Monday, November 22, 2010

A gift from Randal O'Toole?

So I was having this Twitter exchange with Virginia Postrel, whose work I've found very helpful in the past. It started with a column she wrote for the Wall Street Journal on the glamour of wind turbines and high-speed rail. I responded:
capntransit: @vpostrel Of course, highways have NEVER been oversold with glamour. And they always pay their own way. Right?

vpostrel: @capntransit Highways were glamorous in mid-20th century, w/ the self-deception that implies. But they're funded by gas taxes.

capntransit: @vpostrel No, highways are not completely funded by gas taxes.

vpostrel: @capntransit Gas taxes + tolls=not subsidized. Sounds like Wisc isn't consistent, but the TX example doubly proves the pt.

capntransit: @vpostrel I don't think taxes + tolls covers it. Maybe in Texas, but definitely not here in NY.

vpostrel: @capntransit In CA, where I live, they haven't built non-toll highways since the 1970s.

So I went looking for figures about California and came up with this one from a guy named George Crissman almost $7 billion in 2004. The funny thing is that it's part of an anti-transit campaign. As far as I can tell, the reason he went to the trouble to compute it is that he divides it by vehicle-miles to get 2¢ per vehicle-mile, and then by 1.6 to get 1¢ per passenger-mile. Still, wow, 6.9 billion dollars in highway subsidies in 2004.

It gets weirder though: this guy got his formulas for calculating subsidies on the basis of Federal highway statistics from Randal O'Toole! That's right, one of the biggest road-widening apologists in the country has given us this formula in connection with his self-published 2000 book, The Vanishing Automobile. O'Toole writes:
Get all the data you need to calculate highway subsidies in your state from Highway Statistics. For any given year, you will need to download tables HDF, SF-1, LGF-1, LGF-2, and VM-2. From table HDF, add together the following for your state:

* The amount of federal highway user fees that are diverted to mass transit;
* The amount of federal highway user fees diverted to (from) other states or for general purposes.

These are the diversions. Ignore state diversions because they are accounted for in table SF-1. Then add together the following:

* Appropriations from state general funds from table SF-1;
* Other imposts from table SF-1;
* Miscellaneous from table SF-1;
* Federal funds from other agencies from table SF-1; and
* Total dispersements by local governments from table LGF-2.

From this sum, subtract the following from table LGF-1:

* Motor fuel and vehicle tax revenues;
* Tolls;
* State highway user imposts; and
* Federal FHwA funds.

What is left are the supplementary funds. Subtract the diversions from the supplementary funds. If the result is less than zero, there is no subsidy: Road users are subsidizing something else. If the result is positive, there is a subsidy. To calculate the subsidy per vehicle mile, divide the subsidy by the total number of vehicle miles driven in your state shown in table VM-2. To get the subsidy per passenger mile, divide again by 1.6.

So I've taken all those tables and put together a spreadsheet that calculates the subsidy for each state. There are some interesting figures in there! According to the 2008 numbers, the biggest total road subsidies come from California ($11 billion), Texas ($8 billion), New York ($8 billion) and Florida ($5 billion). Tennessee pays for 91% of its road expenses with user fees, mostly gas taxes; no other state comes close. In actuality, $260 million in federal gas tax money is diverted to other states or uses, but that's made up by $522 million in local government funding. Tennessee is followed by Indiana (66%), Ohio (65%), South Carolina (63%) and Maryland (62%).

Again, I'm still not quite sure why O'Toole wants to help us calculate the billions of dollars that subsidize roads. He seems very fond of the per-vehicle-mile and per-passenger-mile figures, probably because those measures favor a system that involves lots of long trips. I'd like to see a figure of subsidies per trip, but I can't find a figure for trips in the National Highway Statistics. Table HM-62 gives weighted average daily traffic per lane on principal arterials, but it doesn't tell us how many lanes of principal arterials there are. I could probably figure it out, but I'm just about ready for bed.

The other thing that I'm not sure about is that O'Toole tells us to get the motor fuel and vehicle tax revenues from Table LGF-1, but Crissman gets his from Table LDF. These are very different figures, and I don't know which to go with and why. The spreadsheet that I've linked to uses LGF-1.

You do get some interesting figures per capita, though. In big empty states like Alaska, Wyoming, Nebraska and South Dakota, roads are subsidized to the tune of more than $500 per person. Good ol' frontier self-reliance! Sadly, New York isn't far behind South Dakota at $403 per person. The lowest subsidies per capita are Tennessee, South Carolina, Ohio, Maryland and Connecticut.

