Wednesday, January 4, 2017

Self Driving Cars and Cities

If there's one technology everyone thinks will impact cities in the future right now, it's self driving cars.  Now living in the Bay Area, I hear and see constant hype about these things, actually I've seen the real things, with Google logos, moving 10 mph by the Mountain View Caltrain station.

So how will they impact cities?  Will they repeat the automobile patterns of suburban development?

It doesn't matter.

Wanna know why?

Well, first, most technologies that really impact cities aren't foreseen years in advance.  Few engineering feats impacted cities in the early 19th century like the Cotton Gin.  It allowed New York to surpass Philadelphia before the Erie Canal was built, as Lower Manhattan became the premier port for transferring cotton from Southern plantations to New England mills.   Philly couldn't compete due to the cost of going up and down the Delaware River, not to mention the longer distance to New England mill towns than New York had.  The reason those New England towns, like Pawtucket and Fall River, grew was their position on waterfalls, which Philly also enjoyed, yet wasn't prepared to take advantage of because city leaders didn't see the impact coming.    And neither did anyone else.

The gin increased the country's cotton output by 100-fold from its invention in 1793 to 1825, and port traffic in New York soared as a result.   Cotton accounted for nearly half of New York's exports prior to the Civil War.  There was no Gartner or IDC report predicting hockey stick growth for Mr. Whitney's machine in the 1790s, but it had a massive impact on American cities.

There's another reason self-driving cars won't impact cities as greatly as everyone thinks.  People-driving Ubers are already doing it.  And like the Cotton Gin, no one saw them coming.   In DC, many of the top nightlife spots, like Georgetown, 14th Street, H Street NE, and Adams-Morgan, are not especially close to the Metro.  But no problem getting an Uber in these neighborhoods.   Much of SoMa in San Francisco is in a similar situation.

The reality is that you are far more likely to sit in the back seat of an Uber between now and 2030 than in the backseat (or passenger seat) of a self-driving car.  Why?  The replacement cycle for cars is long, and going up, with the average car lasting 12 years now, about 50% higher than 30 years ago. And all those embedded semiconductors that allow a car to drive itself also allow any car to distribute gas precisely through fuel injectors, to avoid drifting into adjacent lines, and to spark the ignition without a distributor.  This is why if you own a car built in the last 20 years, you're wasting money on any sort of "tune up".

The above is all the result of dramatically declining prices for embedded chips, which in turn have produced self-driving capabilities, and have captured the press's imagination much more than distributorless ignitions, because they sound a lot more exciting.  These cheap chips have also improved the reliability of cars, and extended their lives, making it possible for modern cars to last much longer than their 1970s ancestors which relied on carburetors, hydraulics and mechanical linkages to distribute fuel, spark the engine, and drive the car.   So are you going to trade in your reliable, electronically-controlled Lexus ES300 Hybrid, with 30 payments left, to get a more expensive self-driving version?

Self-driving cars are not a surprising development for cities the way the Cotton Gin was, and the underlying technology trends supporting them are already at work in existing cars.  On top of that, people-driven Ubers are already providing an efficient way to get around cities that no one saw coming ten years ago, and have provided an alternative to public transport.   Just ask anyone who tried to catch a cab in the Mission, Adams-Morgan, Belltown, or the Back Bay ten years ago,

Monday, January 2, 2017

Close Correlation Between City Growth and MSA Growth this Decade

While there is all kind of debate regarding city vs. suburb growth, this decade has seen metro health increasingly tied to core city health.  The correlation between population growth in core cities and their respective MSAs for the 2010's is .90, which is exceptionally high.   This holds up for the Sunbelt, the West, the Northeast, and the Midwest, though small population changes there distort some of the city growth/MSA growth ratios.

From 1950 to 2000, the two were often inversely correlated, with many cities declining in growing metros.   But this trend has reversed, with one of the most impacted cities of the post-war period, DC, now growing more than 1.5x faster than its suburbs.   Overall, most cities are now growing between .8x and 1.5x the rate of their MSAs.

