48(8) 1581–1604, June 2011 0042-0980 Print/1360-063X Online © 2010 Urban Studies Journal Limited DOI: 10.1177/0042098010375323 Edward Goetz is in the Humphrey School of Public Affairs, University of Minnesota, 301 19th Avenue South, Minneapolis, Minnesota, MN 55419, USA. E-mail: egoetz@umn.edu. Gentrification in Black and White: The Racial Impact of Public Housing Demolition in American Cities Edward Goetz [Paper first received, October 2009; in final form, April 2010] Abstract The gentrification that has transformed high-poverty neighbourhoods in US cities since the mid 1990s has been characterised by high levels of state reinvestment. Prominent among public-sector interventions has been the demolition of public housing and in some cases multimillion dollar redevelopment efforts. In this paper, the racial dimension of state-supported gentrification in large US cities is examined by looking at the direct and indirect displacement induced by public housing transformation. The data show a clear tendency towards the demolition of public housing projects with disproportionately high African American occupancy. The pattern of indirect displacement is more varied; public housing transformation has produced a number of paths of neighbourhood change. The most common, however, involve significant reductions in poverty, sometimes associated with Black to White racial turnover and sometimes not. The findings underscore the central importance of race in understanding the dynamics of gentrification in US cities. as part of multimillion dollar redevelopment projects. This form of state-supported gentri- fication has displaced hundreds of thousands of low-income families, most of them African American. In this analysis, I examine the racial dimension of the direct and indirect displace- ment induced by public housing demolition. The more prominent role of public invest- ment in recent gentrification efforts reflects the fact that neighbourhoods reached by this ‘third wave’ are more physically remote from Cities in the United States up and down the urban hierarchy have experienced significant levels of gentrification1 since the national econ- omy emerged from the recession of the 1990s (Wyly and Hammel, 2004). This period has differed from previous waves of gentrification in the degree of public-sector investment that has driven neighbourhood transformations (Hackworth and Smith, 2001). Prominent among these state interventions has been the demolition of public housing, in some cases 1582 EDWARD GOETZ traditional centres of capital and seen as too risky and challenging for speculative private investment (Hackworth, 2002). In many cases, the gentrifying neighbourhoods have been the home to older public housing communities that have suffered from disinvestment and neglect for decades. The neighbourhoods are not only forbidding to private inves- tors because of the scale of deprivation and the accompanying problems of crime and disinvestment, they are also dominated by the super-blocks that typify public housing estates and thus lack the type of housing typi- cally exploited by gentrifiers. The demolition of large public housing estates clears away significant concentrations of poverty from these areas and allows for the conversion to a housing stock and land use pattern more accommodating of private-sector investment. In the US, the centrepiece of state-led efforts to deconcentrate poverty and transform inner-city neighbourhoods is the HOPE VI programme. This programme has been used in dozens of cities to demolish public housing developments and to create new mixed-income communities in their place. The programme thus redefines public hous- ing policy and serves as the main vehicle through which the state has triggered inner- city revitalisation (Newman, 2004; Wyly and Hammel, 1999). The major transformation of public housing, in fact, reflects several dimen- sions of neo-liberal urban policy in the US over the past 20 years. While the elimination of the physical structures of public housing eliminates visual references to New Deal and welfare state policies no longer dominant, the removal of concentrations of very-low- income people of colour allows a reimaging of urban spaces critical to the national and international competition for private invest- ment (Newman and Ashton, 2004). Further, the policy prescriptions imposed upon former public housing residents, both relocation to private-sector housing through the use of Housing Choice Vouchers and the increased behav ioural monitoring and screening techniques employed at the mixed-income communities, reflect newfound emphases on choice, individualism and market discipline that are central to the neo-liberal governance paradigm (Jones and Popke, 2010). The dismantling of public housing is not limited to HOPE VI projects, however. PHAs in cities like Atlanta, Memphis and Las Vegas have plans to demolish most or all of their public housing. Chicago’s Plan for Transformation calls for the demolition of more than 20 000 units and a net reduction of 13 000, while in post-Katrina New Orleans, the local housing authority has demolished thousands of units that were kept vacant after the hurricane and flood. In these ways, housing policy and, more specifically, public housing demolition and dispersal have been employed as economic development strate- gies by local governments intent on finding and forcing new paths of neighbourhood change and gentrification (Newman, 2004). Cities in which market pressures for gentri- fication have been the strongest, for example, have been the most aggressive in tearing down public housing (Goetz, forthcoming). Public housing projects such as Earle Village in Charlotte, NC, Techwood Homes in Atlanta, Allen Parkway Village in Houston, St Thomas in New Orleans, Cabrini-Green in Chicago and Ellen Wilson in Washington, DC, have all been demolished to make way for new devel- opment that has ignited significant private- sector investment in housing and commercial markets nearby (Kingsley et al., 2003; Jones and Popke, 2010; Keating and Flores, 2000; McGhee, 2004; Fosburg et al., 1996; Bennett and Reed, 1999). In fact, HUD has emphasised the potential for additional investment and neighbourhood change when evaluating HOPE VI propos- als, looking for projects that could catalyse significant neighbourhood transformation (Zielenbach, 2002). The emphasis on leverag- ing private capital is in practice an incentive GENTRIFICATION IN BLACK AND WHITE 1583 for projects located in neighbourhoods ripe for private investment. By fiscal year 2002, local housing authorities were required to demonstrate how their proposed HOPE VI redevelopment would “result in outside investment in the surrounding community” (US GAO, 2003, p. 9). Thus, HUD has targeted projects they felt had the greatest potential to spur additional public and private investment in the form of new or rehabilitated hous- ing, commercial investment, new jobs and improved public infrastructure. Race and Gentrification Although gentrification has typically been seen as “class-based colonization of urban land”, it has a clear racial dimension as well (Moore, 2009, p. 138). Demographic transforma- tions produced by gentrification are nearly as frequently racial as they are class-based. The predominant racial reality of gentrification has been one of White gentrifiers displacing low- income Black incumbents. This is certainly an element of third wave gentrification, just as it was in previous waves. However, the extended reach of current gentrification efforts, and the contemporaneous efforts to extend homeown- ership, have increased the prevalence of ‘Black gentrification’ (Moore, 2009). Recent ethno- graphic research has explored the dynamics occurring in Harlem, Chicago’s southside Bronzeville neighbourhood, and Philadelphia in which the gentrifiers are Black middle-class homeowners and the class conflicts produced are experienced entirely within the Black com- munity (see Boyd, 2008; Patillo, 2007, Hyra, 2008, Moore, 2009). Moore (2009) argues that the dynamics of Black gentrification are distinctive. The phenomenon itself, she argues, is conditioned by larger patterns of racial segregation and exclusion, and the constrained mobility of the Black middle class. Hence, Black gentri- fication depends upon the size of the Black middle class within a city, prevailing patterns of segregation and exclusion in the local housing market, and to some extent upon the identification of ‘historically Black com- munities’ to which gentrifiers can lay claim. According to Hyra the black middle class fiercely wants to ‘restore’ these communities to safe, prosperous, and tranquil places (Hyra, 2008, p. 130). These gentrifiers find themselves investing in communities with high concentrations of public housing and poverty. They contend that their potential for wealth generation is limited by public policy decisions that concentrate subsidised housing in minor- ity neighbourhoods. They argue for a ‘fair share’ approach to public housing that would locate fewer subsidised units in Black neigh- bourhoods and more in middle-class White neighbourhoods that are largely free of such housing. Thus, members of the Black middle class support the demolition of public hous- ing as a necessary step in creating livable com- munities (Patillo, 2007). On the other hand, displacees see themselves as members of the same community that the Black middle class