Content uploaded by Mary Becker
Author content
All content in this area was uploaded by Mary Becker on Oct 25, 2022
Content may be subject to copyright.
Journal of Economic Studies
Obesity and motor vehicle deaths: A panel-data analysis
Journal:
Journal of Economic Studies
Manuscript ID
JES-03-2019-0097.R2
Manuscript Type:
Research Paper
Keywords:
Obesity, Motor Vehicle Fatalities, Seat Belt Laws, Injury Prevention,
Highway Safety
Journal of Economic Studies
Journal of Economic Studies
The authors thank an anonymous referee and Dr. Mohsen Bahmani-Oskooee, Editor of the Journal of Economic
Studies, for helpful comments and suggestions and assume responsibility for any errors.
Obesity and motor vehicle deaths: A panel-data analysis
Abstract
Purpose – This paper analyzes the impact of obesity on the probability of a motor vehicle
fatality (highway death rate) and on its component probabilities: the probability of a fatality,
given a crash (vulnerability rate), and the probability of a crash (crash rate).
Design/methodology/approach -- Using state-level data for 1995-2015, the paper estimates
models explaining all three rates. Explanatory factors include obesity and a representative set
of potential determinants.
Findings – Results indicate that obesity has a statistically significant positive relationship with
the highway death rate and the crash rate. Also having a statistically significant positive
association with at least one of the three rates are the proportions of young and old drivers,
alcohol consumption, the ratio of rural to urban vehicle miles, and temperature. Factors with a
statistically significant negative relationship with at least one of the rates include primary seat
belt laws and precipitation. In 2016, 928 traffic fatalities could have been avoided if obesity
rates decreased by one percentage point.
Practical implications –Seat belts and crash dummies should be better designed to fit and
represent those with higher BMI’s, and education efforts to increase seat belt use should be
supplemented with information about the adverse impact of obesity on highway safety.
Originality/value – This paper uses 21 years of state-level information, including socio-
economic and regulation data, and contributes to the existing research on the relationship
between obesity and highway safety.
Keywords Obesity, Motor Vehicle Fatalities, Seat Belt Laws, Injury Prevention, Highway Safety
Article Classification Research paper
Page 1 of 21 Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
2
1. Introduction
The relationship between the body weight of drivers and the risk of fatality from motor
vehicle crashes has been the subject of increasing scrutiny from various disciplines, including
the fields of medicine, public health, psychiatry, economics and environmental health services.
Although motor vehicle fatality rates in the United States have been trending downwards for two
decades, obesity rates have grown during that same period, as shown in Figure 1. In 1995, the
first year of data for this study, the average percentage of a state’s population that was obese
was 15.6 percent, and the average state’s rate of motor vehicle deaths was 17.8 per billion
vehicle miles travelled. By 2015, the average state’s obesity rate had increased to 29.6 percent
and the average state’s motor vehicle death rate had decreased to 11.6 per billion vehicle miles
travelled. However, for the most recent time period of this study, from 2011 to 2015, the state’s
average rate of motor vehicle deaths has reversed course and increased from 11.4 to 11.6
deaths per billion vehicle miles travelled. This has occurred while obesity rates have continued
to rise from an average of 27.8 to 29.6 percent.
Figure 1: Mean motor vehicle fatality and obesity rates for U.S.
The chart depicts the average number of state motor vehicle deaths per billion vehicle miles
travelled (solid line) and the average state obesity rate, as a percent (broken line) for the years
1995 – 2015.
Page 2 of 21Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
3
This paper examines the relationship between rising obesity conditions and motor
vehicle fatalities. Contrary to the pattern suggested above, much, if not most, of the recent
research has identified a positive relationship between the obesity rate of drivers and motor
vehicle crashes or fatalities. These papers have used either Crashworthiness Data System
(CDS, https://www.nhtsa.gov/national-automotive-sampling-system-nass/crashworthiness-data-
system) or Fatal Accident Reporting System (FARS, https://www.nhtsa.gov/research-
data/fatality-analysis-reporting-system-fars) data, which report data from individual crash
scenes. These data sets include information relating to crash characteristics (car make, vehicle
age, time of day, weather conditions, road type, airbag deployment, etc.) and driver
characteristics (age, race, gender, seat belt use, body mass index (BMI)[1], severity of injury,
and alcohol or drug use). A gap in the research exists, however, as these studies do not
include socio-economic and government regulation data, such as income and seat belt laws.
This paper fills that gap and broadens the discussion of the costs to society of obesity.[2]
Using a panel of annual state-level data for the years 1995-2015, the results of this
paper confirm a positive relationship between state obesity rates and motor vehicle death rates
in the U.S. The data suggest this relationship is driven by increasing crash rates, a result that
has not been found previously in a multi-year study. In fact, the results suggest that an increase
of obesity by one percentage point is associated with an additional 928 traffic fatalities each
year.[3]
The remainder of the paper is organized as follows; section 2 provides a literature
review, section 3 presents the model and defines the variables, section 4 discusses results,
section 5 provides robustness tests and section 6 concludes.
2. Literature Review
There is a substantial body of literature on the factors that influence U.S. motor vehicle
safety. Loeb et al. (1994) review much of the earlier empirical research in this literature and
Blattenberger et al. (2013) summarize findings that are more recent. The reviewed studies
typically estimate the effects of potential causal factors on measures of highway safety (e.g.,
fatality rates such as deaths per vehicle mile) utilizing statistical methods (e.g., multivariate
regression) on observational data. Causal factors considered include characteristics of vehicle
occupants, economic considerations, vehicle characteristics, and aspects of the driving
environment. Some of the research also examines the impact of governmental policies (e.g.,
speed limit laws, seat belt laws). Among this literature, Simmons and Zlatoper (2010) uniquely
account for the occupant characteristic of body weight. Controlling for other causal factors and
Page 3 of 21 Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
4
using state-level data for 2005 only, they find that highway fatality risk increases with the
percent of a state’s population that is obese. This paper extends their work by examining for the
first time the relationship between highway safety and body weight utilizing state-level data for
over 20 years.