For those of you who are more familiar with this data than I am, is there a reason why I've never seen this before? Is the data suspect, or the formula?


Alon Levy said...

Remember, Cap'n: passenger-miles don't vote. When a mode of transportation is so low-capacity it forces you to travel three times as far, it's inappropriate to compare anything based on distance traveled.

If you want, you can find some sort of farebox operating ratio for roads here. No need to trust O'Toole's calculations; you can straightforwardly see that, excluding property tax-funded local streets, highways generated $122 billion in revenue in 2008, while using $193 billion in direct spending plus $4 billion for collection expenses. And that doesn't include depreciation.

Stephen Smith said...

Another thing to keep in mind is that all of these figures leave out local roads entirely. These are near 100% subsidized, and you could argue that they're far more important than highways. Local roads could exist with far fewer highways, but highways certainly couldn't exist with fewer local roads.

The best way to reduce subsidies for local roads is to implement a citywide congestion charge, but that's generally not supported by the Kochtopus.

- Stephen Smith
Market Urbanism

Pantograph Trolleypole said...

Cap'n don't forget that TXDOT actually calculated how much it subsidized roads. It has since disappeared from their site but the evidence remains...

Alon Levy said...

Yes, as I said, the national numbers I linked to exclude depreciation. The Texas numbers include depreciation, hence the lower rates.

I forgot to mention, another problem with talking in terms of cents per passenger-mile is that it's an unfamiliar measure, which can be made to sound low. It's much more straightforward to say "To make non-local highways pay their operational costs, the gas tax would need to be raised by 43 cents a gallon." It tells drivers exactly what they can expect.

BruceMcF said...

The biggest load of bullocks in his formula is that it counts all gas taxes as being generated by the use of the roads funded by gas taxes, when in many cases the gas tax revenues are generated by driving on other streets that are not funded, either at all or in proportion, by the gas tax, so he completely ignores the issue of cross-subsidy, which primarily flows from urban motorists who do the largest share of driving on unsubsidized streets, to rural and outer suburban motorists, who do the largest share of driving on (with respect to the Federal gas tax) subsidized Interstate, US, State, County and Township highways.

One reason Ohio ranks so high in terms of degree of user pays in his formula is because of the way that the state gas tax reinforces rather than compensates for the cross-subsidy built into the Federal gas tax. Its not really a 65% of user pays ... it is a smaller percentage of user pays and a healthy dose of a user somewhere else pays.

If it was any other mode of transport, O'Toole would call "a user somewhere else pays" a subsidy ... but not for roads.

saosebastiao said...

I don't really have a problem with O'Toole's calculations, although every type of analytical calculation can fall short, as commenters here have shown.

My problem with his calculation is his interpretation of it. His rhetoric is basically this: "Sure, roads are subsidized by about $.01 per passenger mile, and thats no good...but transit is subsidized by $.50 per passenger mile! That is terrible, and we need to stop funding transit!!!"

The problem with that interpretation is that it ignores very basic microeconomic principles...principles that have been proven empirically a million times over by now.

By economic definition, a car is a Normal Good, and transit is an Inferior Good (check the wikipedia entry on Inferior Good before you go flaming is just a name), and they are substitutes for each other. What this means is that when people become wealthier, they tend toward driving, and when they become poorer, they tend toward transit. Again, this is empirically proven. So it shouldn't be any surprise that when we subsidize the normal good, people will switch from the inferior good. Even if we subsidize the inferior good to an equal amount, there will still be a net shift to the normal good.

Furthermore, transit's costs per passenger drop dramatically with higher usage rates. In fact, unless costs are completely out of control (and in some cases they are), transit can break even at some given load rate. By shifting even a tiny amount of people toward driving, we are explicitly increasing the subsidies to transit.

Fact: If we all use our cars more, we will require more subsidies from the government. If we use transit more, we will require less subsidies for transit. O'Toole understands this...but you will never hear him admit it, because it doesn't fit his agenda. As long as he can convince people that subsidies aren't as bad for cars as they are for transit, he has made his employers happy.

M1EK said...

The difference between Postrel's point and TXDOT's is not really about depreciation, BTW, it's about the claim that if gas taxes pay all or almost all of the cost of highways, they must not be subsidized.

This is bunk, of course, because in Texas gas taxes are constitutionally dedicated to state highways (signed and numbered) but are, as everywhere, accrued when driving on any roadway - including the large networks of major arterials in most urban areas that do not receive any state support whatsoever.

In other words, people driving on 'local roads' (really major arterials without a route shield) are subsidizing 'highways'.

Also, in metro areas, there are occasional large 'donations' from local and county general funds to get highways built as well.