is investing in. In many cases, Black public housing residents have been in the commu- nity longer than members of the Black middle class who might be more recent residents. To public housing residents forced out of their neighbourhoods, Black gentrification is no better than White gentrification. Indeed, in many ways, Black gentrification is seen as a greater betrayal (Patillo, 2007). The current campaign to demolish and transform public housing will primarily affect low-income Black families. Public housing in the US is disproportionately occupied by people of colour, predominantly African Americans, and it is disproportionately located in minority neighbourhoods (Newman and Schnare, 1997). In 2000, 48 per cent of the residents in public housing nation-wide were African American, despite the fact that Blacks make up less than 15 per cent of the national 1584 EDWARD GOETZ population. In larger cities (having more than 5000 units of public housing), 66 per cent of public housing residents were Black. In some cities, notably Birmingham, Detroit, Memphis, New Orleans and Washington, DC, HUD data show that figure to be 99 per cent. Given these residency patterns, any action taken on public housing will have a disproportionate impact on African Americans. Furthermore, the con- centration of public housing in heavily Black neighbourhoods means that indirect displace- ment, should it occur, will also have a dispro- portionate impact on African Americans. It is possible then to conceptualise the neighbourhood change generated by public housing displacement as taking place along two continuums (see Figure 1). In neigh- bourhoods with little or no change in the poverty rate and in racial profile (the area near the intersection of the two axes), the neighbourhood remains largely static despite the public housing demolition. Where racial turnover (Black to White) has taken place, without a change in poverty (lower right quadrant), the neighbourhood has deseg- regated but remained poor. In practice, this has remained a largely hypothetical outcome. When both racial turnover occurs and a significant reduction in poverty takes place (lower left), neighbourhoods can be said to be experiencing White gentrification. When poverty is significantly reduced without racial change (upper left), Black gentrification has occurred. Neighbourhoods that continue to segregate and concentrate poverty are located in the upper right quadrant. The Record of Displacement in Public Housing Demolition and Redevelopment Studies of households displaced from public housing show a distinct pattern of reloca- tion (see the review in Goetz and Chapple, 2010). The evidence on HOPE VI and other instances of forced displacement from public housing suggests that displaced residents typi- cally move nearby and remain in the central city (see Comey, 2007; Goetz, 2003; Gibson, 2007; Clampet-Lundquist, 2004; Varady and Walker, 2003; Trudeau, 2006). Furthermore, displaced residents typically move to other disadvantaged neighbourhoods, with poverty rates considerably above city-wide aver- ages (Buron et al., 2002; Goetz, 2003, 2010; Clampet-Lundquist, 2004; Boston, 2005; Oakley and Burchfield, 2009). Very few dis- placed public housing residents return to Continued segregation and decline Desegregation White gentrification Black gentrification Change in size of African American population Decline Increase Change in extent of poverty Decline Status quo Increase Figure 1. Neighbourhood change in predominantly Black public housing communities. GENTRIFICATION IN BLACK AND WHITE 1585 live in the mixed-income redevelopments sometimes built to replace their public hous- ing communities. Estimates from national studies indicate that the percentage of original residents who return to the redeveloped site generally ranges from 14 per cent to 25 per cent (Marquis and Ghosh, 2008). The rate of return is low for a number of reasons: the redeveloped sites often have fewer public housing units than the projects they replace; new management standards make it dif- ficult for previous residents to pass tenant screening criteria; and the long time-span between displacement and the completion of redevelopment means that many previ- ous residents have resettled into new com- munities and do not wish the disruption of moving again (Wilen and Nayak, 2006; Jones and Popke, 2010). The benefits of being moved away from public housing communities subject to demo- lition are strongest in residents’ perception of reduced crime in their new communities and in their increased satisfaction with the quality of their housing post-move (see, for example, Petit, 2004; Gibson, 2007; Goetz, 2003, 2010). Improvements in housing and neighbour- hood characteristics are also consistently reported by displaced public housing families (see, for example, Popkin, 2006; Buron et al. 2002; Comey, 2007; Brooks et al., 2005; Goetz, 2003), though exceptions are reported in some cases (Gibson, 2007; Manzo et al., 2008). Less positive results are reported in the areas of health and children’s school experiences, and in residents’ economic self-sufficiency. Manjarrez et al. (2007) report no significant or consistent improvements in health for families displaced by HOPE VI. Similarly, children of HOPE VI families showed no health improvements over a five-year period (Gallagher and Bajaj, 2007). School achieve- ment among children relocated by HOPE VI has shown the same pattern. Children’s new schools remain racially and economically segregated, in part because many HOPE VI moves are within the same, underperform- ing urban school systems (Popkin, 2006). Children in households relocated due to HOPE VI or similar public housing redevel- opment show no educational improvements relative to control group members on a range of academic achievement measures (Jacob, 2004; see also, Gallagher and Bajaj, 2007). The evidence is clear and consistent in showing that displacement from distressed public housing projects has had no demon- strable positive effect on employment, earn- ings or income of individuals. The lack of any effect on economic self-sufficiency is repeated across all studies of public housing displacement (see Turney et al., 2006; Levy and Woolley, 2007; Clampet-Lundquist, 2004; Goetz, 2003, 2010; Curley, 2009). In fact, the finding is repeated for all forms of dispersal (for example, Kling et al., 2007; US HUD, 2004; Vigdor, 2007). The impact of displacement on social capi- tal is especially problematic. Movers report difficulty in establishing new social ties, they miss their social milieu from the old neigh- bourhood and worry about isolation in their new places (Curley, 2009; Greenbaum et al., 2008; Trudeau, 2006; Clampet-Lundquist, 2004, 2007; Gibson, 2007; Kleit, 2001). Manzo et al. (2008) find that public housing residents value their communities and regard them as having had a positive impact on their own lives. Two-thirds of the displaced families in their study felt the original public housing project was a good place to live (see also Gibson, 2007, for similar findings from Portland Oregon). These ties to place and to a social network are disrupted by displace- ment, even when demolition removes public housing projects that are unsafe, crime-ridden and physically deteriorated—i.e. extremely dysfunctional and dangerous places that are themselves argued to undermine the social, emotional and financial well-being of their inhabitants (see for example, Popkin et al., 2000; Venkatesh, 2000). 1586 EDWARD GOETZ Data and Methods The analysis proceeds in two parts. The first is an examination of direct displacement induced by public housing redevelopment and demolition between 1996 and 2007 in the 139 largest US central cities.2 The second part is an analysis of indirect displacement triggered in neighbourhoods surrounding HOPE VI redevelopment projects begun during the 1990s. Direct Displacement To analyse direct displacement, it is neces- sary to know the number and characteristics of residents who lived in public housing projects prior to demolition. A list of public housing projects that have been demolished since 1990 was obtained from HUD. The agency has no data on the families directly displaced by public housing demolition and redevelopment projects. As a result, projects on the demolition list were matched with the resident data contained in HUD’s Picture of Subsidized Households databases that provide details about the residents of public housing projects across the country. I estimate the racial impact of displacement by looking at the racial breakdown of public housing units prior to their demolition. HUD’s database is available for the years 1996, 1997, 1998 and 2000. In general, I use the dataset that corre- sponds to the year prior to the demolition of a given project. Thus, for projects demolished in 1997, the 1996 database provides informa- tion on the resident mix. No resident infor- mation is available for projects demolished prior to 1997. For all projects demolished after 2000, I use the most recently available database, the 2000 version. For example, the Bernal Heights Dwellings in San Francisco were demolished in 1997. HUD’s 1996 Picture of Subsidized Households database indicates that in 1996 the 208 units in the Bernal Heights Dwellings were 93 per cent occupied (193 households). In 1996, 69 per cent of households at Bernal Heights were African American and thus 133 African American households (193 * 0.69) were dis- placed by the Bernal Heights demolition. In cases of partial demolition (for example, a project with three high-rise towers in which only one is demolished), I assume that the racial breakdown of tenants in the demolished building is identical to that of the overall development. There are several potential standards against which to judge whether public housing demo- lition has had a disproportionate impact on Blacks, each one corresponding to a more restricted spatial scale. To judge the displace- ment of Blacks from public housing against the representation of Blacks in the national population would show tremendous dis- parity owing to the over-representation of Blacks in public housing. A second possible standard is the proportion of Blacks in public housing across the nation (49 per cent). This too would produce a conclusion of highly disparate impact. Yet, this is also a less than adequate standard because we know that public housing demolition is concentrated in larger cities, places where Blacks typically make up a higher percentage of public hous- ing residents. Restricting the referent standard to large cities would produce a more targeted comparison, but in this analysis I go beyond that to use the overall racial distribution of a given city’s public housing stock as the reference for demolitions that took place in that city. That is, if 69 per cent of the public housing stock in San Francisco was occupied by Blacks (i.e. the same as the Bernal Heights development described earlier), then the Bernal Heights demolition would be judged not to have had a disproportionate impact on African Americans. HUD’s 1996 database indicates, however, that city-wide, public housing in San Francisco was only 49 per cent Black in 1996. Thus, I conclude that the Bernal Heights Dwellings demolition did have a disproportionate impact on African GENTRIFICATION IN BLACK AND WHITE 1587 Americans. For each demolition in the database, a disparity ratio is computed with the numerator being the proportion of the demolished project occupied by Blacks and the denominator being the city-wide percent- age of public housing occupied by Blacks. This analysis is repeated and aggregated for all public housing projects for which data are available.3 This standard allows one to establish, at the level of an individual public housing authority, whether public housing demolition is disproportionately affecting African Americans. Indirect Displacement The analysis of indirect displacement employs a different sample of projects and a different database. The analysis is based on the author’s assembled data on HOPE VI redevelopment projects.4 The projects were geo-coded and census data from the 1990 and 2000 censuses were collected for the areas surrounding the projects. The analysis is therefore based on changes taking place between 1990 and 2000 in these project neighbourhoods. A HOPE VI project is defined as the full set of redevelopment-related activities that take place at public housing developments. A single HOPE VI project may receive multiple HOPE VI grants (several have). The most common pairing of grants is one grant for demolition and a second, later grant for rede- velopment. In other cases, as the demolition and redevelopment of large public housing developments has proceeded in stages over a long period of time, successive stages have received separate HOPE VI grants. Different public housing developments sharing physical space are deemed to be a single project for the purposes of this analysis. So, for example, the ABLA projects in Chicago are four separate public housing develop- ments, the Jane Addams Homes built in 1938, the Robert Brooks Homes completed in 1943, Loomis Courts constructed in 1951 and the Grace Abbot Homes built in 1955. In all, the four developments had 3600 apartments in a single contiguous location. For this study, ABLA, which received eight separate HOPE VI grants, is considered a single project. Neighborhood is defined as the census block groups whose centroids are within a half-mile radius of the HOPE VI project address.5 These trapezoidal areas were truncated wherever significant man-made or natural boundaries occurred, such as rivers or major highways. Once the relevant block groups were identi- fied, the Geolytics Neighborhood Census database was used to collect social, physical and economic characteristics for 1990 and 2000. The Geolytics database standardises census boundaries across the two census years, allowing for comparison of identical spatial areas. In the analysis to follow, I examine the degree to which public housing demolition and redevelopment have triggered the dis- placement of low-income households and African Americans in the surrounding neigh- bourhoods. Displacement is given as a decline in the neighbourhood population (poverty households and the African American popu- lation) at a rate greater than occurring in the city at large. A simple measure of the relative change in the African American population, for example, is computed as follows where, CB2K and CB90 = percentage of city- wide population that is Black in 2000 and 1990 respectively; and NB2K and NB90 = percentage of neighbourhood population that is Black in 2000 and 1990 respectively. This produces a difference-in-difference score in which a positive value indicates greater decline in the Black population at the neigh- bourhood level than at the city-wide scale. Timing the Intervention A HOPE VI redevelopment project con- sists of many events, including relocation, ( ) ( )CB2K CB90 NB2K NB90− − − 1588 EDWARD GOETZ demolition and project completion. One challenge of this analysis is determining how much redevelopment activity is necessary to trigger neighbourhood change. Some of these projects are located in neighbour- hoods poised to gentrify and thus change will occur quickly. Other neighbourhoods with different conditions may need more dramatic effor ts before change occurs. Because of this, we test for three different intervention points. A HOPE VI project is typically announced with much fanfare as local officials herald a multimillion dollar reinvestment effort. It is possible that private investors waiting for a sig- nal to trigger neighbourhood change will take the grant announcement as the time to act. In the analysis to follow, I examine the neighbour- hood change for projects funded before 2000. There are 176 revitalisation grants in that group. The first visual evidence of change taking place, however, is the relocation of the project residents. This event may potentially serve as the critical ‘intervention’ point in the process, triggering changes throughout the neigh- bourhood. Thus, I examine HOPE VI projects in which relocation occurred before 2000, a total of 100 projects (a subset of the 176 projects in the first sample). Finally, the most dramatic visible sign of redevelopment may be the demolition of the old public housing structures and commencement of construc- tion. Taking this as the intervention point leads to a smaller sub-sample of 64 projects. Analysis Direct Displacement There were 394 public housing projects demolished in the 139 largest US central cit- ies between 1995 and 2007. These projects accounted for 163 393 units of public housing (an average of 415 units per project). Of these units, 110 227 (67 per cent) were occupied dur- ing the year for which we have occupancy data. The HUD data contain resident demographic information for 313 cases, or 87 251 house- holds. The average size of the projects in the database is 397 units, although the median is 293. The mean is skewed upward by a relatively few large projects; one-quarter of the projects had more than 515 units prior to demolition. The number of people displaced in the 313 projects for which resident information is available is estimated at 239 844 people.6 This is an underestimate of the total number of people displaced because it excludes 83 projects for which HUD reports no resident information, despite the fact that they were at least partially occupied the year before demolition. The majority of cases in the database are HOPE VI projects (228, or 73 per cent of the projects for which we have resident informa- tion). While HOPE VI projects are on aver- age larger than other projects that have been demolished (a mean of 421 units compared with 327; p < 0.05), the data show that the HOPE VI projects and other demolitions were statistically identical in terms of resident demographics. On seven indicators (percentage of residents earning less than $5000, percentage with wage incomes, with welfare income, per- centage seniors, disabled, minority and African American) there was no statistical difference between the HOPE VI and the other demoli- tions (data not shown). Thus, all demolitions in these cities are analysed as a single group. The overwhelming majority of house- holds directly displaced by public housing demolitions across the country are African American. Of the 87 251 displaced house- holds for whom demographic information is known, 71 373 (82 per cent) households (or more than 192 000 residents, given average household size in these projects) were African American. The average demolition displaced 229 African American households (or 641 African American residents). In half of the demolished projects, African Americans were 95 per cent or more of the households. Are these figures higher than one would expect to find in these cities during these years? Table 1 GENTRIFICATION IN BLACK AND WHITE 1589 Ta b le 1 . D em o gr ap h ic c h ar ac te ri st ic s o f d em o li sh ed p u b li c h o u si n g, 1 9 9 6 –2 0 0 7 P er ce n ta ge B la ck (n = 3 05 ) P er ce n ta ge H is p an ic (n = 3 06 ) P er ce n ta ge M in or it y (n = 3 06 ) P er ce n ta ge e ar n in g le ss t h an $ 50 00 (n = 2 96 ) P er ce n ta ge w it h w ag es (n = 2 96 ) P er ce n ta ge w it h w el fa re (n = 2 96 ) P er ce n ta ge o f fe m al e- h ea d ed h ou se h ol d s (n = 3 05 ) P er ce n ta ge of s en io rs (n = 3 04 ) P er ce n ta ge d is ab le d (n = 3 04 ) D em o li sh ed p ro je ct s 7 9 .5 1 1 .5 9 4 .2 3 2 .2 2 4 .9 2 8 .6 8 2 .8 1 5 .4 1 6 .9 O th er p u b li c h o u si n g p ro je ct s 7 3 .2 1 1 .2 8 7 .5 2 5 .2 2 2 .2 2 1 .5 7 8 .5 2 5 .9 2 0 .2 Si gn ifi ca n ce ** * - ** * ** * ** * ** * ** * ** * ** * N ot e: * ** p < 0 .0 0 1 . 