Training the leaders... of tomorrow.


Executive Summary

The 2000 census confirmed what many in Baltimore had feared. The city's population had continued to decline between 1990-2000, losing 85,000 residents over that decade. Moreover, the city's rate of population loss had accelerated to nearly double its rate between 1980-1990. While such dramatic statistics are typically associated with decline and distress, it is also possible that the changing profile of the population and the changing nature of the city's assets might produce a much smaller--but still viable--city.

This report examines the trends underlying Baltimore city's population loss in 10 Baltimore city neighborhoods, including shifts in the number of households, average household size, and the types of households living in the city. It also examines the extent to which these shifts have played a role in the unexpected situation of apparently healthy neighborhoods whose population is not increasing, and conversely, neighborhoods whose population is increasing and are experiencing some distress. This analysis has important implications for both neighborhoods and city policies.

This report begins by establishing the demographic context for the analyses presented in this report by reviewing Baltimore's trends in population, household size, and household type. Although our analyses focus on a sample of neighborhoods, we highlight- trends in the city as a whole. In addition, we compare these citywide trends with those in selected other cities, and with those in the nation.

Despite city officials' need to understand neighborhood location decisions in order to develop effective tools to both attract and retain residents, few urban policy researchers have proposed explicit theories of neighborhood choice. The most prominent theory, the life-cycle theory, dates back to the 1950s. This theory posits that neighborhood preferences are shaped by a household's life stage: marriage, child bearing, child-rearing, and empty nesting. While accurately describing 1950s households and their preferences, the life-cycle theory no longer explains neighborhood choice in the 21st century.

At least three changes over the last several decades account for the inadequacy of the life-cycle theory. Major demographic shifts include decreasing household sizes, increasing number of single households, and an increasing number of openly gay couples and non-married couples (Lang et al. 2000). Recent literature suggests that gender, race, and even school quality also shape neighborhood preferences. Finally, the life-cycle theory does not address neighborhood choice when families are constrained by their income and race (Shlay and DiGregorio 1985; Spain 1989; Sigelman and Henig 2001; Boehm and Ihlanfeldt 1991). Certain households may not select a neighborhood in which they would be a minority; similarly, some households may be priced out of particular housing markets. Because the life-cycle theory does not fully explain the complexity of neighborhood choice, this study adopts a broader framework. We view the life-cycle stage as just one of many factors--including demographic and social characteristics--likely to influence neighborhood choice.

In this analysis, we examine the neighborhood preferences of three household types: (1) households with children; (2) elderly households; and (3) nonelderly, childless households. We evaluate their choice of neighborhood in terms of six dimensions: (1) physical environment; (2) social environment; (3) school quality; (4) crime; (5) investment value; and (6) image. Each of these six dimensions is translated into specific measures that we obtained from census and administrative data as well as our own surveys. Based on a broader conceptual framework of neighborhood choice, we hypothesize that the three household types differ in the neighborhood characteristics they prize most. We examine this hypothesis by comparing our ranking of characteristics for each household type to the "actual" ranking shown in the data.

This study of population dynamics and neighborhood attributes required both a quantitative and qualitative approach. The qualitative analysis was used to validate the quantitative data and to capture the intangible perceptions and trends that drive neighborhood choice.

We studied five pairs of neighborhoods: (1) Canton and Hampden; (2) Locust Point and New Northwood; (3) Bolton Hill and Howard Park; (4) Seton Hill and Belair-Edison; and (5) Ashburton and Hamilton. The neighborhoods were selected for the study based on three criteria. First, one neighborhood in the pair had a good reputation as being "on the rise," while the other was to serve as a counterpoint, reputed to be suffering some distress. Next, the neighborhoods had generally similar histories and housing stock. Lastly, changes in each neighborhood were largely driven by market forces and not major governmental intervention, such as the Hope VI public housing replacement program.

To identify what people look for in a neighborhood, we started with basic characteristics such as the physical environment, school quality and image, and then identified specific measures of those indicators. The 1980, 1990 and 2000 censuses were the principal sources of data on population and household trends and characteristics. Administrative data from a range of sources provided estimates of crime rates, property sales prices, renovation permits and various neighborhood amenities. School quality was measured by Maryland State Department of Education standardized test scores and by personal interviews with a number of residents. Qualitative data were collected to both complement and supplement the quantitative information. The qualitative data consisted of on-site observations and interviews with neighborhood experts and residents. Finally, we gathered newspaper articles dating back at least five years to see how the study neighborhoods were portrayed in the media.

