Austin Contrarian

Congestion tolls vs. revenue tolls

I think one thing that confuses people about my support for congestion pricing is that they're not clear what I mean by "congestion pricing."   

There are two kinds of road tolls.  One is a flat rate or flat charge per mile or some variation thereon.  This is the kind of toll TxDOT levies.  I've called these "revenue tolls" in the past, just to distinguish them from congestion pricing.

Revenue tolls are inefficient.  If environmental costs, wear-and-tear, and accident risks are handled through gas taxes and insurance, as they should be, a motorist driving down a deserted MoPac at 3 a.m. inflicts exactly zero cost on the rest of us.  Zip.  This means that any toll we charge him exceeds the marginal cost of his trip to the rest of us.  Charging more than marginal cost causes some people to forgo trips even though the benefit to them exceeds the harm they do to the rest of us.  That's a dead-weight loss.  (That loss might be behind some of the opposition to toll roads.)

But revenue tolls can also be too low.  When a road becomes congested, each driver entering the highway imposes a very slight cost on everyone else by increasing congestion.  The marginal cost of each trip is a positive number.  A revenue toll may or may not be high enough to cause the driver to internalize that cost.  If the revenue toll is too low, there will be too much congestion.

Revenue tolls simply aren't a good way to handle the costs that drivers impose on one another.  Thus, tolled roads like Highway 45 in north Austin are not priced right.

Optimal congestion pricing, on the other hand, is efficient.  Congestion pricing forces drivers to internalize the cost they impose on everyone else.  The optimal congestion price equals the incremental cost of congestion each driver inflicts on other drivers.  It so happens that the optimal congestion price is also the lowest price that will eliminate congestion.

In an ideal world, the congestion charge would vary continuously with the demand for the road.  The charge would rise at peak hours.  At off-peak hours, when demand is light, the optimal congestion price would be zero.  But the charge would always be just enough to keep traffic moving.  For I-35, that price would be zero at 2 a.m., but it might be $10 at 4:45 on a Friday afternoon.  Some places are now experimenting with continuously variable congestion tolls.

Congestion pricing monetizes economic waste.  Without congestion pricing, the cost of sitting in traffic or delaying a trip is lost for good.  Congestion pricing turns this loss into money, money that can be used to increase road capacity or build sidewalks or fund better schools.  Congestion pricing also gives us clear signals about the adequacy of road capacity.  When congestion tolls yield "too much" revenue, we know that we need more capacity.  

As I noted in a comment to an earlier post, there are only a few things that virtually all economists agree on.  One is that rent control degrades the quality and quantity of housing in the long run.  Another is that congestion pricing is good policy. 

One other point on rail

I should have made this point, too:  Our failure to congestion price our roads really screws up our decisions on increasing capacity.

When roads are congestion-priced, the revenue tells us when we need more capacity.  On this, see this (pdf):

One important feature of congestion prices is that they not only discourage usage when congestion is present, but they also generate revenue for capacity expansion.  Indeed, it has long been recognized that under certain conditions the optimal congestion prices for a fixed amount of capacity will automatically generate the appropriate amount of revenue to finance capacity expansion.

This means that when optimally priced road generates a stream of surplus revenue, it is time to add more capacity -- and the "right" amount of new capacity (assuming constant cost per unit of capacity) will be whatever the surplus revenue pays for.

If we priced things right, we'd know when to invest in rail.  Take I-35 through central Austin.  If congestion pricing produced a huge surplus (and it would), we could invest that money in additional capacity.  We could then decide whether new roads or rail would give us the best return on our investment.  Since increasing road capacity on I-35 would require tearing down neighborhoods or building a triple-decked highway -- both prohibitively expensive -- I'm confident that commuter rail would be the way to go.  In either case, we would have a firm understanding of the economic return from the investment.

Rail

If you haven't noticed, I mostly blog about urban economics and Austin development.  I rarely blog about transportation issues, though, and when I do I mostly write about congestion pricing of roads.

So where are my posts on light rail?  That ought to be within my bailiwick. 

