April 9, 2019

More Thoughts on Defining Gentrification

Share Posted By:
A neighborhood at risk of gentrification

Key Points At a Glance:

  • Research on gentrification is highly subjective and dependent on the data and methods used
  • Even when conceptions of gentrification are the same, different criteria and variables can produce very different findings about where and how often it occurs
  • Ramifications of these differences for measuring associated impacts of gentrification can be significant
  • Policy responses derived from subjective measures can have unintended consequences

As described in my paper, Gentrification: Framing Our Perceptions, the ways researchers measure neighborhood change influences their findings about which places experience the most or the least gentrification. This applies to recent analyses making headline-grabbing claims about the intensity of gentrification in Washington, D.C., and other places around the country, which are contingent on the specific variables and criteria used to make such a determination.

Neither right nor wrong, these findings are simply the product of their subjectively chosen inputs, and should be viewed as such, rather than taken as gospel truths about the extent and concentration of gentrifying neighborhoods.

New Report Suggests Gentrification Clustered in Large Cities

A report released last month by the National Community Reinvestment Coalition (NCRC) is the latest to wade into the gentrification definition debate. The authors of the report analyzed tract-level data from the 2000 Longitudinal Tract Database (LTDB) and the 2009-2013 American Community Survey (ACS) to evaluate changes in neighborhood characteristics.

Specifically, they looked at changes in household incomes, house prices and the share of residents with college degrees to identify 1,049 tracts hat they interpret as suggestive of gentrification. While these tracts represent only 9 percent of eligible tracts nationwide, that share reached as high as 40 percent in Washington, D.C., and over 20 percent in five other metro areas. Furthermore, the authors find that just seven large metros areas (out of over 900 studied) accounted for nearly half of all gentrified tracts, while more than three-quarters of all metro areas had no evidence of gentrification over their study period.

These findings offer one view on where and how much gentrification occurred over the first decade of the 2000s. What if, however, they authors had made a different choice about their definitions and criteria for gentrified neighborhoods; would their findings be different? One way to assess the importance of their definition is to compare their counts of gentrified tracts to another measure applied over the same period.

How Do These Findings Compare to Other Measures of Gentrification?

In developing their method, the NCRC report authors refer to a 2005 study by Lance Freeman, which used similar variables to measure the presence of gentrification at the tract level in the 1980s and 1990s. Table 1 contrasts the Freeman approach with NCRC's, showing how each defines tracts as eligible to gentrify (e.g. low-income and disinvested enough at start of decade) and the subset of those that exhibit enough change by end of decade to qualify as having gentrified.

Table 1: Comparison of Gentrification Measures

 Definition Freeman NCRC

Types of tracts evaluated

Central city tracts in metropolitan areas Tracts in metro areas with a population of at least 500

Eligible to gentrify (required tract conditions at start of decade)

Median household income and share of housing built in prior 20 years in bottom 50 percent of tracts in metro area Median household income and median house value in bottom 40 percent of tracts in metro area

Gentrified (required change in eligible tract conditions between start and end of decade)

Change in share of residents with college degree in top 50 percent of tracts in metro area, and any real increase in tract median house price Change in share of residents with college degree AND change in real house prices both in top 40 percent of tracts in metro areas, and any real increase in tract median household income

Conceptually, the Freeman and NCRC approaches are similar; both look at incomes and housing market conditions to define eligibility for gentrification, and changes in house prices and share of residents with college degrees to identify gentrified tracts, relative to conditions in the metro area. Freeman, however, uses slightly less stringent thresholds in both sets of criteria (i.e. 50 percent of tracts in metro versus 40 percent in NCRC’s approach). He also requires tracts to be in the central city of the metro area, uses a different variable to capture housing conditions at the start of the decade, and does not consider changes in income over the decade to define gentrification.

So what happens when these two measures are applied to data from the same time period? To compare findings from the NCRC report to those based on Freeman’s method requires extending Freeman’s analysis to tract data from the 2000-2010 Decennial Census. While not exactly the same as the time frame and data source used by NCRC, this application of Freeman’s method offers the closest comparison of the two approaches.

Table 2: Comparison of Findings

Definition Freeman NCRC
Data source and timeframe 2000-2010 Decennial Census 2000 LTDB and 2009-13 ACS
Metro area definition and vintage used Consolidated Metropolitan Statistical Areas (CMSAs) as of 1999 Core-Based Statistical Areas (CBSAs) as of 2010
Total tracts evaluated 73,057 NA
Tracts in metro areas 58,388 67,153
Tracts eligible to gentrify 12,515 11,196
Gentrified tracts 4,858 1,049

Table 2 shows the side-by-side comparison of tract counts based on the two measures. From the start, differences in the methods used by these two approaches are evident; Freeman evaluates tracts using metro-area status at start of the decade, which for 2000 means starting with the 323 consolidated metropolitan statistical areas (CMSAs) as defined in 1999. NCRC, meanwhile, uses the more recent and expansive definitions of core-based statistical areas (CBSAs), of which there were 935 nationwide in 2010. The result is almost 9,000 more tracts in NCRC’s analysis than in Freeman’s. The number of tracts deemed eligible to gentrify, however, is larger under the Freeman definitions, which is expected given the more restrictive thresholds for identifying areas with low enough incomes and house prices in the NCRC analysis. Finally, the number of gentrified tracts is more than four-times greater using the looser Freeman approach to measuring change observed over the decade. Thus, even with similar conceptions of what types of conditions constitute gentrification, these two approaches yield markedly different results.