Individual metro growth rates are notorious for moving up and down, as regional economic strength ebbs and flows, or is impacted by the heavy presence of particular industries.  But as regions become more strongly linked to their core cities, they could increasingly become dependent on further strengthening those cities.  No one 30 years ago would have predicted such a strong resurgence for DC, especially during a period of sequestration and defense cuts.

Another important consideration is that many of the fastest growing cities aren't cheap.  Seattle is now growing faster than Atlanta, at both the city and MSA levels, in spite of median home prices that are twice as high.  Similarly, Denver is growing faster than Phoenix in spite of home prices that are about 50% higher.  

While regions are notoriously poor at coordinating economic development across jurisdictions, suburbs appear to becoming more closely tied to their core cities, and could likely benefit from boosting them further.

While correlation does not prove causation, this is one hell of a correlation.

Core City % Growth 2010-2015
MSA % Growth 2010-2015
City/MSA Growth Ratio
New York
San Diego
San Francisco
Las Vegas
St. Louis

Man Made Locations for Cities

In the last post, I discussed the low densities of American cities below the 36'30" line of latitude. One factor for this is their locations.   Look at a map, San Francisco looks like it should be the spot for a large city, same with New York, Boston, even Charleston.  Moving away from the coasts, the river and water intersections by Cincinnati, Cleveland, Chicago, Pittsburgh, Buffalo, Milwaukee, New Orleans, St. Louis, Seattle, and Portland also make decent sized cities in those locations seem very reasonable.  And then considering the importance of water power to early manufacturing and milling, you can understand how cities developed near waterfalls at Minneapolis, Baltimore, Philadelphia, and Richmond.

Made by Man or by Mickey Mouse
So you had oceans, water intersections, and fall lines account for much of this country's urban growth, especially in its Northern half.  So what happened to the south?  Early San Francisco investors were concerned that San Diego would be their main in state rival, and overlooked LA, why did they miss it? Why did Atlanta, only the fourth largest city in its state before the Civil War, become the largest metro in the Southeast?  Its downtown isn't even on the river!  Charlotte also has the same odd relationship with its river, why did it get so large?  Also, why the hell is there a city where Orlando is?  Shouldn't Tampa Bay be much larger?  Is Mickey really that powerful?

The technologies that defined American city growth up to about 1850 required certain physical traits. Cities had to accommodate the costs of making and moving goods with clipper ships, cotton gins, river barges, and the like.  But as time has gone on, technology has advanced to the point that topography and proximity to water have mattered less.  Essentially, technology has allowed large cities to form at what are essentially man made locations.

In the 1840s, Milwaukee grew over 10x, faster than Chicago, which was booming, but at a slower pace.   Chicago ended the decade with a little under 30,000 people, Milwaukee a little more than 20,000.  Both were still smaller than Cincinnati, Pittsburgh, and St. Louis, but were growing rapidly due to their locations on the Western side of the Great Lakes.    Neither had rail connections to the east, and they could have ended up close competitors.  But the 1850s were to rail construction what the 1990s were to fiber optics and Internet infrastructure, and by 1860, Chicago was sailing past Milwaukee.   Rail essentially reduced the impact of the city's natural setting, and in the South, it allowed less-than-obvious places, like Atlanta and Charlotte, to surpass fall line cities like Augusta and Columbus,  GA, ocean ports like Charleston and Savannah, and ultimately river cities like Memphis and New Orleans.

Ships to the North, Oranges to the South
This move away from natural settings went a step further as technology advanced.   Look at a map of Southern California.   The port by Coronado Island, San Diego, looks like a much more obvious place for a large city than an area of land 15 miles from any ocean, and adjacent to a river that had to be placed on top of concrete by the Army Corps of Engineers, in order to stop it from flooding in the winter, even though it ran dry in the summer.  But that place, LA, grew way beyond its peer to the south.