The analysis was hindered in a number of ways. For example, because only parts of the 2000 census had been released at the time of the study, we lacked some key measures including income and race for each household type. Following the convention of much urban policy research, we relied on census tracts as proxies for neighborhoods, though the boundaries of the tracts often differed from those of the neighborhoods as they are popularly conceived.

The analyses reported here are largely descriptive and suggestive, not definitive or conclusive. However, the rich array of quantitative and qualitative data provide for a better understanding of population trends, neighborhood assets and prospects for the future. It is critical to emphasize that the case study neighborhoods in our analysis are not "treatment" and "control" or "comparison" groups, in the sense that each neighborhood in a pair of neighborhoods in our sample is very similar except for a difference in the issue under study. Instead, they are best thought of as neighborhoods that provide useful counterpoints to each other for the purposes of studying the role of population increases or decreases and a neighborhood's trajectory. Although a case study research design such as this cannot isolate the cause of the difference between neighborhoods, the study does contribute to our knowledge of population dynamics and neighborhood assets, a field which is in its infancy. What is needed at this stage is much richer and more detailed descriptive information of what is actually occurring on the ground in neighborhoods. The case study approach is designed to provide such descriptive information.

Although Canton and Hampden share similar histories, critical events in the last two decades have caused them to change at different rates and in different ways. Both Canton and Hampden were historically white, working-class neighborhoods with strong industrial bases. During the last two decades, spillover effects from gentrifying neighborhoods that border the Inner Harbor of Baltimore, and an infusion of private development, have transformed the character of Canton. Canton, located along the harbor, is currently considered one of the hottest neighborhoods in Baltimore, largely due to its popularity among yuppie newcomers. Hampden, a larger neighborhood located in north Baltimore, has had a somewhat less favorable reputation, but has experienced some revitalization during the last decade, particularly along its main shopping street. Hampden's quirky reputation and affordable housing have made it a popular neighborhood among artists and less affluent nonelderly, childless households. Despite offering different qualities and amenities, both neighborhoods are finding new life by attracting different types of nonelderly, childless households.

During the last two decades, Canton's population increased by 19 percent while Hampden's population decreased by 17 percent. At first glance, the population trends in both neighborhoods from 1980 to 2000 are consistent with the common perception that neighborhoods succeed when population increases and are in decline when population decreases. However, a closer examination of population trends by decade, and of trends in household size and type, calls this conventional thinking into question. Indeed, we found that Hampden's population decline and relatively stable number of households masked important increases in its nonelderly, childless households. Hampden may be in the process of a favorable transformation and revitalization that belies its perception as a neighborhood in some distress.

Both Canton and Hampden have shifted to neighborhoods composed primarily of nonelderly, childless households. During the last two decades, Canton attracted 576 new nonelderly, childless households (an increase of 168 percent), while it lost households with children and saw its elderly population remain fairly constant. Similarly, between 1980 and 2000, Hampden attracted 1,590 new nonelderly, childless households, while it lost both households with children and elderly households. By 2000, 74 percent of Canton's households were nonelderly, childless households. While this trend may have been anticipated in Canton, given its reputation as a yuppie neighborhood, we were surprised to note a similar trend in Hampden. By 2000, 73 percent of Hampden's households were nonelderly, childless households--essentially the same proportion as in Canton.

Since nonelderly, childless households comprise the majority of households in both Canton and Hampden, our conceptual framework led us to expect the same assets being prized by such households in both neighborhoods: proximity to work, restaurants, nightlife, a "specialness" to the neighborhood, and a greater emphasis on physical environment than on school quality and crime. Our research supports this hypothesis but also points to interesting distinctions between the nonelderly, childless households in Canton and those in Hampden.

Canton and Hampden differ in important ways. For example, the median residential sales price in Canton is 47 percent higher than that of Hampden, the median income of Canton residents is 15 percent above that of Hampden, and Canton offers the harbor and nightlife as important amenities, while Hampden offers a main street district and a small town feel. Perhaps even more importantly, the character of the new residents differs across neighborhoods: Canton is reputed to attract yuppies while Hampden draws artsy, and less affluent, nonelderly, childless households. This finding of diversity within nonelderly, childless households reemerges in other neighborhood analyses in this report.

Both the Locust Point and New Northwood neighborhoods lost population at roughly the same rate as Baltimore city over the past two decades: 16 percent in Locust Point and 17 percent in New Northwood. Yet, both neighborhoods lost only four percent of their households since 1980. In both neighborhoods, the proportion of nonelderly, childless households grew, while the proportion of households with children declined. Locust Point lost elderly households while New Northwood gained them.

