Self-driving cars versus pedestrians

For many, it’s all but a certainty that our world will soon be full of self-driving cars. While Google’s self-driving vehicles have an impressive safety record in their limited testing, it’s just a matter of time until one is involved in a serious crash that injures someone in a vehicle, or a pedestrian.

So, in a way, it’s good news that Google is devoting some of its considerable intellectual energy to figuring out ways that we might lessen the seriousness of pedestrian injuries in the event of such collisions. Earlier this month, Google unveiled plans for a novel plan to coat the exterior of self-driving cars with a special adhesive that would cause any pedestrians the vehicles struck to adhere to the car rather than being thrown by the impact.

Whether it would be better to find oneself stuck to the car that struck you, rather than being pushed aside, is far from clear. But pedestrian safety in a world of self-driving cars is clearly an issue that needs to be dealt with.

Here at City Observatory, we’ve come up with our own concepts for, if you will, lessening the impact of autonomous cars on pedestrians. In the interest of safety and advancing the state of the art, we’re putting our ideas into the public domain, and not patenting any of them.

Pedestrian Shock Bracelets. Most pedestrians are already instrumented, thanks to cell phones, and a large fraction of pedestrians have fit-bits, apple watches and other wearable, Internet-connected devices. We propose adding a small electroshock device to these wearables, and making it accessible to the telematics in autonomous vehicles. In the event that the autonomous vehicle’s computer detected likelihood of a car-pedestrian collision, it could activate the electroshock device to alert the pedestrian to, say, not step off the curb into the path of an oncoming vehicle.

google car diagrams-01

Personal airbags. Airbags are now a highly developed and well-understood technology. Most new cars have a suite of frontal impact, side curtain and auxiliary airbags to insulate vehicle passengers from collisions. The next frontier is to deploy this technology on people, with personal airbags. Personal airbags could have their own sensors, inflating automatically when the pedestrian was in imminent danger of being struck by a vehicle.


google car diagrams-02

Rocket Packs. While a sufficiently strong adhesive might keep a struck pedestrian from flying through an intersection and being further injured, perhaps a better solution would be to entirely avoid the collision in the first place by lifting the pedestrian out of the way of the collision in the first place. If pedestrians were required to wear small but powerful rocket packs, again connected to self-driving cars via the Internet, in the event of an imminent collision, the rocket pack could fire and lift the pedestrian free of the oncoming vehicle.


google car diagrams-03

We offer these ideas partly in jest, but mostly to underscore the deep biases we have in thinking about how to adapt our world for new technology.

It has long been the case with private vehicle travel that we’ve demoted walking to a second class form of transportation. The advent of cars led us to literally re-write the laws around the “right of way” in public streets, facilitating car traffic, and discouraging and in some cases criminalizing walking. We’ve widened roads, installed beg buttons, and banned “jaywalking,” to move cars faster, but in the process making the most common and human way of travel more difficult and burdensome, and make cities less functional.

Everywhere we’ve optimized the environment and systems for the functioning of vehicle traffic, we’ve made places less safe and less desirable for humans who are not encapsulated in vehicles. A similar danger exists with this kind of thinking when it comes to autonomous vehicles; a world that works well for them may not be a place that works well for people.

Consider this recent “Drivewave” proposal from MIT Labs and others to eliminate traffic signals and use computers to regulate the flow of traffic on surface streets. The goal is to allow vehicles to never stop at intersections, but instead travel in packs that create openings in traffic on cross streets that allowed crossing traffic to flow through without delay. Think of two files of a college marching band crossing through one another one a football field.

It’s thoroughly possible to construct a computer simulation of how cars might be regulated to enable this seamless, stop-free version of traffic flow. But this worldview gives little thought to pedestrians—the video illustrating drivewave doesn’t show any pedestrians, although the project description implies they might have access to a new form of beg button to part traffic flows to enable crossing the street. That might be technically feasible, but as CityLab’s Eric Jaffe pointed out, “it would be a huge mistake for cities to undo all the progress being made on human-scale street design just to accommodate a perfect algorithm of car movement.”

Not all of our problems can be solved with better technology. At some point, we need to make better choices and design better places, even if it means not remaking our environment and our communities to accommodate the more efficient functioning of technology.

Thanks to Matt Cortright for providing the diagrams for our proposed pedestrian protection devices.

California’s latest affordable housing proposal gives insight into housing politics

At first blush, it’s a bit confusing: Why, in a region that desperately needs more affordable housing, would there be so much opposition to a proposed law that would make it easier to build affordable housing?

The proposal in question was offered up last week by California Governor Jerry Brown as part of the state’s budget. It would streamline the approval process for new residential buildings with at least 20 percent affordable units (or 10 percent near transit) that already meet existing zoning by exempting them from an additional layer of environmental review, called CEQA, or the California Environmental Quality Act.

Credit: Daniel Hohern, Flickr
Credit: Daniel Hohern, Flickr


In essence, this is a kind of inclusionary zoning measure—something that many Bay Area affordable housing groups have historically supported—that promises regulatory advantages in exchange for below-market units. But unlike many inclusionary zoning laws, which allow developers to build taller or denser, Gov. Brown’s proposal would cede not one extra inch of height or additional apartment over what local cities have already designated.

Okay, so what’s the problem again? Well, virtually everywhere else in the country, if a developer wants to build something that matches their lot’s existing zoning, they can do so “as of right”: that is, as long as some local administrative body certifies that the plans actually meet zoning requirements, they get a construction permit.

But in California, developers can be required to go through the extra step of preparing a CEQA impact study. Essentially, that gives local bodies the ability to slow or block developments—even if they meet all existing zoning requirements.

A study late last year found that this layer of environmental review can add years to projects that would be routine in other parts of the country, and may well exacerbate urban sprawl, and its attendant environmental effects, because four out of five CEQA-related lawsuits focus on infill development. That gets to another central problem with CEQA as an environmental law: because it only considers impacts at the location of the proposed development, it can’t weigh the full tradeoffs of blocking housing construction in one location and thereby pushing it to another location where it may do even more environmental harm.

In other words, CEQA appears to be as much a special tool for obstructing development that meets local zoning requirements—or extracting more concessions from the developer—as a means of protecting the environment. It institutionalizes local discretion over development to an even greater extent than zoning.

Gov. Brown’s proposal, then, would reduce local discretion, acknowledging the statewide need for more housing, and especially relatively low-cost housing. While that may be good for California’s housing problems, it removes some of the negotiating power local authorities and organizations currently wield—hence, perhaps, their opposition.

This tension between local and regional or statewide power over development should be familiar. As we’ve written, when decisions are made at a local level, very place-specific costs like blocked views, competition for on-street parking, or “undesirable” neighbors are given high priority, while broader benefits to housing affordability, transportation access, and economic opportunity are often given short shrift. Not surprisingly, research has found that places where states exercise more power over development decisions have better housing outcomes, in the form of less segregation, than places where power is held more locally.

The battle in California reflects this dynamic. Local groups are put in the position of opposing a measure to speed the construction of affordable housing because it institutionalizes a broader kind of cost-benefit analysis, removing local discretion over every single project. (Of course, local governments still have the power to downzone, if they believe that the zoning they have already decided on is unacceptable, and the state law would not override that decision.) It’s also another data point suggesting that solutions to our “shortage of cities,” and resulting housing crunch, might be most likely to come from states, rather than city governments.

The Week Observed: May 27, 2016

What City Observatory did this week

1. Last month, we released the Storefront Index, a report that catalogued the nation’s retail clusters and provided a window into the spatial organization of an important part of Jane Jacobs’ famous “sidewalk ballet.” This week, we lifted the curtain a bit to explain how we built the index, hoping to give others who might wish to repeat or modify our methodology for their own research purposes a head start.

2. The growing economic strength of city centers is one of the most important facts of life in American urban policy today. This week, we updated our previous report, “Surging City Center Job Growth,” with three years of additional data that was unavailable when it was written. Our findings: although outlying areas have improved their standing since the depths of the recession, the pace of job sprawl has declined considerably in this economic expansion compared to the previous one.

3. We’re not the only ones finding strength in urban cores. Two other recent studies, in addition to our own, point to the same conclusion. The Washington think tank the Economic Innovation Group found that just 20 counties—all large urban counties—accounted for fully half of the country’s new business formationbetween 2010 and 2014. They also accounted for a disproportionate share of all new jobs. That’s a reversal from previous decades, when relatively smaller counties grew faster than larger ones. In addition, the Brookings Institution finds that urban core cities have continued to close the gap in their population growth rates with outlying parts of metropolitan areas.

4. A new affordable housing proposal in California is shedding light on some of the dynamics of housing politics in that state. Governor Jerry Brown has floated allowing developments that contain at least 20 percent below-market units and meet existing local zoning requirements to bypass an additional, only-in-California level of local discretion, called CEQA. But local governments and even some affordable housing advocates have come out against this fast-tracking of affordable units, because it reduces the bargaining power of local interests. That lines up with previous research that we’ve highlighted showing that regions where states exert more control of the development process are less segregated than places with more local control.

The week’s must reads

1. The share of new people in the rapidly growing Houston metropolitan area in the city proper has increased from just 12 percent from 2000 to 2010 to 28 percent from 2010 to 2015, according to Rice University’s Urban Edge blog. Of course, the city of Houston includes everything from burgeoning 21st century urban townhouse and apartment flat neighborhoods to classic late 20th century suburbia, so it won’t be clear exactly where these new residents are going until we have better tract-level data. But as Houston Tomorrow points out, there are payoffs even just to more centrally-located sprawl: the average person in the Houston metro area drives nearly 23,000 miles per year, as opposed to just over 19,000 in the city proper. In denser inner neighborhoods, that drops to 14,000 miles.

2. More from Texas: At D Magazine, Patrick Kennedy uses our Storefront Index to correlate downtown destination density with parking prices. Not surprisingly, the more downtown storefronts, the higher parking prices are. Kennedy finds similar patterns for job and residential density. Does that mean places with lots of people, jobs, and stores need more parking? Kennedy says no—it means that their land is valuable, and reserving lots of it for car storage doesn’t make sense.

3. A while ago, we highlighted an Orlando suburb that became the first municipality in the country to subsidize ride-hailing apps, like Uber, as a kind of transit service. Now, a much larger city, Philadelphia, has announced a limited-time partnership with Uber to solve the “first and last mile problem,” with discounts of 40 percent for trips to and from several regional rail stations. If the pilot is successful, it could be extended and expanded. This may be another sign that predictions from Yale Law professor David Schleicher about the inherent incentives of local governments to promote ride-hailing services as a public good were prescient.

New knowledge

1. The Seattle-based Frontier Group released a report this week, “A New Way Forward,” on the possibilities of a zero-carbon transportation system. It would rely on electric vehicles powered by renewable energy; more urban patterns of growth that allow for more trips to be taken without powered vehicles; more reliance on public transit; better pricing of transportation options to reflect their real social costs; and a suite of other measures.

2. School test scores in Washington, DC are up. Kristin Blagg and Matthew Chingos at the Urban Institute ask whether that’s just a consequence of changing demographics, or if there seems to have been genuine improvement. The answer: demographics can’t explain all of the test score improvements. Because their analysis only looks at district-wide changes, it’s less clear if the improvements might be tied to the benefits of integrated schools, or whether even schools that have remained racially and economically segregated have seen gains as well.

