Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 27
2
World and U.S. Safety Trends
C hapter 1 explained that the safety programs of other countries seized the attention of U.S.
safety professionals and advocacy groups because of impressive declines in numbers and
rates of traffic fatalities relative to U.S. experience. In this chapter, the first section compares
traffic safety trends of the United States and other countries over the past 40 years. The second
compares trends among U.S. states, since the performance of the best states might also be a
useful benchmark for judging U.S. safety programs, along with the best performances among
other countries. The third section reviews studies that used statistical methods to explain why
some countries and states have performed better than others. The final section presents a more
detailed characterization of the U.S. traffic safety problem, describing how risks differ among
categories of roads, vehicles, regions, and drivers.
WORLD FATALITY RATE TRENDS
Nations differ greatly in traffic fatality rates (per capita and per vehicle kilometer) and in trends
in rates over time. They differ also in practices with regard to driver and vehicle safety
regulation and enforcement and road construction. The relative success of the different policies
cannot be inferred by examining the aggregate fatality rate data alone because many factors other
than government policies affect the trends. Nonetheless, the trends measure overall progress in
reducing risk and naturally have led policy makers to ask whether lessons applicable to the less
successful jurisdictions can be learned from the experiences of those that are more successful.
Most of the comparisons in this chapter are in terms of fatality rates per kilometer of
vehicle travel. Comparisons of rates of injuries and total crashes would also be valuable, but
comparable international data on these measures do not exist. Box 2-1 explains why rates per
vehicle kilometer are useful measures for comparisons.
When fatality rates for high-income and low-income countries over many years are
compared, a pattern emerges of rising per capita fatality rates in the earlier stages of motorization
of transport, followed by falling rates in the later stages. Because motorization rises with
income, fatalities per capita tend to increase with increasing income among countries with low to
medium income per capita, and then to decline with increasing income among countries with
medium to high average incomes (Figure 2-1). For example, from 1975 to 1998, reported road
traffic deaths per capita declined by 43 percent in France and 27 percent in the United States but
rose by 79 percent in India (1980–1998) and 243 percent in China (Kopits and Cropper 2005a,
170). In the poorest countries, only a small proportion of trips is by motor vehicle, and deaths
are relatively rare. However, fatality rates per vehicle kilometer of travel are high for several
reasons: the condition of infrastructure and vehicles may be poor; road users and authorities lack
experience; and on roads where motor vehicles mix with many pedestrians and cyclists, deaths of
pedestrians and cyclists are a large share of the total.
27
OCR for page 28
28 Special Report 300: Achieving Traffic Safety Goals in the United States: Lessons from Other Nations
Box 2-1
Measures for International Comparisons of Safety Performance
Some analysts have argued that total fatalities or casualties or per capita rates are more suitable
measures than rates per vehicle kilometer for benchmarking safety performance or for defining
safety goals. For example, the Organisation for Economic Co-operation and Development’s
Working Group on Achieving Ambitious Road Safety Targets avoids reporting crash rates per
vehicle kilometer, explaining (OECD and International Transport Forum 2006, 8):
The relative progress in road safety depends somewhat on what one uses as a measure of
exposure to risk (i.e., population, registered vehicles, distance travelled). There has been a
considerable debate in the past about which measure is most appropriate as an exposure
measure. Those in the health sector prefer the use of population as the denominator since it
permits comparisons with other causes of injury or with diseases. As the health and transport
sector increase their level of co-operation, fatalities per 100 000 population are becoming
more widely used.
In the transport sector, it has been common, where data are available, to use fatalities per
distance travelled (e.g. fatalities per million vehicle-kilometres) as a principal measure or
fatalities per 10 000 vehicles. Fatalities per distance travelled has traditionally been favoured
by road transport authorities as it implicitly discounts fatality rates if travel is increased.
Objections to the use of rates per vehicle kilometer to measure safety have been strongly
stated, for example as follows (Richter et al. 2001):
The use of [deaths per vehicle mile] as the criterion implicitly endorses an ethically
problematic paradigm that weighs the benefits of transportation—time saved—against the
losses—deaths and injuries. If we use absolute numbers, we hold that individuals should not
be sacrificed for collective benefits. . . . The use of time trends in [deaths per vehicle mile]
within one mode of travel precludes examining alternative strategies based on shifts to public
transport, a mode usually with much lower risks.
In this report, international and interstate comparisons are expressed in terms of rates per
vehicle kilometer and of total numbers of fatalities. One of the goals of public policy concerning
road safety is to reduce the risk of road travel. The road-using public expects government
authorities to provide safe roads. Crash and fatality rates per unit use of the road system (e.g.,
per vehicle kilometer) are measures of this risk. (In contrast, few people would argue that
reducing tobacco-related fatalities per cigarette smoked should be a goal of health policy.)
Observing rates, and not just numbers of crashes, is essential in determining the effectiveness of
most of the safety measures that road authorities have at their disposal. The reductions in total
annual fatalities in the benchmark nations are the consequence of declining rates of fatalities per
vehicle kilometer, not of declining use of the roads in those countries. This rate decline is
therefore the phenomenon that must be understood if the United States is to take advantage of
other countries’ experiences.
(continued)
OCR for page 29
World and U.S. Safety Trends 29
Box 2-1 (continued)
The number of fatalities per vehicle kilometer is an imperfect measure of road travel risk.