LA's location was chosen by Mexican farmers who needed access to the river, and it was setup as a small agricultural community, not a booming Ocean port like San Francisco.  And LA's original industry wasn't movies, Disney, or oil, but oranges.  This was the same industry that launched another city in an odd spot for a city, Orlando.   The underlying technology, refrigerated boxcars, allowed these oranges to be shipped to population centers across the country.   These refrigerated boxcars, in turn, encouraged more investment in rail links to these cities, pushing their growth even higher.  In 1890, 20 years before the first movie studio opened, and two years before the discovery of oil in Southern California, Los Angeles was already up to 50,000 people, 3x the size of nearby San Diego, which had been expected to become Southern California's answer to San Francisco.

From Refrigerated Boxcars to the Disney Monorail
Orlando never grew like LA did, but the mere fact that oranges and refrigerated boxcars allowed a city to pop up in such an odd place did impact Tampa, just as LA drew people away from San Diego. Tampa was a dense city in 1950, with over 6,500 ppsm, and without a growing city two hours east, could have blossomed into a much larger area.  But regardless of what happened to Tampa, the location of Orlando, especially after Mickey and Goofy arrived, could barely be described as a natural setting for a city.

If you look at most growing regions and cities today, from Charlotte to Phoenix to Atlanta to Dallas to Orlando, they rely very little on water or their natural setting for growth, unlike New York, Boston or San Francisco in their boomtown days.   It's easy to say this is all because of cars or something Congress did, but when people chose to haul Citrus to the Midwest or to haul the kids to Disney, I doubt they wrote the House of Representatives.

Sunday, January 1, 2017

Why are There So Many Low Density Cities Below 36'30"?

Have you ever wondered why so many cities in the South and Southwest have such low densities?  Is it history?  The heat?  Agricultural tradition?  Growth during times of car domination?

I've been looking into this a bit more to figure some of this out.

Low Density and Low Latitudes
If you look at the core cities within the top 50 MSAs, only 19 have densities over 5,000 people per square mile.   And of these, just two, LA and Miami, are below the 36'30" line of latitude.  19 others are below this line and have densities under 5,000 ppsm.  Meanwhile, among the top 50 MSAs north of the 40th parallel, 11 of 14, or nearly 80%, have core cities with densities over 5,000.  Chart below shows this, and please don't ask why I use airport code abbreviations.

Amount Latitude Core City Density Cities Represented over > 5000 ppsm in 1950
ranked 1 to 50 and North of 40th Metros with Core City > 5000 people per sq mi LGA, BOS, PVD, BDL, BUF, PIT, CLE, MDW, MKE, MSP, SEA LGA, BOS, PVD, BDL, BUF, PIT, CLE, MDW, MKE, MSP, SEA
ranked 1 to 50 and North of 40th Metros with Core Under 5000 people per sq mi PDX, DTW, SLC PDX, DTW
Between 36’30” and 40 Metros with Core City > 5000 people per sq mi DCA, BWI, PHL, STL, SFO, SJC DCA, BWI, PHL, STL, SFO
Between 36’30” and 40 Metros with Core Under 5000 people per sq mi RIC, ORF, CVG, SDF, MCI, DEN, SMF, IND, CMH RIC, ORF, CVG, SDF, MCI, DEN, SMF, IND, CMH
ranked 1 to 50 and South of 36’30” Metros with Core City > 5000 people per sq mi LAX, MIA MIA
ranked 1 to 50 and South of 36’30" Metros with Core Under 5000 people per sq mi RDU, CLT, ATL, BHM, TPA, MCO, JAX, MSY, BNA, MEM, OKC, DFW, HOU, AUS, SAT, PHX, SAN, LAS, ONT ATL, TPA, BNA, PHX, JAX

Now why is the 36'30" line so important?