3. A new paper from Karen Chapple and Miriam Zuk at UC–Berkeley looks at the relationship between housing production—both market-rate and below-market—and low-income displacement in the Bay Area. They find that both types of housing are associated with reduced displacement, with below-market housing having roughly twice the per-unit effect as market housing. They also find that both kinds of housing appear only to work as anti-displacement measures at relatively larger geographies, which they suggest is a result of the incredibly intense housing pressures at smaller, block-group-sized neighborhood levels. While the authors position the paper as a counterpoint to an earlier report from the CA Legislative Analyst’s Office that emphasized the importance of market construction, Zuk and Chapple do reaffirm the importance of more market housing as part of the solution to the Bay Area’s housing problems. One issue: their models only manage to explain less than 20 percent of all the variation in displacement, suggesting other major unobserved factors that need to be sought out in future research.

The Week Observed is City Observatory’s weekly newsletter. Every Friday, we give you a quick review of the most important articles, blog posts, and scholarly research on American cities.

Our goal is to help you keep up with—and participate in—the ongoing debate about how to create prosperous, equitable, and livable cities, without having to wade through the hundreds of thousands of words produced on the subject every week by yourself.

If you have ideas for making The Week Observed better, we’d love to hear them! Let us know at,, or on Twitter at @cityobs.

City center job growth continues strength; suburbs rebounding from recession

As recently as the years 2002 to 2007, outlying urban neighborhoods and suburbs experienced much faster job growth than urban cores. But as a February 2015 City Observatory report, “Surging City Center Job Growth,” documented, that pattern reversed from 2007 to 2011, with urban cores overtaking more peripheral areas and maintaining positive job growth through those recession years. Since that report, however, the Census’ Longitudinal Employer-Household Dynamics database has released three more years’ worth of jobs figures, allowing up to update these trends for 2012, 2013, and 2014.

Using the same methodology as the 2015 report on the 51 largest metropolitan areas, we find that the relatively increased strength of city centers (which we define as areas within three miles of the central business district, or CBD) has held well into the post-recession recovery. While cores are now growing more slowly than the peripheries, the gap has narrowed substantially compared with the last economic cycle—a notable shift after decades of consistent job sprawl.


From 2002 to 2007, average annual job growth in urban cores was 0.26 percent, and 1.09 percent in peripheral, non-core areas. (These numbers differ slightly from those in the original report because of revisions from LEHD itself, as well as a data quality screen we implemented, which is explained at the bottom of this post. After applying the data screen, we based our findings on 19 of the 51 largest metropolitan areas. Numbers from all 51 MSAs are included at the end of this post.) From 2007 to 2011, this pattern reversed, with average annual  job growth of 0.49 percent in urban cores and just 0.10 percent in peripheries. And from 2011 to 2014, while annual employment growth in non-core areas rebounded from the depths of the recession to 2.04 percent, it also improved in city centers, growing to 1.97 percent.

In other words, while city centers lagged metropolitan peripheries in average annual job growth by 0.83 percentage points from 2002 to 2007, from 2007 to 2014, urban cores have actually grown 0.19 percentage points faster than peripheries. And although peripheries have grown slightly faster since 2011, urban cores remain in a much stronger position than they found themselves in during the previous economic expansion. The persistence of this pattern suggests that the dramatic decline in job sprawl we found from 2007 to 2011 was not simply a temporary result of the recession, but is enduring through the current economic recovery.

One of the biggest challenges in the perennial discussions of city versus suburb job growth is how to define the core and the periphery, and where to get accurate date.  In this post we describe why we think the three mile radius measure is a better indicator of the health of metropolitan cores, particularly for comparisons.  We also discuss the strengths and limitations of data available to measure core employment.  Despite the improvements in the geographical detail of data sets, there are still important limits.

These findings have major implications for American cities. We believe the evidence suggests the decline of job sprawl is a positive development, for at least three reasons.

First, many of the nation’s leading economists now agree that dense, walkable employment centers lead to improved productivity and economic growth. When firms cluster geographically such that the cost of travel between them is reduced, they are able to share resources such as physical infrastructure, labor pools, information, and technological or organizational innovation. According to research by Harvard professor Ed Glaeser, per capita productivity increases by four percent as population density increases by 50 percent—a difference roughly equivalent to the gap between Dallas (at about 3,600 people per square mile) and San Jose (at about 5,600).

Second, when people and jobs relocate to urban centers, they reduce carbon emissions in at least two ways. The first is by replacing some car trips with more emissions-efficient modes, like public transit or carpooling, or with zero-emissions modes, like walking or biking. As we have noted at City Observatory, jobs in central cities, which tend to be public transit hubs, make it more likely that even workers who live in outlying suburbs will use transit. In addition, even when commuters continue to drive, they are likely to drive fewer miles when jobs are located in central areas, reducing their emissions.

Finally, by allowing commuters to use less costly forms of transportation—public transit, walking, or biking—the movement of jobs to central cities can be a significant boon to social equity. Low-income workers are particularly sensitive to the high costs of car ownership and use, and researchers such as those at the Center for Neighborhood Technology have shown substantial differences in average transportation costs between denser, closer-in neighborhoods and more outlying communities. In many cases, the option of not using a car is worth several thousand dollars a year, a crucial increase in disposable income for many households.

This neighborhood in Atlanta has affordable housing—but its transportation costs make it unaffordable. Credit: Center for Neighborhood Technology
This neighborhood in Atlanta has affordable housing—but its transportation costs make it unaffordable. Credit: Center for Neighborhood Technology


But policymakers cannot trust that this trend will simply continue on its own, or that it will not pose challenges that will need to be addressed. In many places, the growth of demand for living and working in city centers has outstripped the growth of actual room for people to live and work in urban environments. This may be a result of some combination of a slowness of market actors to respond to demand; physical constraints on the growth of living and working space; or regulatory constraints on that growth.

This dynamic has produced a shortage of cities—demand for real estate in these central cities exceeds supply, driving up prices, excluding those who cannot afford them, and slowing the growth of these in-demand urban cores. Addressing this shortage, and making sure that residents outside the core have high-quality access to employment there with sustainable, low-cost transportation like public transit, biking, and walking, is key to leveraging this trend for maximum benefit in American cities.

How to measure the “core”?

While this report uses a three-mile radius around a city’s central business district as a proxy for “city center,” there are other ways to aggregate employment data at the sub-metropolitan level. One such method is to use counties, for which data is updated more regularly than the LEHD database our method relies on.

A recent example of this type of county analysis was published by former Trulia chief economist Jed Kolko in January 2016. Kolko’s data shows that while counties at the center of large metropolitan areas have seen job growth since 2007, they are outpaced by suburban counties in the same regions—although central counties have improved the most compared to patterns observed in 2000-2007, and appear to be continuing that acceleration. He further argues that to the extent that employment growth in these urban counties does appear to be gaining steam, it’s too early to tell whether that trend is cyclical or more long-term.

This analysis provides another useful perspective, but as we’ve pointed out, it also has limitations. In many instances, county-level data are too geographically coarse to detect shifts in growth patterns within metropolitan areas. Central counties seldom correspond to the central city of a metropolitan area, and even more rarely to what locals would understand as the “urban core.”  In many cases, central counties include substantial suburban job centers: Microsoft’s suburban campus in Redmond, Washington, is in Seattle’s King County, for example, and the suburban office parks around O’Hare Airport are part of Chicago’s Cook County. Moreover, county sizes vary substantially from region to region.

Even so, Kolko’s county-level data shows that—consistent with the data reported here—central areas’ job growth performance has improved in 2007-15, compared to the economic expansion of 2001-07. It also shows that job growth is accelerating faster in urban counties than in suburban counties in large metropolitan areas.

A key question raised by Kolko is whether the relative improvement in job growth in urban cores is a temporary cyclical change—one which will disappear as the economy normalizes—or a long-term structural shift in the relative fortunes of central and peripheral locations. While we may not have a definitive answer for several years, both the data presented here and Kolko’s county-level data suggest the answer is the latter. Far from being a temporary artifact of the recession, the improvement of urban cores is continuing into the current economic expansion.

Methodology and data notes

The data for this report are drawn from the Census Bureau’s Local Employment and Housing Dynamics program which combines administrative data to estimate employment levels by street address for workers and employers. Compiling these data is a complicated process and hinges on the consistency with which administrative records are compiled.  In the course of looking at employment data for 2012-2014 from LEHD, we discovered a number of instances of year-to-year job changes in both core and non-core areas that were so large as to raise questions about the reliability of the data. In some cases, there were periods of one to two years in which data for a particular geography exhibited a major departure from its historical pattern and then subsequently reverted to values close to the previous baseline.

For example, Detroit’s core is reported to have 140,507 jobs in 2002, but only 115,318 in 2003, and then 130,076 in 2004. Salt Lake City’s core displayed an even more curious pattern. In 2008, it reported 123,859 jobs; that fell to 110,649 in 2009. While a ten percent drop seems extreme, it did coincide with the Great Recession—but subsequent years are harder to explain. In 2010, reported employment increased again to 120,521; then fell to 104,823 in 2011; then grew again to 114,323 in 2012. These examples are not atypical for the metropolitan areas that failed our data quality screen.

These metropolitan areas are also typical of those with data anomalies in another way. Because LEHD breaks down employment by sector, we can determine that nearly all of the net job changes in the cores of Detroit and Salt Lake City occurred in “Educational Services.” (In Detroit, for example, LEHD reports the number of such jobs as 27,784 in 2002; 7,212 in 2003; and 27,797 in 2004.) Both of these cities have large universities within their cores.

These problems may be inherent in utilizing disparate sets of administrative records to establish workplace and residence locations. LEHD data are based on personal and business tax records which were not primarily designed for the purposes to which they are currently being adapted. In particular, it seems possible that year-to-year changes in these records at large organizations with multiple office sites switch between associating subsets of employees between the central office and some sort of satellite office. In most cases, large changes in one category—educational workers in the core, for example—are offset by changes in the other direction in another category—educational workers in the periphery. This hypothesis is also supported by the tendency for these changes to fall in the educational sector, or other categories—such as “Health Care and Social Assistance” and “Administration and Support”—likely to be part of large organizations. Our hypothesis, then, is that these administrative record changes, which do not necessarily reflect any actual physical rearrangement in the real world, are behind the odd data patterns we see.

Administrative data like that used to construct the LEHD database are not primarily intended to serve as a resource for time-series analysis. The Census Bureau constructs its estimates of data each year separately, so changes in reporting locations or disaggregations from year to year can alter results. As the Census Bureau warns: “The LEHD program produces each year of LODES independently, so there may be time series inconsistencies due to updates and methodological changes that can complicate longitudinal inferences.”

Validating all of the LEHD data is far beyond our resources. Instead, we chose to create a data quality screen focusing only on employment totals at levels of geography that might meaningfully affect our final analysis. The screen flagged metropolitan areas with large one-year changes in total employment: greater than eight percent in either direction in core areas, and greater than five percent in either direction in non-core areas. (The cutoffs differ because of the larger variance in growth figures for cores.) Metropolitan areas with such changes were excluded, unless the changes a) were consistent with multi-year trends, or b) were large declines during the Great Recession. As in our earlier report, we excluded the category of public administration from all totals. Federal employment was added to the LEHD program only after 2010, making later year data in this category non-comparable with earlier year data.

The Census’ LEHD program represents an invaluable source of data for social science research on any number of subjects: economic development, commute patterns, social networks, and many others. It has already been widely used for urban policy research, including by Elizabeth Kneebone and Natalie Holmes at the Brookings Institution (Kneebone & Holmes, 2015); researchers at the Federal Reserve Bank of Cleveland (Hartley et al., 2015), and others (Meltzer & Ghorbani, 2015). We are enormously grateful to the program for doing the work of compiling this information. However, we do want to flag these issues as a potential area of improvement, and suggest that other researchers using LEHD data should keep them in mind while performing their analysis.