Data on rates for all crashes and for injury crashes by severity would be more useful in
examining the effects of safety programs, but these data are not available on a consistent basis
internationally. In addition, aggregate annual rates for entire national or state road systems hide
important geographical and temporal differences.
In wealthier countries, most trips are by motor vehicle, and thus deaths of persons who
are not motor vehicle occupants are a smaller proportion of total traffic deaths than in low-
income countries. Also, vehicle occupant fatalities per vehicle kilometer decline, presumably
because infrastructure and vehicles become safer, drivers become more skilled, traffic regulation
becomes more effective, and increasing vehicular congestion in cities slows speeds and thus
reduces crash severities. Eventually fatalities per vehicle kilometer decline enough that fatalities
per capita begin to fall. The negative correlation between degree of motorization and national
traffic fatality rate is known as Smeed’s law and has long been a subject of study and
controversy (Adams 1987).
Fatality rates per vehicle kilometer have declined greatly in every high-income country in
the past several decades (Figure 2-2a, Table 1-1), and the absolute disparity of rates among
countries has lessened (Figure 2-3). A comparison of the U.S. experience with that of 15 other
high-income countries for which 1975–2008 data are available shows that the U.S. fatality rate
was less than half the aggregate rate in the other countries in 1975 but has been higher since
2005 (Figure 2-2c). Consequently, total annual traffic deaths in the 15 countries fell by 66
percent in the period, while U.S. deaths fell by only 16 percent. The U.S. fatality rate was
among the best before 1990 but has been below the median rate of the group every year since
2001.
0 5,000 10,000 15,000 20,000 25,000
Per Capita GDP, 1985 dollars
FIGURE 2-1 Traffic fatality rate per capita versus income, 88 countries, 1963–1999.
(SOURCE: Kopits and Cropper 2005a.; copyright, Elsevier; used with permission.)
OCR for page 30
30 Special Report 300: Achieving Traffic Safety Goals in the United States: Lessons from Other Nations
10
9
8
7
fatality rate (per 100 million vehicle-kilometers )
6
5
4
3
2
1
0
1965 1970 1975 1980 1985 1990 1995 2000 2005
Australia Austria Belgium Czech Republic
Denmark Finland France Germany
Great Britain Japan Netherlands Norway
Slovenia Sweden Switzerland United States
1.8
1.6
1.4
fatality rate (per 100 million vehicle-km)
1.2
1
0.8
0.6
0.4
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
Australia Austria Belgium Denmark Finland
France Germany Great Britain Japan Netherlands
Norway Slovenia Sweden Switzerland US
(a)
OCR for page 31
World and U.S. Safety Trends 31
70,000 7000
60,000 6000
50,000 5000
annual vehicle-km (billions)
annual fatalities
40,000 4000
30,000 3000
20,000 2000
10,000 1000
0 0
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
year
fatalities, 15 countries (not including US) fatalities, US
vehicle-km, 15 countries vehicle-km, US
(b)
5
4.5
4
3.5
fatalities/100M vkmt
3
2.5
2
1.5
1
0.5
0
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
year
fatalities/100M vkmt, 15 countries fatalities/100M vkmt, US
(c)
FIGURE 2-2 (a) Fatality rates per vehicle kilometer, selected high-income countries, 1965–
2005 and 1997–2008. (b) Annual traffic fatalities and vehicle kilometers, United States and
15 other high-income countries, 1975–2009. (c) Fatalities per 100 million vehicle
kilometers, United States and 15 high-income countries, 1975–2008. Note: Countries
included in Figures 2-2b and 2-2c are Australia, Austria, Belgium, Denmark, Finland, France,
Germany, Great Britain, Israel, Japan, Netherlands, Norway, Slovenia, Sweden, and
Switzerland. (SOURCES: OECD n.d.; OECD and International Transport Forum 2010.)
OCR for page 32
32 Special Report 300: Achieving Traffic Safety Goals in the United States: Lessons from Other Nations
7
6
number of countries
5
4
3
2
1
0
0-0.2 0.2-0.4 0.4-0.6 0.6-0.8 0.8-1.0 1.0-1.2 1.2-1.4 1.4-1.6 1.6-1.8 1.8-2.0 2.0-2.2 2.2-2.4 >2.4
fatalities per 100M vehicle-km
1994 2007
FIGURE 2-3 Distribution of fatality rates of 16 high-income countries, 1994 and 2007.
Note: Countries are as in Figure 2-2, including the United States. (SOURCE: OECD n.d.)
The roughly exponential shapes of the fatality rate time trends and the bunching of
national fatality rates in the 0.6 to 1.0 range in recent years (Figure 2-3) suggest the possibility
that, as rates become lower, it becomes more difficult to obtain further reductions comparable in
absolute terms with the reductions of earlier decades. According to this interpretation of the
trends, U.S. improvement has been slow because the U.S. rate was already low 30 years ago, and
other countries have been able to improve more rapidly because improvement is easier when the
starting point is a relatively high fatality rate. These curves suggest at least that some underlying
universal phenomena have driven fatality rate trends toward convergence. It may be speculated
that the improvement reflects a learning process by all the agents—drivers, nonmotorized road
users, road authorities, health services, and law enforcement and public safety agencies—within
the road transportation system as that system develops and matures in a country. In the 1960s,
U.S. highways, vehicles, and travel patterns differed greatly from those of most of the
benchmark countries. Today, the differences persist but have narrowed.
