Earlier this week, The New York Times published an op-ed from Jed Kolko–”Seattle Climbs but Austin Sprawls, The Myth of the Return to Cities”–offering up another iteration of his long running argument that the urban rebound is overstated. His key point: in the aggregate suburbs are growing faster than cities, and that the density of most metropolitan areas continues to decrease.
The article is surely refuting something of a straw man: not everyone is moving back to cities everywhere. And some big metro areas in the sunbelt continue to sprawl rapidly. But even if they aren’t absolutely outperforming suburbs everywhere, cities are relatively much stronger than they were a decade (and especially several decades ago), and are disproportionately attracting the best and the brightest. The chief reason they’re not doing even better–as Kolko concedes–is that we’re not building housing in city centers fast enough.
Kolko’s indicated that the move back to city centers is real:
That revival is real, but it has mostly been for rich, educated people in particular hyperurban neighborhoods rather than a broad-based return to city living. To be sure, college-educated millennials — at least those without school-age kids — took to the city, and better-paying jobs have shifted there, too. But other groups — older adults, families with kids in school, and people of all ages with lower incomes — either can’t afford or don’t want an urban address.
And coincidentally, at the same time The Wall Street Journal summarized a new report from BuildZoom’s Issi Ronnem which emphasized the growing demand for housing in cities. Ronnem’s work shows that home sales have accelerated closer to the urban core in most cities, and remain depressed in the suburbs. The takeaway:
While new home sales within 5 miles of the centers of 10 of the country’s priciest and most densely populated metropolitan areas have surpassed levels from 2000, they remain more than 50% below where they were in 2000 when you go more than 10 miles out.
As Justin Fox pointed out at Bloomberg, the we’re bumping up against the limits of supply in cities. More and more Americans, especially young adults want to live in cities, but the supply of housing there is constrained, and growing only slowly–which is why rents have spiked in cities. And while more households may be locating in less dense neighborhoods, the price of housing in suburbs has fallen relative to that in city centers.
We’ll weigh in soon with our own City Observatory analysis of these data, but in the mean time, we think that several of the points that we raised a few weeks back about Kolko’s last round of research are worth remembering. There are several key reasons why no one should be glum about the continuing prospects for urban growth and revitalization.
First, employment and population data aggregated nationally by county don’t reveal much about the most urban neighborhoods. We’ve made this point before with Kolko’s analysis of job growth. County boundaries correspond poorly, if at all, with cities or urban density.
Consider Chicago, which gets tarred with the “Biggest Loser” title in earlier Kolko’s tabulations. While Cook County saw an overall decline in population, the most urban parts of the city, including its Downtown Loop and Northside neighborhoods, continued to rack up big gains.
A more finely grained geographic analysis shows that the closer you get to the city center in most metros, the stronger has been the performance. While its true that the more outlying parts of some cities are losing population, their cores are becoming increasingly vibrant. As we’ve noted, that notion of critical mass at the neighborhood level is one of the defining characteristics of urban growth.
Second, national aggregations conceal local patterns. To be sure, these data show a renewed growth in the sunbelt. And in the biggest sunbelt metros, a big chunk of the growth is in their sprawling suburbs (Houston, Dallas). But the growth in these metros is not representative of what’s happening in many other places.
Third, there’s a baseline issue here. City growth has decelerated from the past year or two. But city growth this decade looks far different than it did a decade ago. While Kolko’s FiveThirtyEight.com post just shows the change in city and suburb growth rates over the past few years (and emphasizes the one-year change between 2015 and 2016), his longer blog post on his own website shows the change in population by type of county since 2001. Taking this longer view, its apparent that growth rates in suburbs have declined sharply since the last decade, while growth rates in urban counties were up.
Between 2001 and 2005, low density suburban counties (gray line) were growing more than 2 percent per year–their growth rates since 2011 are about half that. During that same time, the most urban counties (blue line) were growing sluggishly or losing population–and they’re now growing. While urban counties dramatically underperformed suburban ones last decade, the growth patterns have converged dramatically since then. As a result, when judged against the baseline of the previous decade, more urban counties have accelerated, while more suburban ones have faded.
Fourth, in many places we’re bumping up against the (policy-induced) limits to meeting the market demand for urban living. In the early stages of growth, cities can add population by filling vacant housing. But as vacancy rates fall, the only way to accommodate more people is to add more housing, which is a process that (a) takes time, and (b) is too often unfortunately encumbered by NIMBY building restrictions. That suburbs might grow faster than cities for a time is not necessarily evidence of a declining demand for city living, so much as it is evidence that we haven’t expanded the supply of urban living opportunities to meet that demand.
