USA Today reports dramatically more expensive tolls - lousy data (ANALYSIS)
USA Today had an article (2008-01-28) saying tolls are getting "dramatically more expensive." There are certainly some large increases in US tolls, but in line with industry practice many tolls have been frozen for quite a number of years so it is unclear what it means on an annual basis.
The USA Today article highlights the proposed New Jersey increases under Gov Corzine's with their four sets of 50% increases at four year intervals plus indexation. It also cites toll increases at the PANYNJ, the Golden Gate Bridge, Indiana, Massachusetts, NYSTA, Pennsylvania.
The USA Today story is still a bit thin. Corzine's proposals are an outlier, and they are just proposals. They face an uphill battle to gain support in the state legislature as part of a huge reorganization based around a public-public toll concession. Our guess would be that Corzine's grand monetization scheme will either crash completely or be adopted in a heavily moderated form.
As for the rest of the toll increases it is unclear if they indicate that use of toll facilities generally in the US is getting much more expensive.
We strongly suspect that there's some truth to the theme of increasingly expensive tollroads - based on the greater frequency of toll increases - but there aren't any decent data to substantiate that.
Journalists aren't statisticians and can't be expected to do the math involved in constructing an index of toll rates in America, so we shouldn't beat up on a fellow hack.
A national toll rate index
It would be nice to have a national toll rate index along with all the other price indices, or even toll rate indices for individual tollroads and toll networks.
Of course toll rate increases are rarely simple - like everything up 10%. They are 22% up for cash, 17% for ET for cars, and many other differentiated increases.
How to make that intelligible to people? One way is to focus on the toll increase for the most common user - the car with a transponder say. Another is to see how projected revenue with no change in volume of traffic or distance traveled increases on past revenue.
A national toll index would not be easy to compute given all the different classes of vehicles, the different charges for electronic tolling and cash, the special commuter deals, the different rates by time of day. Maintaining it would be expensive too since stuff's changing all the time and it would need regular re-weighting as the relative importance of different facilities changes.
All price indices face these kinds of problems, so they are not insurmountable.
In one way a national toll index would be easier to construct than other price indices. All the data is public. Collection costs would be low.
Of course the index wouldn't have any necessary consequence for toll rate setting. Some toll rates are too low and need to be raised faster than others. Toll rates that are deterring traffic may be too high and need to be lowered, regardless of what's happening all over.
A toll rate index would at least provide the basis for a more informed debate as to whether proposed increases are really our of line with the rest of the industry, and with the prices of other goods and services.
The Hardy - we never knew she was way up there, Uh Oh
The USA Today piece also has a listing of the nation's "busiest" tollroads. There are problems with that too. The newspaper credited IBTTA as being the source. It is horrible data.
Someone in Houston must have been pulling legs at IBTTA to make the Hardy Toll Road the second busiest tollroad in America. Whoever put that out hasn't ridden the Hardy.
It is a modest, mostly 2x2 lane tollroad going north out of Houston. It has always struggled to compete with the parallel free interstates not far to the east and the west.
The latest annual report of Harris County shows it doing 92k/day, not 979k. Order of magnitude error guys.
The busy tollroad in the Houston area is the Sam Houston Tollway, much of it eight lanes and always humming with traffic. It's right up there with the other big tollroads of America: 740k tolls/day. (We think the mistake in the Hardy number was to put in numbers for all three of the Houston area tollroads - the Sam 740k, the Hardy 92k and Westpark 83k - and attribute the whole lot to Harris Co's second ranker.)
Tolls not vehicles
There's another problem with that table. It purports to be counting average vehicles per day on each of the tollroads. It does nothing of the kind. It is actually a count of vehicle passes of toll points, best called toll transactions. That is very different from vehicle numbers or trips.
Take a vehicle traveling on the Sam Houston Tollway - which should be up there with the New York State Thruway in the table (why use the purely honorific name Gov Thomas E Dewey Thruway which term no New Yorker has ever heard of?)
A car coming from the northwest, from say College Station and headed for Hobby Airport on the southside of Houston will pass through the following toll plazas:
- Sam Central
- Sam South
- Sam Southwest
- Sam Southeast
That trip will count as four toll transactions. That isn't four vehicles on the tollroad. It's one vehicle, making one trip, but happening to do four toll transactions because it passes four tolling points.
If the tolling points were on the ramps or at side plazas as in ticket system or trip based tolling - as in the New Jersey Turnpike and most of the NYS Thruway - that same trip would register as a single toll transaction instead of four transactions.
The best that table can do is measure the busyness of the toll processing system not the busyness of the tollroad as such. It adds apples and oranges as far as the roadway is concerned - point tolls and trip tolls.
Another problem. Measuring the Garden State Parkway by toll transactions (let's pretend they didn't mislabel it vehicles/day) can produce ridiculous results when the toll system is reorganized.
The GS Parkway has been converting major 2-way barrier plazas to one-way tolling to lower toll collection costs and it has been doubling tolls at those points to sustain revenue. That halves the number of toll transactions at each toll point along the road.
By this measure the GSP is getting less busy!
But the road isn't any less busy, just the tolling system.
Even trips are not much of a measure of how busy a road is. Most tollroads have a heap of entries and exits. The shortest trips only load the segment between the adjacent on and the off, whereas long trips load the road along much more of its length.
Need VMT or VKT data
The best measure of the busyness of the tollroad is vehicle-miles (or vehicle-kilometers) traveled/year or per day. That measure also gives a sense of the earning capacity of a road because most toll rates are expressed per km or per mile.
Vehicle miles traveled can be inferred from daily flow data in the various segments of the road multiplied by the distance between interchanges, or by summing every trip distance.
Some tollroads generate that data, others don't unfortunately.
You can't really blame the journalist when you don't generate good data.