1590 EDWARD GOETZ compares demolished projects with the rest of the public housing stock in the same cities. The data suggest a disparate racial impact of public housing demolition across more than 300 demolitions in these large American cities. The average project demolished was 79.5 per cent African American, while other projects in the same cities were on average 73.2 per cent African American. For Hispanic residents, however, there was no disparate impact; the average demolished project was 11.5 per cent Hispanic compared with 11.2 per cent city-wide. The data in Table 1 also provide evidence of other statistically significant differences between projects that have been demolished and other public housing. In the average demolished project, 32.2 per cent of residents had incomes less than $5000, compared with only 25.2 of residents in comparison projects. Conversely, demolished projects had higher relative populations of wage-earners and resi- dents with welfare income. This is likely to be due to the fact that demolished projects also had significantly fewer seniors and disabled households than public housing that was not demolished. These averages mask a wide range of out- comes across projects. Disparity ratios were derived by dividing the percentage Black in a given project by percentage Black in the rest of the city’s public housing. For example, the Christopher Columbus Homes in Paterson, New Jersey, demolished in 2000, were 97 per cent Black-occupied in 1999. The rest of the public housing stock in Paterson in 1999 was 70 per cent Black. This produces a dispar- ity ratio of 1.39 (97/70). The ratios range from 0 (in projects that displaced no African American households) to 5.08. A disparity ratio of 1.0 signals that a demolished project exactly matched the racial profile of the rest of the public housing stock in the same city for the same year. Table 2 shows projects with the highest disparity ratios. The highest ratios occurred in cities in which African Americans made up half or fewer of all public housing households. Demolitions in those cities, nevertheless, affected some projects with very large pro- portions of African American residents. The unweighted average disparity ratio for the 305 projects for which all data are available is 1.096, indicating that the average public hous- ing project demolished had 9.6 per cent more African American households as a percentage of all households than other public housing in the same cities in the same year. Twenty- two per cent of the demolished projects had ratios of less than 1.0, meaning that there were Table 2. Ten projects with highest disparity ratios Rank Project, City Disparity ratio Percentage Black Project Other public housing 1 Springview Apts, San Antonio 5.08 61 12 2 Iris Court, Portland, OR 2.76 58 21 3 DN Leathers II, Corpus Christi 2.67 32 12 4 320 – 23rd St, Denver 2.63 71 27 5 Arapahoe Cts, Denver 2.60 65 25 6 Curtis Park Homes, Denver 2.44 61 25 7 Arrowhead Apts, Denver 2.41 65 27 8 Mulford Gardens, Yonkers 1.92 96 50 9 College Hill Homes, Knoxville 1.91 90 47 10 Lonsdale Homes, Knoxville 1.88 90 48 GENTRIFICATION IN BLACK AND WHITE 1591 fewer African American households in those projects compared with other public housing in the cities studied. Thirty-seven per cent of the projects had disparity ratios between 1.0 and 1.10, one-quarter (24.6 per cent) had disparities from 1.10 to 1.25 and the rest (16.7 per cent) had disparity ratios of 1.25 or more. The overall disparity ratio is determined by dividing the total number of Black households displaced in all 305 projects by the expected number displaced, where the expected num- ber is simply the city-wide percentage Black applied to each project. In the Christopher Columbus Homes example, in Paterson, New Jersey, if there had been no disparate impact on Blacks, one would expect that 70 per cent of the 314 households in that project would have been Black (matching the rest of the city’s public housing stock in 1999). This means that 220 African American households would have been displaced. In fact, 97 per cent of the project was African American, or 305 households. Thus, this project displaced 85 more African American households (or 39 per cent more) than would have been expected given a non-disparate outcome. Summing this calculation across all 305 projects produces a weighted disparity ratio of 1.077; in the aggregate, projects that have been demolished in these cities have displaced 7.7 per cent more African Americans than would have been the case had there been no disparate impact. The weighted ratio is less than the unweighted average because of large projects in cities such as Chicago, Detroit and Baltimore, where virtually all public housing residents are Black and therefore the individual-project disparity ratios are close to 1.0. Disparity ratios are bounded on the upper end by the initial over-representation of Blacks in public housing in most of the large cities in this sample. In cities like Washington, DC, Memphis and Detroit, where 98 per cent or more of public housing residents are African American, there is essentially no possibility of a disparate racial outcome as defined here. Since both the numerator and the denomi- nator in the disparity ratio have maximum values of 100, as the denominator approaches 100, the possibility of a ratio above 1.0 dimin- ishes. Thirteen per cent of the demolitions in the sample (or 40 projects) took place in cities in which Blacks make up 99 per cent of all public housing households. In one-third of the demolitions (more than 100 projects), Blacks make up more than 90 per cent of all public housing households city-wide. Figure 2 shows how the disparity ratio is related to the percentage of city-wide public housing occupied by African Americans. When all cases are included in the analysis, the unweighted disparity ratio is near 1.096. When cases are eliminated at the upper end— i.e. cases are removed where the percentage of city-wide public housing is 98 per cent or above, the average disparity ratio increases. The average disparity ratio tops out at 1.18 when the analysis is restricted to cases in which the city-wide public housing population that is African American is 75 per cent or less. The other line in the figure indicates the number of cases that remain in the analysis as cases are removed. The disparity ratio becomes less stable as the number of cases declines. In fact, there is a precipitous decline in the number of cases as one moves to the right in the graph, a reflection of the fact that most of the public housing projects demolished are in cities where a large majority of public housing residents are African American.7 Several cities have sizeable disparities in the racial make-up of demolished public housing projects compared with their overall public housing profiles. Table 3 lists cities with a disproportionately high or low percentage of African American residents in demolished projects. In Corpus Christi, Texas, the public housing that was demolished had 2.96 times as many African American households (as a percentage of all households) than did the rest of the city’s public housing stock. The tenth-ranked city on the list, Denver, has 1592 EDWARD GOETZ torn down public housing that had 51 per cent more African American households as a percentage of all households than the rest of its public housing stock. Other cities had fewer African American households in demolished projects than would be expected given the overall racial make-up of public housing in the city. Two examples of this are El Paso, Texas, and Los Angeles, CA, where public housing demolition has affected projects with higher Hispanic populations. Cities in the west and south-west account for seven of the 10 cities with the lowest disparity ratios for African Americans. This suggests that disparity ratios might be an artifact of region and perhaps dependent on whether Blacks make up a small percentage of city-wide units. However, the bivariate 0.6 0.7 0.8 0.9 1 1.1 1.2 100 95 90 85 80 75 70 65 60 55 50 45 40 35 30 25 20 15 Pct of city-wide public housing occupied by African Americans D is p a ri ty R a ti o 0 50 100 150 200 250 300 350 N u m b e r o f c a s e s Disparity ratio n Figure 2. Disparity ratio as a function of percentage Black in city-wide public housing. Table 3. Cities with disparate displacement impact on African American households Cities with greatest disparities Cities with lowest disparities Rank City Disparity ratio Rank City Disparity ratio 1 Corpus