As part of the Missouri Compromise, no new slave state could be admitted above this line.   It was also the existing border between Virginia/North Carolina and Kentucky/Tennessee when the law was passed in 1821.   But even then, just three of the country's 50 largest cities (Charleston, Savannah, and New Orleans) were South of 36-30.  And this did nothing to stop growth in existing slave states, where Memphis shot up into the top 50 by 1860, having only been incorporated in 1826, while Nashville and the Georgia fall line cities of Augusta and Columbus grew into the top 100.  (Atlanta did too, but it was still smaller than those two in 1860)

If we jump ahead to 1950, 41 of the 50 largest cities in the country had densities over 5,000 ppsm (aka people per square mile), far greater than the 19 of top 50 core MSA cities that have these densities today.   But the nine that were under 5,000 ppsm, were all south of the 36'30".   1950 was a peak population year for many traditional cities, and the top 50 MSAs still had over 50% of their aggregation populations living in core cities.   And this was before widespread air conditioning and suburbanization.   It was also two years before Atlanta nearly tripled its land by annexing Buckhead and much of the western half of today's city, pushing its density back under 5,000.

Annexing Cities vs. Annexing Farms
Annexation below the 36'30" was very different than it had been in the North.  When Philly annexed the Northern Liberties and Southwark in 1854, they were already existing cities in their own right. Same when Boston picked up Charlestown, Roxbury, and Brighton, and when New York added Brooklyn.    Densities of the core jurisdictions kept rising.   But Memphis, Charleston, Savannah, Nashville, Jacksonville, Phoenix, and Atlanta added rural places where new housing was planned, but often not yet built, leading to major reductions in densities.  All seven cities were over 5,000 ppsm at some point before 1960, but haven't gotten back to that level since.

Annexation of adjacent rural areas has been far less common above 36'30", occurring primarily in Columbus, Indianapolis, and Louisville.   One technique used to do this, county-city consolidation, first occurred through a de-annexation, when San Francisco County was broken off of San Mateo County in 1856.

One of the distinguishing features of these low density Southern cities was that they had no adjacent cities or towns to absorb like their larger Northern counterparts.   There was no equivalent to Minneapolis-St. Anthony, Boston-Charlestown, Manhattan-Brooklyn, Cleveland-Ohio City, Washington-Georgetown, or even Seattle-Ballard.   One reason is that the Northern cities were built directly on river banks, with different cities on each side, or as a string of towns near an ocean or a lake, each with their own port.  Moreover, these places often had to come together for survival. Beyond just the cities it annexed, Boston had a long competition with nearby Salem for international shipping traffic.   It needed to get bigger to finance port expansion.  Minneapolis and St. Anthony fought for years across the Mississippi's only waterfall, and were forced to merge to help preserve it.   Portland had to surpass nearby Oregon City, while Cleveland and Ohio City fought for dominance on the Cuyahoga before coming together.

Not all Northern annexations were about regional battles.  Chicago acquired Lake View Township (today's Lincoln Park and North Side neighborhoods) because its residents wanted better sewer services.  And they joined in 1889 with a density of about 5,000 ppsm.  Meanwhile, when Atlanta acquired Buckhead and its Northwest neighborhoods 63 years later, it wasn't because voters wanted services, but the city wanted to add wealth, and the impacted citizens were told they had to join by the state.  Moreover, this added 80 square miles of land with barely more than 1,000 ppsm to the city.

The high capital cost of sewer and water hookups remains a primary driver for annexation today, and has driven mergers of rural areas throughout the southern half of the country since WW2.   Moreover, there is a strong history of urban density in low latitudes by the ocean, as LA and Miami have shown, though LA certainly wasn't that way 50 years ago.   In 1910, prior to major annexations, Charleston and Savannah had over 10,000 ppsm.   Memphis and Nashville were over 5,000 ppsm at this time as well.  But most of today's major Southern and Southwestern metros are not on the Ocean nor on a river with heavy barge traffic, and in the cases of Atlanta and Charlotte, their downtowns aren't even on the river.   Where would you see such a thing in the North?

Cities above 36'30", even those in the South, like Richmond and Norfolk, were built on fall lines or ocean ports, much like most Northern cities.   But you look at a map, and you can see a city could have formed around the peninsula that's Charleston, the area between Lake Pontchartrain and the Mississippi that's New Orleans, or even along the fall line at Augusta, GA.     But Orlando?  Really, Mickey Mouse is that powerful?  And then what about San Diego?  Shouldn't it be larger than LA, where downtown is 15 miles from the port?

More on this in an upcoming post.