The short answer is "comparative advantage."  There are lots of other people who know more about this stuff and who write about it regularly.  MIEK and the Overhead Wire cover light rail all the time (the latter multiple times a day), and they both know far more than I about the technical details -- stuff like acceleration rates and track widths and catenaries.  I could try to learn that stuff, but why? 

Rail sparks ferocious arguments, and I don't particularly want these to hijack my blog.  But for the pro side, see MIEK and the Overhead Wire.  Also check out Ryan Avent, who writes about the economics of rail almost daily.  Tory Gattis writes more skeptically, although I don't think he's dogmatic.  For hard-core opposition, there is the Antiplanner.  There are many others, of course, and I don't mean to sleight slight anyone.

My big-picture thoughts: 

1.  One "duh":  Rail should only be built when a cost-benefit analysis warrants.  (It's warranted in New York City, not warranted in McComb, Mississippi; the hard cases are the cities in between.)  This doesn't mean that rail should be required to pay its own way.  Since highways are underpriced, pricing rail at marginal cost would be inefficient.  Since there are subsidies all around, I don't think it's helpful to argue about which mode of transportation gets the most subsidy.

2.  Mobility is a good thing.  Cars provide more mobility than rail --  unless, that is, the roads are overly congested and cannot feasibly be widened or congestion priced.  When estimating the demand for rail along a congested route, it is important to count not only the actual vehicle trips, but the vehicle trips never taken because of the congestion.  The trips lost due to congestion are the main deadweight loss from congestion.

3.  Rail really does foster dense development along routes because, unlike buses, it represents a more or less permanent commitment by the city.  For this reason, it increases property values along the route.  One could argue that if the net increase in property value exceeds the cost of construction, the rail ought to be built.

4.  High-speed rail.  I suspect it's cost-justified in the northeast (especially if airport landing rights were priced properly), and probably along the congested San Diego-San Francisco corridor.  I don't know about Texas.  I would benefit tremendously from high-speed rail; I fly to Dallas or Houston several times a month and rail would make me much more productive.  I just don't know whether that marginal increase in productivity would offset the billions and billions in cost.

5.  Austin.  I don't think light rail from downtown to the airport is worth much.  I won't take it because when I get back to Austin after a day-long deposition in Houston, I want to drive directly home.  I think most other business travelers will be like me.  Tourists:  perhaps not, although I think they'll have to get a car anyway.

I think Cap Metro's commuter rail will be a bust for the reasons MIEK has been arguing for years.

Light rail.  I imagine a light rail line from South Congress to downtown, past the state government complex, and up Guadalupe to the Triangle would work.  South Congress, downtown, and West Campus are rapidly densifying.  The congestion will get much worse, or, more precisely, reach an equilibrium in which lots of trips are forsaken.  Getting from downtown to UT is already a big hassle; I'm regularly deterred from making the trip.  (I voted for the 2000 light rail plan.  If I'm going to shell out for a light rail line, though, I'd want it to go down South Lamar, too.)

Buses.  I like them.  And I ride them.  I get a little bit of exercise, plus I don't like to drive.  I think Cap Metro needs to increase frequency of service on the busy routes.

Those are my big-picture thoughts, for what they're worth.

Subsidies don't matter

A recurring debate between the highway camp and the mass transit camp is, "Who gets the bigger subsidy?"  The highway advocates argue that gas taxes cover most of the cost of roads, while mass transit is almost entirely dependent on government subsidies.  Transit advocates respond that gas taxes really cover only a small fraction of total highway cost; highways are a lot more expensive than the other side admits.  

Personally, I think this is the wrong debate. What matters is that a new piece of infrastructure produce a good return on the investment.  Who pays for that investment is irrelevant.  A road or rail line can be cost-justified even though the government is footing the bill.  Or a road or rail line can be a big pile of crap, in which case it's a big pile of crap regardless of who pays.  Obviously, when the public is footing the bill, it has an incentive to vet the cost-benefit analysis, but the amount of the subsidy, again, has nothing to do with that analysis.

The "highways are already paid for" argument is particularly off the mark.  I could care less whether gas taxes cover all or just part of the cost of roads. Gas taxes are (or should be) levied to internalize externalities. What we do with the money after that is up to us; there’s no iron law that says the money must be plowed back into roads. If gas taxes earn a higher return somewhere else, they should be spent somewhere else.