Included Excluded—failed data quality screen Excluded—incomplete data
Atlanta, GA Austin, TX Boston, MA
Baltimore, MD Buffalo, NY Phoenix, AZ
Birmingham, AL Charlotte, NC Washington, DC
Cincinnati, OH Chicago, IL
Cleveland, OH Columbus, OH
Denver, CO Dallas, TX
Hartford, CT Detroit, MI
Houston, TX Indianapolis, IN
Kansas City, MO Jacksonville, FL
Las Vegas, NV Los Angeles, CA
Louisville, KY Memphis, TN
Nashville, TN Miami, FL
New York City, NY Milwaukee, WI
Pittsburgh, PA Minneapolis, MN
Portland, OR New Orleans, LA
Rochester, NY Norfolk, VA
San Antonio, TX Oklahoma City, OK
San Diego, CA Orlando, FL
San Francisco, CA Philadelphia, PA
Providence, RI
Raleigh, NC
Richmond, VA
Sacramento, CA
Salt Lake City, UT
San Jose, CA
Seattle, WA
St. Louis, MO
Tampa, FL

To test whether our data screen biased the overall results, we also tabulated aggregate data with no exclusions. Although the totals differ somewhat, the trends are broadly similar: average annual employment growth in urban cores is 0.00 percent from 2002 to 2007, 0.42 percent from 2007 to 2011, 1.60 percent from 2011 to 2014, and 0.92 percent from 2007 to 2014. Average annual employment growth in non-core areas is 1.23 percent from 2002 to 2007, -0.10 percent from 2007 to 2011, 2.01 percent from 2011 to 2014, and 0.81 percent from 2007 to 2014.

How we did the Storefront Index

We’ve received many questions on how we did the analysis behind our Storefront Index. This post will describe our dataset, our method, and how we created our visualizations. We hope that this will spur future research and new forms of visualizations, similar to the way in which the release of our Lost In Place data led to amazing reinterpretations of the dataset.

Screen Shot 2016-05-20 at 11.28.32 AM

We used a database of businesses from Custom Lists American Business Directory. The Directory contains 2014 records on US businesses, their industry classification, and their address. Our aim was to understand how clusters of these quasi-private storefront spaces contribute to active streetscapes and generated steady flows of people — so we filtered our business dataset based on three criteria: 1) businesses in the largest metro areas; 2) businesses that have storefronts; and 3) lastly, a spatial filter based on clustering.

For the first filter, we simply chose business in the largest metro areas based on the 2013 definitions of CBSA. For the second filter, we selected businesses that fit into one of 44 industry classifications that would typically have customer-serving storefront. These include businesses like grocers, bookstores, and salons. A full list of our categories can be found on page 18 of the report.

Armed with a storefront business dataset, we next sought to find clusters of storefronts. Thus, for each business with a storefront, we needed to know the distance to the next closest storefront. Our first task was to geocode the addresses, turning the address into a latitude and longitude that we could map. (Luckily, we performed this geocoding in ArcMap just before they cut off access to their geocoding API.) We then used the NEAR function in GIS software, allowing us to calculate the distance to the next closest storefront in meters. To apply our filter, we then chose only storefronts that had another storefront within 100 meters, allowing us to identify clusters of destinations that would be easily walkable.

With our three filters applied, we created a set of map images (for the report) and an interactive map. We used a 3-mile buffer around the central business district of our metro areas of interest. For the images, we highlighted these buffers in white, and used only the points of our clustered storefront locations and the US Census Bureau’s Roads shapefile. For the interactive web map, we used a circles to represent the 3-mile buffer, the points of our clustered storefront locations, and the Mapbox library with the Stamen Toner basemap.

Our list of industry classifications (using Standard Industry Classifications) can be found here and GeoJSON shapefiles for each metro area (using FIPS codes for each metro) can be found here.

The Week Observed: May 20, 2016

What City Observatory did this week

1. What’s the relationship between urban sprawl, income segregation, and economic opportunity? A recent study by Reid Ewing and colleagues at the University of Utah used an innovative new measure of sprawl to correlate with economic outcomes of low-income children, and found a strong positiveassociation between compactness—that is, un-sprawl—and more economic mobility. A doubling of the “compactness index”—roughly the difference between Nashville and St. Louis—was linked to a 41 percent increase in the likelihood of a child born to the bottom fifth of the income distribution reaching the top fifth as an adult.

2. How did San Francisco’s housing market get so crazy? For many, the answer begins with the tech boom of the last decade. But new research by Eric Fischer shows that rental prices there have been steadily increasing at about 2.5 percent faster than inflation for about 60 years. What happened 60 years ago? San Francisco ran out of easily buildable open land. Unwilling to allow for major redevelopment, that means that they’ve been “building up” their housing shortage since the 1950s. Fischer also finds that you can predict housing prices with startling accuracy using just three variables: wages, employment, and housing stock. In his model, what would it take to get prices back down to a reasonable level? Either drop half the city’s jobs, cut wages by nearly half, or build 200,000 new homes—in a city that permitted just over 3,600 last year.

3. Nationally, it appears that rental supply may be catching up to demand. So says market analyst REIS, which finds that apartment vacancy rates have ticked up recently for the first time in years. Does that mean the end of price growth? Perhaps—but of course, apartments aren’t rented in a national market, and there’s no evidence of rising vacancy or a supply-demand match in places like San Francisco, New York, or most other cities making headlines for their housing prices. That said, even on that front there is some good news: the Boston Globehas reported that rents there are growing at the slowest rate in five years as a burst of new apartments came online last year, and another 5,000 are slated for this year.

4. While things might be starting to look better in Boston, over the last several years, there has been a clear pattern of faster rent growth in larger, urban citieswith strong economies. And while these cities’ rents are pulling away from the rest of the country—the top nine metro areas saw rent growth of over 12 percent between 2008 and 2015, compared to just over two percent in the bottom half of the hundred largest metros—high-end units within these metros are also pulling away from the rest of the market. These nuances are important for policymakers and advocates in trying to grapple with housing affordability.

The week’s must reads

1. A common refrain from housing advocates on the land use law side is that the best bet for better housing policy is to look at state governments, since local governments generally focus on the local costs of housing development, like congestion or negative effects on housing prices, rather than the broader regional benefits. Now, California Governor Jerry Brown is taking up the challenge,proposing new rules to force local governments to speed housing development proposals that meet existing zoning and contain at least 20 percent below-market units on site. A handful of other states, like Massachusetts, already have state-level override policies for local housing permits.

2. Another common housing advocate wish: Allowing “accessory dwelling units” or “granny flats,” backyard cottages that add housing without dramatically changing the face of a neighborhood. That’s exactly what Durango, Colorado has done, as described in CityLab. In fact, the city had already had ADUs, built before zoning—but they had been illegal for decades. Now the city is legalizing them again, perhaps bringing hundreds of new housing units online (and above board) without a single new building. It’s a tactic that many cities should be able to copy—especially older ones that, like Durango, already contain many ADU-type buildings constructed before modern zoning codes.

3. You’re in a driverless car, approaching a tunnel. A child steps into the road, and there’s no time to stop. Is your car programmed to continue ahead, killing the child, or swerve into the wall, saving the child but killing you? That’s one question. Another question is: Why is the car going fast enough to kill either of you? At n+1, Daniel Albert examines “the ethics of cars,” both in the future world of self-driving automobiles, as well as in our own era of human-driven ones.

New knowledge

1. A new GAO study has alarming numbers on racial and economic segregation of schools. Between 2001 and 2014, the proportion of all American schools with student bodies that were both at least 75 percent Black and Hispanic and 75 percent low-income grew from nine to 16 percent. In 2001, 14.2 percent of American students attended high-poverty schools; in 2014, more than 25 percent did. These figures reflect, in part, broader trends about growing economic segregation and the concentration of poverty in housing and neighborhoods as well as schools.

2. Many people are familiar with the unfortunate history—and, in many ways, present—of affordable housing being used to keep low-income people, and people of color, concentrated in certain neighborhoods, and out of others. But the University of Minnesota Law School’s Institute on Metropolitan Opportunity released a report this week that identified a different kind of below-market-housing-driven segregation: high-end affordable housing, often pitched as artists’ housing, that serves nearly exclusively white tenants in higher-end neighborhoods. These units often cost far more to build, and so receive more subsidies per unit, than other kinds of affordable housing.

3. The Urban Institute looks at a kind of “missing middle” housing whose production has fallen off since the housing bust of the last decade: two-to-four-unit residential buildings. They identify one of the problems as changing financing standards that have reduced risk tolerance for these types of buildings, perhaps below the levels tolerated for single-family homes. Given that these small multifamily buildings make up a significant proportion of moderate-cost housing in many metropolitan areas, addressing this sudden decline in production may be important.

The Week Observed is City Observatory’s weekly newsletter. Every Friday, we give you a quick review of the most important articles, blog posts, and scholarly research on American cities.

Our goal is to help you keep up with—and participate in—the ongoing debate about how to create prosperous, equitable, and livable cities, without having to wade through the hundreds of thousands of words produced on the subject every week by yourself.

If you have ideas for making The Week Observed better, we’d love to hear them! Let us know at,, or on Twitter at @cityobs.

Cities are adding people, jobs and businesses

A trio of reports released in the past week provide new data showing the economic strength of the nation’s cities.

Whether we look at population growth, new business formation, or job creation, big cities, urban centers and close-in urban neighborhoods are big drivers of national growth. While the data are drawn from different sources and use slightly different geographies, the messages are quite similar.

The Economic Innovation Group, a Washington-based think tank, used county-level data from the Census Bureau’s County Business Patterns program to chart net new business formation and job growth. Over the period 2010-2014, the 20 counties with the largest increase in new businesses (all big urban counties) accounted for a majority of all net new business formation in the current economic expansion. The 20 counties with the largest job increases accounted for 28 percent of all new jobs. This represents a dramatic turnaround from previous expansions in the 1990s and early 2000s, when smaller, less populous counties tended to grow faster. The dwindling rate of new firm formation is a topic of growing concern as we think about the nation’s long term growth prospects; in the current recovery, a few large metropolitan areas have played a dramatically disproportionate role in fueling new business activity.

Credit: Economic Innovation Group


In the 1990s, there was a strong negative correlation between county population and job growth rates, meaning that less populous counties grew much faster than more populous ones. But in the past four years, that relationship has reversed. Small counties are growing the most slowly; larger counties are growing more rapidly. Counties with a million or more residents grew only half as fast as those with fewer than 100,000 residents in the mid-90s (7.7 percent vs. 16 percent); and the larger counties grew more than twice as fast (9.9 percent vs 4.4 percent) since 2010.

The Brookings Institution’s Bill Frey crunched the numbers from the latest (2015) Census population estimates to track population growth in the nation’s 50 largest metropolitan areas, comparing growth rates in the principal city in each metropolitan area, with the remaining jurisdictions. As Frey notes, principal cities—the largest municipality in each metropolitan area—are growing faster than the remaining portion of metropolitan areas, reversing a long-standing pattern of suburban growth outpacing city growth. Cities have grown faster than suburbs since 2010.



And at City Observatory, we’ve released our updated figures on central city job growth. Using fine-grained establishment level data from the Census Bureau’s Local Employment and Housing Dynamics (LEHD) database, we’ve plotted employment change for the three mile radius surrounding the center of the central business district in the nation’s large metropolitan areas. While job growth in urban centers lagged well behind suburbs a decade ago, job growth rates urban centers are today very similar to those in more suburban locations. Bloomberg View columnist Justin Fox addressed the findings of our report in a recent column, and highlighted the pattern of change over time:



While these three reports draw on different data sources, and use somewhat different geographies (large counties, principal cities and a three-mile radius) they tell very similar stories about the persistence of urban-led population and economic growth in the US at least through the middle of the present decade.