Christi 2.96 1 El Paso 0.17 2 Yonkers 2.52 2 Los Angeles 0.22 3 Lubbock 2.36 3 Topeka 0.61 4 Portland, OR 2.03 4 Sacramento 0.61 5 Ft Wayne 1.74 5 Hartford 0.69 6 Syracuse 1.70 6 Boston 0.70 7 Knoxville 1.65 7 San Antonio 0.76 8 Miami 1.62 8 Oakland 0.82 9 St Petersburgh 1.58 9 Tucson 0.85 10 Denver 1.51 10 Tacoma 0.87 GENTRIFICATION IN BLACK AND WHITE 1593 correlation between disparity ratio and percentage of the total public housing stock that is occupied by African Americans is small and statistically insignificant (r = -0.063). There are two possible alternative explana- tions for the racial disparities found in the preceding analysis. First, the disparate racial impact of demolition may be an artifact of the fact that ‘family’ public housing is demol- ished at a greater rate than projects that house seniors, and that senior public housing is less racially segregated. The second counter- explanation for disparate racial impact is that demolition has been targeted to the most dysfunctional public housing developments and that these are disproportionately occu- pied by African Americans. The assumptions behind these alternative explanations are only moderately supported by the data. The pro- portion of a project’s population that is Black is moderately correlated with the percentage that is below the age of 62 and with vacancy rate (a measure of building quality) (r = 0.28 and 0.23 respectively). A logistic regression analysis of all public housing projects in the 139 cities of this sample demonstrates that, even when controlling for building quality and for the presence of seniors, developments that were predominantly (more than two- thirds) Black were 71 per cent more likely to be demolished than projects that were not mostly Black (see Table 4). The analysis shows that building quality (as measured by percentage of units occupied) is an important predictor of whether or not a public housing project was demolished. Additionally, senior buildings (defined here as projects in which more than half of the resi- dents were older than 62) were significantly less likely to be demolished than other proj- ects. Yet, even accounting for those factors, public housing developments that were pre- dominantly occupied by African Americans were significantly more likely to come down than projects with a more integrated profile. The data in Table 4 show that all three of these explanations are accurate in distinguishing demolished public housing from the projects left standing. Indirect Displacement The analysis of indirec t displacement caused by HOPE VI is based on the expec- tation that neighbourhoods surrounding Table 4. Importance of race, controlling for building quality and senior occupancy: binary logistic regression (dependent variable: demolition; N = 1926) Characteristics of public housing development β Exp (β) Significance Number of units 0.000 1.000 *** Percentage units occupied -0.036 0.964 *** Median rent -0.017 0.983 *** Percentage incomes less than $5000 0.000 0.999 - Percentage with wages -0.003 0.994 - Percentage under 25 years old 0.001 1.001 - Senior buildinga -1.693 0.184 *** Predominantly Blackb 0.539 1.714 ** Percentage units with 3 + bedrooms 0.002 1.002 - Log likelihood 1102.15 Percentage predicted correctly 88.6. a 50 per cent or more of the occupants are 62 years of age or more. b 66 per cent or more of the occupants are African American. Notes: *** p < 0.001; ** p < 0.01. 1594 EDWARD GOETZ HOPE VI projects will change due to the redevelopment that occurs on the public housing site. The focus here is on the degree of racial change and poverty reduction in HOPE VI neighbourhoods. In order to control for racial and poverty changes tak- ing place more broadly in the local housing market, neighbourhood change is calculated relative to city-wide change. The average population in these HOPE VI neighbourhoods was 5701 for the 176 that received their funding prior to 2000, 6246 for neighbourhoods that began relocation during the 1990s and 6857 for the neighbourhoods of projects that had been demolished (see Table 5). These neighbourhoods were 60 per cent African American on average and 12–14 per cent Hispanic. The HOPE VI neighbour- hoods were characterised by a high poverty rate (45 per cent), low levels of owner-occupancy (24–28 per cent) and vacancy rates of close to 15 per cent. Because the basic patterns of neighbourhood change are the same across these three nested sub-samples, the focus is on projects that relocated residents prior to the 2000 census (n = 100). Poverty Reduction in HOPE VI Neighbourhoods The commonly accepted threshold for ‘con- centrated poverty’ is an area in which more than 40 per cent of the population is below the poverty line (Jargowsky, 1996). Thus, on average, HOPE VI neighbourhoods were above that threshold when the 1990s began. By 2000, the 100 neighbourhoods that had already experienced HOPE VI relocation averaged 36.1 per cent poverty, a reduc- tion of more than eight percentage points. The average decline in poverty relative to changes taking place at the city level was 7.6 percentage points. That is, the average HOPE VI neighbourhood saw a decline in poverty that was 7.6 percentage points greater than their respective city-wide changes during the 1990s. Most neighbourhoods saw a decline in poverty rate that was significantly greater than the secular trends taking place city-wide (see Table 6). Looking at the projects that moved to relocation during the 1990s (the middle column in the data table), one sees that poverty declined faster city-wide than in the HOPE VI neighbourhood in 25 projects (25 per cent). In 15 cases, the neighbourhood reduction in poverty was slightly greater (less than five percentage points) than what was experienced city-wide. In close to one- half of the cases (46 per cent), however, the neighbourhoods saw a reduction in poverty that was at least 10 percentage points greater than what was happening city-wide. Table 6 indicates that projects that were farther along (i.e. had moved to demolition by 2000) were even more likely to show significant poverty reduction (52 per cent declined in poverty at a rate at least 10 percentage points more than the city). Table 5. HOPE VI project characteristics Neighbourhood characteristics First grant in 1990s (n = 176) Relocation in 1990s (n = 100) Demolition in 1990s (n = 64) Mean population, 1990 5701 6246 6857 Mean percentage Black, 1990 59.7 61.1 60.3 Mean percentage Hispanic, 1990 11.9 12.3 14.1 Mean percentage Poverty, 1990 45.0 45.0 45.6 Mean percentage owner-occupied housing 28.4 25.6 24.3 Mean percentage vacant housing 14.9 14.9 14.7 GENTRIFICATION IN BLACK AND WHITE 1595 Racial Turnover Only one-quarter of the HOPE VI neighbour- hoods saw a reduction in Black population more than 10 percentage points greater than the city-wide rate of change. In comparison with poverty reduction, the impact of HOPE VI on the Black population is more moder- ate (i.e. relatively more cases in which the reduction in Black population outpaced the city-wide trend by 5–10 percentage points). Table 7 summarises the displacement effect for African Americans across all of the HOPE VI projects begun in the 1990s. The data show that HOPE VI redevelopment projects that received funding in the 1990s were, on aver- age, located in cities that were experiencing a relative decline in their African American populations and the projects were located in neighbourhoods that lost population over the decade. Thus, although the neighbour- hoods saw sizeable reductions in their Black populations, one would have expected some reduction given overall city trends during this decade. For projects that received their first grant before 2000, their neighbour- hoods were 59.7 per cent Black in 1990 and 55.9 per cent Black in 2000, a decline in the Black population that is an average of 3.17 percentage points greater than what occurred at the city scale. The per project displacement effect is 159; the average project displaced 159 more African Americans from the neighbour- hood than would have been expected given city-wide trends. The displacement effects increase in projects that went to the reloca- tion or demolition stages during the 1990s. For projects that saw relocation, the average neighbourhood went from 61.1 per cent Black to 56.1 per cent; for projects in which demoli- tion took place in the 1990s, the neighbour- hoods fell from 60.3 per cent Black to 55 per cent Black. Among both groups of projects, the reduction in Black population in these neighbourhoods was greater, on average, than what took place in the rest of the city. Some portion of the neighbourhood-wide reduction in Black population is due to the relocation/demolition of the public housing Table 6. Changes in poverty and in African American population in HOPE VI neighbourhoods relative to changes taking place city-wide, 1990–2000 Projects that received grant in 1990s (n = 175) Projects that relocated families in 1990s (n = 100) Projects that were demolished in 1990s (n = 64) Poverty African American Poverty African American Poverty African American Neighbourhood reduction that trailed the city-wide rate of reduction 44 (25) 51 (29) 25 (25) 25 (25) 12 (19) 15 (23) Decline that exceeded the city-wide rate by less than 5 percentage points 29 (17) 49 (28) 15 (15) 25 (25) 12 (19) 19 (30) Decline from 5–10 percentage points greater than city-wide decline 28 (16) 43 (25) 13 (13) 27 (27) 7 (10) 15 (23) Decline of more than 10 percentage points greater than city-wide 74 (42) 32 (18) 46 (46) 23 (23) 33 (52) 15 (23) Note: percentages are shown in parentheses. 