Speed prediction models for multilane highways: Simultaneous equations approach

Past speed-related research has focused on the operational effects of roadway geometrics along rural two-lane highways using ordinary least-squares regression models. More recent research has focused on the association of traffic flow characteristics on vehicle operating speeds along multilane, limited access highways using a simultaneous equations approach. Few research studies, however, have been conducted to determine the combined association between various geometric design features and traffic flow on operating speeds along multilane highways. This research considers both geometric design and traffic flow parameters, in a simultaneous equations framework, to model the mean operating speed and speed deviation on four-lane highways (two lanes in each direction). Models for both left- and right-lane mean speeds and speed deviations were estimated. The three-stage least-squares estimator was used to investigate the possible endogeneity of mean speed and speed deviation in the system of equations and to account for the contemporaneous correlation between the disturbances across the equations. The results indicate that different geometric design features are associated with mean speed and speed deviation in the left- and right-lane models. As such, it is recommended that future multilane highway speed models consider using a simultaneous equations framework.