However, the experience of the past decade no longer appears to fit this description of
convergence to similar, stable fatality rates. In a group of countries that includes the United
Kingdom, Sweden, Norway, Finland, the Netherlands, Switzerland, West Germany, and
Australia, the fatality rate per vehicle kilometer was close to or lower than the U.S. rate in 1997,
yet each achieved a greater percentage improvement in its rate than did the United States in the
1997–2007 period (Figure 2-2a). In this period, every high-income country shown in Figure 2-2
has reduced its fatality rate by a greater percentage than has the United States. Improvement in
fatality rate in the decade is only weakly correlated with the level of the 1997 rate among high-
income countries (Figure 2-4).
OCR for page 33
World and U.S. Safety Trends 33
Fatality rates of 16 countries: average annual percent change 1997-
2007 versus 1997 rate
1997 fatality rate (fatalities/100 million vehicle-km)
0
0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4
average annual % change in rate, 1997-
-1
USA
-2
-3
2007
-4
-5
2
R = 0.28
-6
-7
-8
FIGURE 2-4 Fatality rates of 16 countries: average annual percentage change for 1997–
2007 versus 1997 rate. The countries included are as in Figure 2-2. (SOURCE: OECD n.d.)
U.S. STATE FATALITY RATE TRENDS
If fatality rate trends can be used as indicators of jurisdictions with relatively successful
government safety programs, then comparisons of trends among the U.S. states might have at
least as much relevance as comparisons of the United States with other countries. The states
independently manage their traffic safety programs [although with a degree of central control
through federal-aid highway program rules and National Highway Traffic Safety Administration
(NHTSA) regulations] and are diverse with respect to demographics, geography, and
transportation system characteristics.
The pattern of fatality rates among the states in some ways mirrors that of the high-
income nations. The 2007–2008 average rate varied among the states from below 0.5 deaths per
100 million vehicle kilometers in Massachusetts and Rhode Island to 1.3 in Louisiana and 1.4 in
Montana (Figure 2-5). Similar to the distribution of national rates, the distribution of state
fatality rates (Figure 2-6) shows a shift toward lower rates and a bunching of rates in the 0.6 to
1.0 range over the past decade. The rates of four states (Massachusetts, Rhode Island,
Minnesota, and New Jersey) were lower in 2008 than that of any of the countries of Figure 2-2.
It is in the speed of improvement in highway safety that the experience of the states
differs from performance abroad. Few states could match the 4 to 6 percent annual reductions in
fatality rates that many high-income nations achieved in the period 1994–2008 (Figure 2-7).
Figures 2-8 and 2-9 show fatality rate trends for selected states that improved more slowly
(Figure 2-8) and more rapidly (Figure 2-9) than the U.S. average in the past decade. The five
states included in Figure 2-8 are those with the smallest percentage declines in the period among
all states with above-average 2008 fatality rates, excluding states with fewer than 300 traffic
deaths in 2008. The five states included in Figure 2-9 are those with the greatest percentage
declines in the period among all states with below-average 2008 fatality rates, excluding states
with fewer than 300 traffic deaths in 2008.
OCR for page 34
34 Special Report 300: Achieving Traffic Safety Goals in the United States: Lessons from Other Nations
FIGURE 2-5 State fatality rates per 100 million vehicle kilometers, 1994–1995
and 2007–2008. (SOURCE: NHTSA n.d.)
18
16
14
number of states
12
10
8
6
4
2
0
0-0.2 0.2-0.4 0.4-0.6 0.6-0.8 0.8-1.0 1.0-1.2 1.2-1.4 1.4-1.6 1.6-1.8 1.8-2.0 2.0-2.2 2.2-2.4 >2.4
fatalities per 100M vehicle-kilometers
1994-1995 avg. 2007-2008 avg.
FIGURE 2-6 Distribution of U.S. state fatality rates, 1994–1995 average and 2007–2008
average. (SOURCE: NHTSA n.d.)
OCR for page 35
World and U.S. Safety Trends 35
25
20
number of states 15
10
5
0
7
annual percent reduction
5
number of countries
4
3
2
1
0
7
annual percent reduction
FIGURE 2-7 Distribution of average annual percent reductions in fatality rates of U.S.
states (top) and of 16 high-income countries (bottom), 1994–2008. Note: In bottom graph,
countries are as in Figure 2-2, including the United States. Values for Great Britain and
Netherlands are for 1994–2007. (SOURCES: NHTSA n.d.; OECD n.d.)
SOURCES OF DIFFERENCES IN THE TRENDS
Safety researchers have attempted to understand the sources of differences in safety performance
among countries and among the U.S. states by looking for correlations between crash
frequencies or rates and the characteristics of the jurisdictions (including road conditions, safety
policies, and demographic and economic factors) that are suspected to influence crash risk. A
second research approach to this question is to measure the impacts of particular safety
interventions directly and then to judge whether the measured program effects are large enough
to explain the overall trends. Studies taking the latter approach to evaluate safety programs in
France, Australia, and the United Kingdom are described in Chapter 3.
