Planning for an uncertain future

By Chris McCahill

Traffic forecasts and other projections are often presented as a single line on a graph or number in a chart. But we know—now more than ever—that these predictions are full of uncertainties. The Sacramento Council of Governments (SACOG), for a new study in JAPA, puts hard numbers to some of those uncertainties in order to plan better for them.

“Although forecasts conform to standard practice and offer an illusion of certainty about the future,” according to the study, “they have a poor track record.” Instead, SACOG tested an approach called robust decision making (RDM), which let them consider outcomes under a wide range of uncertainties and several policy assumptions.

Employment growth is assumed to be 49 percent in their long-range plan, for instance, but they considered outcomes ranging from 21 to 61 percent. Possible gas prices ranged from $1 to $8 per gallon. Other uncertainties included average fuel efficiency, electric vehicle sales, Millennial attitudes about driving, and VMT elasticity with respect to driving costs and economic growth.

They focused on two main outcomes for this study: emissions and mobility (measured as person-trips). By modeling thousands of scenarios in several iterations, they found their current long-range plan meets both outcomes—low emissions and high mobility—in only 12 percent of cases. The emissions standards outlined in California’s SB 375 are met in about 80 percent of cases, but SACOG’s mobility goals are met in only about 30 percent, and the two are often at odds with each other.

 

Figure 1. Share of future scenarios in which policy goals are met. Source: Lempert, et al. 2020

 

Beyond those findings, the modeling offered useful insights about specific kinds of scenarios and policy levers. For instance, faster growth, high gas prices, or lower fuel efficiency make it even harder to meet all policy goals. In the latter case, however, policies supporting higher electric vehicle penetration could double to around five percent the share of scenarios in which those policy goals are met.

In order to run thousands of necessary scenarios, the RDM model is relatively simple and therefore limited in the types of questions it can answer (the authors recommend TMIP-EMAT). The Sacramento Regional Activity-Based Simulation Model (SACSIM), which takes a long time to set up and run, was used only to provide initial inputs for the analysis. Nonetheless, “even a simple model can reveal potential vulnerabilities in proposed plans, providing planners an opportunity to address them,” according to this study.

Chris McCahill is the Deputy Director at SSTI.