More jobs, more businesses, more people. These three reports add to a growing body of data suggestion that large metropolitan areas and urban centered economies are increasingly driving national economic prosperity.

As we’ve noted before, using city and county boundaries to measure differences between “cities” and “suburbs” and particularly to make comparisons across metropolitan areas can be problematic. City and county boundaries often don’t correspond well to patterns of urbanization, and the scope of a largest county or principal city varies widely across metropolitan areas.

The demand for city living is behind the urban rent premium

The US faces a shortage of cities. More and more Americans, especially talented, young workers with college degrees, are looking to live in great urban locations. As we’ve explored at City Observatory, the demand for urban living has increased faster than the supply of great urban spaces—with the predictable result that the price of land is appreciating faster in cities. We’ve pointed to a growing body of data—stronger residential price growth in urban cores, faster appreciation for homes in more walkable areas, a more rapid growth of office rents in walkable locations—all of which signal the growing market value of living and working in cities.

This trend also clearly manifests itself in the residential rental marketplace. New data from multifamily market analyst RealPage shows that apartments in big cities have seen higher rent increases than in smaller ones.

The RealPage data enable us to divide the US top 100 US markets into three broad groups: the nine hottest coastal markets, the remaining 41 of the 50 largest metropolitan areas, and 50 smaller metropolitan areas. Data shown here are for the fourth quarter of 2008 and 2015 and are expressed in inflation-adjusted dollars. Percentages shown are the percent increase in real, inflation-adjusted rents over the entire seven-year period.


The most obvious point is that the large, hot markets are showing the biggest increases, at all price points. The nine markets are all on the East Coast (New York, Boston, Washington) and West Coast (Southern California, the San Francisco Bay area and Seattle). In the nine hot markets, median rents increased about 13 percent, compared to 8 percent for other metro areas in the top 50 and just 2 percent for metros ranked 50th through 100th in size.

As interesting as the pattern of change is across markets, the change within metro markets is also instructive. In every size market, rents are increasing fastest in the highest priced segment of the market place. RealPage reports the average rents for the median apartment, as well as the apartments representing the 25th and 75th percentiles of the marketplace. Think of the 75th percentile as representing the cutoff for the top one-quarter “nicest” apartments, and the 25th percentile representing older, less desirable and more affordable apartments. (We explored the importance of looking at all segments of the rental marketplace to understand affordability a few months back; in many ways, having a broad range of price points within the market is a better indicator of affordability than the median).

The most expensive apartments in the most expensive cities are seeing the fastest rate of price appreciation. For example, in the top nine coastal markets, rents for the 75th percentile apartment increased by about 23 percent, while apartment rents for the median and 25th percentile apartments increased only about half as fast (10 percent and 12 percent respectively). What this signals is that the demand for urban living is being fueled by the preferences of high income households rather than simply a generalized increase in rents that affects every segment of the marketplace equally.

The next 41 largest metropolitan areas–those rounding out the list of the fifty largest metropolitan areas have lower rates of price appreciation overall, but exhibit the same within metro pattern of greater price increases at the higher end of the market. Rents at the 75th percentile increased about 13 percent between 2008 and 2015, but only about 8 percent for the median apartment and 7 percent at the 25th percentile.

What this signals is robust demand for high-end apartments in the nation’s largest and hottest markets. As RealPage economist Shane Squires points out in his commentary on this data, much of the demand for new, high-end product is in walkable locations in cities, where it doesn’t compete directly with more affordable (but far less accessible) suburban locations. This is also consistent with the hypothesis offered by Edlund, Machado and Sviatchi, that time-pressured, dual income households are increasingly willing to pay higher rents in urban centers for better accessibility to jobs and consumption opportunities.

The RealPage data provide nuanced insights into rental price trends that aren’t fully reflected in the usual measures that talk only about the “median” apartment. They show that rental prices are changing at very different rates in different markets, and at different price points with metropolitan markets. This more detailed view of where–and why–rents are increasing should be a useful guide to policy makers as they consider how to define and grapple with housing affordability problems.

(RealPage makes its estimates based on data drawn from lease transactions covering about 10 million apartments nationwide.)

Nationally, apartment supply may be catching demand

There’s more evidence that housing market supply is beginning to catch up to demand in a way that is likely to moderate rent increases. Nothing, it seems, is more infuriating to those caught in a market of steady rent hikes that being lectured by some economist that what is needed to resolve the problem is an increase in supply. Nice to know, but that’s not going to pay the rent any time soon.

But just as after a long winter there are some early signs of spring, there are a few hopeful indicators from housing markets that the long promised relief from increased supply is starting to show up, at least in a small way. Today we look at two recent market reports, one national, and one quite local, that are beginning to indicate a market shift.

Credit: Kelly Sims, Flickr
Credit: Kelly Sims, Flickr


There’s no denying that rents in the US have escalated over the past several years. Overall rents are up 4.6 percent in the past year, and the national rental vacancy rate has plunged from more than 10 percent to about 7 percent, signaling that their are relatively more tenants bidding for every available apartment. As a result the share of households spending more than 30 percent of their income on housing has increased.

For the past several years—and for a variety of reasons—we’ve seen a surge in demand for rental properties. Some of that had to due, especially initially with the collapse of the housing bubble, and several million households being moved, quite involuntarily from the ownership market, into renting. At the same time, younger adults have been much more likely to rent that previous generations, and seem especially enamored of centrally located, walkable apartments in great urban environments. The net effect is that the demand for rental housing has risen steadily.

At City Observatory, we think there are two fundamental, and widely under-appreciated facts about housing markets. First, that when supply does catch up to demand, rent increases soften. Second, supply almost always moves much slower than demand. The supply of rental housing has responded only slowly; and has mostly struggled to keep up with increasing demand.

In the past few months there’s growing evidence that supply is starting to catch up. The market analysts at REIS follow national trends in apartment construction tracking delivery (the completion of new apartments) and absorption (how many newly completed apartments get leased. Absorption is a “net’ figure: the difference between the number of previously vacant apartments that get leased and the number of previously occupied apartments that become vacant over any time periods. The difference between the completions and absorptions drives vacancy rates. When lots more apartments get leased than new apartments are built, vacancy rates fall; when completions outpace absorptions, vacancy rates rise.

REIS tracks these numbers on an annual basis, and their estimates for the past three decades are shown here:



REIS analysts are reading the data to suggest that construction of new apartments is finally starting to have an impact on the market. REIS’s Scott Humphreys says:It’s official: developers are finally building more apartments than there are renters to fill them.

Similarly, at the blog Calculated Risk, Bill McBride reports on data from the National MultiFamily Housing Council (NMFC) which tracks vacancy rates for apartments around the country. Their data show that “market tightness” has been trending downwards for the past couple of years, and leads McBride to conclude that “it appears supply has caught up with demand—and I expect rent growth to slow.”

These data illustrate an important fact about supply and demand: Demand is highly volatile and can change quickly, while supply responds only slowly. Consider: the first decade of the 2000s was mostly a pretty bad time to be a landlord. A steady supply of new apartments was being built, but net absorptions fell short of the number of completions. In fact, at the height of the housing bubble, net absorptions were negative: more households moved out of apartments than moved in.

In the last two quarters of 2015, according to REIS data, completions outstripped net absorption by about 13,000 units in the third quarter and by nearly 15,000 units in the fourth quarter, with the result that the national vacancy rate ticked upward.

While national trends provide a helpful background, like politics, all housing is local. In an important sense, there really is no “national” apartment marketplace: apartments built in Cheyenne Wyoming really aren’t good substitutes for apartments in San Francisco. While there are broad national trends, the trajectory of supply and demand can play out differently in different local markets. The collapse of oil prices, for example, has dramatically altered the demand/supply balance in the oil patch town of Williston, ND, with the result that rents are off more than 20 percent from a year ago.

The evidence from one of the nation’s tightest housing markets, Boston, suggests that supply may be getting closer to catching up with growing demand. The Boston Globe reports that in the fourth quarter, rents in Boston increased by just one-tenth of one percent from the previous year, the smallest increase in five years. About 3,800 units came on line in 2015, and about another 5,000 are in the development pipeline. The news from Boston echoes reports from markets as diverse as Seattle, Denver, and Houston, that the growing number of new properties being completed is producing at least temporarily higher vacancy rates and more favorable rental offerings for tenants.

Nobody’s predicting a glut of unoccupied apartments—either in Boston, or nationally—that will push rents down. But slowly, and inexorably, the supply of housing is catching up to the demand, both in the aggregate and in the specific places where demand has grown most rapidly. It’s a reminder that if policy enables housing supply to expand, relief from higher rents can be delivered through the market.

The long road to San Francisco

Every once in a while, someone writes something that makes a murky, complicated, frustrating issue seem crystal clear.

This post by Eric Fischer is one of those. Doing yeoman’s work, Fischer transcribed decades’ worth of San Francisco housing prices and other data. Among his findings:

  • Though we talk about the Bay Area’s housing crisis as if it were a recent phenomenon, Fischer finds that housing prices have been appreciating at a steady 6.6 percent pace for the last 60 years. In real terms—that is, adjusting for inflation—they’ve been appreciating at about 2.5 percent annually. That doesn’t sound like much, but when it happens for 60 years in a row, it means that housing prices have quadrupled, after inflation, since 1956.
  • We know that housing prices are the result of exceedingly complex market and regulatory systems. But Fischer shows that variations in price over the last four decades can be predicted pretty well using just three variables: employment growth, wage growth, and housing construction.

Both of these have obvious follow-up questions—which Fischer asks and (mostly) answers.

First, if San Francisco’s housing crisis actually started 60 years ago, what triggered it? Well, the housing construction cycle that ended in 1954 was the last in which the city had large tracts of greenfield land. After that, housing could really only be added at scale with infill—but by then, zoning had already been enacted to make that much more difficult. In other words, because it has been unwilling to allow large-scale redevelopment, San Francisco has been steadily under-building basically since it ran out of open land.

Screen Shot 2016-05-16 at 6.02.13 PM


Second, if you can mostly predict housing prices with just three variables, how much influence does each have? Of course, correlation is not causation, and as Fischer admits, small tweaks to the model could produce notably different results—so these should be treated as broad trends, not precise measurements. But in Fischer’s model, a one percent increase in employment is associated with a 0.95 percent increase in rents; a one percent increase in wages is associated with a 1.74 percent increase in rents; and a one percent increase in the housing stock is associated with a 1.70 percent decrease in rents.

If we treat those as roughly accurate, that suggests that to return to 1981-era prices, when inflation-adjusted rents were about two-thirds lower, employment would have to fall by 51 percent, or wages would have to fall by 44 percent, or the housing stock would have to grow by 53 percent.

Michael Anderson of Bike Portland took a look at these numbers and despaired. And he’s not wrong: clearly losing half of all the jobs in San Francisco, or cutting everyone’s income in half, would be disastrous; and building 200,000 new units of housing in a city that permitted just over 3,600 last year seems more than a little out of reach. (And of course, every year that employment or wages rise, that needed housing number goes up, too.)

But Anderson also points out that even if there were no regulatory constraints on building housing, development would surely grind to a halt way before prices could fall down to semi-reasonable levels, as developers realized they could no longer count on the rental revenue they had based their loans on.