More debate on city revival
Is the urban renaissance over?
Earlier this week, The New York Times published an op-ed from Jed Kolko–”Seattle Climbs but Austin Sprawls, The Myth of the Return to Cities”–offering up another iteration of his long running argument that the urban rebound is overstated. His key point: in the aggregate suburbs are growing faster than cities, and that the density of most metropolitan areas continues to decrease.
The article is surely refuting something of a straw man: not everyone is moving back to cities everywhere. And some big metro areas in the sunbelt continue to sprawl rapidly. But even if they aren’t absolutely outperforming suburbs everywhere, cities are relatively much stronger than they were a decade (and especially several decades ago), and are disproportionately attracting the best and the brightest. The chief reason they’re not doing even better–as Kolko concedes–is that we’re not building housing in city centers fast enough.
Kolko’s indicated that the move back to city centers is real:
And coincidentally, at the same time The Wall Street Journal summarized a new report from BuildZoom’s Issi Ronnem which emphasized the growing demand for housing in cities. Ronnem’s work shows that home sales have accelerated closer to the urban core in most cities, and remain depressed in the suburbs. The takeaway:
As Justin Fox pointed out at Bloomberg, the we’re bumping up against the limits of supply in cities. More and more Americans, especially young adults want to live in cities, but the supply of housing there is constrained, and growing only slowly–which is why rents have spiked in cities. And while more households may be locating in less dense neighborhoods, the price of housing in suburbs has fallen relative to that in city centers.
We’ll weigh in soon with our own City Observatory analysis of these data, but in the mean time, we think that several of the points that we raised a few weeks back about Kolko’s last round of research are worth remembering. There are several key reasons why no one should be glum about the continuing prospects for urban growth and revitalization.
First, employment and population data aggregated nationally by county don’t reveal much about the most urban neighborhoods. We’ve made this point before with Kolko’s analysis of job growth. County boundaries correspond poorly, if at all, with cities or urban density.
Consider Chicago, which gets tarred with the “Biggest Loser” title in earlier Kolko’s tabulations. While Cook County saw an overall decline in population, the most urban parts of the city, including its Downtown Loop and Northside neighborhoods, continued to rack up big gains.
A more finely grained geographic analysis shows that the closer you get to the city center in most metros, the stronger has been the performance. While its true that the more outlying parts of some cities are losing population, their cores are becoming increasingly vibrant. As we’ve noted, that notion of critical mass at the neighborhood level is one of the defining characteristics of urban growth.
Second, national aggregations conceal local patterns. To be sure, these data show a renewed growth in the sunbelt. And in the biggest sunbelt metros, a big chunk of the growth is in their sprawling suburbs (Houston, Dallas). But the growth in these metros is not representative of what’s happening in many other places.
Third, there’s a baseline issue here. City growth has decelerated from the past year or two. But city growth this decade looks far different than it did a decade ago. While Kolko’s FiveThirtyEight.com post just shows the change in city and suburb growth rates over the past few years (and emphasizes the one-year change between 2015 and 2016), his longer blog post on his own website shows the change in population by type of county since 2001. Taking this longer view, its apparent that growth rates in suburbs have declined sharply since the last decade, while growth rates in urban counties were up.
Between 2001 and 2005, low density suburban counties (gray line) were growing more than 2 percent per year–their growth rates since 2011 are about half that. During that same time, the most urban counties (blue line) were growing sluggishly or losing population–and they’re now growing. While urban counties dramatically underperformed suburban ones last decade, the growth patterns have converged dramatically since then. As a result, when judged against the baseline of the previous decade, more urban counties have accelerated, while more suburban ones have faded.
Fourth, in many places we’re bumping up against the (policy-induced) limits to meeting the market demand for urban living. In the early stages of growth, cities can add population by filling vacant housing. But as vacancy rates fall, the only way to accommodate more people is to add more housing, which is a process that (a) takes time, and (b) is too often unfortunately encumbered by NIMBY building restrictions. That suburbs might grow faster than cities for a time is not necessarily evidence of a declining demand for city living, so much as it is evidence that we haven’t expanded the supply of urban living opportunities to meet that demand.
Related Commentary