OCR for page 36
36 Special Report 300: Achieving Traffic Safety Goals in the United States: Lessons from Other Nations
1.6
1.5
1.4
1.3
fatalities per 100M vehicle-km
1.2
LA
1.1
WV
KY
1.0
SC
PA
0.9
USA
0.8
0.7
0.6
0.5
0.4
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
year
FIGURE 2-8 Fatality rates, selected states with 2008 rate higher than the U.S. average and
with smaller than average rate declines since 1994. (SOURCE: NHTSA n.d.)
1.6
1.5
1.4
1.3
fatalities per 100M vehicle-km
1.2
MI
MN
1.1
IL
CO
1.0
NY
USA
0.9
0.8
0.7
0.6
0.5
0.4
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
year
FIGURE 2-9 Fatality rates, selected states with 2008 rate lower than the U.S. average and
with greater than average rate declines since 1994. (SOURCE: NHTSA n.d.)
OCR for page 37
World and U.S. Safety Trends 37
In general, the statistical studies take the following factors into consideration in their
crash risk models:
• Traffic characteristics, including the mix of pedestrians and vehicle types sharing the
roads, the degree of congestion, and speeds;
• Demographics: higher crash rates are expected among younger populations;
• Land use: urban and rural areas may have differences in risks;
• Road design standards and maintenance standards;
• Motor vehicle characteristics and condition, including the average age of the fleet and
the presence of passenger restraints;
• Prevalence of alcohol abuse in the population of the jurisdiction;
• Driver behaviors: the prevalence of drunk driving, the rate of seat belt use, speed,
and respect for speed and other traffic laws;
• Quality of medical services; and
• Government safety policies, including vehicle and road design standards, traffic
regulations, enforcement practices, and education and communication activities, which may
influence all of the factors listed above.
The high-income countries are diverse with respect to geography, population density, and
transportation habits. These differences affect the risks that road users confront. As one
example, in Japan and the Netherlands, pedestrians and cyclists make up a greater share of all
persons killed in crashes than in the United States (Table 2-1 and Figure 2-10).
Although exposure data are not available, it is likely that the differences shown in the
table and figure primarily reflect differences in exposure: a much larger share of all road travel
occurs on roads where motor vehicles are mixed with high volumes of bicycle travel in the
Netherlands than in the United States. Such differences are likely to affect trends in fatality
rates, but in complex ways. Trends will be affected by changes in transport habits (e.g., trends in
the relative use of bicycles and motor vehicles), and the differences will affect the relative
magnitudes of the impact of various interventions. For example, the emphasis in the Netherlands
on pedestrian and bicycle safety reflects the high share of deaths in those user categories.
TABLE 2-1 Fatalities by Category of Road User (Percentage of Total Traffic Fatalities)
Japan 2005 Netherlands 2007 United States 2007
Motor vehicle occupants 40 46 74
Bicycle riders 12 24 2
Motorcycle and moped riders
and passengers 17 8 13
Pedestrians and other
nonoccupants 31 12 12
SOURCES: Cabinet Office 2006, 9; SWOV n.d.; NHTSA 2008.
OCR for page 40
40 Special Report 300: Achieving Traffic Safety Goals in the United States: Lessons from Other Nations
measures of these factors would be difficult but might allow this kind of analysis to shed more
light on the importance of policy interventions.
The World Bank study findings are consistent with those of an earlier statistical
comparison of traffic fatalities among OECD countries using annual data for 21 countries from
1980 to 1994, which related deaths in each year in a country to demographic characteristics,
vehicles per capita, and alcohol consumption per capita (Page 2001). Fatalities were found to
increase with the percentage of young people in the population, alcohol consumption, and
percentage of the population employed, and to decrease with the percentage of the population
that is urban. The author proposes that the difference between a country’s actual trend in
fatalities over the period and the trend predicted by the statistical model is an indicator of the
effectiveness of the country’s safety interventions. Because the analysis does not include data on
safety effort, conclusions from its results concerning the effectiveness of country safety
programs are speculative. Interpretation of the statistical results is problematic because data on
vehicle kilometers of travel were not included in the analysis.
Sources of Differences Among Fatality Rates of States and Local Areas
The Insurance Institute for Highway Safety study used statistical methods to search for causes of
the disparity in highway fatality rates among U.S. states (O’Neill and Kyrychenko 2006). As
described above (and shown in Figure 2-5), the states with the highest rates have more than twice
as many fatalities per kilometer of travel as the states with the lowest rates.
The data examined were total fatalities and passenger vehicle occupant fatalities per
billion vehicle miles of travel for 3 years combined (2001 to 2003) in each of the 50 states. The
study tested whether the differences in fatality rates (annual state total traffic fatalities per
vehicle mile) among the states could be accounted for by differences in characteristics of the
populations and transportation systems: population density, the percentage of the population that
is urban, percentage age 16 to 20, median income, percentage with college degree, school
spending per pupil, highway traffic density, and average vehicle age. For example, since rural
road fatality rates are higher than urban rates nationwide, a state with a high percentage of urban
travel would have a lower total fatality rate than a more rural state, even if the two states had
identical rates on urban roads and on rural roads.
The analysis showed that most of the variation in fatality rates among the states could be
explained by differences in these characteristics and that statistical models using the
characteristics could fairly accurately predict the fatality rate ranking of each of the states. States
with a higher percentage of urban population, higher population density, higher traffic density,
higher incomes, and fewer young people had lower fatality rates. The authors conclude that
“crash death rates are strongly influenced by factors unrelated to highway safety
countermeasures. Death rates should not be used . . . to assess overall highway safety policies,
especially across jurisdictions. There can be no substitute for the use of . . . scientific evaluations
of highway safety interventions that use outcome measures directly related to the interventions”
(O’Neill and Kyrychenko 2006, 307).
