By Eric Sundquist
As commute times increase, married women work fewer hours or even drop out of the workforce according to a forthcoming article in the Journal of Urban Economics. The finding, summarized for a general audience recently in Atlantic Cities, helps explain differences in women’s workforce participation across various metro areas.
This insight may seem less important to transportation practice than to labor-market studies. But in its focus on travel time as a driver of economic outcomes, the article has clear relevance to transportation agencies that are wrestling with setting meaningful, outcome-based performance measures.
The paper, by Dan A. Black of the University of Chicago and others, is only the latest to look at economic outcomes of travel time. Questions such as whether people have a fixed “budget” for travel time and how they value their time spent traveling are mainstays of the literature, and of practice. For example, value of time is central to mode-choice modeling. The idea is that time spent traveling is a cost, and travelers would prefer to reach their destinations with less time cost, i.e., in less time.
What this literature does not address is precisely what transportation practice usually focuses on: speed of travel, which often is framed in terms of “delay”, or the difference between observed traffic speeds and some ideal speed. The literature focuses on time rather than delay for the precise reason that it matters how long it takes to get somewhere, not how fast the traveler is going on the way.
The Urban Economics paper tends to conflate these concepts, discussing travel time, delay, and travel distance as if they were all the same thing (and though travel time for a metro area is partly determined by mode split, they do not address mode at all). The authors are economists, so they get a pass on this lack of clarity, and soundness of their finding is not affected since it is based only on travel time. Transportation practitioners working on metrics, however, need to think clearly about the concepts and what we can learn by tracking them.
Delay certainly is of interest, despite the claims above. Daily or hourly variations in delay, for example, are a measure of reliability or uncertainty. High levels of uncertainty clearly are an important cost to travelers.
But delay per se is just one element in determining travel time. Others are distance, mode, and speed itself without regard to delay–even if there is no delay, but the normal speed of traffic or transit is slow, it will take longer to get around than if it is faster.
If delay were the key variable in determining travel time, we would see travel time increase in places where delay was growing and vice versa. But as a previous article in this newsletter pointed out, in a follow-up to the latest edition of the Texas Transportation Institute’s Urban Mobility Report, we don’t see this. There is no statistically significant relationship between change in delay (as measured by TTI) and change in commuting time (as measured by the Census Bureau). However, TTI still refers to its delay measure as the “travel-time index,” a label that makes little sense from a traveler perspective.
How can delay be so different from travel time? One reason may be alleged methodological problems in the TTI report. But even if the TTI measure is valid, lower speed doesn’t always mean longer travel times. For example, we know people are driving less on average, meeting their needs with fewer vehicle-miles traveled. Aggregate traffic hasn’t dropped much, due to population growth that offsets the per capita VMT decline, so we still have delay. But with lower per capita VMT the average traveler is exposed to less time in traffic, simply by driving fewer miles.
The relationship between delay and travel time is more complicated than this brief account.
Even in the 1990s when per capita VMT was moving up, there was no statistical relationship between delay and travel time (see page 34 in this report). Suffice it to say that there is a disconnect between what we conventionally measure (delay) and what matters to travelers (travel time). As agencies adopt measures that help guide their thinking, these distinctions are critical. Delay should not be ignored, especially as a reliability measure, but by itself it is not a particularly robust measure of system performance.
Eric Sundquist is Managing Director of SSTI.