Recent research on the relationship between obesity and motor vehicle fatalities has
focused on the determinants of death or serious injury to drivers given the event of a motor
vehicle crash. The consensus from this research is that obesity is associated with a greater risk
of fatality in the event of a motor vehicle crash. (See, in particular, Carter et al. (2014),
Desapriya et al. (2014), Dunn and Tefft (2014), Jehle et al. (2012), Joseph et al. (2017), and
Rice and Zhu (2013)). Although earlier research indicated that heavier drivers and passengers
might be less at risk of dying once a crash has occurred because of the so-called cushion effect
associated with the distribution of subcutaneous fat, more recent analyses have found little
evidence in support of that hypothesis (Dunn and Tefft 2014). The suggestion of a U-shaped
pattern (lower BMI and very high BMI are associated with a greater risk of fatality) has been
reported by a number of studies, and others find no statistical relationship between death rates
or fatality risk and being overweight (as determined separately from obese) (Jehle 2012; Zhu et
al., 2006; Simmons and Zlatoper 2010; Sivak et al., 2010; Zhu et al., 2010).
There are numerous explanations for the higher risks faced by obese drivers. Obese
individuals are more prone to momentum effects, suggesting that their weight is more likely to
propel them with greater force into the steering wheel, dashboard, or the car window (Desapriya
et al., 2014; Rice and Zhu 2013; Zhu et al., 2010). There are serious co-morbidities associated
with obesity, including fatigue due to sleep apnea and impaired driving and braking response
due to diabetes and the associated nerve damage to the feet (Desapriya et al., 2014; Karimi et
al., 2014; Lavalliere et al., 2012). In addition, there are potential side effects and drug
interactions from the medications often taken by obese individuals (Desapriya et al. 2014).
Because seat belts are designed for the average driver who is assumed to have a BMI
of 25 (Joseph et al. 2017), most restraint systems offer a poor fit for those who are overweight
or obese. As a result, seat belts may be less effective when deployed and obese drivers may be
less likely to wear them (Dunn and Tefft 2014; Reed et al., 2011; Zhu et al., 2010). Further,
there is significant evidence of emergency and post-operative complications among obese
individuals including the higher risk of respiratory failure or pneumonia (Desapriya et al., 2014;
Joseph et al., 2017).
3. Model
Page 4 of 21Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
5
In analyzing the relationship between obesity and motor vehicle deaths, this paper
employs the theoretical framework of Simmons and Zlatoper (2010). Specifically, it includes
obesity as a determining factor in a model explaining these deaths. The model also
incorporates the following explanatory factors typically found in studies of motor vehicle
fatalities: driver and passenger characteristics besides obesity, economic conditions, rural vs.
urban roads, government regulations, and weather conditions. The general form of the model,
as adopted from Simmons and Zlatoper (2010), is:
Motor vehicle death rate = f (driver and passenger characteristics, economic conditions,
locational factors, government regulations, weather conditions, other factors) (1)
The dependent variable is measured by motor vehicle deaths per 100 million vehicle
miles (death rate). The source for this information for 1995-2010 is the Federal Highway
Administration (FHWA, various years), and the source for the 2011-15 data is the Insurance
Institute for Highway Safety (IIHS, 2017a).
It is worth noting how the model in Equation 1 compares to the theory underlying the
empirical research surveyed in Section 2. As mentioned, the prior literature has either
estimated the determinants of highway death rates (deaths per vehicle mile), or the fatality risk
given a crash has occurred. This comparison can be demonstrated by decomposing the
probability of a fatality as follows:
Prob [fatality] = Prob [fatality | given a crash] * Prob [crash] (2)
This study examines the probability of a fatality (Prob [fatality]) and also analyzes the
components of this probability as provided in Equation 2. To do so, it restates equation (2) in
terms of the death rate, a vulnerability rate, and the crash rate:
Death rate = Vulnerability rate * Crash rate (3)
where the death rate is the probability of a fatality, the vulnerability rate is the probability of a
fatality given a crash, and the crash rate is the probability of a crash.
Though the focus of this paper is on state fatality rates (Equation 1), estimates are also
provided of the component model, for completeness. The estimates of the latter model consist
Page 5 of 21 Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
6
of regressing both vulnerability rate and crash rate on the same set of independent variables
included in Equation 1.[4]
The vulnerability rate is measured by (deaths divided by crashes) times 100, and crash
rate is approximated by crashes divided by 10,000 vehicles miles. The source for both the
death and vehicle mileage information for 1995-2010 is the FWHA (various years), and the
source for 2011-15 data is the Insurance Institute for Highway Safety (2017a). Crashes
(accidents) are approximated using information from NHTSA (2017).[5]
3.1. Model 1 – Highway Death (Fatality) Rate
The relationship between driver or passenger body weight and highway death rates is of
particular interest in this study. Thus, the model includes two weight measures reported by the
Centers for Disease Control (CDC, 2016): the proportion of the population that is overweight
and the proportion of the population that is obese.[6] These measures for the overall state
population approximate the analogous variables for motor vehicle occupants, which are
unavailable at the state level.
Based on the amassed literature as related in the previous section, the probability of a
fatality, given a crash, and the probability of a crash are both expected to be higher for obese
individuals due to co-morbidities, seat belt fit, and the problems associated with sleep apnea,
etc. As a result, the expected impact of the obesity rate on the death rate is positive. The
expected impact of overweight individuals on the death rate, however, is uncertain a priori given
the ambiguity of the earlier results pertaining to the risk to overweight drivers.
