What impact does city size have on airport use?


Plane lands at SFO (photo via flickr user Dahlstrom)

On a recent trip to New Orleans, I wondered how city size correlated with the busyness (or lack thereof) of that city’s airport. How strong is the population-to-passenger correlation and how much do other factors, like being a hub and/or being a vacation destination, impact this correlation?

To answer this question, I first compiled a list of the 20 most populous Metropolitan Statistical Areas (MSAs) (source). I modified this list to merge MSAs that overlap and/or share airports (LA + Riverside; Washington + Baltimore; and San Francisco + San Jose). I also expanded the list to include a few smaller MSAs that I thought would be interesting to look at, like Las Vegas and New Orleans.

Next, I compiled airport usage statistics for the 60 busiest airports in the US and matched them with their respective MSA (source). Specifically, I used passenger boardings in 2017 (Chicago Midway and O’Hare and Honolulu’s airport only had 2016 data available). For some cities, there is only one major airport (e.g. Boston with Logan Airport) but for others, there are multiple (e.g. New York’s EWR, LGA, and JFK). Some cities have other, smaller airports that were not included because they were not on the top 60 list and therefore were statistically less relevant.

Finally, I performed a simple linear regression in Excel and added a trend line. As the chart (and R square of 66%) show, MSA size helps explain much, but not all, of the variation in airport usage.

Cities that overperform the most, relative to their population, include Atlanta, Chicago, Miami, San Francisco, Dallas, and Denver. Cities that underperform the most include LA, Philadelphia, New Orleans, St. Louis, Tampa, San Diego, and Portland.

St. Louis’ MSA is roughly the same size as Denver’s (~2.8m) but has less than 25% the passenger boardings! Meanwhile, Denver is half the size of Houston but has more boardings!


An important variable might be whether a city has a hub airport (or multiple).  However, most of the MSAs profiled, and nearly all of the largest ones, have at least 1 hub of the “Big Four” largest US airlines. These US airlines – American, Delta, Southwest, and United –account for the vast majority of domestic travel (~80%). Boston is the largest MSA that isn’t a Big Four hub (but it is a hub for JetBlue – the sixth largest US carrier) which may help explain why it underperforms relative to population. Every other MSA on the list has at least one hub except San Diego, Tampa, St. Louis, Honolulu, Portland, and New Orleans. Obviously, some hubs are more critical spokes than others (e.g., Delta’s Atlanta hub is a far bigger centerpiece in their network than American’s DCA hub is), but hub-status alone is an insufficient explainer of the remaining variation in the data.

Other factors like (new) business activity, population growth, and tourism should be considered, however on the last point, I was somewhat surprised by how several popular tourist destinations underperformed (e.g. Honolulu, New Orleans, Tampa, and San Diego).

Miami and Philadelphia are interesting to compare and contrast. Although they both serve as hubs and have a similar MSA population (~6 million), Miami has nearly three times the number of air passenger boardings (Miami includes West Palm Beach and Fort Lauderdale airports). Miami’s status as a vacation destination and gateway to Latin America probably play an important role as does Philadelphia’s sluggish recent population growth and close proximity to two larger/busier MSAs, Washington and New York.

Rental Cost Burdens in Major US Cities

Housing affordability is, and probably always will be, a hot topic. There are a number of ways to measure affordability but one of the most popular is by measuring “rental burden”. This measure, estimated yearly by the US Census’s American Community Survey, looks at gross rent as a percentage of renting households’ income over the last 12 months.

With a rule of thumb that no more than 30% of your income should go to housing expenses, there are two categories of cost burden:

  • Moderately-cost burdened households spend between 30% and 49.9% of their income on rent.
  • Severely-cost burdened households spend over 50% of their income on rent.

I decided to look at rent burdens in 2015 (the latest year with data) across the 10 most populous US cities, plus my hometown of Washington, DC and famously expensive San Francisco.

What does rent burden look like in these cities?

