Did Raising Freeway Speed Limits Affect Traffic Safety?

By Bennet K. Langlotz, J.D., B.S.E.
Senior Analyst, National Motorists Association Foundation
safetyengineering@langlotz.com
March 1999

In 1995, the U.S. Congress repealed the National Maximum Speed Limit (NMSL), which had in 1973 established a national maximum of 55 mph in response to the energy crisis of that era; the limit was raised to 65 mph in 1987. Leading up to the repeal was debate about its eventual effects, with some organizations predicting an additional 6400 traffic deaths annually if the NMSL were repealed. Following the repeal, 38 states have raised limits on at least some of their highest-speed roadways, typically including stretches of rural interstate highways. (14 states and the District of Columbia did not change speed limits during this period.) All but 9 of the limit-raising states established limits of 70 mph or greater, previously prohibited under the NMSL.

This study analyzes state-by-state fatality data published by the National Highway Traffic Administration (NHTSA), derived from the Fatal Analysis Reporting System (FARS) and the Federal Highway Administration (FHwA). These data are collected and reported using a Federal Fiscal Year (FFY) beginning on September 1 of each year.

This study compares the changes in overall fatality rates between two groups: a test group, those states that raised speed limits, and a control group, those that did not. The changes in fatality rates in these groups are calculated as a difference between FFY 1995, the first full year before limits in the test group were raised, and FFY 1997, the first full year after limits were raised. States raising limits after FFY 1997 are included in the control group, and three states raising limits late in FFY 1997 were excluded from the comparison because of insufficient exposure to higher limits, and are included only in national totals.

The study examines fatality rates per mile driven, which controls for the effects of increased driving in one group relative to the other. In this period, the increase in miles driven was greater in limit-raising states, perhaps because higher limits made highway travel marginally more appealing to some motorists. Importantly, this study also examines complete statewide fatality rate data. The changes in speed limits must be judged in terms of their overall effect on traffic safety, if any, including possible effects on other roads with unchanged limits. Such effects can occur when some traffic shifts from less safe undivided highways to faster, safer freeways. In addition, it has been argued that increases in limits has allowed law enforcement resources to be shifted to more dangerous roads, presumably leading to overall improvements in traffic safety. Also, while total fatality data is available for different road types, including those that raised limits, reliable freeway fatality rate data is not readily available for all states. Looking at statewide figures uses the more uniform Federal Accident Reporting System (FARS), and avoids the mistake of reporting fatality increases due to the likely increased popularity and usage of the higher speed roads (and commensurate but uncounted reductions on other roads) and misattribution of any increases to increased speed limits.

The federal data shows that in 1997, after the majority of states increased their maximum highway speed limit, the total national fatality rate reached an all time record low of 1.64 fatalities per 100 million vehicle miles traveled (VMT), reflecting a decades-long downward trend. Between the comparison years, the national fatality rate dropped by 4.99%, with limit-raising states in the test group dropping by a greater 5.00% (a statistically insignificant amount.) Unchanged states in the control group experienced a fatality rate reduction of 5.38%, with the difference between groups of 0.38% being statistically insignificant compared to expected year-to-year variations of more than 2% between randomly selected groups of the sizes used.

Table 1 shows a list of the test group: all states raising speed limits during Federal Fiscal Year 1996 (Sept 1, 1995-August 31, 1996). States are included when limits are raised on some highway portions to a speed that is the highest in that state. These are generally interstate highways, but in some cases include other high-speed highways. This includes states raising limits to 70 mph or greater as permitted under the repeal of the NMSL, in addition to nine states raising limits to 65 mph, a speed that was legalized in 1987 when the U.S. Congress amended the NMSL to raise the limit from 55 to 65. Thus, this study does not analyze the effects of particular speed limits, but looks at the effects of raising any maximum speed limit.