Two age categories of drivers are accounted for in the model: young drivers (proportion
of licensed drivers aged 24 years or younger) and old drivers (proportion of licensed drivers
aged 65 years or older). FHWA (various years) is the source for the information used to
construct both measures. Loeb et al. (1994, 23-25) reported that fatal crash involvement rates
are highest for the youngest drivers and decline with age until they increase for the oldest
drivers; however, there is mixed statistical evidence on the nature of this relationship. Given the
latter results, the expected associations of death rate with the two driver-age measures are
uncertain.
Alcoholic intoxication is approximated by per capita alcohol consumption (in gallons) for
all beverages, as reported by the National Institute on Alcohol Abuse and Alcoholism (2017).[7]
According to Loeb et al. (1994, 20-21), several empirical studies have found a significant
positive linkage between proxies for driver alcohol consumption and highway deaths.
Page 6 of 21Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
7
Economic conditions are represented by income as measured by real per capita Gross
Domestic Product (GDP, chained 2009 dollars in thousands), as reported by the Bureau of
Economic Analysis (2017). Peltzman (1975) noted that income’s effect on motor vehicle crash
rates is uncertain a priori. Assuming safety and driving intensity are both normal goods, higher
income would be associated with the purchase of safer cars, but this may be offset by increases
in driving speed as the value of one’s time increases with income. Some studies utilizing pooled
cross-sectional, time-series data (e.g., Saffer and Grossman, 1987a, 1987b) have found a
statistically significant inverse association between highway mortality measures and income,
while other research reported a significant positive effect (e.g., Blattenberger et al. 2011). Given
these mixed results, the anticipated impact of income is uncertain.
Travel location is a consideration in highway safety as vehicle speeds tend to be higher
in rural settings than in urban environments therefore increasing the chance of death in a rural
crash. According to Loeb et al. (1994), several empirical studies have found a significant
negative relationship between total highway fatality measures and proxies for the proportion of
urban travel. Given these findings, the locational measure utilized here—the ratio of rural to
urban vehicle miles—is expected to have a positive association with the death rate. FHWA
(various years) is the data source for this measure.
The model also accounts for the government regulation of seat belt laws. These are
approximated by the proportion of the year that a primary seat belt law is in effect, as reported
by IIHS (2017b).[8] Rivara et al. (1999) reviewed 48 studies to ascertain the relative
effectiveness of primary and secondary seat belt laws on the outcomes of restraint use and
crash-related mortality and injuries. They concluded that primary laws tend to be more
efficacious than secondary statutes. Based on this, primary seat belt laws are anticipated to
have a life-saving effect.
The model includes two weather conditions: annual average temperature, in degrees
Fahrenheit, and annual precipitation, in inches. The data source for 1995-2009 is the National
Climatic Data Center (2010). The National Oceanic and Atmospheric Administration (2017)
provided monthly data that was used to approximate annual information for 2010-2015. Prior
empirical research has found motor vehicle fatality measures to have a significant positive
linkage with temperature and a significant negative association with precipitation (Loeb et al.
1994). It may be that motorists drive less carefully when weather conditions are more favorable
(e.g., warmer) but more carefully when conditions are less favorable (e.g., wetter). This study
expects the same relationships.
Page 7 of 21 Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
8
3.2. Models 2 and 3 – Vulnerability and Crash Rates
Much of the recent obesity-related research reported in the previous section relates body
weight to the risk of fatality in the event of a motor vehicle crash (Model 2) and has implications
for the probability of a crash (Model 3). For all the reasons given above, it is anticipated that
higher rates of obesity will be associated with higher vulnerability rates as well as higher crash
rates. At the same time, because of the conflicting results of previous research and the
evidence of U-shaped patterns of body weight and fatalities, the expected relationship between
overweight and the vulnerability rate or the crash rate is uncertain.
For the most part, the expected signs for the explanatory variables in the vulnerability
rate equation (Model 2) are similar to those in the highway death rates equation, with a few
exceptions, noted here. Whereas the evidence relating death rates and driver age has been
mixed, both younger and older drivers are expected to have a higher probability of a fatality,
given a crash (vulnerability rate). Younger drivers are less experienced and less likely to wear
seat belts (Carpenter and Stehr 2008; Carter et al., 2014; McCartt and Northrup 2004), while
older drivers and occupants face a higher risk because of the complications related to
diminishing bone density and other age-related problems (Carter et al. 2014). Higher alcohol
consumption (Zhu et al. 2010) and a greater proportion of rural highway vehicle miles (where
speed limits are typically higher) are both expected to be positively related to the vulnerability
rate. As with highway fatalities (Model 1), the relationship between income and the vulnerability
rate is uncertain a priori. Given the evidence that drivers and other occupants wearing seat
belts are better protected in the event of a crash (Carpenter and Stehr 2008, among others), it is
anticipated that the seat belt variable would be negatively related to the probability of a fatality,
given a crash. Environmental factors may also affect death rates and vulnerability rates
differently. Whereas prior research has suggested death rates are positively and negatively
related to temperature and precipitation, respectively, there is no a priori evidence that these
environmental factors would significantly affect vulnerability rates.
In the equation explaining the probability of a crash (Model 3), as with the vulnerability
rate equation, the coefficients on the young and old variables are expected to be positive.
Teenagers have a higher crash risk due to their inexperience and their propensity for risky
behavior (Hahn and Prieger 2006, Cicchino and McCartt 2015). Older drivers have a higher risk
of crashing typically because of driver error (Cicchino and McCartt 2015). Contrary to the
expectations in the previous two models, income is expected to be positively related to the
crash rate in light of recent evidence suggesting that crashes may be pro-cyclical as drivers
work longer hours, sleep less, and may have higher levels of stress when economic conditions
Page 8 of 21Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
9
are particularly good (Hahn and Prieger 2006, Lam and Pierard 2017). Because Jacobson et al.
