It’s telling that Governing calls gentrification the “g-word”—it’s become almost impossible to talk about neighborhood revitalization without objections being raised almost any change amounts to gentrification. While we applaud the attempt to inject some rigor and precision into a debate that has been too often fueled by emotion and anecdote, Governing’s analysis serves only to muddy the waters of this contentious issue.

The Governing team explains that there is no agreed-upon definition for gentrification, and then go on to choose a definition and use it to measure number of neighborhoods that have, and haven’t, gentrified. The report notes that gentrification is not the same thing as displacement; but then repeatedly describes the harm of gentrification as being the displacement of the existing population.

Is Gentrification About Displacement? Or isn’t it?

The underlying problem confronting Governing’s analysis is the confusion of the terms “gentrification” and “displacement”. The Governing team begins by explaining that their definition of gentrification has nothing to do with displacement, but they then go on to detail how “gentrification signifies displacement of the poor, mostly people of color.”

The reason policy analysts and public officials are concerned about gentrification, to the extent it happens, is because it holds the potential to displace the poor from their longtime neighborhoods. As Governing acknowledges, the original definition by British sociologist Ruth Glass was that the middle class “invade” a neighborhood “until all or most of the working class occupiers are displaced,” and the social character of the neighborhood is changed.

Our own work has shown that over four decades, relatively few high-poverty neighborhoods have seen their poverty rates decline to below the national average. It’s also the case that far more of the long-suffering poor move out of high-poverty neighborhoods that stay poor than move away from high-poverty neighborhoods that see a significant reduction in poverty.

A striking omission from the Governing article is any more than a passing mention of the robust academic literature on the extent of displacement in gentrifying neighborhoods. Columbia University’s Lance Freeman reports that outmigration rates for low-skilled black residents of gentrifying neighborhoods are lower than in otherwise similar, non-gentrifying neighborhoods. The University of Colorado’s Terra McKinnish and Kirk White conclude the demographic flows associated with the gentrification of urban neighborhoods are not consistent with displacement or harm to minority households. New York University’s Ingrid Gould Ellen and Katherine O’Reagan write “. . . original residents are much less harmed than is typically assumed. They do not appear to be displaced in the course of change, they experience modest gains in income during the process, and they are more satisfied with their neighborhoods in the wake of the change.”

It’s also discouraging that this entire discussion of poor neighborhoods isn’t placed in a broader context of income segregation of the U.S. population. Income inequality has achieved a new level of visibility in public discussions in the past few years, thanks to the work of Thomas Piketty and others. A number of scholars—Brown University’s John Logan, Rutger’s Paul Jargosky, the University of Missouri’s Todd Swanstrom, and others—have carefully traced out how income inequality has played out in the form of greater spatial separation by income in the nation’s metropolitan areas. The latest research from Stanford’s Sean Reardon and his colleagues shows that income segregation is increasing, driven by the increasing secession of the rich from neighborhoods of lower- and middle-income households. The focus on gentrification—the very limited and small-scale movement of some higher-income and better-educated households into lower-income communities–completely misses the fact that income segregation is being driven by the decisions of higher-income families to increasingly isolate themselves in higher-income enclaves, often in exclusive suburbs and established high-income areas.

One of the ironies of the fear of gentrification is that is often used as excuse to steer limited dollars for assisted housing predominantly or exclusively into low-income neighborhoods, a practice the University of Minnesota’s Myron Orfield has shown to reinforce historical patterns of racial and economic segregation.

Levels vs. Change

In quantitative analysis, we can measure the level of something (whether incomes are high or low, for example). We can also measure the “change” in something—whether incomes are increasing or decreasing.

The limitation of the Governing definition of gentrification is that it is all about change, and it largely ignores levels. Even if education levels or incomes increase in a poor neighborhood, it still may be much less well-educated and lower-income than the average neighborhood in the metropolitan area of which it is a part. This absence of clearly defined thresholds is a perennial omission of gentrification discussions: if one wealthier, whiter, better-educated person moves into a neighborhood, does that constitute gentrification?