Which Metros are Gentrifying? It Depends…

The NCRC report makes much of the concentration of gentrified tracts (based on its definitions) in a handful of large metro areas. Indeed, the two metro areas with the most gentrified tracts were also the two with the most tracts overall – New York and Los Angeles – though third-ranked Washington DC has the fourth-most tracts overall, while the third-largest metro (Chicago) is seventh by number of gentrified tracts. When ranking metros using Freeman’s definition of gentrification, however, Chicago appears in second place (after New York), and Washington DC in sixth.

The NCRC report’s other metric for gentrification “intensity”, or the share of eligible tracts that gentrified, ranked Washington, D.C. as having the most intense gentrification (40 percent), followed by San Diego (29 percent), New York (24 percent), Albuquerque (23 percent), Atlanta and Baltimore (22 percent each). Applying the same calculation to metros using Freeman’s definition not only finds much higher shares overall (given the larger number of gentrified tracts identified), but also a much different ranking (Table 3).

Table 3: Top 10 Metros by Gentrification “Intensity”

NCRC Freeman
Metro % Gentrified Metro % Gentrified
Washington, D.C. 40.3 Portland 70.3
San Diego 29.0 Sacramento 68.3
New York 24.4 Washington, D.C. 65.2
Albuquerque 22.6 Atlanta 61.6
Baltimore 22.2 Seattle 60.8
Atlanta 22.4 Fort Lauderdale 57.1
Portland 20.4 Denver 56.9
Seattle 20.0 New Orleans 56.6
Pittsburgh 20.3 St. Louis 53.7
Philadelphia 17.2 San Diego 53.3
Note: Only metros with at least 200 total tracts were used in this ranking.

Clearly, the choice of measure makes a big difference in how many and which places appear as gentrified. Indeed, other research has shown that within a metro area, different approaches can produce similar numbers of gentrified neighborhoods, but still differ on the specific areas that are gentrifying.

Why This Matters

The differences in findings based on the definition of gentrification are important, as where gentrification is found determines which areas are evaluated for measuring the consequences of neighborhood change. The NCRC report, for example, looks at changes in the racial/ethnic composition in gentrified neighborhoods, concluding that in just under a quarter of them displacement (which is also subjectively based on their choice of variables and criteria) of Black and/or Hispanic residents occurred. Had the authors of the report used a different method to measure gentrification, it is likely they would have identified different neighborhoods, and potentially found different patterns of displacement within them. The same is true of observed changes in crime, educational outcomes, and community cohesion linked to purported gentrifying places – it all depends on how you measure and classify different types of neighborhood change.

Different definitions of gentrification also have implications for policy, of which government officials should be aware. Reliance on one report or approach to identify places that have gentrified risks drawing conclusions about the best ways to mitigate the assumed negative impacts of changing conditions on residents that may be ineffective or even counterproductive. For example, restrictions or regulations that inhibit new housing development may encourage preservation of historic structures and community character, but will also discourage investment in blighted neighborhoods. Moreover, with fewer new units coming on line, demand for existing housing will rise and add to affordability and displacement concerns, rather than alleviate them.

In Gentrification: Framing Our Perceptions, I listed three guiding principles for policymakers to use when addressing the causes and consequences of neighborhood change that do not rely on subjective assessments of where and why gentrification occurs. These include:

  • Respond to local concerns with local solutions: one-size-fits-all strategies not only misappropriate resources, but may also overlook opportunities to address multiple aspects of community change, such as housing development, transportation, commercial/retail investment, and community safety. Engaging residents in this process also ensures better outcomes tailored to specific community needs.
  • Proactively address housing and community needs: data on gentrification is necessarily back-ward looking, which limits ability of policymakers to respond. By anticipating changes and preparing solutions that can be rolled out quickly and nimbly, officials are better able to head off the conditions that may precipitate gentrification, including neighborhood decline.
  • Promote affordability and opportunity in all communities: one lesson learned from most studies of gentrification is that the majority of neighborhoods deemed low-income and disinvested enough to potentially gentrify do not experience upgrading and improvement in their economic conditions, but remain mired in persistent disadvantage. By tackling needs in these areas, officials can ensure a wider range of affordable and desirable communities exist across their cities. They should also promote affordable options in existing high-opportunity neighborhoods, meet some of the demand that drives middle-income households to low-income communities in the first place.

Finally, it is incumbent on producers and consumers of housing research to be more precise in how we describe neighborhood change, rather than defaulting to the lazy language of gentrification to describe any form of neighborhood improvement we observe. Doing so will not only clarify the myriad ways communities evolve over time, but will prescribe better policy responses to those changes and ensure better outcomes for current and new residents.

Want to know more about how definitions of gentrification impact policy and perceptions of neighborhood change? Read Gentrification: Framing Our Perceptions or watch the short video below with a summary of the report. Additional research on this topic from Enterprise is also in process, so be sure to check back on our blog for the next release announcement.

Posted in:
Opp360 logo

For full access to our tools and resources, please provide the information below.

We use this data to better understand our users; we do not sell or share this data. By providing this information, you can expect to receive newsletters and other updates from Opportunity360.