Let's Talk About Canada

When advocates of lowering the drinking age bring up Europe for comparison, they often paint themselves into a corner.  America and Europe are very different, so in many ways it's apples and oranges.  But European countries are not the only ones that have lower drinking ages.

We do in fact have a good yardstick for what would have happened had the drinking age not been raised to 21 in the 1980s.  It's called Canada.  Their drinking ages have remained at 18 or 19, depending on the province, for the past three decades.  And it is the country that most resembles America in many ways, especially in terms of its car culture.  So let's talk about Canada then.

AGE LIMITS BY PROVINCE/TERRITORY:

Drinking age:  18 in Alberta, Manitoba, and Quebec, all others 19.

Smoking/vaping age (tobacco):  Same as drinking age, except Saskatchewan is 18, and Prince Edward Island recently raised it to 21 effective March 2020.

Toking age (cannabis):  Same as drinking age, except Manitoba is 19, and Quebec raised it to 21 effective January 1, 2020.

Driving age:  16 generally in most provinces (14 for learner's permit in Alberta).  Note that the GDL learner and probationary period can be cumbersome and can last as long as two years depending on province.  Note also that the laws were much more lenient during the 1980s and early 1990s, which was before GDL was implemented.

ALCOHOL-RELATED TRAFFIC FATALITIES

It is often claimed by proponents of the 21 drinking age that raising the drinking age saved lives.  While alcohol-related traffic fatalites did decline, correlation does not prove causality.  First of all, the trend began in 1982, two years before the National Minimum Drinking Age Act that forced all states to raise their drinking ages to 21 by 1987.  Perhaps the trend began even earlier, as total 18-20 year old fatalities began declining in 1979-1980, but 1982 is the first year that FARS has reasonably reliable data for alcohol-related fatalities.  And Canada saw a remarkably similar trend, as you can see in the graphs below (courtesy of NHTSA).

Percent Change from 1982-1997 US: drivers age 16-20 in fatal crashes with positive BAC (FARS) Canada: driver fatalities age 16-19 with positive BAC (TIRF)	Percent Change from 1982-1997 US: percentage of drivers age 16-20 in fatal crashes with positive BAC (FARS) Canada: percentage of driver fatalities age 16-19 with positive BAC (TIRF)

Interestingly, the decline in alcohol-related traffic fatalities since 1982 occurred at about the same rate in both countries, with no evidence of divergence in the expected direction despite the fact that Canada did not raise the drinking age to 21.  In both countries, drivers under 21 saw some of the largest declines of all compared with other age groups, though all ages saw some decline over the long run.  Moreover, progress continued for Canadian teens from 1997-2005, while unfortunately it stalled for their American counterparts during that time, only resuming after gas prices began to skyrocket (which Americans were not used to) and the economy began to sag.  In 2005-2006, the rate of total 15-24 year old traffic fatalities (per 100,000 people) for the USA was 25.5, and 16.9 in Canada, the latter being 33% lower than the former.  And while the percentage of traffic deaths labeled as "alcohol-related" tends to be higher for Canadians of all ages than their American counterparts as a rule, that is likely due to a combination of differences in definitions and testing rates, Canada being somewhat more rural than the USA, and also higher gas prices and unemployment rates in Canada vs the USA (both of which reduces non-alcohol related traffic casualties more than alcohol-related ones), especially since the 1990s.  On an absolute per-capita basis, both types of traffic deaths are still lower in Canada than the USA. 

In other words, the downward trend in fatalities can be explained entirely by other factors, which likely include, inter alia:

• Tougher laws and penalties for DUI
• Better DUI enforcement
• More education and awareness of the problem of impaired driving
• Designated driver programs
• Seat belt laws
• Safer cars and roads due to improved engineering
• Demographic changes
• Changes in gas prices

In fact, some things, such as the 0.08 BAC limit, were already in place in Canada well before 1982.  The BAC limit in the US was still 0.12-0.15 in most states in 1982, and since then all 50 states, DC, and Puerto Rico lowered it to 0.10 and eventually to 0.08.  Zero-tolerance laws for younger drivers, which were enacted in all 50 states and DC by 1998 (beginning in the 1980s), were nonexistent until well into the 1990s in most Canadian provinces, and until very recently no province's law was as strict as in the USA.  Also, graduated driver license rules for younger drivers had eventually become stricter than Canada's in several states.  If anything, fatality rates for all ages should have declined faster in the US relative to Canada, but for some reason they did not.

