IBTTA 98 Wonky forecasts discussed

IBTTA 98

Wonky forecasts

discussed

Florida’s Turnpike director James Ely, speaking at IBTTA, the annual tollsters conference in Houston TX Oct 12, described how in 1990 the state legislature passed a bill (SB1316) allowing his agency to embark on $1.4 billion of new toll roads. One of the provisions of that bill was that proposed new turnpikes had to be “economically feasible,” which was defined by the legislators as generating net revenues after operating costs paying at least half their debt service by 5th year, and self-sufficiency by their 15th year. Traffic and revenue studies would therefore determine whether projects were to be built, or be set aside as “infeasible.” The provision for economic self-sufficiency, Ely said, was intended to “safeguard the turnpike system from political influence.”

Having the financial viability of new toll roads so central to the decisionmaking process — normally government toll agencies are content for their whole toll system to be financially able to carry new roads — the Florida legislature was suggesting the importance of new pikes being self-funding. In the 1990s 192km of new turnpike will have been built by Florida’s turnpike district, Ely said, at a cost of $2.1b ($11m/km, or $2.7m/lane-km.)

Lucky for the Turnpike that the state legislators did not hold its officials too strictly to the terms of the legislation, for as Ely recounted: “The survival and credibility of the Turnpike system was on the line early in 1995.” The first two new pikes opened following the 1990 law, the Seminole Exwy in northeast Orlando and the Veterans Exwy north of Tampa, each were producing first year revenues barely half those forecast at the time they were given the go-ahead.

Releasing the news of the major shortfalls in traffic and revenue was “an unpleasant task,” Ely said, quipping that they had a “birthday without a party.”

The top Florida official called for a “professional relationship” between the toll agency, bond raters, and the traffic & revenue consultant saying: “Any firms that feel that they are serving their clients by offering an overly optimistic forecast are doing their clients and themselves a disservice. Millions of investor dollars are on the line as well as the toll agency’s reputation. The investors and the toll agency must have a reasonable level of confidence that the estimated toll revenue will be realized.”

Seminole Exwy

A panel from the three leading traffic & revenue (T&R) firms discussed some of their crock-forecasts. Art Goldberg, representing the Florida Turnpike’s T&R consultant URS Greiner, said that a post mortem on the Seminole forecast showed up many errors. The local MPO traffic model used to generate trip numbers overpredicted travel. The extent of peakhour congestion in Orlando was overestimated, based on a false analogy with other developed areas where congested conditions were more intense and lasted longer. Development lagged behind expectations along the route of the Seminole Exwy in this northeast fringe area of Orlando. The abutting Orlando Co authority’s Greeneway also lagged in its traffic and revenue. But the largest mistake was failure to get the slope of ‘ramp-up’ correct, as seen by the subsequent convergence toward the forecast.

Goldberg explained this as the forecasters’ inexperience in working on toll roads within builtup areas: “Looking back we were entering a new era of startup facilities which now appear, in hindsight, did not have the same characteristics as those financed earlier. The earlier projects often connected major cities and passed through rural areas with few competitive routes. Today’s toll roads often compete with interstate (motorways) and other non-tolled expressways.”

In builtup areas with many alternative routes it takes longer to win drivers to the toll road, he suggested. Applying the lessons from the early failure, the company had managed to forecast a later toll road (the Greeneway Southern Extension) within 2% in its first year.

SJ Hills pike

Ed Regan of Wilbur Smith spoke about his company’s over-forecasts on the San Joaquin Hills pike in Orange Co CA. A severe prolonged regional recession that reduced jobs in the area and stopped traffic growth on area freeways was the most important factor, according to Regan. Poor signage on the approaches contributed too. Weekend usage and truck usage was severely overestimated. But as on the Seminole, traffic growth has been rapid since opening, and the project was successfully refinanced based on revised forecasts and strongly growing traffic. At over 80k trips/day by now, the SJ Hills has considerably closed the gap between actual and forecast T&R.

Dulles Greenway

Dan Greenbaum represented Vollmer Assoc, forecasters for the Dulles Greenway, which after opening had less than a third of the traffic forecast.

“An analysis made shortly after the opening, using the original model and the actual travel times on the improved Route 7 (VA-7), resulted in estimating toll traffic very close to actual levels. The problem therefore was not the model or the methodology, but the information regarding the schedule of the VA-7 improvements. In the original esimates, by the time the VA-7 improvements were assumed to occur traffic was expected to have increased substantially and as a result VA-7 travel times, even with the improvements would have been substantially slower than they were when the toll road opened. The important value of time factor was estimated accurately, but the time advantage of the new toll road was not as anticipated.”

