The turning point in the number of traffic fatalities: two - - PDF document

The turning point in the number of traffic fatalities: two hypotheses about changes in underlying trends

ICTCT Workshop, Karlsruhe, October 16-17, 2014 Submitted to Accident Analysis and Prevention Ulf Brüde (VTI) and Rune Elvik (TØI)

Page 2

0.0 20.0 40.0 60.0 80.0 100.0 120.0 1940 1950 1960 1970 1980 1990 2000 2010 2020 Relative number of traffic fatalities - all-time high = 100

Relative changes in the number of traffic fatalities in six motorised countries - all-time peak number = 100

United States Great Britain Netherlands Sweden Denmark Norway United States Netherlands Norway Denmark Sweden Great Britain

Page

Two hypotheses about the turning point

Hypothesis 1 (due to Oppe):

There never was any real turning point Long-term trends were the same before-and-after

Hypothesis 2:

The underlying trends changed from before to after the turning point In particular, fatality rate declined at a faster rate

Which are the trends we are talking about?

Annual traffic growth Annual decline in fatality rate (fatalities per billion vehicle kilometre)

3 Page 4

0.000 2.000 4.000 6.000 8.000 10.000 12.000 14.000 16.000 0.000 0.200 0.400 0.600 0.800 1.000 1.200 1.400 1.600 1.800 2.000 10 20 30 40 50 60 70 Relative traffic volume (first year = 1.0) Relative fatality rate and number of fatalities (first year = 1.0) Year

A turning point following from stable underlying trends

Fatality rate Fatalities Traffic volume Fatality rate declines by 5 % per year the whole period Traffic growth slows down by 3 % each year Year 2 = 15.0% Year 3 = 15.0 • 0.97 = 13.8% Year 4 = 13.8 • 0.97 = 12.8% Year 5 = 12.8 • 0.97 = 11.8% Fatalities reach peak values in years 17 and 18

Page

How can we determine the stability of trends?

Fit functions to determine trends before turning point Predict number of fatalities after turning point based on trends before turning point If predictions are accurate, trends before the turning point continued after the turning point Conclusion: trends were stable If predictions are not accurate, trends before the turning point did not continue after it Conclusion: trends were not stable Further conclusion: turning point could not have been predicted based on trends before it occurred

5 Page 6

100 200 300 400 500 600 700 1940 1950 1960 1970 1980 1990 2000 2010 2020 Annual number of fatalities Year

Number of traffic fatalities in Norway 1952-2013 and two models describing long-term development

Model fitted to data for 1952-1970 (solid line) Model extrapolated to 1971-2013 (solid line) Model fitted to data for 1952-2013 (dashed line)

Page 7

• 2.0

0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 1950 1960 1970 1980 1990 2000 2010 2020 Percent annual growth in vehicle kilometres of travel Year

Percent annual growth in traffic volume in Norway and two models describing long-term development

Model fitted to data for 1952- 2013 (dashed line) Model fitted to data for 1952-1970 (solid line) Model extrapolated to 1971-2013 (solid line)

Page 8

0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00 1940 1950 1960 1970 1980 1990 2000 2010 2020 Fatalities per billion vehicle kilometres of driving Year

Fatality rate in Norway 1952-2013 and two models describing long-term development

Model fitted to data for 1952-1970 (solid line) Model extrapolated to 1971-2013 (solid line) Model fitted to data for 1952-2013 (dashed line)

Page

Study of six countries

Denmark (1950-2012) Great Britain (1949-2011) Netherlands (1948-2010) Norway (1952-2013) Sweden (1950-2012) United States (1948-2012) These countries were selected because:

They have all experienced a turning point around 1970 The have data on vehicle kilometres going back to about 1950

9 Page 10

200 400 600 800 1000 1200 1400 1940 1950 1960 1970 1980 1990 2000 2010 2020 Annual number of fatalities Year

Annual number of traffic fatalities in Denmark and two models describing long-term development

Model fitted to data for 1950-1971 (solid line) Model extrapolated to 1972-2012 (solid line) Model fitted to data for 1950- 2012 (dashed line)

Page 11

2000 4000 6000 8000 10000 12000 14000 16000 18000 20000 1940 1950 1960 1970 1980 1990 2000 2010 2020 Annual number of fatalities Year

Annual number of traffic fatalities in Great Britain 1949-2011 and two models describing long-term development

Model fitted to data for 1949-1966 (solid line) Model extrapolated to 1967-2011 (solid line) Model fitted to data for 1949- 2011 (dashed line)

Page 12

500 1000 1500 2000 2500 3000 3500 4000 4500 1940 1950 1960 1970 1980 1990 2000 2010 2020 Annual number of fatalities Year

Number of traffic fatalities in the Netherlands 1948-2010 and two models describing long-term development

Model fitted to data for 1948-1972 (solid line) Model extrapolated to 1973-2010 (solid line) Model fitted to data for 1948- 2010 (dashed line)

Page 13

200 400 600 800 1000 1200 1400 1940 1950 1960 1970 1980 1990 2000 2010 2020 Annual number of fatalities Year

Number of traffic fatalities in Sweden 1950-2012 and two models describing long term development

Model fitted to data for 1950-1966 (solid line) Model fitted to data for 1950- 2012 (dashed line) Model extrapolated to 1967-2012 (solid line)

Page 14

20000 40000 60000 80000 100000 120000 140000 1940 1950 1960 1970 1980 1990 2000 2010 2020 Annual number of fatalities Year

Annual number of traffic fatalities in the United States 1948-2012 and two models describing long term development

Model fitted to data for 1948-1972 (solid line) Model extrapolated to 1973-2012 (solid line) Model fitted to data for 1948- 2012 (dashed line)

Page 15 10.6 2.0

• 5.8
• 5.5

4.2

• 3.6

8.0 2.3

• 4.3
• 5.3

3.4

• 3.1

10.2 2.4

• 4.6
• 6.1

5.1

• 3.8

9.7 3.1

• 2.8
• 5.2

6.6

• 2.3

11.9 2.1

• 6.7
• 5.0

4.4

• 3.0

4.9 2.2

• 1.9
• 3.3

2.9

• 1.2
• 10.0
• 5.0

0.0 5.0 10.0 15.0 Traffic growth before Traffic growth after Fatality rate before Fatality rate after Net trend before Net trend after Mean annual percentage change Mean trends before and after turning point

Changes in trends from before to after turning point

Denmark Great Britain Netherlands Norway Sweden United States

Page

Concluding comments

The turning point in the number of traffic fatalities is common to most highly motorised countries In theory, a turning point could occur even if the long-term trends determining the number of fatalities remain unchanged For the countries included in this study, there is stronger evidence for a change in trends than for stable trends The most important contributing factor is a slowdown of traffic growth The decline in fatality rates has increased in some countries, but not all

16