IBTTA DENVER E-toll penetration tough over 40% — too many infrequent users, Vollmer

IBTTA DENVER

E-toll penetration tough over 40% — too many infrequent users, Vollmer

Beyond 30 to 40% of patrons, electronic transponders (e-tags) are an increasingly expensive toll collection medium, because of the huge proportion of total trips made by occasional users. This theme from Daniel Greenbaum, of Vollmer Associates generated much interest at the IBTTA meeting in Denver mid-Oct.

"To increase market penetration (of e-tags) from 34.5% to 50% the cost more than triples. Assuming only weekday use, the average cost per transaction with 34.5% usage is 4.5c while at 50% the average cost is over 10c. More telling is the incremental cost for the additional 15.5% usage. The cost per transaction for these ETC users is 23c."

Greenbaum's calculations of the increasing diseconomies of electronic toll collection (ETC) beyond a third share are based on a generalization of data from surveys of toll-payers in New York and Florida, he says — typical urban toll facilities. In his example the facility is carrying 10k vehs inwards during 3 peak hours in the mornings weekdays (6am-9am.) These peak direction vehicle flows generate most attention because of their congestion potential. And they offer the greatest opportunity for electronic tolling because they contain a lot of regulars. 70% ETC use is feasible among these 10k/weekday — 7k. But the 10k only represent typically about 23% of daily trips of 44k, because there are normally around 16k vehs 9am-4pm and 8k vehs contra-peak flow (4pm-7pm) and about 10k vehs at night 7pm-6am.

The tween-peaks, contra-peak and night drivers constitute 3/4ths of trips and comprise a far more difficult market for e-tags because so many are infrequent users of the facility, Greenbaum says the data indicate. Commuters may drive a bit outside the peak hours and contra-peak ('reverse commutes') but they are unlikely to be more than 30% of total weekday trips and probably c25% of trips weekend travel included.

Trip frequency distribution on two facilities surveyed by Vollmer & Associates indicate approx equal three way split between 4-or-more trips per week drivers, once-to-thrice a weekers, and less-than-once/week drivers. About half of total trips are made by those who drive the facility once a week or less. Put another way, only half of trips are made by drivers who drive the facility twice or more/week. (All the above are one-way statistics.)

In the peaks the regular users dominate. 66% of morning peak-hour patrons inbound use the facility 4 times or more/week, compared to 22% out-of-peak.

When it comes to supplying transponders to patrons the tollster faces another arithmetical challenge, Greenbaum argues. He needs to look at the frequency distribution of users over say a year, and the dispersion becomes larger again by an order of magnitude. The 4-or-more times/week users (14k) that are seen as prominent in rush hours, and about a third of trips on a daily basis, diminish into near insignificance viewed among the enormous pool of individual users seen over a year. Counted over that period the same 44k veh/day facility/one-direction will encounter a distribution of individual users predominantly skewed toward the occasionals — 6.6k/day less-than-monthly users become 396k individual users over a year, Greenbaum estimates, and 71% of the total 554k individual users of the facility over a year. The approx one-third of less-than-weekly users encountered on a single day become nearly 90% of individual users over a year.

Greenbaum postulates that a certain proportion (he guesses 15%) of even the most frequent users may not establish a toll account and obtain an e-tag, whether because they are concerned about privacy, don't want to make the commitment, or just don't get around to it. So 85% is probably the potential usage for 4 times/week-and-greater users. For 2 to 3 times-a-week users he assumes 30% to be the potential e-tag transaction share, 10% for once-a-weekers and 5% for those using the facility more than once monthly. Given his frequency distribution of uers he gets a potential of only 34.5% for ETC day-round and 60% in the morning peak, and he says these percents are consistent with the results on facilities without a discounted toll for ETC.

Of the 44k vehs/day, 15.2k transactions would be e-toll and given the frequency distribution of usage 24.3k e-tags on the windshields of vehicles would be needed. Greenbaum says that an early rule of thumb in ordering e-tags was to say 70% of the rush hours patrons (7k) are commuters who could get e-tags and 10% (3.4k) of the others (34k) so with a 20% contingency you should order say 12.5k (7k+3.4k+20%) e-tags. But the experience typically is that you quickly run out of e-tags even while the share of e-toll/total tolls is a disappointment.

Says Greenbaum: "This is typical of what has occurred on all toll facilities when ETC was introduced."

