11	7	14	138	34
Largely White	2	0	0	3	8	23	30	22
Largely White-Black	9	3	4	1	0	1	79	17
Total	5	1	1	6	4	11	247	25
West
Melting Pot	0	0	0	8	10	13	202	49
Largely White	0	0	0	5	4	6	54	81
Total	0	0	0	8	9	12	256	53
Total U.S.
Melting Pot	2	1	2	7	5	8	692	42
Largely White	4	2	1	3	4	14	154	21
Largely White-Black	10	3	5	2	2	8	130	13
Total	4	2	2	5	4	9	976	29
Source: Neighborhood Change Database, 1990, 2000. Note: Excludes tracts with less than 200 population in both 1990 and 2000 and tracts with zero population in 1990. High-poverty tracts are those with 1990 poverty rates of 30 percent or more.

For Hispanics, the pattern was different. Tracts with significant Hispanic share increases were found in all regions and types of cities; they were notably more common in Melting Pot cities (7 percent of all tracts in this category nationally) than the other types (2 to 3 percent). Another difference is that significant increases in Hispanic share were almost as prevalent in tracts with high poverty rates (4 percent) as they were in general (5 percent).

Population Growth in Poor Neighborhoods

We have learned so far that while urban growth expanded over the past decade, it did so selectively. Some types of cities and some types of neighborhoods within cities did much better than others. High-poverty neighborhoods, particularly those with predominantly black populations, generally saw the deepest losses in population, as they did in the 1980s. Generally, however, does not mean always. We have looked at differences in median rates across groups, but, like all measures of central tendency, medians mask variety.

How much variation was there in population growth rates within these categories? Quite a bit in most cases. Among all central city tracts, 24 percent lost population at a rate of 5 percent or more in the 1990s. Not surprisingly, among high-poverty tracts (poverty rate of 30 percent or more) a much higher share lost population: 51 percent. But even among low-poverty tracts (poverty less than 10 percent) a non-trivial 15 percent lost at least 5 percent of their populations over the 1990s. This variation also worked in the other direction. As would be expected, a larger share of low-poverty tracts (50 percent) than all tracts (44 percent) grew by 5 percent or more in the 1990s. But even high-poverty tracts had a sizeable share, 29 percent, in the 5 percent or more growth category.

This group, high-poverty neighborhoods with significant growth, represents a small part of the universe (976 tracts, 7 percent of all tracts in the 100 cities), but they are important. They demonstrate that it is possible to attract growth to at least some high-poverty environments. Are they all in the West, or are some of them in the older big cities in the East, that is, the kinds of neighborhoods urban policy has long been hoping to turn around?

The last two columns of table 3 show that high-poverty neighborhoods that are growing significantly exist in all regions and types of cities, but their concentration varies. It is indeed in the West where they represent the highest proportion of all high-poverty tracts: 49 percent in Melting Pot cities and 81 percent in Largely White cities in that region. But their concentrations are also high in the other type of city with the highest growth rates generally: the Melting Pot cities of the other three regions (49 percent in the Northeast and about a third in both the Midwest and the South). These growing high-poverty tracts make up much smaller proportions of all high-poverty tracts in the Largely White and Largely White-Black cities of the Northeast, Midwest, and South (range of 9 to 22 percent).

The third column in table 3 shows that significant increases in black population share were not common in tracts of this type (only 2 percent of these tracts, compared with 4 percent of all tracts). The sixth column, however, shows that significant increases in Hispanic share were much more prevalent among these higher-growth high-poverty neighborhoods, appearing in 9 percent of all such tracts, compared with only 5 percent of all tracts. This comparative concentration of Hispanic share increases held in almost all regions and city types.

Implications

Full interpretation of how the well-being of our urban neighborhoods changed in the 1990s is not possible until the economic and social indicators from the 2000 census are available for analysis. The data presented here, however, do offer an interesting preview. Results are mixed, but the situation is clearly a step up from the almost universally bleak assessments in studies of 1980s trends, particularly for the northern regions. While some cities in the Northeast and Midwest still suffered declines, others, especially in Melting Pot metros, have rebounded significantly.

In these cities, it is clear that overall growth did not imply turnarounds for high-poverty neighborhoods in general. Indeed, most such neighborhoods continued to lose population, particularly those that were predominantly African-American at the start of the decade. Nonetheless, growth was not confined solely to the wealthier parts of town. A sizeable minority of high-poverty neighborhoods did grow significantly. The fact that they were found in all regions and types of cities indicates that change is indeed possible, even in places where conditions have seriously deteriorated.

