Instructions
ECON7012/8012 – Topic 5a
Evaluating outcomes from policy reform
A/Prof Jordi McKenzie, Department of Economics
Paper Details
• Carpenter, C. S., & Warman, C. (2019), ‘What do bicycle helmet laws do?
Evidence from Canada’, Economic Inquiry, 57(2), XXXXXXXXXX.
Abstract
Twenty-one states and the District of Columbia require youths to wear helmets when
iding a bicycle, and there has been a push to extend such laws to adults. We provide
new evidence on helmet laws by studying Canada using difference-in-differences
models and restricted area-identified public health survey data with information on
cycling and helmet use for nearly 800,000 individuals from 1994 to 2014. We first
confirm prior patterns from the United States that laws requiring youths to wear
helmets significantly increased youth helmet use. We then provide the literature’s first
comprehensive evidence that “all-age” bicycle helmet laws significantly increased
oth adult and youth helmet use by 50%–190% relative to pre-reform levels, with
larger effects for younger adults and less-educated adults. All-age helmet laws had
modest effects at reducing cycling and increasing in-home exercise during winter
months among adults but did not meaningfully affect weight. Overall, our findings
confirm that all-age helmet laws can be effective at increasing population helmet use
without significant unintended adverse health consequences.
Introduction
• Cycling accounts for largest number of head injuries in emergency rooms
• Governments have increasingly adopted laws requiring children to wear
helmets when riding a bicycle
• Previous research using staggered timing of adoption of mandatory youth
helmet laws across U.S. states has shown these laws had robust effects at
educing bicycle related deaths and injuries among youths
― However, this occu
ed not only through significantly greater helmet
wearing but also through reduced bicycling participation
Introduction
• Local and state governments in the United States have debated extending
helmet laws to apply to all individuals, including adults
• The “all-age” helmet laws have been met with strong opposition from cycling
activists, and thus far have not been adopted by any U.S. state
― Handful of places outside US have adopted them, including Australia
and New Zealand
• As public bike-share programs proliferate, there is concern that policies
equiring adults to wear helmets would substantially limit their popularity and
effectiveness
Preview of Results
• This paper provide quasi-experimental evidence on the effects of bicycle
helmet laws by examining Canada where many provinces have adopted
these policies over the past two decades
― The large samples of survey data (over 775,000 respondents) allow test
for treatment effect heterogeneity
• Use the staggered timing of adoption of the provincial helmet laws to
estimate difference-in-differences models of the effects of laws on outcomes
• Use two-way fixed effects models with controls for province and year fixed
effects, individual characteristics, and other relevant public policies
• Confirm prior research that youth-targeted helmet laws in Canada were
associated with at most modest estimated declines in youth cycling and
extremely large increases in youth helmet use
Preview of Results
• Examine all-age helmet laws and find no relationship with youth cycling
• All-age helmet laws did, however, significantly increase youth helmet use by
around 30 percentage points or nearly 200%
• Find all-age helmet laws unrelated to adult cycling participation or intensity
• Also find all-age helmet laws significantly increased probability that adults
eport always wearing helmets by around 19.7–24 percentage points, or
about 50–65% relative to the average helmet use prior to adoption
• Finally, also examine helmet law effects on other exercise behaviors
― Evidence that during winter months in Canada all-age helmet laws are
associated with reductions in cycling and substitution toward in-home
exercise
Institutional Background
• Laws generally
equire
individuals to
wear a helmet
when cycling on
public roads
• Table 1 provides
a list of the
provinces and
cities in Canada
that adopted
different types of
helmet laws
• Use two data sources to estimate the effects of mandatory helmet laws on
helmet use and bicycle riding behaviou
― First data are restricted-use versions of the NPHS from 1994 to 1999
― Second data from the CCHS from 2000 to 2014
• The NPHS and CCHS contain several questions about helmet use, bicycle
iding, and other types of exercise used to create key outcome variables
Data Description and Research Design
• Create two bicycling outcomes based on these questions
― Any Past 3 Month Cycling. Indicator variable equals one if respondent
eported any leisure cycling in past 3 months (and zero otherwise)
― Total # Minutes Leisure Cycling Past 3 Months. Combine information
from questions about the number of times of leisure cycling and length
of cycling on each occasion (using midpoints of ranges and assigning a
value of 1.5 hours to respondents who say they cycle for more than an
hour on each occasion) to create variable representing cycling intensity
Data Description and Research Design
• Also create two helmet use outcomes based on responses to question about
frequency of helmet use
― Frequent Helmet Use. Equals one if the respondent reported wearing a
helmet “always” or “most of the time” when riding a bicycle (and zero
otherwise)
― Always Helmet Use. Equals one if the respondent reported always
wearing a helmet (and zero otherwise)
• The CCHS and NPHS data also allow direct measures of possible
unintended adverse consequences of mandatory helmet laws, as prior work
has found that youth helmet laws reduced youth cycling in the United States
• In addition to cycling outcomes, also examine other exercise, weight, and
obesity outcomes as further tests of possible unintended consequences
Data Description and Research Design
• Estimate two-way fixed effects models separately for youths and for adults
― Yipt are outcomes of interest for individual i in province p in year t
― Xipt is vector of individual characteristics that includes age, sex, race,
education, and marital status
― Zpt is vector of province/time-varying covariates including the provincial
unemployment rate and an indicator for whether the individual lives in a
place that adopted a public bike-share program such and when a
provincial graduated license policy came into effect
― Pp and Tt are full sets of province and year dummies, respectively
Data Description and Research Design
Results – Descriptive Patterns
Results – Descriptive Patterns
Descriptive Statistics
Provincial Trends in Bicycling Participation
Provincial Trends in Helmet Use
Empirical Results
Empirical Results
Robustness Checks
Heterogeneity Effects
Exercise Participation
Exercise Substitution Effects
Discussion and Conclusion
• Results show that bicycle helmet laws in Canada worked to significantly
increase helmet use for both youths and adults
• Youth-targeted laws significantly increased youth helmet use and all-age
laws significantly increased both youth and adult helmet use
― In proportional terms effects of helmet laws at increasing youth helmet
use were larger for all-age laws than for youth-targeted laws
• Find little systematic evidence that all-age helmet laws in Canada induced
population-wide reductions in cycling
• Also find limited effects on other leisure-related physical activities, and do
not find adverse effects on population weight
Questions for Class Discussion
1. Explain how the ‘percentage effect’ is calculated for the all-age and youth
law results in the various specification reported in Table 3. Are these results
significant?
