Disclaimer Disclaimer Disclaimer Disclaimer Home           Home Home           Home About  site   About  site About  site   About  site About  me    About  me About  me    About  me Current reading   Current reading Current reading   Current reading CONTACT   CONTACT CONTACT   CONTACT Links                  Links Links                  Links Book shelves      Book shelves Book shelves      Book shelves
Blog Archive    Blog Archive Blog Archive    Blog Archive Recent posts  Recent posts Recent posts  Recent posts
A couple of weeks ago the new update of alcohol sales data in Scotland were released. This provided a first  opportunity to see whether the introduction of ‘Minimum Unit Pricing’ (MUP) in Scotland in May 1 2018 has  had any impact on sales at all.  If you want more information about the MUP in Scotland, here is a good start <link>.  On twitter this sparked a debate about what can be learned from these data, and academics from The  University of Sheffield naturally got involved in these discussions. As you will have seen in recent posts on  this blog  (for example here),   I do some work with natural experiments, timeseries and the creation of counterfactuals (counterfactuals are  the ‘what would have happened has this thing not changed’ situation), so I asked, and received, the trends  data from Sheffield. I put the resulting causal estimates of the effect of the introduction of MUP on the  alcohol sales on Twitter in response to a between one of the Sheffield researchers and a solicitor with a  tenuous grab of Twitter. However, it would be a shame if those results just disappeared, so here a short post  describing them.  * So there are a couple of things to be aware of first:  By way of a bit of background, the minimum price per unit of alcohol was set to 50p. What this  means in practice is that this would affect sales through supermarkets and off-licenses (where  alcohol can be bought cheaply) but not so much drink sold in pubs, restaurants and clubs (because  they already charge more). As such, MUP should help the struggling pub industry in the UK rather  than be a problem. So this is our first hypothesis right there: if causally related, any effects from  MUP on sales of alcohol should be observable in the sales of off-licenses but not in on-trade sales.  We need to create a good comparator, because just a comparison between pre- and post-MUP will  not really work. Other changes that may have happened, such as changes in the population  structure or economic up- or downturns for example, could affect these trends as well. Here this is  difficult because there are not much data around that would be appropriate to use as comparator.  Luckily, and without much work, comparable data for England and Wales (combined) could be made  available. At face value this works because of comparable demographic and economic patterns.  Similar to previous work, I used Bayesian structural timeseries. In general, this is a good method to  tackle these kind of questions because it combines timeseries modelling with tools to create a  counterfactual from a combination of different other (control) timeseries. Here this is a bit of an  overkill because there is only one timeseries (that of England and Wales) to create the  counterfactual with….but I had the scripts ready to quickly run the basic model. So there is that…. A  more detailed description of how this works to estimate the effect of an intervention can be found  here <LINK>. However, compared to the results that made it on Twitter, I had a closer look at the modelling to  see if this could be improved….and of course it could.  So the results of this new modelling. Overall, see the figure below, there has been a 5% stronger decline  in alcohol sales in Scotland since the introduction of the MUP then what was expected based on the  counterfactual, and this effect ranges between an 11% reduction and a 1% incrase with a probability of a  real reduction of 95%; so that is quite convincing.   As per our hypothesis, any MUP effect would be strongest in off-trade sales, and indeed as shown in the  figure below these have declined by 6% more than expected; with a range of the effect of -1% to -12%, and a  probability of a true effect of 99%. So this makes our conclusion that this is likely the result of the  introduction of MUP and not something else stronger. And finally, the evidence for our hypothesis would be even stronger if we would also find no, or less,  evidence of an effect on on-trade sales. These results are shown in the Figure below, and indeed, although  there is also an additional reduction in on-trade sales of 4% (apologies for the typo in the figure), there is  little evidence that this is the result of the introduction of MUP; the range of possible effect sizes ranges  from -26% to and 19% increase and the likelihood of a real effect is only 63% (or pretty much 50/50).  So there you go. These analyses provide some evidence that MUP does have an effect on alcohol sales in  Scotland. It does however, also show that the effect is relatively small (although a 5% extra reduction in  alcohol sales in Scotland equates to quite a lot of drinks), and suggests that the majority of the impact is on  the off-trade; as intended.  Of course, the real test will have to do with whether MUP also leads to a reduction in the incidence of  adverse health effects from alcohol. According to modelling by the University of Sheffield estimated that in  the first year alone, introducing a 50p minimum unit price in Scotland would mean around:  60 fewer deaths 1,300 fewer hospital admissions 3,500 fewer crimes The first data of Scottish alcohol mortality post-introduction of MUP was published recently <link> and  eyeballing it these do not look as promising. Unfortunately, the data for England and Wales are not yet  available, so I cannot do these analyses for health outcomes yet….. 
Serving the community with a smile.......... ................in a Public Healthy kind of way
The impact of the Minimum Unit Pricing (MUP) in Scotland
Back Back Back