  • Of the cities reviewed, the city with the lowest proportion of rental cost burdened households was San Francisco (37%)!!! This number surely comes as a surprise to many however, if we consider a few characteristics of SF, this makes more sense. Chiefly, many lower income residents (more likely to be cost burdened) were priced out a while ago, SF has a high number of rent controlled properties, and although SF is famously expensive to rent in it also has a very high average income.
  • The next lowest cost burdens were in Washington, DC (44%) and in San Antonio, TX and Dallas, TX (each with 45%). In DC, despite high living costs, average incomes are also high. Conversely, in the Texan cities, average incomes are significantly lower but so are housing costs.
  • Over half of renting households in New York, San Jose, Philadelphia, San Diego, and Los Angeles are at least moderately burdened. A whopping 58% of renting households in Los Angeles pay at least 30% of their income in rent.

2015 rent burden by city Continue reading

Tax Increment Financing: A Quick Primer

Cities seeking to spur development in underinvested areas have a number of tools at their disposal to entice developers. These tools include zoning changes, density bonuses, improved infrastructure, tax-related incentives, and more. All carry their own benefits, risks, and drawbacks, but within the broad tax category is a particularly interesting approach called tax increment financing (TIF).

How does it operate?

TIF has been succinctly defined as “a targeted development finance tool that captures the future value of an improved property to pay for the current costs of those improvements.” It does this by:

  1. Creating a TIF district – a TIF district (or authority) is created composed of the property/properties to be developed as well as surrounding lots
  2. Freezing assessed value within the TIF district – the assessed taxable value of property within the district is locked or frozen at pre-development level. This base tax level will continue to flow to the local or state tax authority for the remainder of the district’s life (e.g. 20 years).
  3. TIFDiverting additional value to the TIF – any additional taxable value that the district enjoys during the term of the TIF district is diverted to that authority.
  4. Paying for improvement through the TIF – that new revenue is used to fund improvements within the district, including those the district (or another authority) may have paid for at the start (outright or through a bond) to catalyze the growth.
  5. Benefiting the state/city budget after TIF term ends – After the TIF district concludes, the full taxable value of the property within the district can be taxed by the traditional state/local body.

What can TIF money be used for?

TIF funds are used to finance activities or pay off debt for costs related to “public infrastructure, land acquisition, demolition, utilities, planning, and more. TIF funds have also been used to help support community amenities such as parks, recreational facilities, schools, and network infrastructure.” (source) Continue reading

What makes a region, a region?

A frequent complaint of Consolidated Statistical Areas (CSAs) – a Census description explored in an earlier blog – is that they are so large they have little practical meaning. While “cities” have important political and socio-cultural identity implications, and metropolitan areas are especially important for broader economic reasons (demand and supply) as well as cultural ones (e.g. “I’m from the Bay area“), CSAs are frequently derided as pointless.

One commonly cited exception is the Bay Area. Technically, there are two metropolitan statistical areas (MSAs) in the Bay Area: San Francisco/Oakland and San Jose (Silicon Valley). The Census combines the two MSAs into a single CSA and most people seem to think that’s appropriate because they have well-integrated infrastructure, overlapping labor markets, regional coordination, and a shared identity.

Critics say practically all other CSAs, fall into two easily dismissed categories:

  • Meaningless because they merely add a few far-flung communities in the metropolitan hinterland (e.g. Chicago’s CSA is only 4% bigger than its MSA)
  • Meaningless because they merge two completely distinct MSAs that shouldn’t be merged (e.g. Washington and Baltimore).

But let’s unpack that second category by comparing the Bay Area CSA to the Washington-Baltimore CSA on a few metrics that one might reasonably assume are CSA qualifiers:

Criterion Bay Area CSA

(pop. 8,751,807)

Balto.-Wash. CSA

(pop. 9,665,892)

Adjacent and geographic-ally consolidated MSAs are adjacent.

Combined area: 10,135 sq. miles

Distance from SF to San Jose: 48.4 miles.

MSAs are adjacent.