So we ought to focus on which infrastructure will produce the best return.  Sometimes the better project will be a road; sometimes it will be mass transit.

I know that it is impossible to perform a cost-benefit analysis without a full reckoning of the costs.  To the extent the subsidies debaters are just trying to drag all of the costs out into the open, fine.  But in the end, the subsidies debate cannot answer the big question, "What should we build?"

More on gas prices and home prices.

I've tried to evaluate the impact of gas prices on home prices before.  Richard Green, a GWU real estate/finance/economics professor, comes to a completely different conclusion:

Let's say the average household makes five one-way trips per day--for work, shopping, entertainment, etc. Let's also say that the average car gets 20 mpg in city driving. Each mile of distance to work, shopping, etc. is therefore now 50 cents per day per household more expensive than before. A household living immediately adjascent to work and shopping should then be willing to pay $5 per day more in rent than a household 10 miles away compared with six years ago, all else being equal. This becomes $150 per month, or $1800 per year. Assuming a five percent cap rate for owner occupied housing, this translates to $36,000 in relative change in value. Given that the median house price in the US is about $220k, this is kind of a big deal.

The assumptions here are pretty crude (particulalry the ceteris paribus assumption), but if gas remains at its current real price, we will see the shape of US cities change.

Green is using the standard monocentric model.  Under that model, all employment and city amenities are assumed to be concentrated in the CBD.  With that assumption (Ok, and a few others), the model predicts that land rents are completely determined by commuting  costs.  That is, property values (which depend directly on rents) will decline to offset any increase in gasoline costs.

I am skeptical that home prices will completely offset the rise in gas prices (and, to be fair, Green acknowledges he is making a bunch of assumptions for the sake of simplicity).

First, homeowners can cut their gas bills by switching to more efficient vehicles, combining errands, taking public transit, etc.  This means they won't necessarily respond to higher gas prices by bidding up rents dollar for dollar.

Second, cities aren't really monocentric.  They are polycentric.  Think of Houston, which has its downtown, the Galleria, the Med Center, Greenway Plaza, the Energy Corridor, etc.  A home nine miles from downtown Houston may become less attractive to downtown commuters, but if it is only three miles from the Galleria, it may become relatively more attractive to Galleria commuters.  The push and pull of the different employment centers will snarl the simple rent gradient predicted by the monocentric model.

Ultimately it's an empirical question.  It will be interesting to see what really happens to home prices.  Will (have) home prices in Houston react(ed) differently than home prices in a relatively more monocentric city such as Boston?

I think Green definitely overstates things, though,  when he says "we will see the shape of U.S. cities change."  Even if one assumes that home prices will completely absorb the rise in gas prices, our urban form won't change much, at least not quickly.  Rising gas prices theoretically could drop the value of suburban land below its value as farmland, but it still won't be worth anyone's while to plow under all those houses.  The markets simply will clear at very low prices.  The low land prices may discourage the construction of new suburban housing, but this assumes that central city zoning will allow the density necessary to accommodate the suburban refugees (which it certainly should).  Even then, adding a robust 1-2% per year to existing central city housing stock won't have a dramatic effect on the city's urban form.  For example, if a city starts with 70% of its housing as detached single-family homes, it will take a very long time before most of its housing is made up of condos and apartments. 

In other words, Atlanta won't turn into Berlin during our lifetimes, regardless of what gas prices do.

Where gas prices hit hardest.

According to this neat graphic from the NY Times, it's rural areas: 

Dense cities do better than sprawling cities, but the difference between rural counties and suburban/urban counties dwarfs the difference between suburban counties and core counties.

I've been skeptical of claims that the end of the suburb is upon us.  I'm not so sure about poor, rural areas like the Mississippi Delta.  The reason these places haven't completely emptied out already is that housing is so cheap there.  High gas prices are significantly raising the real cost of housing there.  If already low property values (and hence rents) cannot drop enough to make up the difference, the existing residents will leave -- and they won't be replaced because no one will be willing to compensate newcomers for living there.