Credit: Anh Dinh, Flickr
Credit: Anh Dinh, Flickr


In fact, this brings up a point that is often elided when we talk about places, like San Francisco, where prices have gotten totally out of control: A rapid “correction” to those prices would, in many ways, be an economic crisis. Nearly every homeowner would find themselves tens or hundreds of thousands of dollars underwater on their home, probably causing a massive region-wide foreclosure crisis; renters would also be affected, as their landlords would also be underwater or even bankrupt, many stuck with mortgages based on valuations that anticipated rents three or four times higher than their new, more reasonable levels.

In America, no democratically elected government could ever propose, let alone enact, a policy with these consequences. Rather, San Francisco’s best hope is to build enough to prevent more price increases above inflation, and maybe even keep price appreciation below inflation, so that over the coming decades prices can float down to a more reasonable level without ruining every property owner in the city—and, in the meanwhile, produce as much non-market housing as possible. (Universal housing vouchers or housing tax credits would help.)

Fischer’s work, then, might best serve as a warning to other cities: housing crises, driven in part by supply shortfalls, build up over decades—and once they’ve been going on that long, the path back down the mountain is perilous and slow.

USDOT to shut down nation’s roads, citing safety concerns

WASHINGTON, DC – Citing safety concerns, today Secretary of Transportation Anthony Foxx announced he was contemplating the closure of roads to all private vehicles in nearly every city in the country until he could assure the nation’s drivers that they would be safe behind the wheel.

The announcement comes on the heels of comments by Secretary Foxx that the Department of Transportation may shut down the Washington Metro heavy rail system because of ongoing safety issues.

Since 2009, 14 Metro riders and employees have died in collisions, derailings, and other incidents. On an annual basis, that translates to about 0.48 fatalities per 100,000 weekday riders.*

However, Secretary Foxx noted that this is exceeded by the fatality rate of car crashes in every single American metropolitan area for which data was compiled in a recent report from the National Highway Traffic Safety Administration.

In San Francisco, 3.75 people died in automobile crashes per 100,000 residents in 2014, a rate 7.8 times higher than the fatality rate on Metro. In Raleigh, NC, the automobile crash fatality rate was 7.50 per 100,000, or about 15.6 times higher than the fatality rate on Metro. And in Dallas, the automobile crash fatality rate was 12.02 per 100,000, or about 25.0 times higher than the fatality rate on Metro.

A partial list of other cities in which Secretary Foxx is threatening to shut down automobile traffic includes:

Screen Shot 2016-05-11 at 2.05.36 AM

Each year, more than 30,000 Americans die in automobile crashes, at a rate higher than nearly every other industrialized nation, even accounting for higher vehicle miles traveled rates.

“This carnage is unacceptable,” the Secretary said. “Until we can assure America’s drivers and pedestrians that they are no more likely to die on the road than they are on the most dysfunctional heavy rail system in the country—a feat that, in many cities, will require a 90 to 95 percent reduction in road fatalities—I cannot in good conscience allow a single motor vehicle to menace our cities.”

*Methodology and sourcing: Road fatality rates are taken from the National Highway Traffic Safety Administration. WMATA Metro fatality rates are from news reports on fatalities since 2009; the denominator is half of the average weekday ridership from the most recent APTA ridership report, from Q4 of 2015. (We divided total ridership in half to estimate the total number of individual riders taking two trips per day.) This is designed to create a relatively high fatality rate for WMATA—making a relatively small denominator, of only the number of people who use WMATA on a daily basis—compared to the road crash fatality rate, which uses a relatively large denominator, the total number of people living in a metropolitan area.

Sprawl, segregation, and mobility

This is the fourth in an ongoing series of posts about income segregation, urban planning, and economic opportunity. In the first, we examined three different ways of looking at income segregation: the proportion of people living in low-income neighborhoods, high-income neighborhoods, or both “extremes.” In the second, we looked at another kind of income segregation, measured along the entire income spectrum, and distinguished between segregation and inequality. In the third, we examined how income segregation has changed, both since 1970 and since the Great Recession.

Over the last two weeks, we’ve written about how income segregation is really many different kinds of sorting; how to measure several of the most important kinds; how and why to distinguish between segregation per se and inequality; and how income segregation has changed over the last 40 years.

But we’re not just interested in analyzing and diagnosing income segregation for its own sake. We’re interested in how it intersects with outcomes we care about—notably economic opportunity for people with low incomes, and the strength of common civic culture—as well as policy levers that cities and other governments can use to improve those outcomes.

That makes a recent study led by Reid Ewing of the University of Utah particularly valuable. Ewing et al’s paper is one of the first to rigorously analyze the relationship between economic mobility, income segregation, urban sprawl, inequality, and other potential correlates of economic mobility.

The study builds off of an innovative “compactness index” developed by Ewing and Shima Hamidi to compare levels of sprawl between metropolitan areas. Previously, researchers like Raj Chetty and his team at the Equality of Opportunity Project had used commute times as a proxy for sprawl, but that’s obviously related to a number of other factors beyond the spread-out-ness of the urban environment.

The big takeaway from Reid’s study is that in metropolitan areas that are more compact—that is, less sprawl-y—children born into the lowest fifth of the income distribution are much, much more likely to reach the top fifth as adults. On average, if City A is twice as compact as City B, low-income children in City A will be 41 percent more likely to reach the top fifth of the income distribution than low-income children from City B. (In the real world, “twice as compact” is roughly the difference between sprawl-y metropolitan Nashville and less sprawl-y metropolitan St. Louis.)

St. Louis. Credit: Ron Reiring, Flickr
St. Louis. Credit: Ron Reiring, Flickr


Reid et al also find that lower levels of income segregation—specifically, segregation of the poor—is associated with greater mobility, confirming the findings of other researchers.

But when it comes to the interaction of sprawl with income segregation, things are a little more ambiguous. The study actually finds a negative relationship between compactness and segregation—that is, more compact metropolitan areas tend to be somewhat more segregated by income.

That raises a few questions. First, if compactness is associated with more income segregation, and more income segregation is associated with lower mobility, then how can compactness itself be associated with higher mobility? The answer to that, according to Reid et al, is that other effects—perhaps most notably access to jobs—overwhelm the income segregation effect and cause the net effect to be positive.

Second, why would more compact metro areas tend to be more segregated? There are a number of possibilities here. One is that there is some inherent connection between relatively dense built environment and segregation—but it’s not totally clear what the mechanism there would be. On the other hand, there are a number of possible third factors that might lead to the appearance of such a relationship. One is that in the US, more compact urban areas tend to be older urban areas—and older urban areas are ones where larger proportions of neighborhoods were around to be affected by the more blatant policies that promoted racial segregation, including redlining, restrictive covenants, and widespread racist violence. These historically racially segregated neighborhoods are, today, very disproportionately likely to be areas of concentrated poverty. In that way, the mechanism wouldn’t be that compact development leads to segregation, but that older cities are both more compact and more segregated.

Another possibility is the “modifiable aerial unit problem.” Reid et al measure segregation by Census tracts, which are drawn to have very roughly equal numbers of residents. That means that they’re much smaller in denser cities—and, perhaps, more sensitive to block-by-block sorting than in more sprawling regions, where tracts can include many different blocks that aren’t really in the same neighborhood.

Census Tracts at the same scale in Brooklyn and suburban New Jersey. Credit: Social Explorer

Screen Shot 2016-05-13 at 5.56.30 PM
Census Tracts at the same scale in Brooklyn and suburban New Jersey. Credit: Social Explorer


Or perhaps there’s something distinct that Reid’s compactness index is picking up that other definitions of sprawl don’t include. After all, previous research has in fact found a connection between looser zoning regulations—that is, ones that allow more compact multi-family development—and less income and racial segregation, presumably because a wider variety of housing types is more likely to include a wider variety of housing prices.

But in any case, clearly more research is needed. In the meanwhile, Reid’s study provides more evidence both that more compact urban areas provide more economic opportunity to the low-income, and that income segregation is a key lever of opportunity as well.

The Week Observed: May 13, 2016

What City Observatory did this week

1. A new study from Stanford Business School claims that society reaps the greatest benefits from low-income housing when that housing is built in the lowest-income neighborhoods—as opposed to integrating it within higher-income neighborhoods. But there are a number of caveats and concerns we have with the study. For one, it looks at a very specific form of low-income housing, LIHTC, with income targets 50 percent or more greater than the median income in the neighborhoods where the Stanford study finds benefits. Second, the “costs” to higher-income neighborhoods appear to mostly be the result of discrimination—which we wouldn’t give much weight in policymaking. And finally, the benefits to integration go far beyond what was measured by the study.

2. One of those benefits of integration: It can create its own positive feedback loops that reduce the tendency to re-segregation. Research from the University of Minnesota’s Institute on Metropolitan Opportunity shows that in metropolitan areas with regional school desegregation initiatives, mixed-race neighborhoods are much less likely to re-segregate than in areas without school desegregation initiatives. In other words, if white households aren’t as able to find segregated schools, they’re less likely to want segregated neighborhoods, and less likely to take part in destructive cycles of flight and reconcentration. Intentional housing desegregation—by, in part, putting low-income housing in higher-income neighborhoods that lack it—might be able to pull the same trick.

3. On Tuesday, Secretary of Transportation Anthony Foxx announced that he was considering shutting down the Washington, DC Metro because of safety concerns. On Wednesday, Foxx announced that, having reviewed the numbers, it turns out that road fatality rates in virtually every metropolitan area in the country exceed fatality rates on Metro—in many cases, by factors of 25 or more—and the USDOT would look into shutting down all roads to private motor vehicles until he could guarantee drivers and pedestrians that they would be as safe on a typical American street as they are on the most dysfunctional heavy rail system in country. Unfortunately, that appears to be unrealistic—most cities would need to reduce road fatalities by 90 to 95 percent—so motor traffic may be kept out of American cities indefinitely.

4. Last week, we explored four different ways of measuring income segregation, and the implications of each one for broad-based economic opportunity. This week, we take a look at how each kind of income segregation has changed, both over the long term and just the last few years. In both cases, the answer is: it’s up. But there are some important nuances: since 2007, nearly the entire rise in income segregation appears to come from increased sorting among people in the “middle,” between the 10th and 90th percentile of income, according to researchers Kendra Bischoff and Sean Reardon. We also created an interactive tool for you to look up how income segregation has changed in your region since 1970.

The week’s must reads

1. The nation’s most successful public transit system, the New York City Subway, has old, often less-than-attractive stations, crowded trains, plain plastic seats, spotty cell service, and none of the frills that civic leaders often claim are necessary to get people—especially middle-class people—out of their cars. So why do people use it? Because it’s fast, convenient, and reliable. At The Transport Politic, Yonah Freemark uses the example of the NYC Subway to commend Boston’s plan to save a rail extension by cutting costs on everything that won’t affect the qualities that riders actually care about: the frequency, speed, or reliability of service. It’s a great case study for anyone thinking about how to prioritize transit change in their city.

2. Many people believe that the purpose of zoning is to preserve neighborhood character. But Seattle’s The Urbanist blog points out that there are actually hundreds, if not thousands, of multi-family buildings in that city’s “single-family” zones—that is, places where multi-family buildings have been outlawed as “incompatible.” How can that be? These apartments and condo buildings were mostly constructed before the advent of zoning turned many already-existing communities into “illegal neighborhoods.” Many of these neighborhoods—especially closer to downtown—are among the city’s most popular, reaffirming yet again that the lack of housing diversity that zoning codes often prescribe is neither necessary nor desirable.