Table 1 shows that the test group of limit-raising states experienced fatality rate reduction of 5.00%, comparing the first full year prior to raising limits, and the first full year after the transition year. The total number of fatalities was nearly unchanged in these states, even as the total number of miles driven increased 5.82% during this period. Of these 33 states that raised limits, 10 happened to experience an increased fatality rate, and 23 experienced a decreased fatality rate. These numbers reflect the normal year-to-year fluctuations that can be expected to occur in each state due to random chance, although the predominance of fatality rate reductions is consistent with the aggregate reduction in fatality rate.

Table 1. Test Group: States raising highway limits during FFY 1996

VMT = Vehicle Miles Traveled (thousands)
FR = Fatality Rate
DLI = Date of Limit Increase
CML = Current Maximum Limit

Table 2 shows a list of control states in which highway speed limits remained unchanged during the transition year, as well as the comparison years before and after the transition. Two states that raised limits after the period for which data is gathered are included among the control states.

Table 2 shows that the control group of unchanged states experienced fatality rate reduction of 5.38%, comparing the first full year prior to limit increases in the test group, and the first full year after the transition year. The total number of fatalities was nearly unchanged in these states, even as the total number of miles driven increased 5.68% during this period. Of these 15 control group, 5 happened to experience an increased fatality rate, and 10 experienced a decreased fatality rate. These numbers reflect the normal year-to-year fluctuations that can be expected to occur in each state due to random chance, although the predominance of fatality rate reductions is consistent with the aggregate reduction in fatality rate. The fact that the test group states experienced a greater increase in miles driven compared to the control group (a difference of 0.65%) is consistent with the explanation that increased speed limits makes freeway travel more appealing compared to other modes of transportation, although further study and additional data collection would be helpful in establishing whether this is statistically significant.

Table 2. Control Group: States not raising highway limits during FFY 1996

Table 3 shows a list of states excluded from the comparison because they experienced a speed limit increase during a comparison year. In all three cases, the limit change occurred late in FFY 1997, providing only about three months exposure to the new limits. This is believed insufficient exposure to provide meaningful conclusions that would justify their inclusion in the test group, and makes inclusion in the control group improper because it may partially mask or diminish any observed effects.

Table 3. Excluded States, raising highway limits during comparison year of FFY 1997

VMT = Vehicle Miles Traveled (thousands)
FR = Fatality Rate
DLI = Date of Limit Increase
CML = Current Maximum Limit

Table 4 shows the aggregate of all states in the nation, including test group states, control group states, and excluded states. Between the first full year prior to limit increases in two-thirds of states, and the first full year following, the nation fatality rate dropped by 4.99% to an all-time record low of 1.64 fatalities per 100 million vehicle miles traveled. Thus, the limit-raising states in the test group experienced a slightly better result as did the nation as a whole, although the difference lacks statistical significance. The total number of fatalities was nearly unchanged in the nation, even as the total number of miles driven increased 5.68% during this period.

Table 4. National Totals

VMT = Vehicle Miles Traveled (thousands)
FR = Fatality Rate
DLI = Date of Limit Increase
CML = Current Maximum Limit

In the comparison between states in the test group and control states, a small difference of 0.38% in fatality rate reduction is observed. This difference, whether or not caused by speed limit changes or random chance, is a small number dwarfed by the order-of-magnitude greater fatality rate reduction in limit-raising and other states. Whether this small difference is actually the effect of increased speed limits or is simply due to random chance requires a determination of statistical significance. This study calculated the expected typical random difference between two groups of randomly assigned states, and determined the chance that the small difference was due to random chance.

To determine expected random variation, this study conducted 50 different trials in which states were randomly assigned to a test group of 33, and a control group of 15 (corresponding to the group sizes of the study data), using the 1995 and 1997 data. In each of the 50 trials, randomization was provided by assigning a random number to each state, and sorting the states based on the random values. In the resorted list, the first 33 were assigned to the test group, and the next 15 to the control group. For these two groups, a fatality rate change was calculated, and a difference between the two fatality rates was recorded. For each of the 50 trials, a new set of random numbers was assigned to all states, which were resorted, and calculations made to generate a new fatality rate difference. The absolute values of these 50 fatality rate differences are shown in Table 5. The average of the values is 2.80% and the median is 2.53%. These indicate that the observed fatality rate difference of 0.38% is well below typical random variations, only about one seventh of expected variations.

