Rambling About Cities and Transportation Costs

A few weeks ago I visited a client's offices, which are within walking distance of my own office.  However, I was running late, and I chose to take a taxi.  Big mistake!  It ended up taking me 40 minutes to go 1.25 miles, a distance I could easily have covered in 25 minutes walking.  (Walking is slow because of stoplights etc.  For this particular trip, there is no subway that could meaningfully reduce travel time relative to walking the whole way.)

This got me thinking.  Big, dense cities seem to have the following characteristic:  people are simultaneously slowed down and pushed closer together, with ambiguous effects on travel time.  This has interesting consequences for transportation in a city.

The biggest winner from the strange "physics" of city travel is grade-separated mass transit.  Subways are not that fast in absolute terms.  They also have the disadvantage that their travel route is inflexible—I have to walk about 10 minutes at each end of my commute.  But subways have the crucial advantage that they are about as fast during rush hour as they are during off-peak hours.  Cars are much faster during off-peak hours, and they can take you directly from point to point.  But these advantages are reduced or eliminated by (A) rush-hour traffic, and (B) reduced distances.  (So for instance, late at night I sometimes take a taxi home, spending about $25 but saving about 25 minutes in travel time.  When traffic is light, cars are incredibly fast relative to subways.)

The other big consequence of "city physics" is that land values are higher and more dependent on location.  Living a mile further from your job is not a big deal in a small city where it means perhaps an extra 5 minutes and 25 cents in gasoline per day.  But living a mile further from the subway station is a very big deal—that's maybe an extra 25-30 minutes of commuting per day, sometimes in very unpleasant conditions.  The effects are much larger for retail businesses—being a mile further from your customers would essentially destroy a lot of New York City businesses, but in small cities everyone drives anyway, and it makes only a small difference.  As a result, land values in small cities are "low amplitude."  Sure, some areas are going to be more expensive than others, but the differences will generally be muted.  In New York City it's not uncommon to see a very expensive neighborhood within half a mile of a slum.  (Though even slums are relatively expensive.)

And this, in turn, has consequences for inequality.  Below a certain level of income, small cities are terrible, because car travel is prohibitively expensive and very few jobs or stores are within walking distance.  This is one reason places like Ferguson, Missouri are hellholes for poor people.  But at a surprisingly low income, the situation reverses itself and small cities become much better for lower-middle-class people.  Once you can afford a car, basically, you are likely to be far better off in a small city.  If you can drive, then you can work anywhere in the city at relatively low cost.  Housing is cheap.  You can go to the park or the wilderness without paying insane amounts to live near them.  Retail goods and services are cheap, with a few exceptions.  (Nail salons are apparently quite cheap in New York relative to smaller cities, though this might not be for the best of reasons.  Also, while food is generally less expensive in small cities, certain kinds of ethnic food would be hard to find.)  And by the way, cars are not that expensive.  In New York City you pay about $1,400/year for a Metrocard.  A car probably costs more than $1,400/year in a small city, but not by a tremendous amount.  Certainly a couple that shares a car in a small city is not going to pay much more than they would pay for transportation in New York City.  And of course housing should more than make up for the difference.

Manufacturing jobs are also much more plentiful in small cities than in New York, because modern factories use a lot of space and can't profitably be located in areas where land is expensive.  Schools are a mixed bag.  Some schools in New York City are excellent, but overall the public schools are either horribly expensive (via the cost of living in a good school district) or mediocre at best, and the private schools are also very expensive.  That said, if you have high-school aged kids who are extremely good at taking tests, New York City's magnet schools are probably better and cheaper than almost any other option.

So what are big cities good for?  Basically, they are good for very poor people, very rich people, and gay people.  Very poor people, because cities do a lot of redistribution and don't require a car to get around.  Very rich people, because cities support high-paying jobs and abundant opportunities for consumption and socializing.  (And so poor people, again, because the presence of all those rich people creates a lot of low-skill jobs as servants etc.)  Gay people, because big cities are tolerant and contain enough other gay people to offer a lot of potential matches.  (As the country becomes more tolerant, medium-sized cities will become more and more competitive, and I expect big cities' allure for gays to decline.)

But here's the thing.  The pattern of development that is appropriate for a big city is very different from the pattern of development that is appropriate for a small- or medium-sized city.  Qualitatively different.  This is a huge problem, because (A) before cities become big, they generally go through stages during which they are small- or medium-sized, and (B) it is tremendously expensive to retrofit a city with the infrastructure that a big city requires.  This is not just a matter of building subways, though that is important.  It involves lot size, house size, building height, parking lots, parks, the size and variety of retail stores, etc.  Some of those things are relatively easy to fix, but some are prohibitively expensive.  One that I didn't mention, but that is prohibitive by itself, is street layout—cities somewhat counter-intuitively need to devote more space to streets, and need the streets to be laid out in a tight, interlocking grid.  But re-routing streets doesn't just cost a lot in construction terms, it requires re-allocating land ownership (the new streets have to go through what is currently private property, and lots will be bisected or worse by the new streets).

My view is that the government should consciously set out to develop a few more big cities along the lines of New York and Chicago (and it should probably spend money making both of those cities, especially Chicago, more urban).  This could be done in a few ways, all of them expensive and risky.  The government could simply build subways in the middle of nowhere (this has the advantage of being quite cheap), zone for density, and then subsidize the rapid development of a new city.  Alternatively, the government could select a few medium-sized cities, rezone large swathes for density, and then spend a bunch of money adding subways and other big-city infrastructure.

I somewhat prefer the former option because starting from scratch allows a tremendous amount of flexibility, and the politics seem much less complicated.  (Just dealing with the suburbs of an existing medium-sized city would be very taxing.)  But if we are going to enjoy the benefits of urbanism at a reasonable cost, we need more than our handful of legacy urban centers, and we can't rely on natural growth to deliver.  At best we can hope for the gentrification of a few old, somewhat-dense cities (like Philadelphia and Baltimore), all of which are in the eastern part of the country and none of which is truly equipped to be a New York-style urban center.