Worse still, according to a 2004 book by Leonard Evans, former safety researcher for General Motors, America has been lagging behind several other countries in terms of traffic safety.  The table below shows the change in the number and rate of total traffic fatalites (all ages) over time in the US and three other countries that maintained lower drinking ages since 1979.

Country	MLDA	1979 Fatalities	2002 Fatalities	% Change (raw)	% Change (per vehicle)	% Change (per VMT)
USA	21	51,093	42,815	-16.2%	-46.2%	-52%
UK	18	6,352	3,431	-46.0%	-67.1%	-70%
Canada	18 or 19	5,863	2,936	-49.9%	-63.5%	N/A
Australia	18	3,508	1,715	-51.1%	-79.1%	N/A

As for all-cause deaths, as well as traffic deaths specifically, regression discontinuity studies in the USA invariably find a significant jump in deaths upon turning 21, while in Canada there is a jump at 18 or 19 depending on the provincial MLDA, albeit a smaller one than the USA sees at 21, and even then only among males.  This simply confirms the "powder keg theory", and also militates against the idea that a higher drinking age results in fewer NET excess all-cause deaths in the long run.

TEEN DRINKING AND "BINGE" DRINKING

Of course, highway fatalities are not the only concern raised about the drinking age.  Proponents of the 21 drinking age also claim it reduced teen drinking and "binge" drinking (5+ drinks in an occasion).  But that trend, as measured by the Monitoring the Future (MTF) survey, began in 1979, which was several years before most states adopted the 21 law.  And raising the drinking age may lead to reduced reporting in surveys even in the absence of actual behavioral change.  So all teen surveys ought to be taken with at least a grain of salt, if not a pound.

The province of Ontario (with a drinking age of 19 since 1979) has a similar survey (OSDUS) going back to the 1970s, though not all the measurements are the same.  The following table, again courtesy of NHTSA, shows the changes in Ontario during the most relevant time period (1979-1991 unless otherwise stated) compared with the USA.  The American data are for grade 12 only, while the Ontario data are for grades 7-13 combined, so they are not directly comparable.  The trends, however, are strikingly similar.

Drinking Behavior	1979	1991	% Change, 1979-1991
Annual drinking: USA	88.1%	77.7%	-12%
Annual drinking: Ontario	76.9%	58.7%	-24%
Daily drinking: USA	6.9%	3.6%	-48%
Daily drinking: Ontario	0.9%	0.4%	-56%
5 or more drinks: USA (past 2 weeks)	41.2%	27.9%	-32%
5 or more drinks: Ontario (past 4 weeks)	27.0%	21.9%	-19%
5 or more drinks: USA (past 2 weeks, 1979-1993)	41.2%	27.5% (1993)	-33%
5 or more drinks: Ontario (past 4 weeks, grades 7, 9 and 11 only, 1979-1993)	24%	15% (1993)	-38%
Drive after drinking: USA (past 2 weeks)	31.2% (1984)	18.7%	-40%
Drive after drinking: Ontario (annual)	43.2% (1983)	20.1%	-53%

Of course, that is only one province.  What about the rest of Canada?  Unfortunately, most Canadian provinces do not have longitudinal data going back that far, or even before 1996, so we are stuck with doing a crude cross-section using current data for our international comparison.  The following table consists of the past-month prevalence of "binge" drinking (5+ drinks in an occasion) for high school seniors as reported in recent (2005-2008) student surveys, in selected states and provinces.  All American data were taken from the national Youth Risk Behavior Survey, while Canadian data were taken from various provincial surveys. 