Combined area: 11,954 sq. miles

Distance from Baltimore to DC: 38.3 miles

Both CSAs sprawl but their core areas are relatively adjacent and much more compact.
Integrated transportation infrastruc-ture They don’t directly share an airport although SFO is the undisputed busiest within the CSA, drawing San Jose users. The commuter rail network, CalTrain is frequent and includes off-peak bi-directional trains. A BART  (San Francisco’s suburban heavy rail network) extension will reach San Jose in the next few years. Frequent commuter buses between the two. In addition to sharing an airport (Baltimore-Washington International is the busiest airport within the two MSAs), they have an integrated transportation network with MARC providing frequent commuter rail and off-peak bi-directional trains. Frequent commuter buses between the two. The transit cards of each system is usable on the other. Both CSAs have internally linked and mutually dependent transportation networks.
Integrated labor markets (source: commuting data from American Community Survey) About 8.7% of workers in the CSA live in one MSA and work in the other (total of 253,700 workers). About 5.5% of workers in the CSA live in one MSA and work in the other (total of 234,895 workers). For each CSA, about a quarter million people live in one MSA but commute to the other every day, representing an overlapping labor market.
Regional Coordination Leaders from both MSAs have cooperated in the past on policy goals of mutual interest (e.g. housing), and there have been calls for improved business coordination. However, no permanent forum exists for regional cooperation between the two. In addition to transit coordination (Maryland in one partner in DC’s mass transit system), various partnerships occur across the two MSAs. For example, the DC and Baltimore MSAs submitted a joint bid for the Olympics. No permanent forum exists for regional cooperation between the two. No large region in the U.S. really does this well but both CSAs exhibit some degrees of coordination, perhaps more so in the Balto-Wash CSA where the State of Maryland has at times helped convene.
Social identity There is a strong pop cultural recognition of a single “Bay Area”, typically tied to features such as mild pleasant weather, expensive real estate, and the thriving tech industry. However, there is strong variation within the CSA, within distinct communities and cultures. Baltimore and Washington developed over a couple centuries as neighboring but distinct cities. Over the years, this divide, between white collar Washington and blue collar Baltimore has faded, and there has been cultural cross-pollination throughout, but there remain lingering differences in perception and reality. This is a very subjective measure but arguably one of the most important when deciding what feels like an “area.” While there is some common identity between Baltimore and DC (especially among Marylanders), the Bay Area certainly has a more clearly integrated identity, at least in popular perception.


Highlights from the American Housing Survey


Source: Flickr User Pasa47

Earlier this month, the US Census Bureau released statistics from its 2015 Housing Survey. The survey provides incredible info about the characteristics of occupied housing units in 25 large metropolitan areas and the US as a whole.

There isn’t too much “new” in this latest survey, but it’s given me an opportunity to revisit the overall data.I thought I’d highlight some interesting stats that demonstrate the differences between two metro areas particularly important to me (DC and NYC) and the nation as a whole (see table below).

  • The most common housing type in the DC and NYC metro areas is the same as for the country as a whole – the ubiquitous single family detached home.
  • The second most common housing type shows each area’s uniqueness. For DC, it’s the familiar townhouse and for NYC it’s large buildings with over 50 units.
  • While nearly 6% of housing units in the US are mobile/trailers, only 0.5% of housing units in the DC or NYC areas are.
  • The busiest decade for housing construction was the 1950s for NYC, the 1970s for the US, and the 2000s for DC. Perhaps as a result of this age, fully 71% of units in NYC require step(s) to access (it’s about 55% for DC and the US).
  • While a third of all US housing units are in 1-story structures, only 6% and 5% are in DC and NYC, respectively.


Source: Author, via 2015 American Housing Survey (Census Bureau), Accessed 1/25/2017

Note: While I say “DC” and “NYC”, the data is looking at their metropolitan areas, not just the city proper. 

Does Gentrification Cause Displacement?

Gentrification is a heated word. It hangs ominously over every conversation on the return of investment and residents to the urban cores of many of our metropolitan areas over the past decades. Ominous because it represents the reverse of the coin of prosperity. The downside. The negative impacts. It’s dark and something vilified. A candidate for mayor of San Francisco in 1999 went so far as to declare “war on any and all gentrification.”

But why is it so often and easily criticized? Gentrification is a scapegoat for numerous processes, some recent and others decades-old, that better deserve the ire of the urban poor and their allies. Indeed, anger towards gentrification itself is often misplaced and obscures these real concerns. Perhaps none is greater than “displacement”. That gentrification causes displacement is commonly believed, yet beyond anecdotal evidence, is it true?

Continue reading

MSA vs. CSA: What’s in a Name?

