In the first article on the Luminant Analytics’ blog in January, we noted how US road fatalities have been increasing overall. In this second article, we want to drill deeper and explore whether the trend was general, or whether a regional component is at play. To understand this better, we look at fatality trends in the four census regions. As Figure 2-1 indicates, much of the rise in overall road fatalities, after being adjusted for vehicle miles traveled, has been driven by the Western and Southern regions. In Western US, road fatalities increased by 10.2 % from 2010 to 2016, while in the South there was an increase of 4.8%. A very different pattern of still declining fatality rates is seen in the Midwest and Northeast, though of a lower magnitude than of the rates in 2004-2009.
Sources: Fatalities-FARS 00-16; Vehicle miles traveled (SA), US Bureau of Transportation Statistics, 00-16; Divisions-US Bureau of Census
However, considering the size of the US, it is important to note that not all divisions in the Southern and Western regions have fared equally poorly, as seen below.
Sources: Fatalities-FARS 00-16; Vehicle miles traveled (SA), US Bureau of Transportation Statistics, 00-16; Divisions-US Bureau of Census
The states in the Pacific region in Western U.S. and South Atlantic in the South stand out with the largest rise in fatality rates. However not all states in those divisions have the same experience. Arkansas, California, Oregon, Delaware, DC and Florida have encountered a rise in fatalities by more than 30% in the past six years, exhibiting a divergence from their deep decline in 2004-2009. But not all is catastrophic, as seen is the still declining fatality rates in Maryland and West Virginia
Sources: Fatalities-FARS 00-16; Vehicle miles traveled (SA)-US Bureau of Transportation
Statistics, 00-16; Regions-US Bureau of Census
Sources: Fatalities-FARS 00-16; Vehicle miles traveled (SA)-US Bureau of Transportation Statistics,00-16; Regions-US Bureau of Census
These trends beg the answer to two key questions: why do fatality rates vary across regions and divisions, and why are neighboring states often so different? In general, while a growing economy and cheaper gas prices lead to more vehicles on the road, the extent to which it does often varies by other state-level characteristics. However, that variation is more or less captured by the vehicle miles traveled, and these fatality trends are evident even after normalizing the fatality rates by increased vehicle travel. There is also the hypothesis of more speeding -related crashes leading to higher fatalities. We have tested this hypothesis using FARS data, where we looked at how posted speed limits prior to a fatal crash have changed over time, for the most problematic states identified above.
Sources: Fatalities-FARS 00-16; Vehicle miles traveled (SA)-US Bureau of Transportation Statistics, 00-16
We find mixed evidence for the speeding hypothesis for this subset of high fatality states. While some of the high fatality states (CA, FL, DC) posted higher speed limits from 2010-2016 than from 2004-2009, in some others (GA, NC, DE) the speed limits dropped recently. In the better performing states, the evidence is again mixed as while speed limits have declined in West Virginia, they have risen in Maryland.
What we can elicit from the data above with a certain degree of confidence is that while speeding has a role in increasing road fatalities, it is certainly not the silver bullet answer to the question. Changes in driver behaviors like increasing incidence of distracted driving with the ubiquitous use of smartphones, and relatively higher numbers of inexperienced drivers could explain just part of the puzzle. Yet another contributing factor could be variation in enforcement of speed limits and distracted driving laws across the different states. Fundamental differences in transportation infrastructure across states and across time could also explain some of the regional trends. Luminant´s blog articles will tackle some of these hypotheses with a critical data-oriented overview in the coming months.
Whatever the reasons for the current trend, it is clear that unraveling this puzzle will not be an easy task as it requires gathering the big picture view of evolving driver behavior, vehicle features, economic conditions and road characteristics, among other elements.