The study shows how demographic factors influence state-level accident rates, but its
results are not conclusive on the question of whether differences among the states in safety
policies have affected their relative success in improving highway safety, and the study certainly
is not intended to imply that safety policies do not matter. The inclusion of policy-related factors
(e.g., the quality of the state’s roads or the intensity of enforcement) in the statistical analysis
OCR for page 41
World and U.S. Safety Trends 41
might reveal that such factors account for a measurable share of the fatality rate differences
among the states.
The second study of differences among the states (Noland 2003) focused on how
improvements in road infrastructure have affected traffic fatalities and injuries and considered
the effects of demographics, seat belt use, alcohol consumption, and quality of medical services.
Road improvements have always been an important element of U.S. safety programs. Roads
built to high design standards (for example, the Interstates) have lower average fatality rates than
roads of lower classes, so the expectation has been that upgrading the road system would
improve safety.
The study used data on annual injuries and fatalities and on various explanatory factors
for each of the 50 states for 1985–1997. Road infrastructure was measured by data on lane miles
by lane width and road class, excluding local roads. The statistical analysis also considered
measures of seat belt use (belt use rates reported by NHTSA and whether a primary seat belt use
law was in effect), demographics (state population by age cohort), quality of medical services
(infant mortality rate and hospitals per square mile), and per capita alcohol consumption.
The study concluded that there are no consistent safety benefits from improving road
infrastructure, as measured by extent, functional class, and lane width. Adding lane miles
increased fatalities. Upgrading the functional class distribution had little effect on fatalities or
injuries. A higher percentage of arterial and collector lanes with widths of 12 feet or greater was
associated with an increase in fatalities and injuries. The author notes that all of these
conclusions conflict with engineering conventional wisdom about the safety effects of geometric
improvements but are consistent with other statistical studies. For example, an earlier statistical
study (Fridstrøm and Ingebrigsten 1991, 370) using county-level data in Norway found that
when traffic expands and road capacity remains constant, casualty crashes increase by only half
the increase in traffic and so the crash rate declines, but when traffic volume and road capacity
both expand at the same rate, crash rates are unchanged.
This study, as did the World Bank study, used very approximate measures of some of the
explanatory factors because no direct measure was available. The analysis did not use vehicle
kilometers of travel as an explanatory variable because, the author explains, vehicle kilometers
are highly correlated with population, which was included. The omission of vehicle kilometers
from the model means that a plausible alternative explanation for the findings cannot be
excluded—that is, that a larger stock of infrastructure is observed to be related to higher fatalities
because more infrastructure indicates more travel rather than because more infrastructure
increases the risk of travel.
The age distribution of the population was found to have a large effect. When the
percentage of the population between ages 15 and 24 years increases, fatalities and injuries
increase. When the percentage of the population over age 75 increases, fatalities and injuries
decrease, perhaps because this age cohort travels less by road. An increase in seat belt use and
the existence of a primary seat belt law both are found to reduce fatalities, but seat belt usage
does not affect injuries. Lower alcohol consumption reduces fatalities but not injuries.
Improvement in the quality of medical services, as approximated by the infant mortality
rate in the state, reduces fatalities but does not have a significant effect on injuries. This result
reinforces the conclusions of other research (Zwerling et al. 2005), which found by a different
analysis method that, when crash severity is controlled for, persons injured in rural crashes have
a lower chance of survival than persons injured in urban crashes, and that this difference
accounts for an important share of the difference between urban and rural fatality rates. The
OCR for page 42
42 Special Report 300: Achieving Traffic Safety Goals in the United States: Lessons from Other Nations
largest positive effects (as indicated by the numbers of 1985 fatalities that would have been
avoided if the 1997 values of the variables had prevailed) in the Noland study were for seat belt
use, age distribution, and alcohol consumption.
The two U.S. studies summarized above are representative of numerous studies that have
used data on fatality or casualty frequency in multiple U.S. states over a period of years to assess
statistically the effects of particular interventions (e.g., seat belt laws) or to explore the possible
causes of interstate differences in casualty frequency and rate. Another recent study in this group
(Babcock and Gayle 2009) includes a literature review. In general, the studies find that external
factors (e.g., demographic and travel characteristics) account for a large share of variation in
casualties over time and among states and that a large share of interstate and temporal variation
is unexplained by the factors considered. Some studies conclude that specific interventions are
effective, but the effects usually appear to be small in comparison with the overall variation
among states and over time.
Concluding Observations
None of the studies offers a satisfactory comprehensive explanation for the general pattern of
declining and converging fatality rates among countries and among the U.S. states shown in
Figures 2-2 and 2-6. However, a small number of factors appear to be important in driving the
trends:
• The aging of the populations of the high-income countries has reduced fatality rates.
• Increasing congestion appears to reduce rates, presumably through its effect on speed.
• Higher alcohol consumption and alcohol abuse in the general population lead to
higher traffic fatality rates.
• Higher seat belt use decreases fatalities.