(2012) find that crash rates are higher in urban areas, in part because of higher levels of
congestion, a negative relationship between the proportion of rural to urban driving and the
crash rate is expected.
The nature of the relationship between the crash rate and the seat belt variable is
uncertain a priori, however. This is partly due to the paucity of studies on crashes per se, given
the lack of consistent state-level data on total crashes. In addition, empirical evidence on the
linkage between crashes and mandated safety measures is inconclusive. For example, using
time-series data for the entire U.S., Chirinko and Harper (1993) found the crash rate to have a
positive but statistically insignificant association with an index of motor vehicle occupant safety
standards. The coefficients of temperature and precipitation are expected to be similar to the
predicted effects in Model 1: crash rates are expected to be higher in states with higher average
temperatures or lower annual precipitation.
To control for influences, such as cultural and regional attitudes about driving and
wellness, which may differ by section of the country but are constant across time, this study
includes regional dummy variables. Previous panel data analyses of U.S. motor vehicle
fatalities have employed such variables (Blattenberger et al., 2013; Blattenberger et al., 2012).
The dummy variables utilized here pertain to the nine Census Divisions defined by the U.S.
Bureau of the Census (2010): New England, Middle Atlantic, East North Central, West North
Central, South Atlantic, East South Central, West South Central, Mountain, and Pacific.
4. Data and Estimation Results
This paper uses time series state-level data corresponding to the 48 contiguous U.S.
states over a 21-year period (1995-2015). A description of the variables, along with means and
standard deviations, are presented in Table 1. The data in Table 1 indicate that over this period,
average motor vehicle deaths per 100 million vehicle miles has been 1.435; the average
vulnerability rate is 0.649 fatalities per 100 crashes; and the average crash rate is 219.643
crashes per 10,000 vehicle miles. Over 36 percent of the population on average is overweight
and slightly less than one quarter is obese. On average, 13 percent of the licensed drivers are
24 years and younger, and slightly over 15 percent of licensed drivers are 65 years and older.
The average annual alcohol consumption during the two decades is approximately two and one
third gallons per person per year. The average real per capita disposable personal income is
over $43,000. The ratio of rural to urban travel miles over the period is approximately one. The
Page 9 of 21 Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
10
mean seat belt value of 0.414 indicates that, on average, a primary seat belt law is in effect for
41 percent of the year. The average annual temperature is slightly over 520 F and the average
annual precipitation in inches is less than 38 inches.
Insert Table 1 here
The estimation results are presented in Table 2 below. Time fixed effects and standard
errors clustered at the state level are accounted for in each specification; regressions 2, 4, and
6 add regional dummy variables. Overall, the adjusted R-squares ranging from 0.72 to 0.91
indicate that the common set of independent variables explain much of the variability in all three
dependent variables. Moreover, the F statistics for no fixed effects (FE), significant at the 1
percent level, indicate that inclusion of the time dummies improves the overall fit of the model.
The remainder of the discussion focuses on regressions 2, 4, and 6, which include regional
dummies.
The results suggest that higher rates of obesity are significantly associated with a higher
motor vehicle death rate and probability of a crash, similar to results found by Simmons and
Zlatoper (2010). However, the relationship does not hold when we consider vulnerability rate.
This indicates that as the rate of obesity in the population increases, so does the rate of death
by motor vehicle crash. The coefficient of 2.88 indicates that for every one percentage point
increase in the rate of obesity, the rate of motor vehicle deaths increases by 0.0288 deaths for
every hundred million vehicle miles travelled. According to the Federal Highway Administration,
Americans drove 3.22 trillion miles in 2016 (FHWA 2017). Holding all other variables constant,
this implies that 928 deaths could have been avoided in 2016 with a one percentage point
decrease in the obesity rate.
The percent of overweight individuals is not statistically significant in any of the
specifications, which indicates that being overweight does not appear to increase the risk of a
fatal motor vehicle crash. This result is consistent with the conclusions of recent literature that
finds evidence of a U-shaped pattern, indicating that normal or lower weight and obese
individuals are more at risk for fatalities in motor vehicle crashes (e.g. Jehle, 2012 and Zhu et
al., 2010).
Insert Table 2 here
Page 10 of 21Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
11
The proportion of young drivers has no detectable statistical relationship with amount of
deaths per motor vehicle mile or the crash rate, but appears to be positively related to the
vulnerability rate. This latter result is consistent with expectations, given that the younger drivers
have less experience, and with Carter et al. (2014), who suggest that younger drivers are less
likely to wear seat belts.
The proportion of older drivers is statistically significant and positive in the equations
explaining the death and crash rates. This indicates that a fatality or a crash is more likely when
the proportion of older drivers is higher. The proportion of older drivers, while having the
expected positive sign, is not statistically significant in the vulnerability rate model. In light of
recent evidence, this is unexpected since Carter et al. (2014) and others report that the risk of
serious injury increases with age presumably because of complications related to diminishing
bone density, among other factors.
As expected, an increasing rate of alcohol consumption is associated with increasing
rates of crashes and motor vehicle deaths (see Loeb et al., 1994). Alcohol consumption is not
statistically related to the vulnerability rate (the probability of death in the event of a crash),
however, which is an interesting result deserving further scrutiny. We find no significant
relationship between income and the dependent variables.
An increasing percentage of rural relative to urban highway travel is positively related to
the highway death rate and the vulnerability rate. The lack of nearby medical facilities and
higher speeds contribute to the increased vulnerability, given the event of a crash on a rural
road. The positive estimated coefficient in the crash rate equation (Model 3) is unexpected, and
the coefficient is insignificant in the appropriate one-tail test, despite its large t-value. Although
Jacobson et al. (2012) report higher crash rates in urban areas, one possible explanation for the
finding to the contrary here is the typically higher driving speeds on rural roads may contribute
to higher crash rates.