Researchers should use the new census data to examine many aspects of neighborhood change in the 1990s, but understanding what lies behind these growing high-poverty neighborhoods would seem to deserve priority. The general growth of Hispanic and Asian populations was no doubt an important factor in these areas, as it was for large cities overall. Gentrification surely also played a role in some cases. We need to understand to what extent (and where) population growth was accompanied by income growth and reinvestment, and where it was not. The answers should provide clues as to what it may take to stimulate neighborhood revitalization more broadly and to do so in a manner that benefits, rather than displaces, original residents.

---

Notes

1. This selection includes the largest 100 PMSAs and MSAs based on their 1990 populations. Primary Metropolitan Statistical Areas (PMSAs) are metropolitan subcomponents of our largest urban agglomerations, Consolidated Metropolitan Statistical Areas (CMSAs). Metropolitan Statistical Areas (MSAs) are separate freestanding metropolitan areas. Since we ranked PMSAs and MSAs by size, some smaller PMSAs within CMSAs are not included. We also excluded suburban PMSAs that did not have large central cities within their boundaries. The Bureau of the Census recognizes several individual municipalities as "central cities" in many metropolitan areas. In most cases, for this analysis we accept only the predominant city as the central city (e.g., Chicago in the Chicago PMSA). There were seven exceptions, however, in which we classified two municipalities as together making up the central city: Anaheim/Santa Ana, CA; Fort Lauderdale/Hollywood, FL; Greensboro/Winston-Salem, NC; Greenville/ Spartanburg, SC; Minneapolis/St. Paul, MN; Tampa/St. Petersburg, FL; and West Palm Beach/Boca Raton, FL.

2. These categories are adapted from typology defined by demographer William Frey (2001).

3. One other research effort on neighborhood population change in the 1990s so far is noteworthy. Berube (forthcoming) analyzes variations in population change by tract location within cities, whereas this paper focuses on variations by tract racial/ethnic composition and poverty level.

4. To avoid outliers, the database for this analysis excludes all tracts with less than 200 population in both 1990 and 2000, and all tracts with zero population in 1990, regardless of their 2000 value.

5. Census tract boundaries do not always conform to municipal boundaries. We define our cities as the aggregation of census tracts (as defined in 2000) that most closely approximates the official city (Census Place) boundaries in 2000 and, as noted, we use those same boundaries for 1980 and 1990. Thus, our totals for a city may differ from the Place totals published by the Bureau of the Census. The Neighborhood Change Database was developed by the Urban Institute and GeoLytics, Inc. The Database is documented in Tatian (2002), which can be found at http://www.geolytics.com.

6. To allocate non-Hispanics who identified more than one race in the 2000 census into the three racial categories, we applied an algorithm (developed by demographer Jeffrey Passel—see explanation in Tatian 2002) that we believe does a reasonably good job of achieving comparability over time. First, we assigned those who indicated black and any other race(s) as black. Among those remaining, we then assigned those who indicated Asian and any other race(s) as Asian. Among those remaining, similar assignments were made for Native Hawaiian and Other Pacific Islander, then for whites, and then for American Indian/Alaska Native. Finally, after this procedure was complete, all groups other than blacks and whites were added together to form the "other race" category.

7. We applied Frey's (2001) framework to our selected metros using 2000 data, as he did in his paper.

References

Berube, Alan. 2001. Racial Change in the Nation's Largest Cities: Evidence from the 2002 Census. Washington, D.C.: Center on Urban and Metropolitan Policy, Brookings Institution. April.

__________. Forthcoming. Patchwork of Population Change: Neighborhood Growth and Decline in Large Cities in the 1990s. Washington, D.C.: Center on Urban and Metropolitan Policy, Brookings Institution.

Frey, William H. 2001. Melting Pot Suburbs: A Census 2000 Study of Suburban Diversity. Washington, D.C.: Center on Urban and Metropolitan Policy, Brookings Institution. June.

Glaeser, Edward L., and Jesse M. Shapiro. 2001. City Growth and the 2000 Census: Which Places Grew and Why. Washington, D.C.: Center on Urban and Metropolitan Policy, Brookings Institution. May.

Jargowsky, Paul A. 1997. Poverty and Place; Ghettos, Barrios, and the American City. New York: Russell Sage Foundation.

Kasarda, John D., Stephen J. Appold, Stuart Sweeney, and Elaine Sieff. 1997. "Central-City and Suburban Migration Patterns: Is a Turnaround on the Horizon?" Housing Policy Debate 8(2): 307-58.

