Study of Minneapolis dynamic priced lanes shows drivers see posted tolls as proxy for congestion - more likely to choose toll lanes when posted prices high
2013-08-07: Researchers in Minnesota studying driver responses to toll rates on the dynamically priced I-394 and I-35W express lanes have found results that confound conventional models. They have found that users of the toll express lanes are attracted to them in greater numbers as prices rise - the reverse of the straightforward economic notion of higher prices turning away users, and lower tolls attracting them.
The research by David Levinson of the Networks, Economics and Urban Systems Group at the University of Minnesota and Michael Janson a masters student there involved three field experiments in which prices were changed without any public announcements and responses studied, plus analysis of two years of traffic and toll rate data.
The other surprising finding was that drivers on the toll express lanes consistently pay $1 to $2 per minute ($60 to $120 per hour) of time saved, far more than conventionally estimated values of time saved.
Levinson in an interview with TOLLROADSnews said that it seems clear that drivers are using the toll prices displayed on variable message signs ahead of entry points as indicators of the severity of congestion ahead in the general purpose lanes.
"In the absence of better information about travel times and traffic speeds drivers see a high toll as a warning of slow traffic and so they are more likely to choose the toll express lanes than at lower prices," Levinson says.
The Janson-Levinson paper postulates that express lanes use is a conventional economic good with a downward sloping demand curve. But they say that higher posted toll rates have the effect of moving the demand curve rightward because the tolls displayed are the dominant indicator of the potential time savings in the express lanes. (Price is on the vettical axis, quantity on the horizontal.)
And similarly when low toll rates are posted on the variable signs, approaching drivers take that as a message that the benefits of the express lanes are less and on balance they will tend to stay out. (Aleftward move in the demand curve.)
MnPASS toll rates are subject to adjustment every three minutes under the dynamic pricing and an algorithm relates prices to measured density of traffic. The denser the traffic gets the higher the posted toll. The MnPASS algorithms were written by Raytheon and are managed by MnDOT. Raytheon also developed the programming for the much more complex dynamic pricing model used by Transurban in the 495 Express Lanes on the Capital Beltway in the Washington DC area.
The Janson-Levinson studies - with MnDOT support - involved deliberately tweaking the pricing models used on the I-394 Express Lanes for two weeks in December 2012 and for 3 weeks in January this year. Their procedure involved lowering the price rate change thresholds of traffic density - setting toll rates higher for any given level of traffic in the lanes (density.) This was done without any public announcement in an attempt to isolate drivers' responses to the toll rate signs.
The higher posted prices consistently produced more traffic in the toll express lanes (greater density) they found.
"Figure 3 reveals that the MnPASS lanes saw a consistent increase in density throughout the peak period during the field experiment. Although less than the price increase, density at nearly every time segment during the analyzed periods was higher. This led to the positive elasticity results displayed in Tables 4 and 5."
SOVs and HOVs too are attracted to the express lanes by higher displayed prices. Statistically significant elasticities ranged between s small +0.03 and a quite high +0.85.
There was no indication in aggregate driver behavior of the expected lower use with higher prices.
Janson-Levinson's report: "The increased demand resulting from higher prices (and decreased demand from lower prices) is likely a result of driver perception of the posted price. Drivers likely view the price as an indication of time savings and GP congestion, suggesting higher prices provide greater time savings. No travel times or congestion levels are made available to drivers entering MnPASS corridors, therefore, the MnPASS price may act as a signal of downstream congestion."
We asked Levinson how the models seem to work as well as they do in managing traffic if the algorithms don't account for displayed prices as proxies for time saved. Levinson demurred saying this will be the subject of followup work but he thinks constraints built in to the pricing may be in play that prevent the proxy effect from having them go completely haywire.
If they can raise toll rates and actually attract more traffic, Transurban would seem to have a ready way of doing a whole lot better financially from the 495 Express Lanes we suggested to Levinson.
Sounds like a toller's dream, a world turned upside down in which the more you charge 'em, the more they'll buy in?
How could they go wrong?
Levinson was quick to add a caution against this happy toller's thought.
"No this (positive elasticity) depends on drivers' perception of price as a proxy for time saved."
If drivers get to understand the price isn't a good proxy, that they are being misled by price as a measure of time savings, he said, their behavior will change.
Levinson and his colleague managed to successfully fool Minnesota drivers for those five weeks by monkeying with the price/density thresholds, but try and do it much more and at some point drivers will catch on he says, and their behavior will change and those demand curves in drivers' minds will cease doing their rightward shifts and the positive elasticities will go poof.
Levinson is working with MnDOT and is reluctant to discuss policy lessons.
Our take is that dynamically priced facilities need to provide drivers with hard real time data on current time savings and relative speeds so they know what they are buying. Price as proxy for service needs to be disowned. Or else they risk unstable performance of their traffic management algorithms and a serious backlash by customers feeling they've been 'had.'
$1 to $2 per minute saved
The second sensational finding from the Janson-Levinson study is the huge price drivers on the MnPASS express lanes are paying for the actual time they save versus travel in the free lanes. Time savings on the I-394ELs were mornings average 1.67 minutes and afternoons 0.8 minutes and on the I-35W lanes 2.9 minutes and 1.2 minutes. Tolls being paid averaged $2.60, $1.37, $2.91 and $2.53.
Motorists were therefore paying from $1 to $2/minute or $61 to $124 per hour of time saved. This is way over the regularly assumed value of time saved of $15 or so per hour.
The researchers note that drivers of express lanes are paying for more than time savings - they are also paying for reliability of travel time and for safety and reduced stress.
But Levinson says he's convinced drivers tend to think they are saving much more time than they actually are.
"There's a real perception problem."
You have to worry a bit how they'll react when they catch on…
Michael Janson and David Levinson, "HOT or not: driver elasticity to price on MnPASS HOT lanes":