Blue sky traffic and revenue forecasts

A toll roads financier, Robert H Muller of J.P. Morgan Securities of New York City has drawn attention to poor traffic and revenue studies, most of which overestimate toll potential. Speaking at the IBTTA meeting in Rome, Muller said forecasting failures had attracted little attention because there has been no financial default. The real financial crocks among the new toll roads have all been bailed out, or quietly taken over, by other state agencies to avoid default on debt.

Muller lists 14 toll road projects financed in the past dozen years and finds a clear tendency to overestimate toll revenues in the prior traffic and revenue studies. Exceptions are the Georgia-400, and the Illinois North-South. GA-400 is the missing link in a radial highway heading south from the Altanta area's peripheral highway I-285 into the central business district. Traffic on GA-400 is running 30% above projections, and the IL N-S is 10 to 15% ahead of expectations. But 12 of 14 are underperforming.

Four of the 12 are moderate underestimates:

• Dallas North extension TX —26% underestimate in its first year but by its fourth year it was back on track

• northern segment of the Central Florida Greeneway FL — at predicted levels first 2 years but now 25% below

• Foothill CA — 12% low

• Sam Houston TX — about 20% low

Unfortunately this leaves 8 disgracefully incompetent estimates, real crocks of forecasts:

• Seminole FL — 47% low

• Veterans FL — 46% low

• Sawgrass FL — 60% low

• Southern CF Greeneway FL — 50% low

• S. Connector of CF Greeneway FL — 60% low

• Hardy TX — 75% down

• Kilpatrick OK — 70% down

• Creek OK — 45% off

Muller could have added the Oseola Parkway FL (down 70%), the Dulles Greenway VA (60% down) which would have added another two to the list of crock forecasts and SR-91 Express CA (my guess only: 20% down) which goes into the middle list of moderate understimates. And the VDOT Dulles Toll Road VA (25% over) could have been added to the two that were underestimated.

Muller lists:

• underestimates 2

• moderate overestimates 4

• blue sky overestimates 8

With my additions it is:

• underestimates 3

• moderate overestimates 5

• blue sky overestimates 10

The pattern is clear. Hire a consultant to do a traffic and revenue study and the odds of an overestimate are over 80% and the odds of getting a 'blue sky' crock are over 50%! At least that's the record.

Muller sees some patterns. The worst overestimates were those based on assumed high local growth, the 'greenfields' development type of highways. The crocks among the forecasts were also predominantly fringe or peripheral highways, bits of beltways. The overperformers (GA-400, IL-N-S, DTR VA) are all congestion-relievers and two are radials. Indeed IL N-S could be regarded as a radial also if you regard Chicago's O'Hare airport, rather than the Loop, as the real center of that metro area. It goes right by the airport, and is close to the geographic center. Muller points out another pattern of errors — where time savings on a new toll road are small, over-estimation is greater. That suggests one mispecified relationship. Maybe there's a threshold level of time savings (say 10 mins?) before people will shift to a toll road, whereas a linear relation is apparently specified (the forecasters regard their algorithms as proprietary so we can only infer their structure.)

The clear pattern in the mis-estimation of traffic revenues suggests that there is something systematically wrong with the models being used to make these forecasts, as well as faulty inputs perhaps. Muller is generous to the forecasters saying they are learning from their mistakes — improving. And he says the poor forecasts aren't always their fault. Their clients sometimes dictate unrealistic assumptions or won't spend the money for a quality forecast. But it might be healthy to open up for scrutiny the methodologies that produced so many crocks? And which could produce some big financial wrecks or at least raise the cost of capital unnecessarily. (Contact Robert Muller, J.P. Morgan Securities 212 483 2323)