Regarding seat belt laws, the negative and significant coefficients on the seat belt law
variable in Models 1 and 3 (regressions 2 and 6) indicate that for states with primary seat belt
laws in effect for longer parts of the year, death and crash rates are lower. Despite evidence
that drivers and passengers who wear seat belts are better protected in the event of a crash
(implying a lower vulnerability rate (Rivara et al., 1999; Sivak et al., 2010)), this paper finds no
statistically significant relationship between the seat belt law variable and the vulnerability rate.
Finally, increasing temperature is statistically significantly associated with increasing
rates of highway deaths and crashes, as expected (Loeb et. al., 1994; Simmons and Zlatoper,
2010), as well as with increasing vulnerability rates. A possible explanation for the latter result
Page 11 of 21 Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
12
is that motorists may drive more recklessly in warmer conditions, causing crashes in these
circumstances to be more deadly. Whereas no significant relationship is found between
precipitation and highway deaths and crashes, a significant negative relationship is found
between precipitation and vulnerability rate, indicating that crashes may be more deadly in drier
conditions.
As indicated earlier, regional dummy variables were used to control for impacts that vary
across sections of the U.S. but are constant over time. It is worth noting that all of these
variables are highly statistically significant in the estimations.[9]
5. Robustness
To assess the stability of the obesity effect over time, the data was divided into three
equal time periods of seven years: 1995 – 2001, 2002 – 2008, and 2009 – 2015. Equations 2, 4
and 6 from Table 2 were re-estimated for each period. Results for the obesity coefficients are
presented in Table 3. In the death rate equation (2), obesity is positive and significant at the 5
percent (1995 – 2001) or 10 percent level (2002 – 2008 and 2009 – 2015). Results for the
vulnerability rate equation (4) indicate that obesity is not a significant predictor in any of the time
periods examined, similar to results for the entire 1995-2015 time period. For the crash rate
equation (6), the obesity variable is positive and significant at the 10 percent level for the 1995 –
2001 period and positive and significant at the 1 percent level for the 2002 – 2008 and 2009 –
2015 time periods. These results together reinforce the findings in Table 2 above that increased
obesity is associated with higher motor vehicle fatalities and crash rates.
Insert Table 3 here
Additionally, OLS regressions were run on the data for each year of the sample period,
1995 – 2015. The obesity variable is positive and significant at the 5 percent (>5 to 10 percent)
level for 9 (4) years for the dependent variable death rate, 2 (2) years for vulnerability rate, and
10 (2) years for crash rate. When the lagged dependent variable is included as an additional
independent variable on the entire panel of data for regressions 2, 4, and 6, the obesity variable
is positive and significant at the 5 percent level for the dependent variables death rate and crash
rate, and insignificant for vulnerability rate.
As a final test, OLS regressions including a time trend variable instead of time dummies
were examined. These regressions included regional dummies and White standard errors. For
the dependent variables death rate and crash rate, the obesity variable is again positive and
Page 12 of 21Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
13
significant at the 1 percent level and the time trend variable is negative and significant at the 1
percent level. Interestingly, the same results hold for the regression on the vulnerability rate
dependent variable. Robustness results are available upon request.
6. Conclusion and Summary
This paper presents evidence that increases in the percentage of obese U.S. drivers are
associated with increases in motor vehicle deaths. After controlling for a set of representative
contributors to motor vehicle fatalities, a panel of state-level data for the years 1995 – 2015
reveals that increases in motor vehicle crash rates as well as fatalities are associated with
increasing obesity. To our knowledge, no other paper has examined this issue at the state level
over this length of time. Further, these results suggest that a one percentage point reduction in
the obesity rate would have avoided 928 traffic fatalities, based on 2016 data. Given that much
of the existing obesity research focuses on the effects of rising obesity rates on health care
costs and productivity, this study adds another dimension to the public health problem of
obesity.
Finally, the results of this paper are consistent with research on individual crashes and
fatalities, and support previous research that alcohol, rural driving and warmer climates increase
motor vehicle fatalities, while primary seat belt laws contribute to reducing such deaths.
Moreover, the results here also support the call for improved seat belt design, as well as the use
of crash dummies that better reflect those of greater body weight. Currently, restraint systems
in the U.S. and crash test dummies used by the NHTSA are designed to reflect individuals with
a BMI of 25. Although companies have developed crash-test dummies representing individuals
with a BMI of 35 (as well as dummies designed to represent the elderly), the newer crash
dummies have not yet been adopted by the NHSTA or IIHS. (Gruley, 2019; University of
Virginia, 2018) Education efforts to increase seat belt use should be supplemented with public
health messages about the role of obesity and its contribution to highway death rates.
Page 13 of 21 Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
14
Footnotes
[1] Body Mass Index (BMI) is a person's weight in kilograms (kg) divided by his or her height in meters
squared.
[2] Cawley (2015) provides an extensive review of the economic research on obesity.
[3] This finding is based on estimates from the Federal Highway Administration (2017) that Americans
drove 3.22 trillion miles in 2016.
[4] The decomposition approach described here and the specification and use of vulnerability and
accident rates are taken from Chirinko and Harper (1993, 281-283).
[5] State-level data on total accidents or crashes is unavailable, although NHTSA (2009, 2017) provides
fatal accidents/crashes by state. According to NHTSA (2017), at the national level fatal crashes
accounted for 0.6% of total crashes in 1995-2012 and 0.5% of total crashes in 2013-15. Assuming the
same percentage for all states in the respective years, total crashes are estimated for each state in 1995-
2012 by dividing its fatal crashes by 0.006; and they are estimated for each state in 2013-15 by dividing
its fatal crashes by 0.005.