• Improved quality of medical services reduces fatality rates. The most important
effect may be the speed and quality of emergency medical services, but the statistical studies
were not refined enough to isolate this aspect of medical systems.
A lesson that all the studies support is that differences in national- or state-level rates are
imperfect indicators of successful safety policies, because differences in these rates reflect to a
great extent differences in fundamental demographic, economic, and geographical
circumstances. Therefore, to find the best international models for the United States to emulate
and to draw the right conclusions from these models, detailed examinations of specific policies
and programs—how they were implemented and the results they produced—will be needed.
FACTORS AFFECTING U.S. FATALITY RATE TRENDS
The previous sections identified characteristics of populations (especially the age distribution)
and highway systems (including the distribution of traffic between urban and rural areas, which
is an indicator of congestion, speed, and timeliness of emergency response, and the mix of kinds
of motorized and nonmotorized vehicles and pedestrians on the roads) that influence fatality
rates and trends. As an aid to interpreting U.S. trends, this section describes coincident trends in
population age distribution, the urban and rural distribution of travel, and the mix of size and
OCR for page 43
World and U.S. Safety Trends 43
types of vehicles on the roads. Chapter 4 will describe the U.S. incidence of high-risk behaviors
(drunk driving, speeding, and failure to use occupant protection) that also influence trends and
differences among countries.
Demographics
Research summarized in the previous section showed that countries with aging populations
experience declines in highway fatality rates. U.S. drivers aged 16 to 20 years are involved in
fatal crashes more than twice as frequently, per licensed driver in the age group, than drivers
over age 35 (Figure 2-11). In the period 1997 to 2001, the fatal crash involvement rate per
kilometer driven for drivers aged 16 to 20 years was 5 times the rate for drivers aged 45 to 54
years, and the rate per kilometer driven for drivers older than 75 years was nearly 4 times greater
than the rate for drivers aged 45 to 54 years (GAO 2003, 18). Similar patterns probably hold in
other countries.
The median age of the U.S. population is lower than in most other high-income nations.
This characteristic probably tends to elevate the U.S. fatality rate in comparison with other
countries. However, the rate of aging of the U.S. population is in the middle of the range for
high-income countries (Figure 2-12); therefore, differences in the rate of aging probably do not
explain much of the difference between the United States and other countries in the rate of
decline of crash rates in recent decades.
Urban and Rural Travel
One factor that can explain part of the variation in fatality rates across U.S. states is differences
in the distribution of travel by road type and by urban versus rural setting. Fatality rates per
vehicle kilometer are 2 to 3 times higher on roads in rural areas than on urban roads of similar
design and function (Figure 2-13). Fatality rates on secondary roads (the collector and local
classes in Figure 2-13) are 1.5 to 3 times higher than on roads built to Interstate highway
standards (limited-access divided highways) (FHWA n.d.).
Since the states differ in the fraction of travel that is urban and in the distribution of travel
by road class, the differences in fatality rates shown in Figure 2-13 account for part of the
variation in fatality rates across states. In particular, rural states tend to have high fatality rates.
Some states in which both rural and urban rates are lower than the national averages have total
rates above the national average because a high proportion of their travel is rural. Similar
differences in the mix of travel by road type and land use, and trends over time in this
distribution, probably account for some part of observed international differences in fatality rates
and trends.
The important policy problems are to determine why these differences by road type exist
and what can be done to reduce fatality rates in the higher-risk road segments. Part of the
difference in risk presumably relates to speeds (e.g., urban Interstates are more subject to
congested, slower-speed operations) and to slower emergency response on rural roads. There
may be other systematic differences among road classes in the frequency of alcohol-impaired
driving, seat belt and helmet use, mix of vehicle types, and driver age distribution.
OCR for page 44
44 Special Report 300: Achieving Traffic Safety Goals in the United States: Lessons from Other Nations
50
45
involvements per 100,000 drivers
40
35
30
25
20
15
10
5
0
16-20 21-24 25-34 35-44 45-54 55-64 65-74 >74
age
FIGURE 2-11 Driver involvements in fatal crashes, per 100,000 licensed drivers, by age,
United States, 2008. (SOURCE: NHTSA 2009, 100.)
FIGURE 2-12 Median age in 2000 (top) and percentage change in median age between
1975 and 2000 (bottom) for various countries. (SOURCE: United Nations 2002, Annex III.)
OCR for page 45
World and U.S. Safety Trends 45
2.5
fatalities per 100M vehicle-k m
2
1.5
1
0.5
0
Interstate Arterial Collector Local
Rural
Urban
FIGURE 2-13 U.S. fatality rates by road class, 2007. Note: Arterials are roads designed to
carry relatively high traffic volumes, usually at high speed. Local roads provide direct access to
developed property and serve local trips; most are designed for relatively low volumes and low
speeds. Collector roads are intermediate in function and design between local roads and
arterials. (SOURCE: FHWA n.d.)
Vehicle Mix
The mix of vehicles in the United States has been changing over time and differs from that in
many other countries. For example, in the United States, travel by light trucks (a category that
includes light vans and sport-utility vehicles) has been growing more rapidly than that for
passenger cars. The number of passenger cars involved in fatal crashes each year has been
falling, while the number of light trucks involved increased from at least the 1970s until 2005
before beginning to decline. The number of motorcycles involved in fatal crashes increased
sharply through 2008 (Figure 2-14). Motorcycle occupant fatalities declined from 2008 to 2009.