As the University of Pennsylvania’s Mark Stern and Susan Seifert note:

Clearly, there is no objective measure of when neighborhood improvement—or, in Jane Jacobs’ striking phrase, ‘unslumming’—becomes gentrification. But if we see neighborhood revitalization as desirable, we cannot afford to label all population change as gentrification. (2007)

There’s something odd about a measurement that purports to examine the extent of “gentrification,” but excludes from its analysis all city neighborhoods with income levels over 40 percent of the metropolitan average. In San Francisco, a city with 196 census tracts, Governing concluded that only 16 were “eligible” to gentrify, and that only 3 tracts did gentrify. Superficially, this creates the impression that gentrification affects only 3 neighborhoods in San Francisco, as opposed to say, 84 neighborhoods in Philadelphia, and 39 neighborhoods in Baltimore. But is it meaningful to conclude that gentrification, income disparities, and displacement are a tenth as widespread in San Francisco as these other cities, simply because San Francisco already had most of its neighborhoods dominated by “the gentry?” Should we put any stock in a measure that says gentrification is more prevalent in Detroit (7 neighborhoods), Cleveland (10) and Fresno (5), than it is in San Francisco (3 neighborhoods)?

The levels of income and education are still very low, and the levels of poverty still very high, in many of the neighborhoods that Governing says have “gentrified.” Can it be said that a neighborhood has gentrified if it has, by comparison to the rest of its metropolitan area, a higher fraction of low-income households than the rest of the metropolitan area? Is this really a helpful way to describe what’s happening in metropolitan areas?

No Apparent Displacement in “Gentrifying” Tracts

The identifiable harm of gentrification is displacement: if Governing has identified tracts that are gentrifying, and if gentrification is a problem, then we should be able to find evidence of displacement. Or, put another way, if neighborhoods are gentrifying and displacement isn’t happening, is it a serious problem?

Let’s compare what happened to gentrifying and non-gentrifying tracts, according to Governing’s definition. Of a total of 4,750 central city census tracts “eligible” to gentrify, 948 gentrified, but 3,802 did not. The tracts that did not gentrify lost 2.4 percent of their population in aggregate. The tracts that did gentrify actually saw their population increase by 6.7%. (This is consistent with the “up or out” pattern we identified in high-poverty tracts nationally: either poverty rates decline and population increases, or high poverty rates persist and population declines).

One recurring theme in overly simplistic gentrification analyses is housing as a zero sum game. If one whiter, wealthier, or better-educated person moves into a neighborhood, that must necessarily mean that one poorer, less-educated person of color must move out. Both Governing’s report and our analysis of high poverty neighborhoods shows that gentrifying or rebounding neighborhoods are actually seeing population increases.

It’s also the case that more poor people live in the Governing’s “gentrified” tracts today than in 2000: the poverty rate in gentrified tracts declined by 0.7%, while the total population increased by 6.5%. Assuming the poverty rate in these tracts exceeded 13 percent in 2000, the population living below the poverty line in these tracts had to have actually increased between 2000 and 2013; hardly evidence, on its face, of widespread displacement.

Another recurring theme in this kind of analysis is the implication that if a neighborhood doesn’t gentrify, that it somehow stays the same. But our analysis of poor neighborhoods showed that places of high poverty aren’t stable; if these neighborhoods don’t see a reduction in poverty, people leave: on average high-poverty neighborhoods that didn’t rebound lost 40 percent of their population over four decades. The same pattern holds for Governing’s “non-gentrifying” neighborhoods—they lost 2.4 percent of their population. This contrasts with the gentrifying places, which were adding population.

Understanding the Nuance of Neighborhood Change

Rather than just a binary classification of neighborhoods as either “gentrified” and “not gentrified,” it’s worth looking at the actual numbers of people involved, poor and non-poor, to get a sense of what’s really happening and what it means.

At City Observatory, we frequently work with tract-level data, and closely follow developments in Portland, where we’re based. We took a close look at Governing’s data for Portland, in part because they concluded that 58% of “eligible” census tracts gentrified in the past decade.