In order to get 50% e-toll usage (22k/day) on Greenbaum's frequency distribution he has to push usage to 90% by 4-and-more/weekers, 67% by 2&3/wk, 40% by once-a-weekers, 15% by 1/wk-to-1/mth, and 7.5% by those using the facility less than once a month. And the last represent 30k individual users needing tags out of almost 80k. 80k tags are needed out there to get 22k e-toll transactions a day on a 44k tolls/day facility.

"As the occasional users are brought into the program, an extremely large number of these individual users are required to impact overall market penetration because of the infrequent use by this group."

Cost: The point of all Greenbaum's arithmetic is to show the rapidly increasing marginal cost (for a single toll facility - TRnl qualification) of enlarging the share of transactions done by ETC. An e-tag costs about $25 and if it is amortized over 8 years costs $3.12/yr. Traditional American billing done monthly costs 32c postage, $3.84/yr. So the direct marginal cost of each account is about $7.00/yr, producing e-toll transaction costs of 3c for the 4/wk users, 7c for the 2-3/wk, 14c for 1/wk, 28c for 1/wk to 1/mth, and $1.75 for <1/mth users. Applying these costs to the frequency distribution of users the cost of transponders and billing at 34.5% e-toll share is $170k and at 50% share is $559k. A tripling of ongoing cost for a 45% increase (50/34.5) in usage! Hard to explain to the board? The extra cost/transaction for the extra ETC users is 23c, the paper says. At 34.5% share the average cost per e-toll is 4.5c while at 50% share the average cost is over 10c.

Comment: Greenbaum's numbers assume one-way tolling, as on New York City and Bay area bridges and tunnels. If tolls have to be collected both ways, as on many other facilities, e-toll usage will halve his per-transaction costs. A further assumption in Greenbaum's work is that each toll facility issues transponders for the one facility, and that facilities toll quite independently of one another. But in the real world of big cities as multiple facilities read the tags issued by different agencies (as in California with TCA and 91-Express and on multiple Bay area bridges, in VA with contiguous toll roads, and in NY/NJ with the Triborough, Thruway and Port Authority reading one another's tags) more motorists will make multiple transactions daily on the same tag, so electronic tolling should be able to slide back down the viciously steep Greenbaum cost curve described above. If, as many in the industry hope, tags can be used for parking, at airports and so forth, the economics can improve further. And perhaps the steep Greenbaum cost incline can be countered by finding ways to reduce billing costs for the infrequent users. There is no reason why, for example, hordes of infrequent users have to be sent monthly statements, just the same as the daily users. Or why some of the fixed cost of the system shouldn't be shifted away from the toll agency to the patrons either with an upfront fee or graduated toll charges which hit the occasional user with a higher toll rate than regulars. 91-Express is effectively doing this already via a three tier pricing based on a mix of high posted tolls and discounts and up-front charges. Express Club members on 91-X who agree to pay $15/month membership get 60c discounts on each toll so they save if they make more than 25 trips/mth; Standard Plan has a $5/mth account fee and a Convenience Plan so-called for the occasional users has no flat rate monthly charge but requires $40 up front for the transponder — and the right to use the toll road. Pfeffer anticipated Greenbaum!

Alan Plain head of the New York State Thruway said recently that his agency estimated the average cost of an e-toll transaction is about one-third of the cost of a manual toll take. If a toll collector costs $30/hr all-up (taking tolls pays one heck of a lot better than writing about them, if I might be pardoned the comment!) and collects an average 150 tolls/hr, the toll collector costs 20c/toll, to compare perhaps with 7c/e-toll, which roughly fits some of the relevant Greenbaum numbers. ETC costs have to be compared with radically differing alternative toll collection costs in different facilities depending on prevailing labor arrangements, sunk costs in toll plaza equipment, and land costs. OTCC in Toronto has said that license plate imaging, billing and collection on 470-ETR will cost around 70c/transaction, the premium set over transponder toll fees.

Greenbaum's paper is extremely valuable in drawing sharp attention to the fact that ETC is no panacea, that the patron base is highly variegated, and that there will come some point when pushing transponders becomes more expensive than alternate means of collecting tolls.

The statistical robustness of Greenbaum's calculations has to diminish the further out on the usage curve he goes, because as he says, rather little solid data has been collected on the infrequent users. The paper ends with the sensible call: "Don't forget about those infrequent users, because there are an awful lot of them." (Contact Daniel Greenbaum, Vollmer & Assoc 212 366 5600x156)