2. What do the results of Table 4 suggest about the effects of helmet laws on
helmet use for adults and for youths?
3. Discuss the results of Table 6 concerning heterogeneity in the effects of all-
age helmet laws on both youths and adults.
4. Highlight and discuss the main results about substitute exercise participation
and seasonal exercise participation as presented in Tables 7 and 8.
ECON7012/8012 – Topic 5a�Evaluating outcomes from policy reform
Paper Details
Introduction
Introduction
Preview of Results
Preview of Results
Institutional Background
Slide Number 8
Slide Number 9
Slide Number 10
Slide Number 11
Results – Descriptive Patterns
Results – Descriptive Patterns
Descriptive Statistics
Provincial Trends in Bicycling Participation
Provincial Trends in Helmet Use
Empirical Results
Empirical Results
Robustness Checks
Heterogeneity Effects
Exercise Participation
Exercise Substitution Effects
Discussion and Conclusion
Questions for Class Discussion
What Do Bicycle Helmet Laws Do? Evidence from Canada
WHAT DO BICYCLE HELMET LAWS DO? EVIDENCE FROM CANADA
CHRISTOPHER S. CARPENTER and CASEY WARMAN∗
Twenty-one states and the District of Columbia require youths to wear helmets
when riding a bicycle, and there has been a push to extend such laws to adults. We
provide new evidence on helmet laws by studying Canada using difference-in-differences
models and restricted area-identified public health survey data with information on
cycling and helmet use for nearly 800,000 individuals from 1994 to 2014. We first
confirm prior patterns from the United States that laws requiring youths to wea
helmets significantly increased youth helmet use. We then provide the literature’s first
comprehensive evidence that “all-age” bicycle helmet laws significantly increased both
adult and youth helmet use by 50%–190% relative to pre-reform levels, with large
effects for younger adults and less-educated adults. All-age helmet laws had modest
effects at reducing cycling and increasing in-home exercise during winter months among
adults but did not meaningfully affect weight. Overall, our findings confirm that all-age
helmet laws can be effective at increasing population helmet use without significant
unintended adverse health consequences. (JEL I18, I12, K32)
I. INTRODUCTION
Bicycling is enjoyed by over 66 million
adults in the United States (Statista 2018)
ut is also associated with substantial health
isks: according to the American Association of
∗The authors are grateful to Mark Anderson and three
anonymous referees as well as Hope Corman and semina
participants at the 2018 NBER Spring Health Economics
Program meeting, the Canadian Economic Association, the
Southern Economic Association, Dalhousie University, Geor-
gia State University, McMaster University, St. Francis Xavie
University, and Saint Mary’s University for very valuable
comments. The results in this paper are based on confiden-
tial versions of the National Population Health Surveys and
the Canadian Community Health Surveys. Readers interested
in obtaining access can contact the authors for directions. The
analysis presented in this paper was mainly conducted at the
Atlantic Research Data Centre which is part of the Cana-
dian Research Data Centre Network (CRDCN). The services
and activities provided by the Atlantic Research Data Centre
are made possible by the financial or in-kind support of the
SSHRC, the CIHR, the CFI, Statistics Canada, and Dalhousie
University. The views expressed in this paper do not necessar-
ily represent the CRDCN or those of its partners. The authors
are also grateful to RDC analyst Heather Hobson for her assis-
tance, as well as Yasmine Amirkhalkhali and Min Hu. The
authors also mention that all e
ors are their own. The usual
caveats apply.
Carpenter: Professor, Department of Economics, Vander-
ilt University, NBER, and IZA, Nashville, TN 37240.
Phone XXXXXXXXXX, Fax XXXXXXXXXX, E-mail christo-
pher.s.carpenter@vande
ilt.edu
Warman: Associate Professor, Department of Economics,
Dalhousie University and NBER, Halifax, NS B3H4R2,
Canada. Phone XXXXXXXXXX, Fax XXXXXXXXXX, E-mail
XXXXXXXXXX
Neurological Surgeons, cycling is the sport