1596 EDWARD GOETZ site, while the rest is the spillover, or indirect displacement effect. The relative size of the direct and indirect is impossible to determine because some of those displaced from the demolished public housing project may have moved elsewhere within the neighbourhood. The previous analysis indicated that the aver- age public housing demolition removed over 200 African American families. Using this as a guide, it appears that on average, HOPE VI redevelopment projects have overall gener- ated little indirect displacement of African Americans at the neighbourhood level. These national averages, however, obscure significant variation across cities and across projects. Table 8 shows the 10 most extreme cases of indirect displacement of African Americans in HOPE VI projects begun in the 1990s. The table lists the change in Black population in the neighbourhood as well as the estimate of displacement, which is the difference between the rate of racial change in the neighbourhood and what took place in the city as a whole during the decade. The most extreme case is the McGuire Gardens project in Camden, NJ, a city that saw a 5 percentage point decline in the Black population between 1990 and 2000. The neighbourhood of the McGuire Gardens project saw a 50 percentage point decrease in the proportion of its population that is African American. Thus, the city-wide trend and the neighbourhood trend diverged by 44.86 percentage points. In the cases of the Schuylkill Falls project in Philadelphia and the Commodore Perry project in Buffalo, the cities actually gained Black population while the HOPE VI neighbourhoods lost Blacks; thus the displacement estimate is greater than the actual level of decline in the neighbourhoods. This pattern applies to seven of the 10 cases shown in Table 8. At the same time, there are a number of HOPE VI projects that produced opposite effects or no racial effects at all. For example, the three HOPE VI projects in Albany, NY, a city that experienced a decline in the African American population of 6.25 percentage points between 1990 and 2000, all saw increases in the Black population in the neighbourhoods sur- rounding the project site. The Fairfield Homes project area in Baltimore saw an increase in the Black population of 4.12 percentage points, while the Black population declined in the city as a whole by 4.88 percentage points. Similar patterns were seen in Hartford, CT, and in single projects in Pittsburgh and Milwaukee. Table 7. Indirect displacement of African Americans in HOPE VI neighbourhoods Neighbourhood characteristics First grant in 1990s (n = 176) Relocation in 1990s (n = 100) Demolition in 1990s (n = 64) Mean Black population, 1990 3422 3671 3953 Mean percentage Black, 1990 59.7 61.1 60.3 Mean Black population, 2000 2594 2643 2763 Mean percentage Black, 2000 55.9 56.1 55.0 Mean expected reduction in Black populationa 670 810 942 Mean reduction in Black population 828 1028 1190 Per-project displacement effectb 159 217 248 Excessive reduction in Black populationc 27 902 21 767 15 889 a The reduction in the neighbourhood Black population if the neighbourhood change had matched the overall city change. b Actual reduction in neighbourhood Black population minus the expected reduction. c Sum of per-project displacement effects. GENTRIFICATION IN BLACK AND WHITE 1597 Taken together, the data presented in this section suggest that, while significant sec- ondary displacement of African Americans occurs in some cases, there are offsetting examples where no such indirect displace- ment has occurred or where African American populations have actually increased in the neighbourhood. HOPE VI projects seem to generate a range of racial outcomes not easily summarised. Figure 3 arrays HOPE VI projects along two dimensions; change in poverty and change in African American population, relative to changes taking place in the city as a whole.8 The most populated quadrant is the lower left which contains neighbourhoods experiencing White gentrification. These are neighbour- hoods that saw a relative and sizeable reduc- tion in poverty and a relative and sizeable reduction in African American population. A smaller number of neighbourhoods saw a reduction in poverty with little to no change in race, or even an increase in the Black popu- lation. These neighbourhoods could be said to be undergoing Black gentrification. The full national sample of HOPE VI proj- ects includes a number of projects located in neighbourhoods that are not predominantly Black, many of which are south-western cit- ies with large Hispanic populations. In these neighbourhoods, HOPE VI projects may have little to no impact on the Black population, which in any case, is not a large portion of the neighbourhood. HOPE VI Displacement Patterns in Predominantly Black Neighbourhoods By limiting the analysis to HOPE VI rede- velopments taking place in predominantly (50 per cent+) Black neighbourhoods, it is possible to focus more directly on the issue of Black displacement. Sixty-two per cent of the neighbourhoods in our overall sample can be classified as predominantly Black in 1990. Table 9 summarises the displacement infor- mation for projects in these neighbourhoods. The data indicate that, in HOPE VI neigh- bourhoods in which Blacks were the largest racial group, they constituted on average 84 per cent of the population in 1990. For all projects that began in the 1990s, the Black population declined four percentage points (84.0 to 79.9). For projects that went as far as demolition, the decline was six percentage points (84.7 to 78.8). In absolute numbers, the average reduction in Black population over the Table 8. Indirect displacement of African Americans for projects begun before 2000: ten highest levels of indirect displacement City Project Neighbourhood change in percentage Black, 1990–2000 Indirect displacement, 1990–2000a Camden McGuire Gardens -49.84 44.86 Philadelphia Schuylkill Falls -27.95 30.88 Buffalo Commodore Perry -23.37 29.67 Tulsa Osage Hills / Country Club Gardens -26.06 27.98 Stamford Southfield Village -24.56 21.98 New Haven Elm Haven -21.10 21.30 Chicago Cabrini Green -22.84 20.41 Charlotte Earle Village / First Ward -18.89 19.54 Birmingham Metropolitan Gardens -8.86 18.74 Portsmouth Ida Barbour Homes -14.25 17.32 a Calculated as the difference between the reduction in percentage Black in the neighbourhood and the city-wide reduction in percentage Black. 1598 EDWARD GOETZ –50 –40 –30 –20 –10 0 10 20 30 –50 –40 –30 –20 –10 0 10 20 30 40 Change in Pct. Poverty C h a n g e i n P c t. A fr ic a n -A m e ri c a n Black gentrification White gentrification Figure 3. Relative changes in race and poverty in HOPE VI neighbourhoods, for projects that relocated residents prior to 2000 (n = 100). Table 9. Indirect displacement of African Americans in predominantly Black HOPE VI neighbourhoods Neighbourhood characteristics First grant in 1990s (n = 108) Relocation in 1990s (n = 62) Demolition in 1990s (n = 39) Mean Black population, 1990 4824 4927 5232 Mean percentage Black, 1990 84.0 84.2 84.7 Mean Black population, 2000 3643 3486 3539 Mean percentage Black, 2000 79.9 78.7 78.8 Mean expected reduction in Black populationa 1005 1175 1372 Mean reduction in Black population 1180 1441 1694 Per-project displacement effectb 175 256 322 Excessive reduction in Black populationc 19 974 11 155 12 524 a The reduction in the neighbourhood Black population if the neighbourhood change had matched the overall city change. b Actual reduction in neighbourhood Black population minus the expected reduction. c Sum of per-project displacement effects. GENTRIFICATION IN BLACK AND WHITE 1599 decade was 1005 persons for projects begun in the 1990s. For projects that completed relocation, the reduction was 1175 and, for projects that moved to demolition, the aver- age reduction in Black population was 1372. The per-project displacement impact isolates the reduction in Black population above (or below) what is expected given city-wide trends. Predominantly Black neighbourhoods with HOPE VI projects initiated in the 1990s saw a reduction in the Black population of 175 more than expected given city-wide trends. For projects that went to relocation, the dis- placement effect was 256 and it was 322 for projects that saw demolition during the 1990s. Once again, the national averages hide a great deal of variation. Figure 4 provides a look at projects that took place in African American neighbourhoods. The predominance of White gentrification is more pronounced in this sample of projects. Twenty-seven HOPE VI neighbourhoods (44 per cent of this sample) saw a significant decline in both poverty and African American population. Black gentrification characterises 18 per cent of the HOPE VI neighbourhoods that began the 1990s with a predominantly Black population. These neighbourhoods are highlighted in the rectangle within Figure 4. While Chicago’s Bronzeville neighbourhood surrounding the Robert Taylor Homes is included in this group, the data reveal that Black gentrification is also occurring in HOPE VI neighbourhoods in Pittsburgh, Charlotte, Columbus (Ohio), Louisville and Wilmington (North Carolina), among other places. –50 –40 –30 –20 –10 0 10 20 30 –50 –40 –30 –20 –10 0 10 20 30 40 Change in Pct. Poverty C h a n g e i n P c t. A fr ic a n -A m e ri c a n White gentrification Black gentrification Figure 4. Relative changes in race and poverty in predominantly Black HOPE VI neighbourhoods, for projects that were demolished prior to 2000 (n = 62). 