Katz, Bruce, and Alan Berube. 2002. "Cities Rebound—Somewhat." The American Enterprise 13(4): 47.

Mincy, Ronald B., and Susan J. Wiener. 1993. The Underclass in the 1980s: Changing Concepts, Constant Reality. Washington, D.C.: The Urban Institute.

Simmons, Patrick A., and Robert E. Lang. 2001. The Urban Turnaround: A Decade-by-Decade Report Card on Postwar Population Change in Older Industrial Cities. Washington, D.C.: Fannie Mae Foundation.

Tatian, Peter A. 2002. CensusCD Neighborhood Change Database (NCDB) Data Users' Guide: Short Form Release. Washington, D.C.: The Urban Institute.

G. Thomas Kingsley coordinates the Neighborhood Change in Urban America series and is director of the National Neighborhood Indicators Partnership.

Kathryn L. S. Pettit is deputy director of the National Neighborhood Indicators Partnership and codirector of a project examining the relationships between neighborhood conditions and health outcomes.

TABLE A1. Characteristics of Central Cities (100 Largest Metropolitan Areas)
	Population 2000 (thousand)	Population change (%)		Percent of tracts by rate of population change, 1990-2000			Median 1990-2000 tract population change by 1990 predominant race/ethnicity
		1980-90	1990-00	<-5%	-5% to +5%	>+5%	White	Black	Hispanic	None
Northeast Melting Pot	8,671	2.7	8.8	13	22	65	11	5	10	13
New York, NY	8,008	3.6	9.4	22	67	0	11	7	10	14
Bergen, NJ	149	2.1	5.9	16	58	0	27	-10	7	9
Jersey City, NJ	240	2.2	5.0	32	45	0	12	-12	134	5
Newark, NJ	274	-16.3	-0.5	21	37	0	15	-12	5	4
Northeast, Largely White	2,794	-2.3	-3.9	47	34	19	-3	-12	-16	-4
Providence, RI	174	2.5	8.0	13	18	68	11	0	0	7
Boston, MA	589	2.2	2.6	23	39	38	2	-1	-34	6
Worcester, MA	173	4.9	1.7	15	63	22	1	0	-2	-4
Allentown, PA-NJ	107	1.5	1.1	19	59	22	2	0	0	-9
Bridgeport, CT	140	-0.6	-1.5	33	23	44	11	-16	-26	0
Springfield, MA	152	3.1	-3.1	43	43	14	-1	-13	-11	-9
Rochester, NY	220	-4.2	-4.9	46	51	4	-3	-8	0	-8
New Haven, CT	124	3.5	-5.2	41	38	21	-3	-16	0	-3
Albany, NY	96	0.3	-5.3	46	50	4	-4	-11	0	-9
Harrisburg, PA	49	-1.7	-6.5	60	33	7	-1	-11	0	-5
Scranton, PA	76	-7.2	-6.6	74	22	4	-6	0	0	0
Pittsburgh, PA	335	-12.7	-9.5	72	25	4	-8	-16	0	-8
Syracuse, NY	147	-3.5	-10.1	68	23	9	-7	-20	0	-11
Buffalo, NY	293	-8.2	-10.8	69	24	7	-7	-16	0	-13
Hartford, CT	122	2.5	-13.0	67	12	21	15	-18	-21	-16
Northeast, Largely White-Black	1,518	-6.0	-4.3	42	39	19	0	-10	-15	-2
Philadelphia, PA-NJ	1,518	-6.0	-4.3	42	39	19	0	-10	-15	-2
Midwest, Melting Pot	3,565	-6.1	-4.1	30	27	42	6	-7	4	8
Minneapolis/St. Paul, MN-WI	670	-0.1	4.6	14	42	44	1	9	0	9
Chicago, IL	2,895	-7.4	4.0	34	24	42	9	-7	4	8
Midwest, Largely White	4,288	0.1	1.5	47	31	22	-2	-14	-5	-6
Wichita, KS	345	6.1	9.7	31	36	34	-1	-19	0	4
Columbus, OH	742	11.1	9.2	40	29	31	0	-13	0	-5
Indianapolis, IN	799	4.2	7.0	39	28	33	0	-14	0	-5
Grand Rapids, MI	198	3.8	5.3	13	52	35	0	0	0	5
Omaha, NE-IA	395	3.1	4.8	28	40	31	0	-6	0	-1
Kansas City, MO-KS	444	-2.9	1.3	49	27	24	0	-13	-5	-12
Akron, OH	216	-6.8	-3.4	40	52	8	-4	-7	0	-5
Toledo, OH	310	-6.1	-5.8	60	33	7	-5	-16	0	-8
Lansing, MI	122	-2.3	-5.9	67	29	5	-7	-13	0	-9
Cincinnati, OH-KY-IN	338	-5.6	-8.6	65	30	5	-7	-13	0	-6
Dayton, OH	172	-7.1	-9.6	78	16	6	-7	-16	0	-14
Flint, MI	125	-11.9	-11.4	76	20	5	-10	-16	0	-1
Youngstown, OH	82	-17.2	-14.3	85	13	3	-8	-25	0	-22
Midwest, Largely White-Black	2,478	-11.4	-7.4	60	26	14	-1	-12	-12	-3
Milwaukee, WI	597	-1.3	-5.0	52	30	18	-2	-18	-12	-2
Cleveland, OH	478	-11.9	-5.4	53	36	12	-1	-10	0	-11
Detroit, MI	951	-14.6	-7.5	64	19	17	7	-9	0	1
Gary, IN	103	-23.2	-11.9	91	9	0	-15	-14	0	0
St. Louis, MO-IL	348	-12.4	-12.2	68	25	7	-4	-23	0	-13
South, Melting Pot	7,468	9.3	13.6	22	26	52	11	-8	2	15
Austin, TX	664	31.3	31.5	3	20	77	16	13	1	23
Fort Worth, TX	532	14.1	19.0	15	22	62	15	-8	16	17
Dallas, TX	1,190	10.5	17.9	17	22	61	13	-8	18	25
Orlando, FL	204	12.5	16.8	35	21	44	10	-9	0	-6
Houston, TX	1,935	3.6	15.3	16	22	62	7	-1	7	19
San Antonio, TX	1,155	18.0	15.0	14	29	56	27	-9	2	9
El Paso, TX	563	19.7	8.8	43	33	24	11	0	-6	-1
Ft. Lauderdale/Hollywood, FL	290	-2.7	6.0	16	35	49	4	-4	0	8
Miami, FL	363	3.1	0.9	30	40	30	5	-12	2	1
Washington, DC-MD-VA	572	-4.9	-5.7	55	30	15	1	-10	0	-1