(Most suburbs are different because (1) they are wealthier than rural counties, so gas prices don't take such a large bite; and (2) declining home prices caused by rising gas prices may hurt existing residents but will attract new ones -- unless gas prices rise so high that the suburban home is a bad deal even when (literally) dirt cheap).

Related post: Is the end of sprawl in sight?

Ben Wear's Great Race

One of us has a good sense of timing. . . .

Ben Wear of the Statesman has conducted an, uh, "experiment" to determine whether SH 130 is faster than I-35 even in rush hour traffic.  His conclusion, based on what he admits is an unscientific sample of one:  I-35 is faster.

He and a colleague both started north of Georgetown at 7:15 a.m.  She took I-35 through Austin to FM 1327 south of town.  He took the new toll road.  He had to get on FM 1327 for the last few miles since SH 45, which will connect SH 130 to I-35, hasn't been completed.

She won.  She drove 43.3 miles in 45 minutes.  He drove 54.8 miles in 54 minutes, nine extra minutes.  Ergo, I-35 is faster (and cheaper).

Wear admits traffic may have been unusally light on I-35 that morning.  I think that's a bit of an understatement, but I'll leave that to others to pick apart.

I have another problem with his analysis, though, one that makes me think he's trying to stack the deck against the toll road.  It's a subtle thing, but it's annoying.

Wear's colleague describes a fight through stop-and-go traffic from Round Rock to Cameron Road, through one traffic snarl after another.  Wear describes a zip down the Autobahn.  Here's his description:

The drive on Texas 130 is predictably uneventful and stress-free. With only one car visible about a quarter mile ahead and none in the rearview mirror, I set the cruise control to 70 mph.

I will have to tap the brakes only once in the next 46.8 miles of toll road. Much of the time there are no cars within 100 yards of me, and I see less than two dozen 18-wheelers the whole trip. The view is mostly of cows, green fields and old farm buildings.

Wear still lost the race.  By nine minutes.  See?  I-35 kills SH 130 even though you get to fly down SH 130 at 70 mph.

But Wear only describes the first 46.8 miles of his trip, the part where he zipped along at 70 mph.  He doesn't really talk about the last 8 miles. 

Here's an algebra question:  Ben drives his Taurus 54.8 miles in 54 minutes.  He averages 70 mph for the first 46.8 miles.  What is his average speed for the last 8 miles?

Be sure to show your work.

If Okies can do it . . .

Oklahoma City plans to re-route an interstate highway away from downtown:

In Oklahoma City, the interstate will be moved five blocks from downtown to an old railroad line. The new 10-lane highway, expected to carry 120,000 vehicles daily, will be placed in a trench so deep that city streets can run atop it, as if the highway weren't there.

The old highway will be converted into a tree-lined boulevard city officials hope will become Oklahoma City's marquee street.

By tearing down the Crosstown Expressway, the city hopes to spur development of 80 city blocks stretching from downtown to the Oklahoma River — an area that contains vacant lots, car repair shops and a few small homes.

"We've always been a good place to live, but we've never had a city we could show off," Oklahoma City Mayor Mick Cornett says. "Moving the expressway makes it possible for a day to come when hundreds or thousands of people will live downtown."

The project will cost $557 million, mostly federal and state funds. The city will pay to spruce up the boulevard, build parks and put a pedestrian bridge over the new below-ground interstate.

If Oklahoma City can do this, why not Austin? 

Housing + Transportation Affordability Index

Check out this totally cool site that maps housing plus transportation costs (plus a whole bunch of other statistics, like average block size) by census block group.  They use census data to estimate the average home and transporation costs as a percentage of median income. The map's creators' point is that those suburban homes are a whole lot more expensive when you factor in transportation  costs. 

I'm fuzzy on their methodology for estimating transportation costs.  I spot-checked the block groups around my neighborhood, and found big differences in transportation costs that I'm skeptical exist.  And bear in mind that housing + transportation costs may take a huge bite out of, say, a Pflugerville household's income, but they would probably be a mere nibble compared to the bite of a central Austin mortgage payment.

Still, this is the coolest thing I've seen in a while.  (I admit that I probably have a warped sense of "cool.")

H/t Smart Growth around America.