1600 EDWARD GOETZ Neighborhoods that are near the intersection of the two axes have not experienced signifi- cant change in either poverty or racial pro- file. In these neighbourhoods, the HOPE VI project has seemingly not triggered any larger neighbourhood change. Finally, the smallest number of neighbourhoods see a sizeable decline in African American residents but no change in poverty (desegregating neighbour- hoods) or see an increase in poverty with little racial change. Discussion Public housing demolition is playing a signifi- cant role in current patterns of gentrification in the US. Local officials have energetically pursued demolition of older public hous- ing projects in many cities to clear away the physical and social impediments to renewed private-sector investment in inner-city neigh- bourhoods. These efforts have displaced hundreds of thousands of very-low-income families since the 1980s and have had a dis- proportionate impact on African Americans. The disparate impact, furthermore, is not merely the result of the fact that Blacks are over-represented in public housing. Faced with a range of public housing projects to sweep away, local housing authorities have systematically chosen projects that, even by the standards of their own city, are dispropor- tionately inhabited by Black families. The racial dimension of gentrification has been, according to Moore (2009), under- theorised. Recent evidence of Black gen- trification has led to closer scrutiny of race in gentrifying neighbourhoods. The data presented here indicate that, in third wave gentrification—in which the state often takes the lead with high-profile redevelopment investments—race-based displacement is a prominent characteristic. The current prac- tice of tearing down public housing and the massive displacement of low-income, African Americans that has ensued is, in many ways, an update of the urban renewal experience of the 1950s and 1960s. The current surge in demolition, however, can be distinguished from the urban renewal story in one important way—the degree of neighbourhood change that it has induced. The data presented here indicate that public housing demolition and redevelopment are generating a wide range of neighbourhood outcomes, the most common of which are patterns of Black or White gentrification. While urban renewal often cleared land that remained fallow for years because of the lax land investment markets that characterised central cities in the 1950s and 1960s, public housing demolition from the mid 1990s through 2007 took place in the context of highly active real estate markets in many cities. Furthermore, agencies often targeted public housing in neighbourhoods where the poten- tial for spillover effects were greatest (Wyly and Hammell, 1999) and, in fact, were asked to do so by HUD. In the aggregate, redevelop- ment projects that began in the 1990s were not associated with a significant amount of neighbourhood racial turnover. Public hous- ing transformation has in some cities led to gentrification that entails significant racial turnover as well as changes in the income pro- file of the neighbourhood. This pattern was found in close to half of the HOPE VI projects in predominantly Black neighbourhoods. The data also show that one in five predominantly African American communities with a HOPE VI project experience Black gentrification— i.e. a significant reduction in poverty without racial change. HOPE VI triggered significant poverty reduction in most neighbourhoods, but was associated with racial turnover in a smaller number of places. The neighbourhood change analysis pre- sented in these pages must be regarded as preliminary given three important consider- ations. First, although the data reveal exam- ples of Black gentrification as well as other paths of neighbourhood change triggered by GENTRIFICATION IN BLACK AND WHITE 1601 public housing transformation, the analysis cannot shed light on why neighbourhoods move down one of these paths or another. Additional research is needed to identify the circumstances that produce, for example, Black gentrification as opposed to White gentrification, or the conditions under which public housing transformation is not able to generate any larger neighbourhood changes. Secondly, the amount of indirect displace- ment is probably underestimated because very few of the HOPE VI projects that form the basis of this analysis had completed the redevelopment process. As a result, the analy- sis more closely reflects the initial population changes associated with demolition and displacement. It is possible, indeed probable, that some if not most neighbourhood change dynamics would begin or accelerate after rede- velopment is complete, producing subsequent settlement patterns that could reflect changes greater than those discovered by this analysis. Finally, the neighbourhood data measure change over a 10-year period. The public housing redevelopment is only one event, albeit a major event, in that 10-year period. Although the analysis controls for broader market changes within the local economy, it is difficult to say, for any given neighbour- hood, whether public housing redevelopment produced the neighbourhood change seen, or was simply a part of a trend that began before demolition. Whether public housing demoli- tion was the cause of neighbourhood change or followed neighbourhood change requires more fine-grained data and analysis. The analysis confirms, however, the observations of previous studies that identify the central importance of public housing transforma- tion to patterns of gentrification in US cities. Notes 1. The definition of gentrification used here is the conventional one that highlights physical upgrading of the housing stock and built environment combined with the displacement of low-income residents in favour of higher socioeconomic groups (see for example, Marcuse, 1985; and Smith and LeFaivre, 1984). 2. The sample was created by combining the lists of the 150 largest US cities in 1990 and in 2000. This produced a list of 169 places, of which 30 were suburban communities. Given the lack of public housing in these suburban communities, they were removed from the analysis. 3. It should be noted at this juncture that the HUD database is dependent on the reporting by local housing authorities. For some cities in some years, no racial occupancy data are reported for any units. 4. Information on project characteristics was obtained from various sources, including HUD and local housing authorities. Address information was obtained from HUD and verified through direct observation, on-line sources and from local housing authorities. 5. The definition of neighbourhood is a difficult issue in large-n research. Jargowsky (1996) used census tracts, arguing that they are ‘the only realistic choice’ for national studies. Zielenbach (2002) defined neighbourhood to include the tracts that contained HOPE VI projects and all other tracts that abutted but did not include the site. Tracts were weighted by the percentage of the projects’ units within each tract, with abutting tracts counted as 5 per cent of the weighted average. The US GAO (2003) defined HOPE VI neighbourhoods as the set of census block groups adjacent to the block group containing the HOPE VI site. Fosburg et al. (1996) allow local research associates to define neighbourhood according to local usage. 6. The number of residents displaced is estimated by multiplying the average household size in each project by the number of occupied units. 7. This suggests another standard against which to judge whether a disparate impact has occurred—the proportion of public housing residents who are African American in the average large US city. In the average large city in the US in 1996, 59.7 per cent of public housing residents were African American. By this standard, there is very large disparate impact, since the average demolished project was 79.7 per cent African American. Thus, demolished projects contained 33 per cent 1602 EDWARD GOETZ more (79.7/59.7 = 1.33) African Americans as a proportion of all residents than would have been expected had demolitions occurred in projects that were representative of public housing in all cities in the sample. 