Here are their headline findings for Portland: According to Governing, 62 of 143 census tracts in the City of Portland were “eligible” to gentrify. Of these, 36 gentrified, and 26 did not, according to their calculations.

We gathered the data on population and poverty for all the City of Portland census tracts for 2000 and 2009-13, and categorized them according to the Governing methodology (we noted the classifications as coded on Governing’s map of Portland tracts). We have one more census tract in the City of Portland (in the not eligible category). We ignore this minor difference for the purpose of our analysis. Here’s what the data show about the gentrifying neighborhoods in Portland in the aggregate:

The total population in “gentrifying” neighborhoods actually increased from about 152,000 to about 165,000. This is consistent with our observation that neighborhood change is not a zero sum game. Neighborhoods can gain new residents without necessarily losing existing ones.

The number of poor persons living in “gentrifying” neighborhoods increased. In 2000, there were 25,037 persons in poverty in these neighborhoods; in 2013 there were 34,499 persons living in poverty, about 9,500 more. Even allowing for the increase in poverty rates in the nation and the region over the decade, it’s hard to argue that there was widespread displacement of the poor from these “gentrifying” neighborhoods if they now have 9,500 more poor residents.

As a result, the poverty rate in “gentrifying” neighborhoods in Portland increased between 2000 and 2013 in the aggregate. In 2000 the poverty rate was 16 percent; in 2013, the poverty rate had risen to 20 percent. Even after “gentrifying,” these neighborhoods had a poverty rate that was higher than the regional average of about 13.5 percent.

Meanwhile, the number of non-poor residents also increased by a net of 3,800, from about 127,000 to about 130,000. The “gentrifying” neighborhoods gained about two-and-a half times more net additional poor residents than net additional non-poor residents. Spread over 36 Census Tracts and about 11 years, (2000 through 2011, the mid-point of the 2009-13 five year census pattern) this works out to a net increase of about 10 non-poor persons per “gentrifying” census tract per year—hardly a sweeping change in typical neighborhood demographics.

Poverty, Population, City of Portland, 2000 and 2009-13

Organized by Governing Magazine Gentrification Typology

This is not to say that all of these neighborhoods experienced this same pattern. By our count, 22 of Governing’s 36 gentrifying Portland census tracts experienced increased numbers of persons living in poverty, and 14 saw decreases in their poverty population. Of these 14, nine had decreases of less than five percent since 2000, four had decreases of between five to ten percent, and only one census tract saw a decrease of more than ten percent in its population living in poverty. Arguably, these tracts may be experiencing a measure of displacement. But for reference, it’s worth noting that the between 1970 and 2010, the typical urban high-poverty tract that stayed high-poverty lost about 40 percent of its population. 31 of the 36 “gentrifying” tracts still had poverty rates above the regional average “post-gentrification.”

Defining the Undefinable

After musing about whether gentrification is synonymous with displacement, Governing concludes that we should treat them as two different things, and cites the Centers of Disease Control as agreeing with this view. The authors quote the CDC as saying: “gentrification is merely the transformation of neighborhoods from low value to high value.” The CDC definition, says Governing, has nothing to do with displacement.

It’s worth looking at the CDC’s work here. It turns out that the Centers for Disease Control hasn’t actually done its own research on gentrification. What you’ll find at the CDC website is one wiki-like page of secondary citations, compiled by an unidentified author: http://www.cdc.gov/healthyplaces/healthtopics/gentrification.htm. The CDC is hardly regarded as an expert or arbiter on the subject: Google Scholar reports a total of four citations to this CDC webpage.

It’s important to note that the CDC website—like the rest of the literature on gentrification—clearly flags the harm of gentrification as displacement of the existing population. They use the term “transformation” in their definition. If you read on in the sentences immediately following the definition cited by Governing, this is clear: “This change has the potential to cause displacement of long-time residents and businesses. Displacement happens when long-time or original neighborhood residents move from a gentrified area because of higher rents, mortgages, and property taxes.”

