Data from the national traffic safety administration shows an ominous trend: traffic related deaths are up 11.3 percent for the first nine months of 2015, as compared to the same period a year earlier.
Although the NHTSA warns that the data are subject to revision, and cautions that it’s too early to discern the causes of this change, those who have been paying attention to the longer trend know that there’s every reason to believe we already have a suspect.
City Observatory noted that the decline in gas prices that started in mid-2014 has led to an increase in driving, reversing a nearly decade-long trend of Americans driving fewer miles per person per day.
What’s striking about the new NHTSA numbers is that road crash deaths are increasing much faster than total miles driven. As a result, the number of deaths per mile driven—which has been declining for decades—jumped up in the first three quarters of 2015, from 1.05 deaths per 100 million miles to 1.10 deaths per 100 million miles.
But this isn’t a new relationship: the same kind of disproportionate change occurred when gas prices increased in 2007-08. At that time, miles driven fell sharply—and traffic deaths fell even faster. In 2008, total vehicle miles traveled declined by 0.7% and in 2009, they declined a further 1.5 percent.
This suggests that the relationship between driving and deaths is non-linear: a one percent increase in driving produces a much larger than one percent increase in deaths—actually, something like a three percent increase in deaths, based on these very partial data.
Why might deaths increase faster than miles driven? There are several reasons. First, we know that some traffic phenomena, like traffic congestion, are very non-linear: the roughly 3 percent decline in traffic in 2008 produced a 30 percent reduction in traffic congestion. Second, it may well be that the additional (or, in economist-speak, “marginal”) miles driven, and marginal drivers driving them, are for some reason less safe than the typical mile driven. If we take longer trips, we may drive on more dangerous roads, or at more dangerous times. Third, it may be that people are driving faster—some research showed that high fuel prices induced motorists to slow down—and speed is strongly correlated with road safety. Over at Streetsblog, Angie Schmitt reports on David Levinson’s theory that price sensitive teen drivers may drive more when gasoline is cheap, and that may account for some of the hypersensitivity of crash rates to apparently small changes in the total amount of driving.
We know that for many reasons, there are structural connections between cheaper gas, more miles driven, and more traffic fatalities. But some analysts want to downplay those structural issues and place the blame on driver behavior. One of the most widely offered explanations for the increase in crashes and fatalities is “distracted driving,” and especially the rise of text messaging. While there’s little question that texting and driving is dangerous, and contributes to many crashes, the numbers simply don’t support this theory.
Available data suggests that text messaging grew much more rapidly between 2008 and 2013—when traffic deaths were declining—than in recent years. One source, citing data collected by Nielsen, reports that text messaging increased by a factor of four between 2007 and 2009, but has grown only about 12-15 percent per year since 2011. Forrester reports messaging increased about 14 percent between 2010 and 2011. While one could wish for much better data about the volume of text messaging, the growth of text messaging seems to coincide with a period of safer driving.
What’s clear though is that the minor changes in the amount of driving that we do seem to have disproportionately large effects on traffic fatalities. Lower gas prices that encourage more driving produce proportionately larger increases in fatalities. And higher gas prices that reduced driving in 2007 and 2008, produced disproportionately large reductions in fatalities.
This feature originally appeared in City Observatory.