The Gentrification Puzzle
Gentrification is a highly charged issue, to say the least. Perhaps that's why a new analysis by Daniel Hartley, a research economist at the Cleveland Federal Reserve, has generated so much attention.
Hartley’s study used Census data to examine the extent of gentrification across America's 55 largest cities over the past decade. (His data track the change between the 2000 Census and the results of the 2005-9 American Community Survey, which he shorthands as 2007). He defines gentrification as a neighborhood (more precisely, a Census Tract) that moved from "the bottom half of the distribution of home prices in the metropolitan area to the top half between 2000 and 2007."
The map below (by my colleague Zara Matheson of the Martin Prosperity Institute) charts these changes based on Hartley's key measures. The pink bar shows the number of tracts with below-median housing prices in 2000. The blue bar shows what percent of those census tracts gentrified.
For all the talk of rampant gentrification, substantial levels are found most in the likely suspects – Boston, Seattle, New York, San Francisco, and Washington, D.C. Most cities experienced much more modest levels. In nearly three quarters of cities, less than ten percent of all neighborhoods experienced gentrification. And in 22 cities – 40 percent of the sample, including San Diego, Charlotte, Buffalo, Pittsburgh and Detroit – gentrification affected 5 or less percent of all neighborhoods.
Boston was one of the major exceptions, where 61 percent of low-price neighborhoods gentrified, encompassing the tracts where about a quarter of the city’s population live. Other significantly gentrified metros were Seattle (55 percent of low-price tracts), New York (46 percent), San Francisco (42 percent), and Washington, D.C. (35 percent).
• • • • •
A common assumption is that gentrification hurts the residents of poorer neighborhoods who are displaced from where they live. But Hartley found that not to be so. His research compared the credit scores (based on the Equifax Risk Scores) of people in gentrifying neighborhoods versus non-gentrifying neighborhoods between 2001 and 2007.
He found that living in a gentrifying neighborhood was “associated with about an 8 point higher increase in credit score compared to living in a low-price neighborhood that did not gentrify." When he parsed the data to account for people who moved out of gentrifying neighborhoods, he found that those who moved away – the theoretically "displaced" – in fact had “a larger increase in credit score (1.5 points more)" than those who stayed put.
Hartley’s findings, as counterintuitive as they may seem, are actually in sync with several other major studies. Looking at the 1990 and 2010 Censuses, Terra McKinnish and her co-authors found that neighborhood changes in predominantly black neighborhoods generally made these areas more attractive for middle-class black households. In his 2011 book There Goes the 'Hood, Lance Freeman of Columbia University found that the way we talk about gentrification "has tended to overlook the possibility that some of the neighborhood changes associated with gentrification might be appreciated by the prior residents."
• • • • •
Hartley's study points to the substantial variation in gentrification across cities. But what accounts for these differences?
My research team and I used Hartley's data to explore the factors that might be associated with gentrification rates. My MPI colleague Charlotta Mellander ran a basic correlation analysis across Hartley's metric, assessing key factors like income, race, density, industry, occupational mix and the like.
We compared Hartley's gentrification measures to economic and demographic characteristics of entire cities, under the assumption that, though gentrification takes place largely in particular neighborhoods, it is shaped by factors that act at the metro level. As usual, I point out that these correlations point only to associations between variables and in no way imply causality. Still, the results are interesting and begin to create a better understanding of the kinds of cities and regions that gentrify and those that don’t.
Gentrification, not surprisingly, is the province of wealthier, more economically advantaged metros. It is correlated with both per capita income levels (.61) and wage levels (.62). Also not surprisingly, gentrification is associated with higher levels of college graduates (.51).
It’s become cliché to say that gentrification is initiated by artists and gays. And indeed Mellander finds close associations between the levels of each of these groups and the number of neighborhoods that have gentrified. Gentrification is correlated with both the concentration of artists, musicians, designers and other cultural creative (.55), and with the concentration of gays and lesbians (.47).
Working class regions were far less likely to gentrify. Gentrification is negatively correlated with the share of blue-collar workers in an area (-.48).
Bigger and denser regions are also more likely to gentrify. Gentrification is associated with both population size (.59) and density (.44). It makes sense that larger, denser metros would gentrify more, given the real opportunity costs of getting around and the critical mass of people looking to move into a convenient part of town.
It also turns on transit infrastructure. Gentrification is much more likely in metros where people use public transportation to get to work (.61). The share of commuters who walk (.36), or bike (.35) to their jobs is also positively correlated. Gentrification is negatively associated with the share of commuters who drive alone to work (-.55).
I’ve written a great deal about the “urban shift” in high-tech industry from traditional suburban “nerdistans” to denser urban locations. Mellander’s findings are in line with this shift. Gentrification is positively associated with the share of science and technology workers (.32) and, even more strongly, the concentration of high-tech industry in a metro (.58).
Surprisingly, we found mixed results when it comes to the cost of housing. Gentrification is positively associated with the median price of housing (.40), median rent (.47), and the monthly costs of housing (.53) at the metro level. But we found no association between gentrification and the share of income devoted to rent, and a negative correlation (-.37) between it and the share of housing costs as a function of median income. This likely reflects the greater affluence and higher incomes of gentrifying metros. Of course this metro-level analysis does not take into account the housing prices or costs of the residents of specific gentrifying neighborhoods, nor is it able to track what happens to those who are displaced.
Interestingly, Mellander found no relationship between gentrification and race at the metro level – measured variously as percent Black, percent Latino or percent White. This likely reflects the scale of analysis. It is likely that race plays a more significant role at the neighborhood level.
Gentrification is frequently seen to go hand and hand with inequality. Here Mellander’s findings are mixed. Gentrification is modestly correlated to wage inequality (with correlations of roughly .33 to both measures of gentrification) but not at all to a broader measure income inequality (based on the Gini coefficient).
• • • • •
A rather clear portrait of a gentrifying city emerges from this analysis: a large, dense, affluent, high human capital, knowledge and innovation-based metro that is well served by transit. Conversely, gentrification is much less likely in smaller, more sprawling metros, those with older industrial and blue-collar economic structures, and where a higher share of the population commutes to work by car. Gentrification thus appears to be connected to the large-scale divides that separate growing from declining cities and metros more broadly.
Gentrification remains one of the most pressing and contentious issues facing cities today. As many of our cities and once-troubled urban neighborhoods revive, urban class divides worsen, displacement occurs, and inequality grows. Hartley's study contributes important data to our grasp of the phenomenon. But more than any other issue, the extent and impacts will require more careful empirical research and analysis like his.