[6] Individuals with BMI values of 25.0 through 29.9 are in the overweight category, while individuals with
BMI values of 30.0 or more are categorized as obese (CDC, 2018).
[7] State-level data on alcohol consumption by drivers is unavailable. This study assumes that such
consumption is positively correlated with alcohol consumption by the general population.
[8] In June 2017, all states except New Hampshire had primary or secondary seat belt laws. Thirty-four
states and the District of Columbia had primary laws, which permit police to stop vehicles solely for
violations of belt laws. Secondary laws in 15 states require another violation before vehicle occupants
can be cited for non-use of a seat belt (IIHS, 2017b).
[9] As an alternative to the regional dummies, state-specific dummy variables were utilized. In these
estimations, many of the explanatory factors that are statistically significant in Table 2 became statistically
insignificant, likely due to multicollinearity. Condition indexes in the estimations including the state-
specific dummies were 230 or higher. According to Gujarati and Porter (2009, p. 340), a condition index
greater than 30 is evidence of severe multicollinearity.
Page 14 of 21Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
15
References
Blattenberger, G., Fowles, R., and Loeb, P.D. (2013), “Determinants of motor vehicle crash
fatalities using bayesian model selection methods“, Research in Transportation
Economics, Vol. 43, pp. 112-122.
Blattenberger, G, Fowles, R, Loeb, PD & Clarke, WA 2012, “Understanding the cell phone effect
on vehicle fatalities: A Bayesian view”, Applied Economics, Vol. 44 No. 14, pp. 1823-
1835.
[dataset] Bureau of Economic Analysis, U.S. Department of Commerce. (2017), Frequently
Asked Questions, Available at: www.bea.gov/faq/index.cfm?faq_id=98&searchQuery=
(accessed August 2017).
Carpenter, C.S., and Stehr, M. (2008), “The effects of mandatory seatbelt laws on seatbelt use,
motor vehicle fatalities, and crash related injuries among youths“, Journal of Health
Economics, Vol. 27, pp. 642-662.
Carter, P. M., Flannagan, C.A.C., Reed, M.P., Cunningham, R.M., and Rupp, J.D. (2014),
“Comparing the effects of age, BMI and gender on severe injury (AIS 3+) in motor-
vehicle crashes“, Accident Analysis and Prevention, Vol. 72, pp. 146-160.
Cawley, J. (2015), “An economy of scales: A selective review of obesity’s economic causes,
consequences, and solutions”, Journal of Health Economics, Vol. 43, pp. 244-268.
Centers for Disease Control and Prevention, US Department of Health and Human Services,
(2018), Defining adult overweight and obesity, Available at:
https://www.cdc.gov/obesity/adult/defining.html (accessed June 2018).
Centers for Disease Control and Prevention, US Department of Health and Human Services,
(2016), Prevalence data: overweight and obesity (BMI) -1995-2015, behavioral risk factor
surveillance system, Available at: www.cdc.gov/obesity/data/databases.html (accessed
August 2017).
Chirinko, R.S., and Harper, Jr., E.P. (1993), “Buckle up or slow down? New estimates of
offsetting behavior and their implications for automobile safety regulation“, Journal of
Policy Analysis and Management, Vol. 12 No. 2, pp. 270-96.
Cicchino, J. B., and McCartt, A.T. (2015), “Critical older driver errors in a national sample of
Serious US crashes“, Accident Analysis and Prevention, Vol. 80, pp. 211-219.
Desapriya, E., Giulia, S., Subzware, S., Peiris, D.C., Turcotte, K., Pike, I., Sasges, D., and
Hewapathirane, D.S. (2014), “Does obesity increase the risk of injury or mortality in
motor vehicle crashes? A systematic review and meta-analysis“, Asia-Pacific Journal of
Public Health, Vol. 26 No. 5, pp. 447-460.
Dunn, R., and Tefft, N. (2014), “Has increased body weight made driving safer? “, Health
Economics, Vol. 23, pp. 1374-1389.
[dataset] Federal Highway Administration, U.S. Department of Transportation, (various years).
Highway Statistics, Available at: https://www.fhwa.dot.gov/policyinformation/statistics.cfm.
Page 15 of 21 Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
16
Federal Highway Administration, U.S. Department of Transportation. (2017), 3.2 Trillion Miles
Driven On U.S. Roads In 2016 [Press release], Retrieved from
https://www.fhwa.dot.gov/pressroom/fhwa1704.cfm.
Gruley, Brian, 2018, “Crash-Test Dummies Are Getting Fatter Because We Are Too”,
Bloomberg Businessweek, https://www.bloomberg.com/news/features/2018-10-25/even-
the-crash-test-dummies-are-putting-on-weight
Gujarati, D.N., and Porter, D.C. (2009), Basic econometrics, Fifth edition, New York, NY:
McGraw-Hill Irwin.
Hahn, R. W., and Prieger, J.E. (2006), “The impact of driver cell phone use on accidents“, The
B.E. Journal of Economic Analysis & Policy, Vol. 6 No. 1, pp. 1-37.
[dataset] Insurance Institute for Highway Safety (2017a), Fatal crash totals. Deaths per 100
million vehicle miles traveled, Available at: http://www.iihs.org/iihs/topics/t/general-
statistics/fatalityfacts/state-by-state-overview/ (accessed June 2017).
[dataset] Insurance Institute for Highway Safety (2017b), Safety belts, Available at:
http://www.iihs.org/iihs/topics/laws/safetybeltuse?topicName=safety-belts (accessed June
2017).
Jacobson, S. H., King, D. M., Ryan, K.C., and Robbins, M.J. (2012), “Assessing the long term
benefit of banning the use of hand-held wireless devices while driving“, Transportation
Research Part A, Vol. 46, pp. 1586-1593.
