Modelling the ability of fare to spread AM peak passenger loads using rooftops

‘Rooftops' originates from theoretical work by Hotelling into location choice. In the 1970s, Hotelling’s ideas were applied to passengers' choice of intercity trains which gave rise to the name ‘rooftops' since the graphs resembled streets of rooftops. Rooftop models have focused on time but including fare is not difficult requiring ‘values of time' to convert fare into minutes. As a demonstration, a rooftops model was applied to a rail corridor to assess the effect of discounts and surcharges on passenger's choice of AM peak trains. The demand parameters such as the value of time and displacement were estimated by Stated Preference market research. Dollar for dollar, the sensitivity to a fare discount was less than for a surcharge and this difference carried through to the value of onboard time. Passengers also valued late displacement higher than early displacement. The model predictions were compared with observed counts of train loads with calibration factors introduced to bring the predicted loads closer to the observed loads. The calibrated model was then used to predict the impact of fare discounts and surcharges. A peak hour surcharge was forecast to reduce peak hour loads more than discounts offered on early and late peak trains. Discounting early peak trains was also forecast to be more effective than discounting late peak trains.