TABLE A1. Characteristics of Central Cities (100 Largest Metropolitan Areas) (Continued)
	Population 2000 (thousand)	Population change (%)		Percent of tracts by rate of population change, 1990-2000			Median 1990-2000 tract population change by 1990 predominant race/ethnicity
		1980-90	1990-00	<-5%	-5% to +5%	>+5%	White	Black	Hispanic	None
South, Largely White	2,256	0.3	6.6	23	39	38	4	-8	9	2
W. Palm Beach/Boca Raton, FL	151	13.9	18.6	13	15	72	15	-8	0	4
Oklahoma City, OK	510	9.6	13.2	19	32	49	8	-15	0	2
Tulsa, OK	392	-0.4	7.1	15	44	41	3	-2	0	16
Tampa/St. Petersburg, FL	547	0.8	5.9	21	42	37	4	-10	9	0
Johnson City, TN-VA	52	-1.7	4.9	17	50	33	1	0	0	0
Knoxville, TN	183	-4.8	2.5	37	33	31	2	-11	0	-9
Chattanooga, TN-GA	162	-8.7	2.0	35	42	23	3	-9	0	-3
Louisville, KY-IN	258	-8.8	-3.7	35	54	12	-3	-7	0	8
South, Largely White-Black	6,552	3.0	4.7	36	30	34	4	-8	0	0
Charlotte, NC-SC	551	22.5	27.5	17	23	60	14	-2	0	10
Raleigh, NC	278	29.3	22.8	12	27	62	10	0	0	8
Jacksonville, FL	737	17.9	15.9	24	31	45	6	-9	0	-2
Greensboro/Winst. Salem, NC	404	6.0	13.8	13	34	53	10	1	0	17
Nashville, TN	541	6.8	11.6	17	37	46	6	-1	0	0
Norfolk, VA	425	51.7	8.2	24	37	39	3	0	0	1
Atlanta, GA	415	-7.2	6.4	27	25	48	17	0	0	2
Little Rock, AR	189	-1.2	3.4	39	32	30	2	-20	0	-1
Baton Rouge, LA	227	-7.8	3.3	30	21	48	6	-7	0	12
Wilmington, DE-NJ-MD	73	1.9	1.6	33	30	37	5	-1	0	-3
Mobile, AL	194	2.1	0.9	28	45	28	0	-4	0	-5
Columbia, SC	114	-8.0	-1.2	51	24	24	1	-9	0	-8
Charleston, SC	102	3.6	-2.0	64	18	18	0	-14	0	1
Memphis, TN-AR-MS	637	-0.9	-2.3	47	28	25	4	-16	0	-2
New Orleans, LA	485	-10.8	-2.5	39	39	22	2	-6	0	1
Richmond, VA	198	-7.4	-2.5	45	35	20	0	-6	0	0
Greenville/Spartanburg, SC	98	-4.2	-7.4	67	28	6	-3	-17	0	-13
Birmingham, AL	231	-11.2	-10.0	62	25	13	-1	-14	0	-1
Baltimore, MD	651	-6.3	-11.5	65	22	13	-4	-17	0	-4
West, Melting Pot	12,670	21.2	14.0	11	32	56	7	6	4	9
Las Vegas, NV	485	52.0	81.9	4	11	85	66	14	0	31
Phoenix, AZ	1,327	23.0	34.0	4	12	84	24	9	14	32
Bakersfield, CA	246	43.6	28.6	7	34	59	7	0	2	4
Fresno, CA	457	34.7	19.2	11	19	70	9	54	13	8
Ventura, CA	166	26.9	19.1	3	7	90	7	0	20	15
Anaheim/Santa Ana, CA	647	31.1	17.9	4	15	82	17	0	13	21
Tucson, AZ	491	15.8	16.4	7	38	55	6	0	4	15
Albuquerque, NM	452	15.1	16.3	26	35	39	-1	0	3	3
Stockton, CA	245	38.9	14.3	15	47	38	1	0	7	2
San Jose, CA	919	20.0	14.1	5	32	64	3	0	15	13
Riverside, CA	254	34.0	12.6	7	41	52	5	0	5	6
San Diego, CA	1,222	29.7	10.1	14	40	46	2	4	2	6
Sacramento, CA	396	34.0	9.9	12	42	46	2	0	0	7
Oakland, CA	399	9.9	7.3	8	38	55	1	12	19	6
San Francisco, CA	777	6.8	7.3	8	39	53	3	17	8	6
Vallejo, CA	119	36.1	6.7	7	34	59	2	0	0	10
Los Angeles, CA	3,697	17.5	6.0	14	38	48	4	3	1	10
Honolulu, HI	372	3.6	-1.3	37	38	25	-9	0	0	2
West, Largely White	2,029	1.4	11.8	3	40	57	6	16	20	15
Denver, CO	555	-4.7	18.7	1	31	67	6	25	20	28
Salt Lake City, UT	181	-2.0	13.6	0	45	55	8	0	0	13
Seattle, WA	563	5.3	9.1	1	46	53	5	6	0	11
Portland, OR	538	1.8	8.8	6	38	56	6	16	0	1
Tacoma, WA	192	11.2	8.4	9	45	45	3	0	0	12
Source: Neighborhood Change Database, 1980, 1990, 2000. Note: Excludes tracts with less than 200 population in both 1990 and 2000 and tracts with zero population in 1990; predominant race = race or ethnic group accounting for 60% or more of tract population; predominant race as of 1990.