Jehle, D., Gemme, S., and Jehle, C. (2012), “Influence of obesity on mortality of drivers in
severe motor vehicle crashes“, American Journal of Emergency Medicine, Vol. 30, pp.
191-195.
Joseph, B., Hadeed, S., Haider, A.A., Ditillo, M., Joseph, A. Pandit, V., Kulvatunyou, N., Tang,
A., Latifi, R., and Rhee, P. (2017), “Obesity and trauma mortality: sizing up the risks in
motor vehicle crashes“, Obesity Research & Clinical Practice, Vol. 11, pp. 72-78.
Karimi, M., Hedner, J., Lombardi, C., McNicholas, W.T., Penzel, T., Riha, R.L., Rodenstein, D.,
and Grote, L. (2014), “Driving habits and risk factors for traffic accidents among sleep
apnea patients – a European multi-centre cohort study“, Journal of Sleep Research, Vol.
23, pp. 689- 699.
Lam, J.P., and Pierard, E. (2017), “Time varying relationship between mortality and business
cycles in the USA“, Health Economics, Vol. 26, pp. 164-183.
Lavalliere, M., Handrigan, G.A., Teasdale, N., and Corbeil, P. (2012), “Obesity, where is it
driving us? “, Journal of Transportation Safety & Security, Vol. 4, pp. 83-93.
Loeb, P.D., Talley, W.K., and Zlatoper, T.J. (1994), Causes and deterrents of transportation
accidents: An analysis by mode, Westport, CT: Quorum Books.
McCartt, A.T., and Northrup, V.S. (2004), “Factors related to seat belt use among fatally injured
teenage drivers“, Journal of Safety Research, Vol. 35 No. 1, pp. 29-38.
Page 16 of 21Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
17
[dataset] National Climatic Data Center, U.S. Department of Commerce (2010), U.S. climate at
a glance, Available at: http://lwf.ncdc.noaa.gov/oa/climate/research/cag3/cag3.html
(accessed July 2010).
[dataset] National Highway Traffic Safety Administration, U.S. Department of Transportation
(2009),Fatal motor vehicle traffic crashes, by state; Fatality Analysis Reporting System
(FARS) 1975-2006 Final & 2007 ARF, Data file provided by NHTSA in June 2009.
[dataset] National Highway Traffic Safety Administration, U.S. Department of Transportation
(2017), Traffic Safety Facts, various years,
https://crashstats.nhtsa.dot.gov/#/DocumentTypeList/12 (accessed June 2017).
[dataset] National Institute on Alcohol Abuse and Alcoholism, National Institutes of Health
(2017), Per capita ethanol consumption for states, census regions, and the United States,
1977-2015, Available at: https://pubs.niaaa.nih.gov/publications/surveillance108/tab4-
1_15.htm (accessed June 2017).
[dataset] National Oceanic and Atmospheric Administration (2017), Center for Weather and
Climate, File with monthly data.
Peltzman, S. (1975), “The effects of automobile safety regulation“, Journal of Political Economy,
Vol. 83 No. 4, pp. 677-725.
Reed, M.P., Ebert-Hamilton, S.M., and Rupp, J.D. (2011), “Effects of obesity on seat belt fit“,
University of Michigan Transportation Research Institute, NHTSA.
Rice, T., and Zhu, M. (2013), “Driver obesity and the risk of fatal injury during traffic collisions“,
Emergency Medical Journal, Vol. 31 No. 1, pp. 9-12.
Rivara, F.P., Thompson, D.C., and Cummings, P. (1999), “Effectiveness of primary and
secondary enforced seat belt laws“, American Journal of Preventive Medicine, Vol. 16,
pp. 30-39.
Saffer, H., and Grossman, M. (1987a), “Beer taxes, the legal drinking age, and youth motor
vehicle fatalities“, Journal of Legal Studies, Vol. 16, pp. 351-374.
Saffer, H., and Grossman, M. (1987b), “Drinking age laws and highway mortality rates: Cause
and effect“, Economic Inquiry, Vol. 25, pp. 403-417.
Simmons, W.O., and Zlatoper, T.J. (2010), “Obesity and motor vehicle deaths in the USA: A
state-level analysis“, Journal of Economics Studies, Vol. 37 No. 5, pp. 544-556.
Sivak, M., Schoettle, B., and Rupp, J. (2010), “Survival in fatal road crashes: Body mass index,
gender, and safety belt use“, Traffic Injury and Prevention, Vol. 11 No. 1, pp. 66–68.
[dataset] U.S. Bureau of the Census. (2010), “Geographic Terms and Concepts – Census
Divisions and Census Regions,” Available at:
https://www.census.gov/geo/reference/gtc/gtc_census_divreg.html (accessed July 2014).
University of Virginia, 2019, “Engineering researcher works to make auto seatbelts safer for
Page 17 of 21 Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
18
obese people”, https://phys.org/news/2019-02-auto-seatbelts-safer-obese-people.html
Zhu, S., Kim, J., Ma, X., Shih, A., Laud, P.W., Pintar, F., Shen W., Heynsfield, S.B., and Allison,
D.B. (2010), “BMI and risk of serious upper body injury following motor vehicle crashes:
Concordance of real-world and computer-simulated observations“, PLoS Med, Vol. 7
No. 3: e1000250, doi:10.1371/journal.pmed.1000250.
Zhu, S., Layde, P.M., Guse, C.E., Llaud, P.W., Pintar, F., Nirula, R., and Hargartin, S. (2006),
“Obesity and risk for death due to motor vehicle crashes“, American Journal of Public
Health, Vol. 96 No. 4, pp. 734-739.
