Stop telling us that minimum alcohol pricing ‘works in Canada’. It’s nonsense

One of the abiding myths about minimum pricing for alcohol is that there is real world evidence from Canada that it ‘works’. This is important to campaigners because the rest of their evidence is based on a questionable computer model from Sheffield University. The best known factoid about the Canadian experience is that a small rise in the minimum price led to alcohol-related deaths falling by a third in the province of British Columbia, thus leading The Guardian to conclude that ‘Canada is proof that state-controlled drinking is good for health’.

Unfortunately, this is nonsense on stilts. Alcohol-related deaths never fell by a third or anything close to a third. In fact, the annual number of deaths directly attributed to alcohol rose from 315 to 443 in British Columbia during the period in question (2002 to 2011). The number of alcohol-related hospitalisations also rose.

Nevertheless, campaigners are doubling down on this claim by reporting a similarly miraculous effect of violent crime. In a study being touted by the Institute for Alcohol Studies, the same increase in British Columbia’s minimum price is said to have reduced violent crime by nine per cent. This claim is illustrated by the graph below, taken from the study:

graph1Unlike the decline in deaths, the drop in crime did actually occur, so that’s a good start, but the graph still raises a number of questions. Firstly, why was $1.17 per unit associated with a lower crime rate in 2010 but not in 2002? Secondly, the price rise was tiny. Can we really believe that a difference of a mere four or five cents per unit is enough to have a dramatic effect on the crime rate?

But the biggest weakness here is the absence of a control group. Without another jurisdiction with which to compare it, all we have is a rough bivariate correlation. Crime, including violent crime, has been generally dropping in most Western countries for twenty years or more. Has the decline in British Columbia been more rapid than in the places which don’t have minimum pricing?

An obvious comparator is Britain, which has so far refused to introduce minimum pricing despite incessant lobbying from ‘public health’ groups. If we take the equivalent data (police recorded crime against the person), look at the same time period (2002-10) and use the same scale, we can see that we have seen a larger decline in violent crime since British Columbia started increasing their minimum prices than British Columbia did.

graph2

The data from England and Wales provide an even better fit for the ‘minimum pricing works’ hypothesis than does the Canadian data. Whereas the crime rate in British Columbia stayed flat when the minimum price went down (in real terms), it rose in England and Wales, as it should do according to the hypothesis. The minimum price then started going up in 2005 and the crime rate immediately started to go down. It would be a beautiful fit if it weren’t for the small problem that we didn’t have minimum pricing. British Columbia did.

Perhaps minimum pricing is such an awesome policy that its effect on violent crime was cosmically transmitted across the ocean. Or, more likely, crime rates are generally declining around the world, so pretty much anything is going to coincide with a fall in crime if you look hard enough.


  • Jon Foster

    It’s hard to know where to start with this rather un-thoughtful and fact-light hatchet job, but let’s take
    things in order.

    Firstly, the claim that previous research in Canada said that alcohol related deaths had gone, when
    actually they had gone up. Chris Snowdon has already been corrected on this,
    but perhaps has forgotten, that the study referred to death rates, not total
    death numbers. There was a significant increase in the population over the
    period of the study, and as a result it is no surprise that total number of
    alcohol related deaths rose. The rate of alcohol related deaths within the
    population decrease however, and this decrease was associated with alcohol
    price changes (please note the word ‘associated’) this is not a very difficult
    concept to understand I think. More details can be found on page 14 here: http://www.ias.org.uk/uploads/pdf/News%20stories/iasreport-thomas-stockwell-april2013.pdf

    Secondly, ‘why was $1.17 per unit associated with a lower crime rate in 2010 but not in 2002?’. There
    could be a number of other explanations/variables however, such as levels of
    policing or demographic changes, or it could be the impact of price changes
    over time; as prices decrease from $1.17 people felt that alcohol was getting
    cheaper, so bought more, but as it increased towards $1.17 they felt it was
    getting more expensive, and so bought less. It’s interesting that despite
    picking on this small detail, Snowdon decided against pointing out any of the
    small details that support the positive impact of minimum pricing.

    Thirdly, ‘can we really believe that a difference of a mere four or five cents per unit
    is enough to have a dramatic effect on the crime rate?’ Just to be accurate,
    over the course of the study minimum prices for packaged beer increased from $3.00
    to $3.54, and for draft beer from $2.05 to $2.22 per litre.

    Rather than dismissingly asking ‘can we believe’ this had an impact, perhaps it would
    be worth looking at what the study actually found. This was that a 10% increase
    in average minimum price for all drinks was significantly associated (note that
    word again) with:

    An 18.81% reduction in alcohol related traffic violations
    A 9.17% reduction in crime against the persons, and
    A 9.39% reduction in total rates of the crime outcomes examined.

    As noted by the authors, some (but not all) of the confidence intervals are quite
    large, so it is hard to be precise about the exact impact of alcohol pricing,
    but they all head in the same direction. Crucially however, there was no
    significant association between minimum alcohol prices and non alcohol related
    traffic violations, which is surely food for thought.

    Snowdon has engaged with none of this, preferring to select one or two misleading, but
    easily explained, issues in an attempt to dismiss the whole study. It’s true
    that a control group would be helpful, but this is something difficult to do in
    large scale studies like this. However, a very wide range of effects were
    controlled for in the study, which, lets not forget, was published in a peer
    reviewed journal, and has no doubt passed before more informed eyes than those
    of Chris Snowdon.

    The next point in the blog post is to introduce a fallacy in order to discredit
    things further, by trying to match the Canadian pricing with falls in crime in
    the UK. This is entirely meaningless so I won’t comment on it further.

    A close reading of this blog shows that there is actually very little to it – an
    old allegation already proved false, a comment on a small detail of the study
    with little relevance, an attempt to discredit the findings without actually
    engaging with them, and then a misleading fallacy thrown in for good measure.
    There is nothing substantive here at all, which is very disappointing coming
    from a magazine like the Spectator.

    • JonathanBagley

      “The next point in the blog post is to introduce a fallacy in order to discredit
      things further, by trying to match the Canadian pricing with falls in crime in
      the UK. This is entirely meaningless so I won’t comment on it further.”

      Pointing out that crimes in various categories have been falling throughout the Western World is certainly not meaningless; and the logical consequence is that any association which can be found has as much validity as your minimum price claim. My favourite was the explanation for US crime that the criminals were never born due to a change in the abortion laws. I believe that may now have been discredited. So perhaps it’s lead-free petrol, or replacing lead water mains with copper?

    • James

      Who cares what contortions bureaucrats perform with their bogus statistics. We like to have a drink of an evening and there’s an end on’t. Leave it to the market, Mr Pecksniff.

  • Jon Foster

    ‘Pointing out that crimes in various categories have been falling
    throughout the Western World is certainly not meaningless; and the
    logical consequence is that any association which can be found has as
    much validity as your minimum price claim.’

    As the study controlled for the general decrease in crime, demographic changes and a whole host of other variables, that statement is a very long way from being true. The associations found were in addition to general decreases in crime.

    As I’ve already mentioned, it was found that minimum pricing was
    associated with, amongst other things, an 18.81% reduction in alcohol
    related traffic violations, but that it was not associated with
    non-alcohol related traffic violations. Yes this is a correlation, but given that you’d expect alcohol pricing to impact only on alcohol related traffic incidents and not on non-alcohol related traffic incidents, minimum pricing seems to be a very plausable explination.

  • oldcobbler

    Wouldn’t other Canadian states make for a better and more obvious comparison than the UK ?