3. In a similar vein, modern American cities often take for granted that businesses and residences need to be kept apart. But CityLab highlights a blog that captures ghosts of Washington, DC’s mixed-use past: old storefronts that have been converted to residential buildings because they ran afoul of zoning laws that found them “incompatible.”

New knowledge

1. The Century Foundation has released a new report laying out principles for national transportation policy. The paper identifies four key challenges, from a lack of a national vision to structural problems with the political system, and proposes four major changes to transportation policy. They include: federal funds should take care of capital maintenance needs before supporting new construction; policy should focus on moving people and goods rather than cars and trucks; a renewed focus on performance management; and use an increased gas tax, vehicle miles traveled tax, congestion pricing, or other transportation-related sources to fund transportation programs.

2. While American urbanists often think of European cities as the kind of walking, biking, and transit paradises we might strive to emulate, in fact, many European cities have made significant progress in the last several decades to become more accessible to non-car travel themselves. A new study investigates these changes in Munich, Berlin, Hamburg, Vienna, and Zurich, showing how coordinated strategies of land use and transportation policy have significantly decreased car travel. The tactics will sound familiar: Parking management, dense mixed-use development, and high-quality transit and biking infrastructure.

3. The consulting firm Urban Spatial and Allan Mallach of the Center for Community Progress have published a massive trove of visualizations of neighborhood change in cities across the country. Among the indicators captured: residents with college degrees, median income, and housing prices. They also look at indicators that correlate with neighborhood change, and say they’re looking to build a predictive model of change.

The Week Observed is City Observatory’s weekly newsletter. Every Friday, we give you a quick review of the most important articles, blog posts, and scholarly research on American cities.

Our goal is to help you keep up with—and participate in—the ongoing debate about how to create prosperous, equitable, and livable cities, without having to wade through the hundreds of thousands of words produced on the subject every week by yourself.

If you have ideas for making The Week Observed better, we’d love to hear them! Let us know at,, or on Twitter at @cityobs.

The positive feedback loop of integration

Yesterday, we critiqued a study that claimed to show that the benefits of putting low-income housing in very low-income neighborhoods greatly exceeded the benefits of putting it in higher-income neighborhoods—especially higher-income and predominantly white neighborhoods—where it might have more of a pro-integration effect.

Among the several points of our critique was that the study severely under-measured the benefits of integration.* While its cost-benefit analysis only counted the income gains based on estimates from Raj Chetty et al’s work, we pointed out that there are many other benefits you might expect from integration: better mental health, school performance, safety from crime, and so on.

But there’s an even bigger issue here that goes beyond any of these discrete benefits. Which is: there is evidence that integration creates positive feedback loops that change the fundamental dynamics of neighborhood change.

Somerville, MA. Credit: Tim Sackton, Flickr
Somerville, MA. Credit: Tim Sackton, Flickr


After all, the study’s authors calculated that a major cost of putting low-income housing in higher-income neighborhoods was that the property values in those neighborhoods declined as a result—not, apparently, because of any problems the new housing caused, as crime did not increase, but simply because their neighbors preferred not to live around the kinds of people who live in low-income housing.

But in many ways, that effect—and those preferences—depend on a steady supply of neighborhoods without any low-income housing. In part, this is the sort of “prisoner’s dilemma” that we’ve talked about before: in a policy context in which segregation creates resource-rich winners and resource-poor losers, any hint that your neighborhood or municipality might be going towards the resource-poor loser end of the spectrum is cause for alarm. The issue isn’t one or two low-income buildings per se—it’s the possibility that once one or two come in, the segregating dynamics of the housing market will bring in so many more that the area will become very disproportionately low-income. And even where the issue is one or two buildings—because a given homeowner happens to just have discriminatory preferences—that homeowner can only act on their preferences and leave the neighborhood if there are other neighborhoods without any low-income housing for them to flee to.

But what if every neighborhood had some minimum level of low-income housing? What if there were a metropolitan area with a regional integration plan that eliminated the option of living in a totally segregated higher-income neighborhood, protected by exclusionary zoning and other anti-poor policies?

Well, we don’t know for sure, because no such metropolitan area really exists. But there are urban regions that have instituted integration policies for public schools. And the evidence from those is pretty encouraging.

Take this 2012 report from the University of Minnesota’s Institute on Metropolitan Opportunity. In it, Myron Orfield and Thomas Luce  look at the trajectories of suburban neighborhoods between 1980 and 2009—asking, for example, how likely it is that a mixed-race community will end up resegregating. The overall numbers are not great: if a Census tract was 23 percent or more people of color in 1980, it was more likely to resegregate than remain diverse by 2009. (The report defined “resegregate” as become less than 40 percent white. Obviously there’s no objective threshold, but the general pattern holds regardless of where you draw the line.) Interestingly—and similar to the results in our “Lost in Place” report—a vanishingly small number of integrated suburban neighborhoods resegregated as white.

Credit: Institute on Metropolitan Opportunity
Credit: Institute on Metropolitan Opportunity


But the report reran these same numbers for 15 metropolitan areas with regional school desegregation initiatives. In other words, these are places where the connection between the demographics of your neighborhood and the demographic of your public school was, to some extent, broken. If a new low-income housing project was built on your wealthy block, then, that wouldn’t necessarily change the demographics of your children’s school, because the desegregation initiative would have already introduced low-income students from other neighborhoods. And by the same token, you wouldn’t necessarily be able to “escape” lower-income students by moving to another neighborhood.

So what’s the result? Well, diverse neighborhoods in these metropolitan areas were much, much less likely to resegregate than similar neighborhoods in regions without school desegregation initiatives. Neighborhoods up to about 37 percent people of color were more likely to remain diverse than to resegregate—and even neighborhoods that were 50 percent people of color in 1980 were only slightly more likely to resegregate, as opposed to having a roughly 75 percent chance of resegregating in regions without school desegregation initiatives.

Credit: Institute on Metropolitan Opportunity
Credit: Institute on Metropolitan Opportunity


And this difference is associated just with ending the “prisoner’s dilemma” in schools, not neighborhoods. That is, even if white people are able to access segregated housing, they appear much less likely to want it if that housing won’t guarantee segregated schools. Imagine, then, what might be possible if segregated housing itself were much harder to come by.

* To be fair, the study’s authors acknowledged at one point that they were doing this. Nevertheless, they didn’t change their cost-benefit analysis or conclusions as a result, so our criticism stands.

The rising tide of economic segregation

Last week, we argued that the problem called “income segregation” is actually several problems, and broke it down with the help of different measurements designed to capture different aspects of the issue.

In particular, we pointed out the need to distinguish between 1) the segregation of poverty, 2) the segregation of affluence, and 3) the segregation of the middle—and 4) the difference between income inequality and income segregation per se.

And—although we’ll go into more detail on why economic segregation matters in a subsequent post—recall that we care enough to dive into this because both a large body of empirical research and on-the-ground experience suggests that economic integration has a major positive influence on economic opportunity, especially for people with low incomes.

Today, we’ll briefly cover how each of these problems has evolved in American cities—both over the long term, and more recently. As before, we’re largely working off of outstanding research by Kendra Bischoff and Sean Reardon, whose report is worth diving into if you want more details.

At the highest level, economic segregation trends are extremely easy to summarize: they’re up. American cities are far more segregated by income today than they were in 1970 by every measure we’re aware of, indicating more “secession of the wealthy,” more concentrated poverty, and even more sorting among the lower-middle and upper-middle income tiers.

Credit: Kendra Bischoff and Sean Reardon
Credit: Kendra Bischoff and Sean Reardon


In 1970, just 8.6 percent of families lived in “poor” neighborhoods (where median income is below 67 percent of the regional median), and 6.6 lived in “affluent” neighborhoods (where median income is more than 150 percent of the regional median). By 2012, those figures had both more than doubled, to 18.6 and 15.7 percent, respectively—meaning that over a third of all families lived in either poor or wealthy neighborhoods, as opposed to just over one in seven in 1970.

But it’s not just segregation of poverty and affluence that have increased substantially over the last few generations. Bischoff and Reardon’s “H” score—which, if you remember, measures segregation along the entire spectrum of income—has also increased, from 0.115 to 0.146. In today’s terms, a shift that size is the equivalent of moving from a metro area of average segregation almost all the way to one in the top ten percent of most segregated metro areas. In other words, it’s a lot.

And what about the question of inequality versus segregation? Is this increase in segregation simply a result of greater gaps in earnings between rich or poor, or are people actually being sorted to live with more people at their relative level? Well, recall that last time we said that the H index was valuable in part because it measures segregation by people’s income rank, rather than absolute level of income—meaning it filters out a lot of the effects of rising inequality. Since it has increased substantially, we can conclude that rising income segregation is not just a result of rising inequality. Rather, there are other factors at work promoting segregation—including some we’ve talked about at City Observatory, like zoning laws that prohibit a mix of housing types in the same neighborhood.

So while rising inequality is correlated with rising segregation, it’s not the whole story. In a previous report, in fact, Bischoff and Reardon found that inequality is mostly associated with the segregation of affluence, and less strongly correlated with the segregation of people with lower incomes.

So that’s how things have changed over the last 40 years. What about the last five?

In their most recent paper, Bischoff and Reardon focus on changes between 2007 and 2012. (For sticklers, these are actually averages of 5-year American Community Survey results from 2005-09 and 2010-2014). Over that period, income segregation has continued its rise, but the trends look somewhat different than they have over the longer term.

A neighborhood in Atlanta. Credit: Chris Yunker, Flickr
A neighborhood in Atlanta. Credit: Chris Yunker, Flickr


Over the last five years, the proportion of families in low- and high-income neighborhoods has continued to increase—but a more sophisticated look at the numbers suggest that’s more about changing income than actual segregation. Rather, Bischoff and Reardon show that most of the rise in income segregation between 2007 and 2012 came from the increasing segregation of lower-middle-income families (those between the 10th and 50th percentile of income) and upper-middle-income families (those between the 50th and 90th percentiles).

How much does that matter? It’s hard to say. Most of the research on the effects of neighborhood income on individuals’ economic, academic, or health outcomes have focused on very poor neighborhoods, or areas of concentrated poverty. But if this trend continues, it may also be worth further investigating the effects of segregation on lower-middle-income, or working class, neighborhoods as well.

Our next installment in this series will examine the relationship between income segregation, urban built form, and sprawl—and how better urban planning might mitigate some of the trends towards increasing segregation that we’ve just discussed.

Until then, you can look up how your region has fared on the interactive tool we’ve created below, based on data from Bischoff and Reardon.


Where should low-income housing go?

A new study has run the numbers, and has concluded that social welfare is optimized by putting affordable housing in very poor neighborhoods, rather than wealthier (and especially whiter) ones.

Authored by Rebecca Diamond and Timothy McQuade of the Stanford School of Business, the study really has two major conclusions. First, building affordable housing in very low-income neighborhoods creates major benefits for the surrounding area. Second, there are major social costs to placing affordable housing developments in higher-income neighborhoods, though they calculate that these costs are outweighed by the income benefits to the affordable housing residents.

While there are some valuable findings here, as you might imagine, we have a few issues.

For one, the kind of affordable housing the study looks at is targeted to people with incomes that are 50 percent or more higher than is typical in low-income neighborhoods. The paper looks at developments funded by the Low Income Housing Tax Credit, or LIHTC. LIHTC buildings generally target households making 60 percent of Area Median Income; the study’s authors estimate that in their sample, that averages about $40,000. By contrast, median income in the low-income neighborhoods they find benefit the most from new LIHTC buildings top out at about $26,000. And since $26,000 is the high end, most of these neighborhoods are actually poorer than that.