Page 18 of 21Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
19
Variable Mean
Std. Dev. Definition
Death Rate 1.435 0.437 Motor vehicle deaths per 100 million vehicle miles
Vulnerability Rate 0.649 0.046 (Motor vehicle deaths divided by total crashes) * 100
Crash Rate 219.643 61.343 Crashes divided by 10,000 vehicle miles
Overweight 0.363 0.014 Proportion of population that is overweight (BMI 25 – 29)
Obese 0.237 0.053 Proportion of population that is obese (BMI ≥ 30)
Young 0.135 0.021 Proportion licensed drivers aged 24 or younger
Old 0.158 0.023 Proportion licensed drivers aged 65 +
Alcohol 2.330 0.492 Per capita apparent alcohol consumption (gallons)
Income 43.080 9.481 Real GDP per capita (000s)
Rural/Urban 1.048 0.83 Ratio of rural to urban vehicle miles
Seat Belt Law 0.414 0.488 Proportion of a year primary seatbelt law in effect
Temperature 52.810 7.637 Average annual temperature (degrees Fahrenheit)
Precipitation 37.619 15.279 Annual precipitation (inches)
Table 1: Variable Definitions and Descriptive Statistics
Page 19 of 21 Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
20
Variable
Exp
Sign
(1) (2)
Exp
Sign
(3) (4)
Exp
Sign
(5) (6)
Intercept ? -1.5148* -2.7684*** ? 0.5033*** 0.5150*** ? -180.033 -373.523***
(-1.65) (-4.91) (24.74) (18.03) (-1.29) (-4.23)
Obese + 2.4472^^ 2.8831^^^ + 0.04921^ -0.0034 + 385.4433^^ 472.9406^^^
(1.94) (3.15) (1.61) (-0.08) (1.97) (3.27)
Overweight ? -0.3844 0.7138 ? -0.0437 -0.590 ? -47.6088 124.6476
(-0.26) (0.84) (-0.87) (-1.14) (-0.22) (0.99)
Young ? 0.3888 0.627 + 0.2122^^^ 0.1384^^^ + -30.8779 29.8528
(0.32) (0.71) (5.85) (3.14) (-0.17) (0.21)
Old ? 1.5456 3.2219*** + 0.0183 0.0008 + 229.6337 492.2733^^^
(1.12) (3.12) (0.43) (0.02) (1.13) -3.11
Alcohol + 0.0831^ 0.11^^^ + -0.0001 0.0008 + 12.7531^ 16.6760^^^
(1.51) (3.58) (-0.08) (0.61) (1.52) (3.42)
Income ? -0.0033 0.0003 ? 0.0001 0.0001 + -0.6251 -0.09367
(-1.07) (0.11) ('0.63) (0.65) (-1.31) (-0.21)
Rural/Urban + 0.2392^^^ 0.2167^^^ + 0.0061^^^ 0.0065^^^ - 33.9128 29.9851
(4.82) (7.00) (4.20) (3.91) (4.52) (6.22)
Seat Belt Law ? -0.0881** -0.0633** - 0.0007 0.0002 ? -13.5749** -9.4392**
(-2.18) (-2.22) (0.43) (0.13) (-2.21) (-2.14)
Temperature + 0.0347^^^ 0.0255^^^ ? 0.0005*** 0.0004** + 5.1201^^^ 3.6787^^^
(9.44) (6.94) (4.05) (2.54) (9.34) (6.54)
Precipitation - -0.0076^^^ 0.0008 ? -0.0005*** -0.0002** - -1.0208^^^ 0.1806
(-4.01) (0.72) (-9.27) (-2.07) (-3.59) (1.11)
n1005 1005 1005 1005 1005 1005
Adjusted R^2 0.7503 0.8479 0.9019 0.9085 .0.7205 0.8308
F-test no FE (Pr > F) < 0.0001 < 0.0001 < 0.0001 < 0.0001 < 0.0001 < 0.0001
Time fixed effects yes yes yes yes yes yes
Regional dummies no yes no yes no yes
Notes: This table presents the results of the fixed effects regression models on motor vehicle death rate (Model 1), vulnerability rate
(Model 2), and crash rate (Model 3). Death rate is measured as deaths per 100 million vehicle miles, vulnerability rate is (number of
motor vehicle deaths divided by the number of crashes) * 100, and the crash rate is approximated by crashes divided by 10,000
vehicle miles. The first column for each model provides predicted signs based on prior research. The second and third columns
present coefficients and t-values in parentheses. Standard errors are clustered at the state level. * indicates significance with 2-tail
test (***1%, **5%, *10%), ^ indicates significance evaluated with appropriate one-tail test (^^^ 1%, ^^5%, ^10%).
Table 2
Panel Regressions of U.S. Motor Vehicle Death Rate Models 1995-2015
Model 1: Death Rate
Model 2: Vulnerability Rate
Model 3: Crash Rate
Page 20 of 21Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Economic Studies
21
Table 3
Robustness Tests: Regression estimates for Obese
1995 – 2001, 2002 – 2008, and 2009 – 2015
Death Rate
(2)
Vulnerability Rate
(4)
Crash Rate
(6)
1995 – 2001
1.723**
(1.97)
0.062
(0.81)
241.939*
(1.70)
2002 – 2008
4.855***
(3.85)
-0.081
(-1.26)
758.305***
(3.93)
2009 – 2015
2.953**
(2.57)
-0.003
(-0.06)
487.79***
(2.60)
Notes: The table shows coefficients and t-values (in parentheses) for the independent variable
Obese for panel data regressions where the dependent variable is Death Rate, Vulnerability
Rate or Crash Rate. The three samples are groups of 7 years based on time periods: 1995 –
2001, 2002 – 2008, and 2009 – 2015. The column headings (2), (4), and (6) correspond to the
same numbered regressions from Table 2. As such, year and regional dummies are included
and standard errors are clustered at the state level. ***, **, and * represent significance at the 1,
5, and 10% levels, respectively, for a two-tailed test.
Page 21 of 21 Journal of Economic Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