Whereas fatal involvement rates for cars and light trucks have been falling, motorcycle
fatal involvement rates have risen sharply since the late 1990s. NHTSA reports that the fatal
crash involvement rate of motorcycles nearly doubled between 1998 and 2005 (from 14.1 to 27.8
involvements per 100 million motorcycle vehicle kilometers), then declined moderately by 2008
(to 23.0 involvements per 100 million vehicle kilometers). In the 1998 to 2008 period, the fatal
involvement rate declined for cars by 30 percent (from 1.2 to 0.8 involvements per 100 million
vehicle kilometers), for
OCR for page 46
46 Special Report 300: Achieving Traffic Safety Goals in the United States: Lessons from Other Nations
35,000
30,000
25,000
vehicle involvement s
20,000
15,000
10,000
5,000
0
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
year
passenger cars light trucks
large trucks motorcycles
FIGURE 2-14 Number of vehicles involved in fatal crashes, by vehicle type, United States,
1994–2008. (SOURCE: NHTSA 2009, 17.)
light trucks by 26 percent (from 1.4 to 1.0 involvements per 100 million vehicle kilometers), and
for large trucks by 29 percent (from 1.6 to 1.1 involvements per 100 million vehicle kilometers).
Thus in 2008, NHTSA reports that the motorcycle fatal involvement rate was 29 times the rate
for cars. Estimates of vehicle kilometers of travel of motorcycles are much more uncertain than
for other vehicle classes because motorcycles make up only a small fraction (less than 1 percent)
of all vehicles on the roads. Consequently, the reliability of the estimated trend of motorcycle
fatal involvement rate per vehicle kilometer is unknown. The 1998–2008 increase in motorcycle
fatal involvements per registered motorcycle was only 15 percent (NHTSA 2009, 17).
The Business Cycle
A 1984 study by a NHTSA analyst showed that U.S. traffic fatalities over the period 1960–1982
correlated closely with trends in population, employment, and unemployment, once adjustments
were made for the 1973–1974 oil embargo and for the imposition of the 55-mph speed limit.
The correlation raised the question of whether any of the slowdown in the growth of fatalities
since the late 1960s could be attributed to the new federal highway safety programs introduced in
the 1960s and 1970s. An update of the analysis (Partyka 1991) found that the model fit to the
1960–1982 data predicted future fatalities poorly: the number of fatalities in 1983–1989 steadily
declined compared with the level that extrapolation of the historical relationship with population
and employment would predict. (The gap was 19,000 fewer fatalities in 1989.) When the
OCR for page 47
World and U.S. Safety Trends 47
original model was refit to data for 1960 to 1989, some correlation remained, but it was much
weaker (R2 = .64 versus .98).
In the 1991 update study, the author speculates that over half of the 1980s decline in
fatalities relative to the prior trend might be attributable to the effects of the increase in the use of
seat belts and the decrease in the incidence of drunk driving between 1983 and 1989. The author
estimates that 9,700 fewer traffic deaths occurred in 1989 than if belt use and drunk driving had
remained at 1983 levels. The study results suggest that external economic factors are important
in explaining safety trends, and in particular trends over shorter time periods, but do not by
themselves fully account for long-term safety trends.
U.S. traffic deaths declined by 9.3 percent from 2007 to 2008 and by 8.9 percent from
2008 to 2009 (NHTSA 2009; NHTSA 2010). These annual declines were two of the largest on
record. The U.S. economy entered a recession in 2007, and the declines are consistent with
experience in past recessions. The largest annual declines in U.S. traffic fatalities in the period
1971–2007 all occurred in the recession years of the period: 7.0 percent in 1991, 9.9 percent in
1982, and 16.4 percent in 1974 (the latter from the combined effects of recession and the oil
embargo). U.S. traffic fatalities increased when economic growth resumed after these past
recessions. In the 15 high-income countries shown in Figure 2-2b (not including the United
States), total fatalities declined by 9.0 percent from 2007 to 2008 and by 5.6 percent from 2008
to 2009, somewhat less than the U.S. annual declines. The employment impact of the recession
that began in 2007 was more severe in the United States than in most other high-income
countries: the number of unemployed increased by 102 percent between 2007 and 2009 in the
United States, compared with 29 percent in the other European Organisation for Economic Co-
operation and Development member countries (OECD 2010). The significance of these short-
period traffic safety trends is difficult to interpret, especially since data on traffic volumes in the
period are not available for most countries. As Figure 2-2b shows, U.S. annual vehicle
kilometers traveled declined from 2007 to 2008; this was the first annual decline since 1980.
U.S. vehicle kilometers traveled rose by 0.2 percent from 2008 to 2009 (NHTSA 2010).
Concluding Observations
Differences in demographics, in the urban-versus-rural distribution of road travel (and the
associated distribution of travel by congested and uncongested conditions), in the distribution of
travel by road class, and in the mix of vehicle types using roads can account for a portion of the
differences in fatality rates between the United States and other countries and among the U.S.
states. However, these factors may not explain a large share of differences in trends in fatality
rates over the past decade or two. Economic cycles and isolated shocks, such as the 1970s
energy crisis, can affect the crash rate trend in the short run.
