We compare three empirical models of airport congestion pricing: a standard
peak-load-pricing model with econometrically estimated demand and delay fu
nctions (Morrison [7], Morrison and Winston [8]); a deterministic bottlenec
k model with traffic adjusting intertemporally to congestion prices (Vickre
y [9], Arnott et al.[2]); and a bottleneck model with time-dependent stocha
stic queuing (Daniel [4]). When implemented using Daniel's airport traffic
data, the models produce similar traffic patterns under weight-based pricin
g, but differ significantly under congestion pricing. We conclude that stru
ctural modeling of stochastic traffic, queues, and intertemporal substituti
on is necessary to produce realistic responses of traffic patterns to conge
stion pricing. (C) 2000 Academic Press.