HOV Lanes and VMT

Caltrans’ guidance on induced travel treats high-occupancy vehicle (HOV) lanes similar to general purpose (GP) lanes in terms of VMT generation.

If a GP lane is converted to HOV, the project is screened as unlikely to induce travel project that adds a new HOV lane generally would be expected to induce as much travel in terms of vehicle miles traveled (VMT) as a new GP lane would.

The latter finding may seem counterintuitive. If travelers are forming carpools, shouldn’t that reduce VMT? It could – and there is room in Caltrans guidance to make this case – but in many instances, carpool formation would seem unlikely.

The latest National Household Transportation Survey (NHTS) finds an average vehicle occupancy (AVO) of 1.67. The Texas Transportation Institute Urban Mobility Report, which focuses on peak-period travel, adjusts that figure downward slightly, to 1.5. These numbers may be surprising in light of the long-term decline in carpool commuting, from 12% in 2000 to 9% in 2017, as reported by AAHSTO. However, as TTI points out, “While work trips make up the largest portion of peak period trips, they are not the majority. Therefore, it would be unwise to have an average vehicle occupancy value that only uses trips to and from work.”

These figures suggest that many carpools are already on the road, and the addition of an HOV lane to an existing multi-lane facility could simply sort the existing traffic rather than prompt new carpool formation. Caltrans guidance reflects this reality. As a policy matter, these findings also suggest that adding HOV lanes often will not meaningfully address congestion.

There could be exceptions, however. Particular corridors may have fewer existing HOVs. For example, an HOV lane added to a single GP lane would be more likely to cause carpool formation than an HOV lane added to a multi-lane facility. And HOV-3+ may be more impactful than HOV-2+. Research published in Traffic Injury Prevention found that 46 percent of HOV trips involve one driver and one passenger, so HOV-3+ could cut eligible HOVs by about half compared to HOV-2+ and are more likely to spur carpool formation.

In summary, the starting point for considering induced travel from a new HOV lane would be to estimate it as we would a GP lane. Any reduction from that estimate would need justification. 

A related note: For managed lanes such as HOV to have any effect they require compliance. Unfortunately, on many corridors violation rates are significant. For example, in South Sacramento, recent traffic counts on SR 99 found violations at 47 percent (northbound) and 35 percent (southbound). Any claims about HOV effects on VMT, then, should be accompanied by support showing that lane restrictions will be enforced. And any discussion of modeling results should disclose the assumed violation rate.