Locust Point's gain of nonelderly, childless households seems to reflect the neighborhood's new image as "up-and-coming" (Gunts 2000). Some experts consider Locust Point "the new Canton," another neighborhood benefiting from the private market development activities of Struever Brothers. The evidence largely supports this perception, with Locust Point mirroring past trends in physical environment, social environment, investment value, and image in Canton. However, differences between the two neighborhoods, including Locust Point's isolation and distance from the inner harbor, may alter Locust Point's eventual outcome.

New Northwood's gain in nonelderly, childless households is attributed to one particular group: single female, childless households. This subgroup appears to be attracted to the neighborhood's suburban feel and affordable housing. The increase in crime rates and poor school quality is consistent with the loss of households with children. The increase of single female households without children may spur future positive changes, but already reflect a neighborhood that is attractive to some new residents.

Both neighborhoods are experiencing increases in nonelderly, childless households and the loss of households with children. Thus, population loss over the past two decades show an important shift in household composition, but may not be a good barometer of the neighborhood's future.

Between 1980 and 2000, Baltimore experienced a 17 percent population loss, while the Bolton Hill neighborhood (census tract 1401) grew by five percent, suggesting that it is a neighborhood with a promising future. Bolton Hill's proximity to downtown and its urban, artsy environment continue to attract and retain nonelderly, childless households. Residents cite the historic architecture, investment value, and a very active civic association, the Mount Royal Improvement Association (MRIA), as the neighborhood's major assets. Median residential sales prices have fluctuated, but they remain twice those of the city. While Bolton Hill continues to support a sizeable elderly population and a small number of households with children, both populations are declining. Despite the fact that the neighborhood possesses one of the best public schools in the city and state, Mount Royal Elementary/Middle School, the neighborhood is not attracting households with children. Other indicators may be preventing households with children from moving in. For example, although crime it declined in Bolton Hill during the 1990s, crime from surrounding neighborhoods may be deterring families with children. In general, Bolton Hill has the key assets a stable neighborhood needs to retain and attract the urban urbane.

Howard Park (census tract 2802) is located along the western border of Baltimore city, adjacent to Baltimore County. Over the last 20 years, this neighborhood has lost 20 percent of its population, outpacing the citywide decline and suggesting it may be in decline. Despite Howard Park's suburban characteristics, such as large, affordable, single-family detached homes with yards, low occurrence of crime, and an improving public school, families with children are the most rapidly decreasing household type in the neighborhood, dropping by 50 percent since 1980. At first glance, this trend is difficult to explain, as key indicators suggest this would be an excellent neighborhood for households with children. One explanation may be that the neighborhood's deteriorating physical environment is no longer attractive to households with children. Furthermore, there may be a lag time in recognizing improving schools and decreasing crime. Conversely, the population of nonelderly, childless households has increased 25 percent and elderly households have grown by 90 percent, and will most likely continue to grow as residents age in place. Howard Park's decline in households with children does not conform to expectations of what these households value most in choosing a neighborhood.

Somewhat unique to Baltimore, the Seton Hill and Belair-Edison neighborhoods both gained population during the 1990s. However, Seton Hill's population gain stems from its ability to attract nonelderly, childless households while Belair-Edison's population gain consisted of households with children. Because Seton Hill is attracting nonelderly, childless households, it gained both population and households in the 1990s. By contrast, because Belair-Edison is attracting households with children, it gained population but lost households.

In both Belair-Edison and Seton Hill, the quality of the physical environment appears to play an important role in location decisions. This is interesting because Belair-Edison is gaining households with children while Seton Hill's growth is among nonelderly, childless households. Part of the explanation is that each group emphasizes different aspects of the physical environment. For households with children, large homes with nearby green spaces are a priority, while for Seton Hill's nonelderly, childless households, the emphasis appears to be on a diverse physical environment and a trendy neighborhood image. Surprisingly, schools and crime did not appear to affect neighborhood choice for either neighborhood.

Both Ashburton and Hamilton have lost population over the past two decades at rates slightly lower than the city. However, population loss is not a good indicator of neighborhood health in Ashburton, where the number of households actually increased between 1980 and 2000. Ashburton's population loss can be explained by the aging of residents and the smaller household size of elderly households. In Hamilton, shifts in the predominant types of households better capture the neighborhood's status than does sheer population loss.