8. The data are shown only for those projects that moved to relocation during the 1990s. The distribution of projects along these two dimensions is similar for projects that only received their funding in the 1990s and for those projects that moved to demolition in 1990s. References Bennett, L. and Reed, A. Jr (1999) The new face of urban renewal: the Near North redevelopment initiative and the Cabrini-Green neighbor- hood, in: A. Reed Jr (Ed.) Without Justice for All: The New Liberalism and Our Retreat from Racial Equality, pp. 175–211. Boulder, CO: Westview Press. Boston, T. D. (2005) The effects of mixed-income revitalization and residential mobility on public housing residents: a case study of the Atlanta housing authority, Journal of the American Planning Association, 71(4), pp. 1–19. Boyd, M. (2008) Jim Crow Nostalgia: Reconstructing Race in Bronzeville. Minneapolis, MN: University of Minnesota Press. Brooks, F., Zugazaga, C., Wolk, J. and Adams, A. (2005) Resident perceptions of housing, neighborhood, and economic conditions after relocation from public housing undergoing HOPE VI redevelopment, Research on Social Work Practice, 15(6), pp. 481–490. Buron, L. (2004) An improved living environment? Neighborhood outcomes for HOPE VI relocat- ees. Policy Brief No. 3, The Urban Institute, September. Buron, L., Popkin, S., Levy, D. et al. (2002) The HOPE VI Resident Tracking Study. Washington, DC: The Urban Institute. Clampet-Lundquist, S. (2004) Moving over or moving up? Short-term gains and losses for relocated HOPE VI families, Journal of Policy Development and Research, 7(1), pp. 57–80. Clampet-Lundquist, S. (2007) No more ‘Bois Ball’: the effect of relocation from public housing on adolescents, Journal of Adolescent Research, 22(3), pp. 298–323. Comey, J. (2007) HOPE VI’d and on the move. Brief No. 1, Metropolitan Housing and Communities Center, The Urban Institute, Washington, DC. Curley, A. M. (2009) Draining or gaining? The social networks of public housing movers in Boston, Journal of Social and Personal Relationships, 26(2/3), pp. 227–247. Fosburg, L. B., Popkin, S. J. and Locke, G. (1996) An historical and baseline assessment of HOPE VI. Volume I: cross-site report. US Department of Housing and Urban Development, Washington, DC. Gallagher, M. and Bajaj, B. (2007) Moving on: assessing the benefits and challenges of HOPE VI for children. Policy Brief No. 4, The Urban Institute, Washington, DC. Gibson, K. J. (2007) The relocation of the Columbia Villa communit y : v iews from residents, Journal of Planning Education and Research, 27(1), pp. 5–19. Goetz, E. G. (2003) Clearing the Way: Deconcentrating the Poor in Urban America. Washington, DC: Urban Institute Press. Goetz, E. G. (forthcoming) Where have all the towers gone? The dismantling of public housing in the US, Journal of Urban Affairs, 33. Goetz, E. G. (2010) Better neighborhoods, better outcomes? explaining relocation outcomes in HOPE VI, Cityscape, 12(1), pp. 5–31. Goetz, E. G. and Chapple, K. (2010) ‘You gotta move’: advancing the debate on the record of dispersal, Housing Policy Debate, 21(1), pp. 209–236. Greenbaum, S., Hathaway, W., Rodriguez, C. et al. (2008) Deconcentration and social capital: contradictions of a poverty alleviation policy, Journal of Poverty, 12(2), pp. 201–228. Hackworth, J. (2002) Post-recession gentrifica- tion in New York city, Urban Affairs Review, 37, pp. 815–843. Hackworth, J. and Smith, N. (2001) The chang- ing state of gentrification, Tijdschrift voor Economische en Sociale Geografie, 92, pp. 464–477. Hyra, D. (2008) The New Urban Renewal: The Economic Transformation of Harlem and Bronzeville. Chicago, IL: University of Chicago Press. Jacob, B. (2004) Public housing, housing vouchers and student achievement: evidence from public housing demolitions in Chicago, American Economic Review, 94(1), pp. 233–258. GENTRIFICATION IN BLACK AND WHITE 1603 Jargowsky, P. A. (1996) Poverty and Place: Ghettos, Barrios, and the American City. New York: Russell Sage Foundation. Jones, K. T. and Popke, J. (2010) Re-envisioning the city: Lefebvre, HOPE VI, and the neolib- eralization of urban space, Urban Geography, 31(1), pp. 114–133. Keating, L. and Flores, C. A. (2000) Sixty and out: Techwood Homes transformed by enemies and friends, Journal of Urban History, 26, pp. 275–311. Kingsley, G. T., Abravanel, M. D., Cunningham, M. et al. (2003) Lessons from HOPE VI for the future of public housing. The Urban Institute, Washington, DC. Kleit, R. G. (2001) The role of neighborhood social networks in scattered-site public hous- ing residents’ search for jobs, Housing Policy Debate, 12(3), pp. 541–573. Kling, J. R., Liebman, J. B. and Katz, L. F. (2007) Experimental analysis of neighborhood effects, Econometrica, 75(1), pp. 83–119. Levy, D. K. and Woolley, M. (2007) Employment barriers among HOPE VI families. Policy Brief No. 6, The Urban Institute, Washington, DC. Manjarrez, C., Popkin, S. and Guernsey, E. (2007) Poor health: the biggest challenge for HOPE VI families? Policy Brief No. 5, The Urban Institute, Washington, DC. Manzo, L. C., Kleit, R. G. and Couch, D. (2008) Moving three times is like having your house on fire once: the experience of place and impending displacement among public housing residents, Urban Studies, 45 (9), pp. 1855–1878. Marcuse, P. (1985) Gentrification, abandonment and displacement: connections, causes and policy responses in New York city, Journal of Urban and Contemporary Law, 28, pp. 195–240. Marquis, G. P. and Ghosh, S. (2008) Housing opportunities for people everywhere (HOPE VI): who gets back in?, The Social Science Journal, 45, pp. 401–418. McGhee, F. L. (2004) How HUD’s HOPE VI program is destroying a historic Houston neighborhood. Paper presented at the Urban Issues Program, GSN Forum, University of Texas at Austin. Moore, K. S. (2009) Gentrification in Black face? The return of the Black middle class to urban neigh- borhoods, Urban Geography, 30(2), pp. 118–142. Newman, K. (2004) Newark, decline and avoidance, renaissance and desire: from disinvestment to reinvestment, Annals of the American Academy of Political and Social Sciences, 594(July), pp. 34–48. Newman, K. and Ashton, P. (2004) Neoliberal urban policy and new paths of neighborhood change in the American inner city, Environment and Planning A, 36, pp. 1151–1172. Newman, S. J. and Schnare, A. B. (1997) ‘... and a suitable living environment’: the failure of housing programs to deliver on neighborhood quality, Housing Policy Debate, 8(4), pp. 703–741. Oakley, D. and Burchfield, K. (2009) Out of the proj- ects, still in the ’hood: the spatial constraints on public-housing residents’ relocation in Chicago, Journal of Urban Affairs, 31(5), pp. 589–614. Patillo, M. (2007) Black on the Block: The Politics of Race and Class in the City. Chicago, IL: University of Chicago Press. Petit, B. (2004) Moving and children’s social connections: neighborhood context and the consequences of moving for low-income families, Sociological Forum, 19(2), pp. 285–311. Popkin, S. J. (2006) The HOPE VI program: what has happened to the residents?, in: L. Bennett, J. L. Smith and P. A. Wright (Eds) Where are Poor People to Live? Transforming Public Housing Communities, pp. 68–90. Armonk, NY: M. E. Sharpe. Popkin, S. J., Buron, L. F., Levy, D. K. and Cunningham, M. K. (2000) The Gautreaux legacy: what might mixed-income and dispersal strategies mean for the poorest public housing tenants?, Housing Policy Debate, 11(4), pp. 911–942. Smith, N. and LeFaiv re, M. (1984) A class analysis of gentrification, in: J. J. Palen and B. London (Eds) Gentrification, Displacement, and Neighborhood Revitalization, pp. 43–63. Albany, NY: SUNY Press. Trudeau, D. (2006) The persistence of segrega- tion in Buffalo, New York: Comer vs. Cisneros and geog raphies of relocation decisions among low-income black households, Urban Geography, 27(1), pp. 20–44. Turney, K., Clampet-Lundquist, S., Edin, K. et al. (2006) Neighborhood effects on barriers to employment: results from a randomized housing mobility experiment in Baltimore, Brookings-Wharton Papers on Urban Affairs, 2006, pp. 137–187. US GAO (US General Accounting Office) (2003) Public housing: HOPE VI resident issues and 1604 EDWARD GOETZ changes in neighborhoods surrounding grant sites. GAO-04-109, Report to the Ranking Minority Member, Subcommittee on Housing and Transportation, Committee on Banking, Housing, and Urban Affairs, US Senate. US General Accounting Office, Washington, DC. US HUD (Department of Housing and Urban Development) (2004) Evaluation of the Welfare to Work Voucher Program: Report to Congress. Washington, DC: US Government Printing Office. Varady, D. P. and Walker, C. C. (2000) Vouchering out distressed subsidized developments: does moving lead to improvements in housing and neighborhood conditions?, Housing Policy Debate, 11(1), pp. 115–162. Varady, D. P. and Walker, C. C. (2003) Housing vouchers and residential mobility, Journal of Planning Literature, 18(1), pp. 17–30. Varady, D. P. and Walker, C. C. (2003) Housing vouchers and residential mobility, Journal of Planning Literature, 18(1), pp. 17–30. Venkatesh, S. (2000) American Project: The Rise and Fall of a Modern Ghetto. Cambridge, MA: Harvard University Press. Vigdor, J. L. (2007) The Katrina effect: was there a bright side to the evacuation of Greater New Orleans?, The B.E. Journal of Economic Analysis and Policy, 7(1), article 64. http://www.bepress. com/bejeap/vol7/iss1/art64. Wilen, W. P. and Nayak, R. D. (2006) Relocated public housing residents have little hope of returning: work requirements and mixed- income public housing development, in: L. Bennett, J. L. Smith and P. A. Wright (Eds) Where are Poor People to Live? Transforming Public Housing Communities, pp. 41–67. Armonk, NY: M. E. Sharpe. Wyly, E. and Hammel, D. (1999) Islands of decay in seas of renewal: housing policy and the resurgence of gentrification, Housing Policy Debate, 10(4), pp. 711–771. Wyly, E. and Hammel, D. (2000) Capital’s metropo- lis: Chicago and the transformation of American housing policy, Geografiska Annaler, Series B, 82(4), pp. 181–206. Wyly, E. and Hammel, D. (2004) Gentrification, segregation, and discrimination in the American urban system, Environment and Planning A, 36, pp. 1215–1241. Zielenbach, S. (2002) The economic impact of HOPE VI on neighborhoods. Housing Research Foundation, Washington, DC.