In other words, they’re talking about neighborhoods so poor that building low-income housing increases the community’s average income. That makes their finding that LIHTC projects increase housing prices by about 6 percent within 0.1 miles (yes, you read that right—more on the geographic range of the effects later) much less surprising. It also means that the finding is less about “affordable housing” per se and more specific to LIHTC, or other subsidies with similar income targets. It’s questionable whether the same results would hold for other kinds of subsidized housing targeted to lower-income households.

Second, it’s worth taking a second to underline exactly what the “costs” of LIHTC buildings are to higher-income neighborhoods. Diamond and McQuade find that LIHTC buildings don’t increase crime. And yet they also find that the average homeowner in such a neighborhood would pay nearly $4,000 to avoid having to live within 0.1 miles of a LIHTC building. But note that we said “homeowner”: renters appear to have no such preference. Even more curiously, this aversion to low-income housing only appears in higher-income neighborhoods with low Black and Latino populations.

What would create such a pattern? The authors have an idea. “If local residents have preferences over the demographics of their neighbors,” they write, “new in-migrants could make the neighborhood more or less desirable.” This may be the world’s politest way of saying “mostly white homeowners appear to be discriminating against Blacks, Latinos, and/or poor people.”

Now, that’s not necessarily a surprise: it confirms many years of research about how racism and the perception of the presence of lower-income people affect housing markets. But it raises a question that anyone in housing policy or urban planning needs to be able to answer: are preferences of advantaged groups for segregation—segregation that we know is harmful for lower-income people and people of color—just another legitimate interest that we need to weigh against the interests others might have in desegregation? As it happens, the authors estimated the gains of integration, and found that they outweighed the costs. But there’s no reason the numbers had to work out that way. If the model’s results had shown that the benefits of segregation to mostly white, mostly higher-income homeowners were greater than the costs to disproportionately Black and Latino lower-income households, would that mean they would have come out in favor of segregation?

Finally, the way the authors do try to quantify the benefits of integration is extremely limited. Their estimates are based on Raj Chetty et al’s findings about the increase in average lifetime earnings for low-income households in higher-income neighborhoods. Of course, we’re big fans of Chetty’s work, and we’ve cited it ourselves extensively. But it’s a huge mistake not to include other potential benefits in a cost-benefit analysis. Other studies, for example, have shown major improvements in mental health; you might also expect better educational outcomes, which arguably have value beyond simply their contribution to future income. There’s also the reduced likelihood of crime victimization; potentially shorter commutes; and so on. None of these are weighed in when the authors conclude that the benefits of building LIHTC in high-income areas are exceeded by the benefits of building in the very low-income neighborhoods we talked about earlier.

So what should we take away from all of this?

  1. Building LIHTC units in very poor neighborhoods may, in fact, be a kind of place-based development strategy with some payoffs, as reflected by rising home prices. But it’s not clear how far it goes as a broad strategy for revitalizing these neighborhoods. For one thing, the strongest gains are in a very small area—just 0.1 miles from the project—with quickly declining improvements beyond that. Moreover, just as their estimates of the benefits of integration are limited, so are their estimates of the costs of segregation. While they do find that LIHTC projects help lower crime, it’s not clear whether there are improvements on other indicators that residents are likely to care about beyond home prices: schools, local retail options, and so on.
  2. Are all preferences made equal? We can use home prices to quantify the preferences of homeowners—but that doesn’t mean we should weigh every kind of preference in the same way. It turns out that for many people, the presence of people of color or lower-income people is enough to cause them to value their homes less. Evaluating policy options always involves value judgments, and econometric models—while often helpful—are not a substitute.
  3. When we do use econometric models, we need to be aware of what’s being left out. This, in fact, is part of our value judgments, whether we’re aware of it or not. Do we value better mental health for low-income people? Do we value education beyond its income effects? Do we value giving people the option to live somewhere they otherwise couldn’t? If none of those are in the model, then we are effectively answering “no.”
  4. Leaving integration up to local governments is unlikely to be successful. This is a point we’ve made before, most recently, perhaps, in citing research by Michael Lens and Paavo Monkkonen that showed that metropolitan areas with more local power in development decisions are more segregated than ones where states play a bigger role. This study underlines that while many people in all sorts of neighborhoods value diversity and integration, some do not, and they are willing to pay thousands of dollars to avoid having low-income neighbors. (This may also be about the kind of “prisoner’s dilemma” of who might get “stuck with more than their fair share” of low-income housing. In fact, research suggests that when integration is widespread, the dynamics of neighborhood change are altered in ways that reduce the incentive for self-segregation fo the advantaged.) When those preferences are combined with hyper-local power over what kinds of housing gets built where, it’s inevitable that many jurisdictions and neighborhoods will create regulatory barriers to low-income people living in their communities: in other words, exclusionary zoning.

The Week Observed: May 6, 2016

What City Observatory did this week

1. At City Observatory, we’re interested in hard numbers—but we’re also interested in the human community and public spaces that cities can create. As we did in April with “Lost in Place,” on Monday we introduced an easy-to-share infographic of our report “Less in Common.” It summarizes many of the facts found in that report, including that the proportion of Americans interacting with their neighbors has declined by a third since 1970, and recreational activities like swimming have been radically privatized. Take a look—and because all City Observatory work is licensed under Creative Commons, you’re free to reuse the graphic in your own presentations or reports with attribution.

2. We often talk about “economic segregation” as if it were one issue—but in reality, that’s a catch-all term for all sorts of separate but related problems. Going off of work by Sean Reardon and Kendra Bischoff, we begin breaking down the concept by explaining some measures of segregation at the extremes: the percentage of families living in high-income or low-income neighborhoods. These measures are particularly relevant if you’re concerned about the effects of resource-hoarding wealthy neighborhoods, on the one hand, and opportunity-starved low-income neighborhoods on the other. But by breaking out these as separate ideas, we can see that some cities have issues with wealthy and low-income neighborhoods in roughly equal proportion, while in others, low-income neighborhoods are a much bigger problem.

3. Having introduced ways to understand and measure the segregation of the high- and low-income, we turn to the H index, a way of measuring income segregation along the entire spectrum, and distinguishing between segregation itself and inequality—different but related issues that may call for different policy approaches. While the High + Low scores suggest that New Orleans is much more segregated than Cincinnati, for example, the H index suggests the issue is actually that New Orleans is much more unequal.

4. One of the best parts of our work at City Observatory is seeing how people across the country apply our findings to their own cities and needs. We did a brief overview of the ways that people reacted and responded to our Storefront Index—from questions about whether businesses in skyways can have the same street-livening effects as traditional storefronts, to examining the inequality of retail and amenity density, to musing about how the Storefront Index could be a tool for planners to identify retail districts that need “room to grow” in the form of more commercial zoning. Let us know if you have your own uses, comments, or questions!

The week’s must reads

1. How do we know how to design the streets near your home, job, or school? Traffic studies, of course! But co.exist breaks down how those studies are themselves designed to almost always end up concluding that every street needs to be wider, faster, and less pedestrian-friendly. That, in turn, encourages more people to drive—so that the next traffic study finds roads need to be even more optimized for traffic, at the cost of safe, pleasant places to walk, bike, take transit, or just hang out.

2. NPR’s Planet Money podcast takes a look at Housing Choice Vouchers, one of the main forms of housing assistance to low-income households in the US. They end up asking some questions we’ve raised ourselves: Why do we make housing assistance a lottery, with many more qualifying households than available assistance, when other important kinds of help—like SNAP, or food stamps—are available to anyone with income below a certain level?

3. At Bloomberg, economist Noah Smith says that major American avenues to “extensive” economic growth, like globalization and the internet, have largely played out their potential for increasing productivity. The next round of US economic growth, he argues, will have to be “intensive”: “getting more output for a given unit of input.” And one way to do that is by urbanizing more. Research suggests that more densely settled metropolitan areas, all else equal, are more economically productive.

New knowledge

1. At City Observatory, we’ve been critical of the mortgage interest tax deduction, which effectively creates billions of dollars in subsidies that largely to go upper-income households. A new paper by Hal Martin of the Cleveland Federal Reserve and Andrew Hanson of Marquette University attempts to measure the effect on housing markets of different kinds of reforms to the MID. They find getting rid of the deduction would cause prices to fall as much as 13.5 percent in Washington, DC and as little as 3.5 percent in Miami. In contrast, swapping out the deduction for a 15 percent refundable credit—which could both reduce subsidies to upper-income households and make more assistance available to lower- and middle-income ones—would tend to increase housing prices—up to 12.1 percent in the Miami metropolitan area.

2. At CityLab, Richard Florida reviews two new studies on the growing economic strength of central cities. The first, by Daniel Hartley of the Chicago Federal Reserve, finds that neighborhoods closer to downtown—but not in downtown—saw faster job growth from 2009-11 than from 2002-2009, outpacing neighborhoods farther out on the periphery. (These results echo our findings in “Surging City Center Job Growth.”) Another, from the Initiative for a Competitive Inner City, finds that cities with stronger industry clusters saw more rapid job growth in those clusters from 2003-2011, underscoring the advantages of agglomeration.

3. Housing assistance can be a lifeline to low-income families, but does it offer benefits beyond immediate housing relief? A paper from the St. Louis Federal Reserve finds that, at least in some cases, the answer is yes: For every year that girls and women between the ages of 13 and 18 live in public housing, adult earnings increase by 9 percent; for each year that their families received housing vouchers, their adult earnings increase by 6 percent. Boys and men, on the other hand, see no such effects.

The Week Observed is City Observatory’s weekly newsletter. Every Friday, we give you a quick review of the most important articles, blog posts, and scholarly research on American cities.

Our goal is to help you keep up with—and participate in—the ongoing debate about how to create prosperous, equitable, and livable cities, without having to wade through the hundreds of thousands of words produced on the subject every week by yourself.

If you have ideas for making The Week Observed better, we’d love to hear them! Let us know at,, or on Twitter at @cityobs.

Income segregation along the whole spectrum

Yesterday, we introduced three kinds of economic segregation, and how you might measure each: the proportion of people in high-income neighborhoods; the proportion of people in low-income neighborhoods; and the proportion of people in either high- or low-income neighborhoods.

Each says something important about how people are sorted by income in a metropolitan area. But these measures also miss some things. For one, they don’t reflect how segregated people in the middle of the income spectrum are—whether working-class and upper-middle-class people live in the same neighborhoods, for example.

More subtly, but importantly, these measurements are also very sensitive to changes in inequality, even if segregation per se doesn’t change. Imagine that in your metropolitan area, you doubled the income of the richest 20 percent of families, and cut in half the income of the poorest 20 percent. All of a sudden, many more neighborhoods would meet the definitions of “rich” or “poor” according to these measurements, and so they would tell you that segregation had increased. But in fact, nobody actually moved—and the likelihood that, say, someone in the 10th percentile of income was living in the same neighborhood as someone in the 90th percentile didn’t change at all. What changed wasn’t segregation, but inequality.

Which brings us to…

The H index

So Riordan and Bischoff created an index, called H, that takes into account everyone. Unfortunately, unlike the previous three indicators, it doesn’t have an easy lay-person interpretation: it’s just a number, varying between 0 and 1, with larger numbers indicating more segregation. Basically, it works by ranking each family household by income across an entire metropolitan area, and then comparing the distributions of rankings in each neighborhood.

An important thing about H is that it is insensitive to changes in inequality. That is, because it depends on ranks and not actual incomes, if the top 10 percent of families doubled their income, or the bottom 10 percent of families cut their income in half, H would not change, even though there would be important implications for economic segregation. (Presumably we care more about high-end segregation, say, if wealthy people are really wealthy. The less different rich people are from everyone else, the less their separation matters.) In part, this is helpful: it means that an increase in H really tells us something about how people are being sorted into different neighborhoods, and not just a change in income inequality in general. But it also leaves out an important part of the story—just how different high-ranked families are from low-ranked families.