Wrong in TRnl backyard

We commented to Greenbaum afterwards that this account misrepresents the history of VA-7 improvements, along the section (VA-28 to US-15) where VA-7 is competitive with the Greenway. Work on VA-7 improvements, a widening from 2x2-lanes to 2x3 lanes, and construction of a third level direct connector ramp at the VA-7/US-15 IC, was only got under way somewhat after the Greenway opened in the fall of 95. Construction work continued on VA-7 through the first two years of the Greenway’s operation and new lanes were only opened the full length of the project near the end of 97, after 27 months of Greenway operation. (We checked the dates with Tom Butler VDOT resident engineer in charge of the VA-7 project.)

If anything, this construction work, which included lane narrowing and lane closures made competitive VA-7 less attractive and increased the travel time advantage of the Greenway as compared to a situation with no construction work as modeled by Vollmer. It is true that the VA-7 work proceeded in stages toward the west and sections opened as completed. What Vollmer says about unexpectedly early competition from an improved competitive free road might be true of some of the shorter trips on VA-28 and VA-7 versus the Greenway in its easterly sections from late 1996, when those sections of improved VA-7 were opening up. But by then Dulles Greenway management had concluded that the Vollmer forecasts on which they had relied in setting a $1.75/$2.00 toll were seriously flawed, and had halved the toll to $1.00 to drum up traffic.

Anger

Rick Froehlich, Dulles Greenway chief financial officer said after the IBTTA session that he was “very angry” about the Vollmer presentation: “I was so angry I just had to walk out. They just won’t take responsibility for their mistakes.”

Poor signage on the appproaches to the Greenway and its omission from maps was also mentioned as depressing traffic. The Vollmer rep noted that traffic numbers have increased considerably and may meet forecast by the time originally assumed for VA-7 improvements.

Left unsaid however is that serious revenue shortfalls remain. Higher traffic is only being attracted on the Greenway at half the toll expected, so it seems there was a major problem with the estimates of time saving and the assumed value of time.

Ed Regan WSA

Ed Regan of WSA speaking generally said that value of time saved is “one of the most important and also one of the most elusive” parts of T&R forecasts. Generally it is in the range 40 to 60% of the wage rate, but the actual number varies by income level and between different places. WSA likes to do value of time studies for each project and has devised an interactive video interview for the purpose. In the Richmond VA area, he said, WSA had found a lower value of time/income ratio in forecasting the 895-Connector toll facility, indicating either a higher resistance to tolls or a population “in less of a hurry.”

All of the panel mentioned the need for a cooperative client prepared to take an active interest in their project, and the fact that a forecast is only as good as the data used in it. Regan said that a reason they are confident in the soundness of their models is that it is “calibrated” so that it accounts for actual existing traffic levels on the different segments of a roadway. (One of the most telling criticisms of the climate forecasting models used to predict catastrophic global warming is that they do not work in reverse and cannot model how temperatures are the way they are now.)

Regan said that the uncertainties of traffic forecasting lie mostly in the “socioeconomic forecasts and future network assumptions, which take place after calibration.” Also the value of time saved is quite difficult to calculate in places where no toll facilities exist to provide empirical data.

He also referred to “stretching the toll rate envelop” as a factor. This refers to new toll roads resorting to substantially higher per-mile toll rates, where he said price sensitivity is greater — see graphs.

Regan was critical of metropolitan planning organizations economic and land use projections. WSA is now relying much more on independent consultants to project the local economy, adding: “We can no longer take the MPO forecasts at face value.” Private sector interests, he said, also can produce rosy scenario forecasts of development which can skew traffic and revenue forecasts skyward.

Network effects that need careful study are the negative effects of enhanced competitive road capacity and of feeder roads to the toll roads that are expected but not built.

Ramp-up has been gotten wrong, it was suggested, because the basic traffic modeling assumes rational decisions by motorists and an informed choice of route. But in the early years of toll facility motorists may not be informed. (TRnl comment: several new toll facilities have had trouble getting permission to install adequate signage where it is most needed — on someone else’s property, that of an unfriendly state highway agency.) Maps in motorists hands take time to show the new facility, while in any case many commuters are creatures of habit, slow to try a new way.

“While we have always been aware of the ramp-up process, we find that the impacts are substantially greater with higher tolls and with the many new projects designed to provide congestion relief in corridors where there are existing competitive routes.”

WSA is urging its clients to do effective promotion, and an introductory “free” period may be an important part of that. Regan urged road developers not to “rush” their T&R studies. And if much time went by and conditions were changing revised studies should be considered. He ended saying “we are continually improving the methodology and many of the reasons that previous projections missed the mark are being addressed.” However the remaining uncertainties need to be better understood by clients. (Contacts: James Ely FL Tpike 904 488 4671, Ed Regan WSA 203 865 2191, Art Goldberg URS Greiner 212 736 4444, Dan Greenbaum Vollmer Assoc 212 366 5600x156, Rick Froehlich Dulles Greenway 703 707 8870x224, Tom Butler VDOT Loudoun Co 703 737 2000)