The age distribution of the population is an external factor that is not directly affected by
transportation policies, and road designs and the urban-versus-rural distribution of travel change
only slowly. However, interventions can be targeted to the segments of road use that are
associated with high risk. For example, licensing and testing requirements can target younger
and older drivers, and highway network screening to identify and treat high hazard locations can
reduce crashes on roads with high crash rates, provided the treatments selected are guided by
sound research and evaluation.
OCR for page 48
48 Special Report 300: Achieving Traffic Safety Goals in the United States: Lessons from Other Nations
REFERENCES
Abbreviations
FHWA Federal Highway Administration
GAO General Accounting Office
NHTSA National Highway Traffic Safety Administration
OECD Organisation for Economic Co-operation and Development
SWOV Institute for Road Safety Research (Netherlands)
Adams, J. G. U. 1987. Smeed’s Law: Some Further Thoughts. Traffic Engineering and Control, Feb.,
pp. 70–73.
Babcock, M. W., and P. G. Gayle. 2009. State Variation in the Determinants of Motor Vehicle Fatalities.
Journal of the Transportation Research Forum, Vol. 48, No. 3, Fall, pp. 77–96.
Cabinet Office. 2006. White Paper on Traffic Safety in Japan 2006: Abridged Edition. Oct.
http://www8.cao.go.jp/koutu/taisaku/h18kou_haku/english/wp2006-1.pdf.
FHWA. n.d. Fatality Rate by Road Function Class Table.
http://safety.fhwa.dot.gov/speedmgt/data_facts/.
Fridstrøm, L., and S. Ingebrigsten. 1991. An Aggregate Accident Model Based on Pooled, Regional
Time-Series Data. Accident Analysis and Prevention, Vol. 23, No. 5, pp. 363–378.
GAO. 2003. Highway Safety: Research Continues on a Variety of Factors That Contribute to Motor
Vehicle Crashes. March.
Kopits, E., and M. Cropper. 2005a. Traffic Fatalities and Economic Growth. Accident Analysis and
Prevention, Vol. 37, pp. 169–178.
Kopits, E., and M. Cropper. 2005b. Why Have Traffic Fatalities Declined in Industrialized Countries?
Implications for Pedestrians and Vehicle Occupants. Policy Research Working Paper 738. World
Bank, Aug.
Kopits, E., and M. Cropper. 2008. Why Have Traffic Fatalities Declined in Industrialized Countries?
Implications for Pedestrians and Vehicle Occupants. Journal of Transport Economics and Policy,
Vol. 42, Part 1, Jan., pp. 129–154.
NHTSA. 2008. Traffic Safety Facts 2007.
NHTSA. 2009. Traffic Safety Facts 2008.
NHTSA. 2010. Highlights of 2009 Motor Vehicle Crashes. Aug.
NHTSA. n.d. Fatality Analysis Reporting System Encyclopedia: Fatalities and Fatality Rates by State,
1994–2008. http://www-fars.nhtsa.dot.gov/States/StatesFatalitiesFatalityRates.aspx.
Noland, R. B. 2003. Traffic Fatalities and Injuries: The Effect of Changes in Infrastructure and Other
Trends. Accident Analysis and Prevention, Vol. 35, No. 4, July, pp. 599–611.
OECD. 2010. Labour Force Statistics (MEI): Harmonised Unemployment Rates and Levels (HURs).
http://www.oecd.org/topicstatsportal/0,3398,en_2825_495670_1_1_1_1_1,00.html.
OECD. n.d. International Road Traffic Accident Database. http://www.swov.nl/cognos/cgi-
bin/ppdscgi.exe?toc=%2FEnglish%2FIRTAD.
OECD and International Transport Forum. 2006. Country Reports on Road Safety Performance. July.
OECD and International Transport Forum. 2010. Press Release: A Record Decade for Road Safety:
International Transport Forum at the OECD Publishes Road Death Figures for 33 Countries. Sept. 15.
O’Neill, B., and S. Kyrychenko. 2006. Use and Misuse of Motor Vehicle Crash Death Rates in
Assessing Highway Safety Performance. Traffic Injury Prevention, Vol. 6, No. 4, Dec., pp. 307–318.
Page, Y. 2001. A Statistical Model to Compare Road Mortality in OECD Countries. Accident Analysis
and Prevention, Vol. 33, No. 3, May, pp. 371–385.
Partyka, S. C. 1991. Simple Models of Fatality Trends Revisited Seven Years Later. Accident Analysis
and Prevention, Vol. 23, No. 5, pp. 423–430.
OCR for page 49
World and U.S. Safety Trends 49
Richter, E. D., P. Barach, E. Ben-Michael, and T. Berman. 2001. Death and Injury from Motor Vehicle
Crashes: A Public Health Failure, Not an Achievement. Injury Prevention, Vol. 7, pp. 176–178.
SWOV. n.d. Casualties by Mode of Transport.
http://www.swov.nl/uk/research/kennisbank/inhoud/00_trend/01_monitor/casualties_by_mode_of_tra
nsport.htm.
United Nations. 2002. World Population Aging: 1950–2050. Department of Economic and Social
Affairs, Population Division.
Zwerling, C., C. Peek-Asa, P. Whitten, S. Choi, N. Sprince, and M. Jones. 2005. Fatal Motor Vehicle
Crashes in Rural and Urban Areas: Decomposing Rates into Contributing Factors. Injury Prevention,
Vol. 11, pp. 24–28.
OCR for page 50