How does H differ from High + Low?

In reality, these two measures are highly correlated, as we might expect. But there are some notable differences. In both San Francisco and Milwaukee, for example, about 35 percent of families live in neighborhoods that are either low-income or high-income. But SF’s H index is 0.14, and Milwaukee’s is 0.19—a very significant jump from the least-segregated third of cities to the most-segregated third. That suggests much more sorting of relatively middle-income families in Milwaukee than San Francisco. It may also suggest that part of San Francisco’s bad showing on the High + Low score is about inequality, rather than segregation—which makes sense if we think that, say, there are many more very high-earning families in the Bay Area than in Milwaukee.

Similarly, Cincinnati and New Orleans score almost identically on the H index, at 0.15. But New Orleans has dramatically more people living in high- or low-income neighborhoods, 39 percent, versus 27 percent in Cincinnati. This is likely because New Orleans has more income inequality than Cincinnati.

In a way, you can think of the H index as a sort of pure description of segregation, while the “High + Low” score captures both segregation and inequality. While we presented that combination as a drawback at the top of this post, it might actually better reflect what many people have in mind when they think about the negative consequences of economic segregation. If harm is caused by extreme neighborhoods—both resource-hoarding rich neighborhoods and opportunity-scarce poor neighborhoods—then it matters how rich the rich are and how poor are the poor.

An increase in the H index might not directly translate to those sorts of ills—rather, it likely suggests that people are increasingly living among people with similar incomes to their own, whether or not they’re creating more rich or poor neighborhoods in the process. That distinction will turn out to be important in our next post about economic segregation: how it’s changing in America today.

How people are using the Storefront Index

For us at City Observatory, one of the most interesting (and fun) parts of our work comes after we’ve finished a Commentary or Report, and we get to watch others react and respond to its findings and arguments. “The Storefront Index,” the report on urban customer-facing business clusters that we released last month, is a great example.

At the Washington Post, Emily Badger focused on what the report had to say about cities as “centers of consumption,” and how a wide diversity of amenities and choices—from different types of clothes to niche bookstores and restaurants that cater both to immigrants from a particular country, province, or even city, as well as interested eaters from everywhere else—are key to the appeal of large urban areas. Badger also noted that the report could help show inequities in urban amenities, pointing out how spare businesses appeared in the Anacostia section of Washington, DC, compared to the rest of the metropolitan core.

Screen Shot 2016-05-04 at 3.43.01 PM


PlanPhilly, in turn, underscored the importance of zoning laws in shaping urban retail districts, arguing that the Storefront maps help “uncover where commercial corridors ‘want’ to expand, and where land use could be more supportive of this.” wondered what sort of effect Minneapolis’ famous (or infamous) skyways have on the relationship between “storefront” businesses and urban vitality. If a shop or restaurant is in a skyway, and inaccessible from the street, does that take away from a city’s public life—from Jacobs’ “sidewalk ballet”?

Skyways in downtown Minneapolis. Credit: photogreuphies, Flickr
Skyways in downtown Minneapolis. Credit: photogreuphies, Flickr


And saw the Storefront Index as evidence of Cleveland’s “downtown business boom.”

For many local outlets, the report was a chance to reflect on their city’s relative standing in terms of urban core retail outlets, whether that was good news (as in Portland, DC, and LA), or not so good news (as in Detroit and Orlando). And Houston’s Swamplot blog eschewed inter-city comparisons to look at which parts of their region were light on storefronts.

All in all, we’re pleased that so many people found the Storefront Index a useful and revealing tool for understanding their cities—and that many are already taking steps towards using it not just for diagnostics, but for its policy implications, from expanding commercial zoning in high-demand districts to questioning the importance of businesses on the street, and not separated in private spaces like skyways.

Have another reaction? Questions? Conclusions? Let us know.

There’s more than one kind of income segregation

Much of the conversation about urban inequality today—from Raj Chetty’s work on intergenerational economic mobility, to issues of concentrated poverty and gentrification—is framed in terms of economic segregation. But it turns out that “economic segregation” isn’t just one thing, and what we mean by the phrase, and how we choose to measure it, has serious implications both for our understanding of urban inequality and the kinds of policies we might design to fix it.

The basic issue is that unlike racial segregation, which has a few (ostensibly) discrete categories into which people fall, income segregation has to divide people based on a continuous spectrum with no obvious objective cutoffs, or even number of categories. Social scientists have come up with a number of different approaches to this problem; in this post, we’ll go through several of the most common and explain why they matter, with the goal of leaving you more able to engage in detailed, thoughtful conversations about inequality, segregation, and opportunity in your own city and beyond. (The examples will be based on work by Sean Riordan and Kendra Bischoff, whose papers on measuring economic segregation over the last several years have been excellent.)


High-income segregation

One approach is to measure how separate upper-income people are from everyone else. You might focus on this if you believe that, especially when the rich have a greater share of total income than they have in generations, what Robert Reich has called the “secession of the successful” threatens to keep an enormous share of society’s resources out of reach of everyone else. You can think of this as a sort of Mossack Fonseca problem: like offshoring wealth to avoid federal taxes, forming clusters of exclusive communities is a way of ensuring that money that might otherwise be used to pay for society-wide benefits will instead be spent disproportionately on the wealthy people themselves.

Riordan and Bischoff measure this by counting the proportion of people in a metropolitan area who live in neighborhoods where the median family income is more than 1.5 times higher than the median family income of the region as a whole. So, for example, the median family income in the Boston metro area is about $96,000 in the 2014 1-year American Community Survey; for that year, this measure would count the number of people living in neighborhoods where the median family income was at least $144,000. Because they use the median, and not the average, this will only capture neighborhoods where at least half of all families meet that threshold of disproportionate income; and because they measure families, it corrects for some of the differences between neighborhoods that result from age differences.

Low-income segregation

Another approach is to measure how separate low-income people are. You might focus on this version if you believe that the main threat from economic segregation is concentrated poverty; indeed, much of the research on economic segregation has focused on the problems associated with neighborhoods where a very large proportion of residents are low-income, including worse economic mobility, educational, and health outcomes.

Riordan and Bischoff’s measure for this mirrors their high-income segregation indicator: the proportion of people in a metropolitan area who live in a neighborhood where the median family income is at least 33 percent below the median income of the region as a whole. So, going back to Boston, this measure would count the number of people living in neighborhoods where the median family income is less than about $64,000.

High + Low

Perhaps you are interested in both of these aspects of economic segregation. An easy way to add them together is to…add them together. Another Riordan-Bischoff index is simply the proportion of people who live in high-income neighborhoods or low-income neighborhoods. This makes sense if you want to see how typical it is for someone to live in a community that is on some extreme, as opposed to being middle- or mixed-income. It makes for a good, quick, intuitive number that captures both high-end and low-end segregation.

On the other hand, it doesn’t necessarily tell you which of these is a problem, or in which proportions. Perhaps one city has a huge problem with concentrated poverty, while another’s issue is concentrated wealth. The policy response would not necessarily be the same to both.

How much of a difference does it make?

So how much do we gain by breaking down these different kinds of income segregation? Quite a bit, actually.


Comparing the prevalence of high-income and low-income neighborhoods by metro area, we can see quite a range of differences. In most cities, these numbers are roughly proportional. In Richmond, VA, 14 percent of families live in high-income neighborhoods, and 16 percent live in low-income neighborhoods. In San Diego, it’s 18 percent in high-income areas, and 21 percent in low-income areas. But in other cities, one side clearly dominates.

In some cities, many more people live in low-income neighborhoods than upper-income ones. In New Haven, for example, 16 percent of families live in high-income neighborhoods—but 25 percent live in low-income ones. In Milwaukee, it’s 13 percent and 22 percent; in Providence, it’s 10 percent and 20 percent. In these areas, concentrated poverty appears to be an even larger problem than in a typical metro area.

Interestingly, there really aren’t many cities where people in wealthy neighborhoods outnumber people in low-income neighborhoods. Partly, that reflects an argument we’ve been making for a while about the relative importance of the issues of gentrification by upper-income people and concentrated poverty. But it’s also, of course, a reflection of the income cutoffs chosen by Bischoff and Reardon. It might be useful if we had a fourth measurement—one that took into account the entire spectrum of income, and didn’t depend on arbitrary categorizations?

Tomorrow, we’ll introduce you to Bischoff and Reardon’s “H Index,” which does just that.

What it means to be in common

When we talk about the costs and consequences of car-dependent urban development, we often talk about hard economics and climate science. Spread-out neighborhoods divided by big, pedestrian-hostile roads force people to spend more on transportation than they would in a place where many trips could be taken by foot or transit. In high-demand cities, relatively lower-density development can lead to a “shortage of cities” that pushes housing prices up, encourages economic segregation, and leads to lower intergenerational economic mobility. And these urban forms are also highly correlated with more greenhouse gas emissions, worsening the threat of climate change.

But people also experience their neighborhoods as communities—as places where people gather, interact, and enrich each others’ lives. In our 2015 report “Less in Common,” we explored the ways in which increasing auto-centric development has degraded this aspect of our urban life. Now, as we did with our report “Lost in Place,” City Observatory and Brink Communication have put together an infographic to make these important ideas easy to share—and as always, this and all of our work is licensed under Creative Commons-Attribution, so feel free to incorporate it in your own presentations or reports.

The infographic illustrates many of the key findings of “Less in Common,” which illustrate ways in which increasing sprawl has weakened our communities, and show how a broader trend of Americans living more widely separated private lives has created a space for smart urban planning to strengthen the public realm.

Click to see the full infographic.
Click to see the full infographic.


Perhaps one of the clearest connections is in recreation: While Americans who went swimming in 1950 would probably go to a community pool, since then, the number of private, in-ground pools has increased from 2,500 to 5.2 million in 2009, as large-lot zoning and the construction of highways far into the suburban periphery has essentially subsidized the consumption of private land, at the expense of public facilities. These trends are mirrored in how we get around, relying more and more on cars cars as a mode of transportation, replacing walking and public transit—modes in which, outside a sealed, private machine, you might actually interact with neighbors or others. In fact, while about 30 percent of Americans reported spending time with their neighbors in 1970, that number was down to about 20 percent today.

This privatizing of public life has also encouraged further segregation of neighborhoods by economic status, a trend that has been well documented, and which we have explored at length at City Observatory.  Rich and poor Americans have become more spatially divided as we sort into high income and low income neighborhoods. While only 15 percent of Americans lived in rich or poor neighborhoods in 1970, by 2012, that figure was up to 34 percent.

The erosion of the civic commons also has a profound impact on economic opportunity: In regions with more economic segregation, children from low-income households are much less likely to be able to improve their income status as adults.

As the rapper Ice Cube told National Public Radio earlier this year, reflecting on the school integration policies of his childhood:

I liked it because I was being bused with a lot of my homies. So we was, like, all going out there, and then it was a lot of different neighborhoods. So it was, like, buses from all these different neighborhoods all converging on this white school. And it was kind of cool because we had a chance to see different things, different people, have different conversations, hear different music and just get a chance to see that the world was bigger than Compton, South Central or, you know, whatever. You know, so we had a chance to really kind of open our horizons…

In other words, the strength of our public spaces and institutions is crucial both for educational and economic opportunity, as well as expanding our sense of collective potential and identities. That’s something we should all be able to get behind.

Click here to see the full infographic.