Ashburton is a well-defined enclave in northwest Baltimore that has established itself as the home of Baltimore's black elite. Ashburton is a "neighborhood of choice," with many long-time residents aging in place. But newer, younger households are also being attracted to the neighborhood by its strong sense of community, architecturally diverse houses, and small town feel. The neighborhood association is actively marketing Ashburton's history and successful image to young potential buyers in the hopes of attracting a new generation of residents.

Hamilton has traditionally been a white, working-class community in Northeast Baltimore, but its racial composition has changed from five percent black in 1980 to 37 percent in 2000. In the 1990s, Hamilton's proportion of households with children grew by four percent--a noteworthy statistic in light of the citywide loss among this group. The attraction of large houses and yards may account for this growth. Nonelderly, childless households, attracted to Hamilton's suburban feel and affordable prices, have increased steadily since 1980, and now account for 50 percent of all households in the neighborhood. Although these households make few demands upon a neighborhood, such as well-performing schools and low crime rates, they also may be less committed to a neighborhood, which could explain the weakened sense of community in Hamilton. Elderly households, often among the most active in neighborhood organizations, decreased by 25 percent in the 1990s.

While Ashburton has a strong sense of community and a legacy that is being transferred, albeit slowly, to a new generation of residents, Hamilton's identity is rapidly changing and its future is less certain.

This research on the precipitants of neighborhood choice based on 10 Baltimore neighborhoods suggests that population change is not a consistent indicator of neighborhood health. While Bolton Hill, for example, a neighborhood perceived to be thriving, experienced an increase in population between 1980 and 2000, Ashburton, Locust Point and Seton Hill lost population though they too were perceived to be on the rise. The primary reason population change does not always serve as a good barometer of neighborhood health is that it obscures trends in household size and type.

In this research, household trends were more telling than overall population figures. The fact that Baltimore lost 28 percent of its households with children between 1980 and 2000 has been much decried. Indeed, nine of our 10 case study neighborhoods lost households with children in absolute terms over the last two decades. Although elderly households were the only household type to increase in absolute terms in the city during the last two decades, six of the 10 neighborhoods in our sample actually lost elderly households. These findings suggest that although some neighborhoods will have to confront the challenges of a graying community in the near future, the aging population is not uniformly distributed across the city.

Although nonelderly, childless households have comprised the majority of Baltimore's households for several decades, the city lost 15 percent of this household type during the last two decades. Nine of our 10 case study neighborhoods deviated from this trend and gained nonelderly, childless households. The influx of this household type breathed new life into Bolton Hill, Canton, Hampden, New Northwood and Seton Hill. One of the most important findings to emerge from this research is that this broad categorization actually encompasses a variety of different subgroups, and that different neighborhoods are successfully attracting different subgroups of nonelderly, childless households.

Another key finding is the importance of the physical environment in neighborhood choice. In all 10 case study neighborhoods, the physical environment emerged as the most important asset, ostensibly attracting and retaining households. Social environment emerged as an important, but not overriding, characteristic of neighborhood choice, playing some role in nine of the 10 neighborhoods. One surprising finding was that a strong sense of community can exist even in a neighborhood with a high percentage of renters.

In many of the case study neighborhoods, particularly those dominated by nonelderly, childless households, the conceptual framework for neighborhood choice used in this study did not predict, nor did we find, school quality to be a strong factor in neighborhood choice. It is important to note, however, that this study only looked at household decisions to remain in, or leave, the city, and did not measure those who would never consider the city because of its poor schools. Keeping this research design caveat in mind, our findings for Belair-Edison and Bolton Hill ran counter to the conceptual framework's prediction that school quality would prove an overriding characteristic for households with children.

This research also suggests that increasing median residential sales prices and the infusion of private development in neighborhoods bolsters neighborhood health. Finally, neighborhood image proved to be an accurate indicator of neighborhood health in Bolton Hill, Canton and Locust Point. By contrast, the purported poor image of such neighborhoods as Hampden and New Northwood were at odds with the trends observed in the quantitative data.

One potentially productive strategy for Baltimore in the immediate future is to focus on attracting and retaining nonelderly, childless households. Such households could be a boon for the city: they would increase the tax base while demanding relatively few city services. In addition, the diversity within the nonelderly, childless household type would allow neighborhoods to market themselves to the particular subgroups most attracted to the neighborhood's existing assets. This study also has implications for the role of research in charting the city's future. Monitoring and understanding population and household trends allows us to anticipate changes, and not just react to them. Such research should be an ongoing effort in Baltimore.