Of the 50 randomized calculation trials, only four instances (8%) generated a lesser difference result, indicating a 92% likelihood that the observed difference is the result of random chance, and that raising speed limits had no effect. Thus, it is extremely unlikely that raising speed limits had even a small effect on fatality rates.

Table 5. Observed Fatality Rate Percentage Differences Occurring in Randomly Selected Populations

This study establishes that increasing speed limits on freeways, particularly rural interstate freeways does not harm traffic safety. While this conclusion does not purport to establish that increased speed limits necessarily have a protective effect, it does disprove the controversial predictions that raising limits would impair traffic safety, and that 6400 deaths per year would result.

A recent unpublished report, released December 1998 by the Insurance Institute for Highway Safety (IIHS), has arrived at a different conclusion. However, this report had several critical limitations. First, it was limited to data from 1995 and 1996, when 1996 was a transition year for most states. By drawing conclusions prior to release of 1997 data, it relies on very limited data. This is of particular concern, because most limit-raising states included in the IIHS report were exposed to higher limits for less than half of one year, some for less than one month; the report appears not to have accounted for the Federal Fiscal year beginning on September 1. Also, the IIHS report omitted 9 of the 33 states that raised limits during FFY 1996, and 8 of the 16 that did not raise limits. At best, this reduces statistical significance by limiting sample size, and at worst introduces bias by omitting states that would have led to a different conclusion. In addition, the IIHS report ignored the effect of increasing vehicle miles driven, and misattributed those effects to higher speed limits.

The IIHS report concluded that the fatality rate on highways with raised speed limits increased even more than did the number of fatalities. This would be possible only if the fatalities increased and the miles traveled decreased. We know this did not happen. Fatalities may have increased, but so did miles traveled. Consequently, the fatality rate could not have increased more than the increase in fatalities. The IIHS report further claimed that fatalities increased on the highways with raised limits, but that there was no corresponding reduction on the rest of the highway system. Again, the actual numbers collected by the state and federal governments contradict this claim. Total fatalities have remained constant or slightly declined, and fatality rates have declined to record lows overall. This could not happen if there were a 15 % increase in fatalities on highways with higher speed limits and no reduction of fatalities on other streets and roads.

Raising speed limits generally stimulates more discretionary travel while attracting travel from roadways with lower speed limits to roadways with higher speed limits. This increased traffic volume on the high-speed highways often results in increased total accidents and fatalities on these same roads. However, there is a commensurate reduction of accidents and fatalities on the rest of the highway system. This pattern is borne out in individual states and the nation as a whole.

With an additional year’s data, and avoiding the problems associated with data from the transition year, this NMAF study establishes that raising limits did not affect traffic safety. This absence of an increase in the fatality rate may be readily explained, even in light of the increasing severity of accidents occurring at higher speeds. In fact, actual speeds did not increase significantly in response to increased limits.

Observations following the 1996 limit increases have shown an increase of only about 2 mph in actual speeds following an interstate speed limit increase of 10 mph. The minimal actual increase would have only a minimal effect on accident severity, and a reduction of accidents due to possible improved traffic flow may offset any severity effect. This is in accord with engineering studies showing that raising or lowering limits has minimal effect on actual speeds. The only significant effect may be that motorists that were previously traveling at illegal speeds are now in compliance with new higher speed limits, while driving at essentially the same speed. Such motorists would be essentially as safe as before, but less likely to be cited for speeding, with associated increased insurance premiums for having received the ticket.

In conclusion, the best available recent data show no effect on traffic safety when interstate speed limits are raised.