---

About the Series

The Rockefeller Foundation has funded the Urban Institute to conduct a project that will advance knowledge about neighborhood change in America's urban areas, particularly as it occurred over the 1990s. The project has had two purposes. The first was to develop the Neighborhood Change Database (NCDB)—the only database that contains nationwide census data at the tract level with tract boundaries and variables consistently defined across the four U.S. censuses from 1970 through 2000 (for more information about the NCDB, visit http://www.geolytics.com). The second was to conduct research on neighborhood change using the NCDB, focusing particularly on changes in the concentration of poverty, conditions in distressed neighborhoods, and racial patterns. This paper is one product of that research.

---

The nonpartisan Urban Institute publishes studies, reports, and books on timely topics worthy of public consideration. The views expressed are those of the authors and should not be attributed to the Urban Institute, its trustees, or its funders.

Usage, posting and reprint of materials on the UI web site:

Most publications may be downloaded free of charge from the web site in PDF format. This information may be used and copies made for research, academic, policy or other non-commercial purposes. Proper attribution is required.

Copyright of the written materials contained within the Urban Institute website is owned or controlled by the Urban Institute. Posting UI research papers on other websites is permitted subject to prior approval from the Urban Institute—contact paffairs@urban.org.

If you are unable to access or print the PDF document please contact us or call the Publications Office at (202) 261-5687.