Location	Binge Drinking (Grade 12)	Drove after drinking (Grades 9-12)	MLDA	Year
USA (overall)	36.5%	10.5%	21	2007
North Dakota	47.0%	18.7%	21	2007
South Dakota	47.3%	13.0%	21	2007
Montana	46.2%	18.5%	21	2005
Vermont	38.0%	9.2%	21	2007
Wisconsin	42.7%	14.3%	21	2007
Alberta	46%	N/A	18	2005
Alberta	41.1%	10.3%	18	2008
Atlantic Provinces	49.7% (44-55%)	N/A	19	2007
British Columbia	43.2%	N/A	19	2008
Manitoba	48.8%	N/A	18	2007
Ontario	48.0%	11.6%	19	2007
Saskatchewan	70%+	N/A	19	2008
Puerto Rico (USA)	33.2%	7.3%	18	2005
Guam (USA)	30.3%	7.8%	18	2007
Northern Mariana  Islands (USA)	36.5%	14.1%	21	2005

Care was taken to compare apples to apples, and that is why the YRBS was used for American data instead of the Monitoring the Future survey.  Canadian surveys and YRBS report past-month "binge" drinking, while MTF reports it for past two weeks (and thus contains lower numbers).  The Manitoba figure was for the past-year, as comparable data for past month were not available, and can thus be considered an upper bound for past-month "binge" drinking.

Note the similarity between the northern states and Canadian provinces which are geographically and demographically similar--they generally tend to be around 50%.  There does not appear to be a significant correlation between the drinking age and "binge" drinking rates.  Remember again that the American data are more likely underreported than the Canadian data due to the drinking age difference and cultural factors.

Also note the below-average numbers for the US territories of Guam and Puerto Rico, both of which have a drinking age of 18 (until Guam raised it to 21 in 2010).  In fact, even the temperance-oriented Robert Wood Johnson Foundation concedes that Puerto Rico was able to reduce both alcohol-related traffic fatalities and underage (under 18) drinking since the 1990s without raising the drinking age.  From 1982 to 2009, Puerto Rico saw a whopping 84% decline in teenage (16-20) drunk driving fatalities, while the nation as a whole saw a 74% drop, in both cases to record-low levels.  Now that's a great American success story.

In other words, it appears that Miron and Tetelbaum (2009) were spot on when they said that the drinking age appears to have "only a minor impact on teen drinking," just like they were right about its lack of a lifesaving effect on the highways. 

OTHER EFFECTS, OR LACK THEREOF

As for allegedly creating a nation of brain-damaged, alcoholic felons by allowing 18-20 year olds to drink, this myth does not hold water either.  In international standardized tests, Canadian 12th graders beat their American counterparts despite the former having similar or lower scores in 4th grade.  In fact, nearly all the countries that beat us set the drinking age at 18 or even lower!  The alcoholism rates in both the USA and Canada are also roughly equivalent, and the adult per capita alcohol consumption rate is actually slightly lower in Canada.  Alcohol-related death rates, both in terms of liver cirrhosis as well as "alcohol use disorder", are also lower in Canada according to the World Health Organization.  In fact, Canadians live on average three years longer than Americans.  And the rates of violent crimes, especially the most serious ones like homicide, tend to be significantly lower in Canada as well.

And while high school graduation rates are slightly lower in Canada compared to the USA, they still exceed most other peer countries, and provinces with a drinking age of 18 are not significantly different from neighboring provinces with drinking age of 19.
 
In short, puritanical America, with our 21 drinking age, appears to be the less healthy society of the two.  And while correlation does not prove causation, the aforementioned statistics certainly won't convince anyone that our illiberal policies are doing much good in reducing alcohol-related problems or improving public health and safety.  We need to see the forest for the trees, something America chronically fails to do in terms of alcohol policy.

Perhaps we can learn a thing or two from our neighbor to the north?

QED