A recent post of mine, pointed out that by 2018, the Washington-Baltimore Combined Statistical Area (CSA) would likely have more people than the Chicago-Naperville CSA. It elicited several responses including that the analyses was pointless because CSA is a meaningless measure – what matters are MSAs (Metropolitan Statistical Areas).

This critique is not new but it reminded me that many are skeptical of CSAs and that how we define (and measure) an integrated urban area, and compare it against others, is both important and contentious.


Continue reading

Average Station Use of (Some) Major Subway Systems


Crowded subway platform in New York City. Source: Flickr user romankphoto

I often travel to New York and, like many urbanists, find myself in unreserved awe at how busy the subway system is and how truly integral the subway is to daily life. While my city’s system (the Washington Metro) is the second busiest heavy rail (aka “subway”) system in the country, the number of people it carries isn’t even close to New York’s: the NY subway transports between 10-11 times more people per year than DC’s metro. Indeed, the NY subway carries 70% of all heavy rail trips in the US!

However, I always had an inkling that on a per-station basis, the stats are a bit closer. New York has an incredible number of stations: 422. While this is undeniably a strength of the system, it also means many stations are small and lightly used.

Continue reading

Wash-Balt area could become nation’s third most populous by 2018


Washington (left) may soon replace Chicago (right) as the third most populous US CSA. Source: DC: flickr user taedc; Chicago: flickr user gravitywave

If recent population growth trends hold steady (see table), the fast growing Washington-Baltimore Combined Statistical Area (CSA), is on track to become the US’s third most populous, passing the Chicago area CSA in 2018.

CSA growth table

If this comes to pass, as seems likely, it would be noteworthy and reflective of the  shift in relative fortunes of the two regions.

Chicago suffers from two ‘curses’, its location in the midwest where most major cities are struggling to grow, and the overall slow growth of large established cities (e.g. New York is also growing slowly – about half as fast as Washington). This could have impacts on Chicago’s psyche: the area has long been the second or third most populous in the country. It is incumbent upon the region’s leaders to take a hard look at what’s working and what isn’t and to explore growth strategies that are inclusive and provide broad-based opportunities far from the shining lights of Chicago’s downtown loop.

On the other hand, the Washington region has a unique opportunity to capitalize on this growth by forging regional strategies that can help sustain it. The CSA continues to diversify and is strong in many attributes that are key to many knowledge economy sectors (e.g. biotech, high-value added professional services, etc.) but stronger links are needed within the two metropolitan areas (Washington and Baltimore) as well as between them to ensure that growth strategies are harmonized. Continue reading

Venture Capital: Relative Distribution Across US Cities

Richard Florida and Karen King at the Martin Prosperity Institute published a report called Spiky Venture Capital, showing how venture capital (VC) clustered around a few metropolitan areas. In fact, they found:

Venture capital investment is concentrated in three broad clusters which account for more than 80 percent of all investment: the San Francisco Bay Area, which spans San Francisco, San Jose, and several smaller metros; the Boston-New York-Washington, D.C. Corridor; and Southern California, spanning Los Angeles, San Diego, Santa Barbara, and Orange County.

For policymakers, the relative clustering of VC also matters. In other words, how much more (or less) VC is attracted to a certain metropolitan area than you would otherwise expect based on its population. For instance, if a metro area attracted 20% of the US’s VC but had only 5% of the US’s population, you could say it outperforms expectations (based on population) by four times.

Drawing from Florida and King’s study, I looked at the 20 largest metropolitan areas for VC, ranked by their share of US VC and compared this to their share of US population (based on 2015 Census estimates):

VC relative

This table is color coded, with cities who underperform VC attraction relative to their population are in red, cities that attract VC 1 to 3 times what their population size alone would predict are in yellow, and cities that attract VC >3 times what their population proportion would expect are in green.

Green cities (super performers) are the “usual suspects” of VC: San Jose (Silicon Valley) outperforming its population by 23.5 times, San Francisco (17.4 times), Boston (6.4 times) and Santa Barbara, CA (5.4 times). Moderate out-performers include some of the country’s largest metropolitan areas, including NYC, LA, and DC, as well as relative start-up hubs like San Diego, Seattle, Austin, Denver, and Raleigh, NC. Some of the under-performers are perhaps unsurprising (e.g. Houston) but for others, like Philadelphia and Chicago, this should be a major wake up call.