By Eric Sundquist
The U.S. transportation field has tried many things to reduce traffic crashes, fatalities, and injuries: drunk-driving and seatbelt laws; in-vehicle safety improvements; wide, straight roads with crash zones; graduated licensing; and more. Yet traffic crashes still kill 35,000-40,000 Americans each year and injure millions.
Todd Litman, the prolific force behind the Victoria Transportation Policy Institute, has a new online resource that helps explain the situation, particularly to non-technical elected officials and policy people who don’t read journals. The tl;dr version: Fatalities are largely a function of miles driven (Figure 1), so you can’t be serious about Vision Zero without also being serious about VMT management.
One implication is that common practices, even those done in the name of safety, often make the situation worse. Litman writes:
Many conventional transportation and land use planning practices tend to increase total vehicle travel and crash risk (DeRobertis, et al. 2014; Dumbaugh and Rae 2009). For example, development policies that separate land uses, minimum parking requirements in zoning codes and unpriced on-street parking tend to increase motor vehicle travel (CARB 2014). Common transport planning practices, often intended to increase traffic safety, often increase total crash risks. For example, since grade-separated highways have low per-mile traffic fatality rates, transportation agencies often justify road widening, straightening, grade separation, hierarchical street systems that force traffic onto higher-speed arterials, and expanded clear zones for safety sake, but such treatments cause motorists to drive farther and faster, which tends to increase total crash casualties (Garrick and Marshall 2011; Karim 2015; Noland and Oh 2004).
For audiences rightly skeptical of descriptive statistics as above, there is a solid literature underlying the VMT-safety finding. A 2017 paper by Hamed Ahangari, Carol Atkinson-Palombo, and Norman W. Garrick at the University of Connecticut, for example, employed gold-standard panel data at the state level to the question:
We found that the variables listed in order of their contribution to the differences in traffic fatality rates between states is as follows (the elasticity values are given in brackets): Vehicle per population (+0.862), VMT per vehicle (+0.840), Infant Mortality (a proxy for the quality of health care in the state) (+0.315), Unemployment (−0.220), Percent of Fatality Caused by Under Influence Drivers (+0.186), Seat belt Usage (−0.170), Gasoline Price (−0.110), and GDP per Capita (−0.105). In other words, vehicle per population and VMT per vehicle are the two factors that contribute the most to the variation in traffic fatality between states.