The geography of gasoline consumption

I've done some more thinking about the Glaser and Kahn paper I bogged about the other day.  Glaser and Kahn examine total energy use among major metropolitan areas in order to compare cities carbon emissions.  But their data also allow us to compare gasoline use city by city.  That's just as interesting a comparison, at least to me.

Glaser and Khan even allow us to make an apples-to-apples comparison because they have constructed a composite "median" household for each city to use as their points of reference.  For example, they don't use per capita gasoline consumption as their point of comparison.  They instead estimate the expected gasoline consumption of a hypothetical $62,000/year household with 2.62 members for each city, using survey data and (I assume) some fancy statistical techniques.

They report pounds of carbon emitted rather than gallons of gasoline consumed, but since they tell us how much carbon they assume one gallon emits (23.47 pounds), we can easily convert their data into gallons. 

Here is the relevant table from their paper:

Here is a chart I compiled from their data depicting relative gasoline consumption:

The thing that struck me when I looked at the numbers was the relatively small difference in gasoline consumption between Houston and most other cities.  (Third column.)  Glaser and Kahn's hypothetical $62,000/year household uses only 55 more gallons per year in Houston than in D.C.  It uses only 135 gallons more gallons per year than the corresponding Sana Francisco household.  The Houston-Chicago gap is only 118 gallons per year.

The gap between Houston and old-line northeastern cities (particularly New York) is larger, but is largely offset by their greater consumption of heating oil.  (Remember Hugo Chavez offering free heating oil to Boston's poor?)

This chart makes the point better:

There is no question that steep gas price hikes would hurt Houston (or Dallas or Austin) households.  At $3/gallon, the Houston household is already spending 5.4% of its gross income on gasoline.  At $4/gallon, it would have to spend 7.2% of gross income on gasoline, assuming no change in gasoline consumption.  But the Sana Francisco household is spending 4.8% of its gross income on gasoline, a figure that would rise to 6.4% with a $1 increase in the price of gasoline.  That 0.8% difference is smaller than I expected.

Glaser and Kahn also compare city and suburban gasoline use:

I summarize the corresponding gasoline consumption numbers in columns 5 and 6 of my chart, and compare suburban Houston households with central city residents elsewhere in column 7.  No surprise that the gap grows significantly.  But it is possible that, given their much higher housing costs, central city residents may be even more susceptible to price variations in commodities like gasoline.  Those suburban Houston households pay a lot less per square foot of housing than similarly situated households in LA, Sana Francisco or DC.

These simple charts don't tell the whole story, of course.  Gasoline consumption is not completely inelastic, especially over the long run.  One could argue that because residents of central cities have access to better mass transit, it is easier for them to substitute away from gasoline use, making their demand for gasoline more elastic than suburb dwellers.  Put simply, city dwellers can cut down on their driving more easily than suburbanites.

That's certainly plausible.  However, city dweller's access to better transit is already reflected in their lower gasoline consumption.  They've already picked the low-hanging fruit, so to speak.  The barrier to additional transit use is often not the difference in monetary cost, but the difference in time and convenience costs. 

Of course, there is some price at which drivers will begin to substitute to mass transit wholesale.  But the suburban drivers have options, too.  A Houston suburbanite who switched to riding the bus or carpooling would save (on average) more gasoline per commute than a central city resident.   Likewise, a Houston suburbanite who combined two errands would save (on average) more gasoline per errand. 

This is ultimately an empirical question.  Like everyone else, I think gas prices are headed up, at least over the next few years.  It will be interesting to see the relative adjustments made by Houston drivers and those in the more compact northeastern and western cities. 

Relative elasticities aside, the numbers are still interesting, especially the aggregate metropolitan area data.  I never would have guessed that a "typical" metropolitan Houston household uses only 5 more gallons per month than its metropolitan D.C. counterpart.

One final point.  There is one area where western and northeastern cities wipe the floor with us, the denizens of the hellishly hot places:  Electricity.  We use fiendish amounts of the stuff.  (There's archaeological evidence that Austin was inhabited before the invention of air conditioning, but some scholars dispute its conclusiveness.)  If electricity prices shoot up like gas prices, we will suffer a lot more than our friends out west and up north, and there'll be nothing we can do about it.