There’s one more wrinkle here. Notice that the CDC definition refers to the transformation of neighborhoods from “low value to high value.” The CDC definition, unlike Governing’s, is about levels, and not just changes. Governing’s definition is not that formerly poor neighborhoods become “high value” neighborhoods, but that they become “higher” value neighborhoods—that their property values increase faster than for the region as a whole. It is entirely possible for a low-value neighborhood to have a higher percentage increase in prices, and still be a low-value neighborhood.

Is there some reason we might want to be cautious about leaning on housing price data from the past decade or so?

A linchpin of the Governing analysis is its reliance on a decade’s worth of housing price data as reported by homeowners to the Census Bureau. They compare the reported value of owner-occupied housing in the 2000 Census, with data from the 2009-13 editions of the American Community Survey, and use this to identify lower income neighborhoods that have experienced higher than average rates of home price appreciation.

As we all recall, that decade represented the period of the biggest volatility in home prices and housing markets in at least eight decades. Home prices were inflated by a giant bubble through the mid-part of the decade, and collapsed in a downturn that wiped out trillions of dollars of household equity, produced millions of foreclosures, and resulted in 17 million more households renting their homes.

Another important development during the past decade was the big increase in gas prices in 2007-08, which persisted until just the past few months. Arguably, higher gas prices had a big impact on the housing market, depressing the price of suburban and exurban homes that were sold to what the real estate community called the “drive-til-you-qualify” crowd.

It’s obvious that this was a decade where a range of market forces were buffeting housing prices. It’s a stretch to assume that gentrification was the sole cause of higher prices in some previously poor neighborhoods.

There’s another deeper technical issue here: Governing’s estimates are based on median home values data that are subject to bias from composition effects. (For a discussion of the volatility and noise that median measures create for housing price inflation estimates, see Jordan Rappaport, “A guide to aggregate house price measures” Kansas City Federal Reserve Bank, 2007.)  The Census Bureau asks for estimates of home values only for owner-occupied homes—not rentals. As long as the number and mix of these homes doesn’t change from one census to another, this isn’t a problem for using the census data to estimate average prices. But the “median” home value is subject to composition effects: if a different number and mix of homes is in the sample in the base year and the final year, the quality of the estimate is compromised. If many formerly owner-occupied low-value homes are foreclosed upon, and converted to rentals, they are no longer included in the median. Because foreclosure was more common among low-value houses this effect drives up the reported median for the remaining higher-value houses. Foreclosure problems are far more common in lower-income neighborhoods than middle- and upper-income neighborhoods. It’s also the case that if new market rate housing gets built in a neighborhood, it is usually at a higher price point than existing housing; this too tends to cause measured median prices to rise, especially in neighborhoods with lots of older, smaller, lower-value homes.

What do we do?

As the Governing team well understands, the metropolitan U.S. is dramatically segregated by income—and income segregation is increasing. An abundance of social science literature shows that concentrated poverty magnifies all of the negative effects of growing up poor (Patrick Sharkey, Jargowsky & Swanstrom). Newly released research on intergenerational mobility shows the devastating effects of concentrated poverty, and that even poor kids that grow up in integrated places have much greater opportunities than their counterparts in neighborhoods of more concentrated poverty (Chetty, Massey and Rothwell).

We live in a nation increasingly segregated by income. Even as racial segregation has waned, income segregation has increased. If we’re serious in our rhetoric about equality of opportunity, we have to do something to tackle the growing spatial class segregation we see in cities.

As Daniel Kay Hertz has observed, critics of gentrification need to say how they expect to achieve a more integrated nation and more integrated cities if they are somehow opposed to some higher-income people moving into what have become lower-income neighborhoods. “The kind of cognitive dissonance that allows someone to decry segregation while they wish to “reverse” the process of integration makes it impossible to articulate a real vision for what a just city might look like. Those who would declare themselves firmly anti-gentrification need to grapple with whether they’re comfortable defending a racial geography born of discrimination and violence.”

Those who raise “gentrification” as an impending threat to American cities owe us a coherent vision of how we can create more just and equitable neighborhoods. Lamenting—and exaggerating—gentrification generates plenty of heat, but precious little light on how cities ought to respond to the twin challenges of income segregation and neighborhood change.