Dr. John Holmes:

I thank the committee for inviting us to give evidence today. As the Chair said, the Sheffield Alcohol Research Group has been conducting work looking at alcohol policies in general and minimum unit pricing in particular since 2008. The Irish Government and the Northern Ireland Executive commissioned us in 2013 to examine the potential effects of minimum unit pricing in their respective countries and we will focus on that work in our presentation today. However, before coming to that, I would like to give the committee a little background on the wider evidence on the effectiveness of using alcohol prices to reduce alcohol-related harm.

A major review in 2009 examined the evidence on the impact of alcohol price changes on alcohol consumption. It found that increases in alcohol prices were consistently and significantly associated with falls in consumption. This was the case for total alcohol and also for each alcoholic beverage such as beers, wines and spirits. We saw the same thing for younger and older drinkers and we saw that binge drinkers were also responsive to price changes. These findings have been replicated across at least two other major reviews. An example finding is that, on average across different times and places, a 10% increase in the price of all alcohol is associated with an average 4.4% fall in consumption.

However, our primary interest is not whether alcohol price increases reduce consumption but whether they reduce the harm caused by alcohol. A further review of 50 studies concluded that indeed they do. Based on the findings of that review, the researchers estimated that a doubling of alcohol taxes in the US would lead to a 35% reduction in alcohol-related mortality, an 11% reduction in car crash deaths and smaller reductions in sexually transmitted diseases, violence and crime associated with alcohol.

Minimum pricing is a specific form of price increase targeting the cheapest alcohol which, as you will see later, is disproportionately purchased by the heaviest drinkers. Therefore, there is good reason to expect it will have an impact on alcohol-related harm. Several Canadian provinces have had this kind of policy in place for many years and recent evaluations have shown that those policies have reduced the harm from alcohol in those provinces. The Canadian policies are not quite the same as minimum unit pricing as proposed in Ireland and the UK, because minimum prices are not directly linked to the strength of the drink. However, the same basic principle of a minimum price, below which alcohol cannot be sold to consumers, does apply.

The Canadian evaluations have shown that, all else being equal, increases in minimum prices are associated with falls in alcohol consumption, in alcohol-related hospital admissions and also in deaths due to alcohol. The graph in figure No. 1 of the opening statement, which has been circulated to members, shows one example study. The black line is the average minimum price of all alcohol in British Columbia between 2002 and 2009. The grey line is the rate of wholly alcohol-attributable deaths in the province. These are the deaths very closely associated with heavy drinking such as alcohol poisoning and alcoholic liver disease. The graph shows that a sharp and sustained increase in minimum prices in 2006 was swiftly followed by a sharp and sustained fall in deaths closely associated with heavy drinking. This prima facieevidence of policy effectiveness was supported by the statistical analysis which estimated that a 10% increase in minimum prices would be associated with a 32% fall in those deaths closely associated with heavy drinking. So there is good evidence from Canada that minimum prices and minimum price increases lead to substantial falls in the harm caused by alcohol, all else being equal.

I will now turn to our own work. We were asked by the Irish Government to estimate the potential impact in Ireland of different levels of minimum unit pricing, MUP, using our Sheffield alcohol policy model, SAPM, as we call it. For a given alcohol policy, our model provides estimates of changes in alcohol consumption, in consumer spending on alcohol, in revenue to retailers and Government, in the rates of various different alcohol related harms, in the costs of those harms to individuals' quality of life and in the direct costs to public services. A key feature of the model is that we do not just estimate these effects at a population level, we are also able to estimate the effects on different groups in the population defined by age, gender, income and how much alcohol people drink.

At this point I will be talking about low risk drinkers, increasing risk drinkers and high risk drinkers. Low risk drinkers are drinking within the Government's drinking guidelines of around 17 standard drinks per week for males and 11 standard drinks per week for females. High risk drinkers, the heaviest drinkers, are those consuming more than 40 standard drinks per week for males and 28 standard drinks or more for females. Increasing risk drinkers fall between those two levels.

The model methodology for SAPM is too complex for this short presentation and we are happy to discuss it in more detail during the questions. The basic idea is summarised in the figure No. 2. The SAPM works sequentially so we first estimate the impact of introducing MUP on prices. We then use those price changes to estimate how peoples' consumption would change. Then we use those consumption changes to estimate how rates of the different harms would change. Finally, we use the changes in the rates of harm to estimate how the costs of harm would change. Although each of these steps involves quite complex statistical work, the methods we use are largely orthodox in scientific terms. What has made our work so influential and impactful is that we look at a very broad range of outcomes, which is unusual for this kind of model. We are also able to look at the impacts on different groups in society and hopefully the committee will see why that is useful.

I will now turn to results. First we estimated the impact of different levels of minimum price on consumption. Figure No. 3 shows that as the minimum price gets higher, the consumption reductions get bigger. This is fairly obvious because one is affecting more of the market if one introduces a higher minimum price. Above a minimum price of about 60 cent per standard drink one starts to get quite large reductions in consumption. At 70 cent per standard drink one gets a consumption reduction of 1.9%. At 80 cent per drink one gets a 3.8% reduction. At 90 cent per standard drink one gets a 6.2% reduction. For the remainder of our presentation I will focus on a minimum price of €1 per standard drink just as an indicative example.

Based on our modelling, we estimate that a €1 minimum unit price introduced in Ireland would reduce total alcohol consumption by around 8.8%. By the 20th year after the introduction of this policy, when we would expect to see the full effects, that reduction of 8.8% would result in around 200, or 16%, fewer deaths per year, and around 6,000, or 10%, fewer alcohol-related hospital admissions. From year 1 of the policy, and every year thereafter, we would expect to see around 1,500 fewer alcohol-related crimes and over 100,000 fewer days absent from work due to alcohol. A cost breakdown is provided in table No. 1 in the handout but the headline figure is that over 20 years, the total reduction in the cost of alcohol-related harm is estimated to be around €1.7 billion. This accounts for costs to the police and health services, and also a financial valuation of improved quality of life.

The impact on retailers is likely to be positive, as minimum unit pricing is not a tax. Money from the higher prices is held by the retailers. We estimate that off-trade retailers, that is, shops and supermarkets, would receive approximately €69 million extra per year from alcohol sales. Although the policy does not directly affect pubs and restaurants because their prices are already higher than the minimum price threshold, we would expect to see changes in consumer behaviour. For example, people who are buying less alcohol in the supermarket might go to the pub more. For that reason, we expect there may be a slight increase in on-trade revenue as well.

Regarding tax revenue, the negative impact on the Exchequer would be modest, as lost duty from falling alcohol sales would be largely offset by rising VAT revenue from higher alcohol prices.

A key feature of the policy of minimum unit pricing is that it does not affect all drinkers equally. The main driver is how much cheap alcohol different groups buy. Low-risk drinkers purchase very little cheap alcohol - that is, one or two standard drinks per week for less than €1 per standard drink - irrespective of whether they are in poverty. Even low-income low-risk drinkers do not buy much of this cheap alcohol. Compare this to high-risk drinkers, who purchase substantial quantities of cheap alcohol. Although high-risk drinkers in poverty buy more cheap alcohol than those not in poverty - 43 standard drinks per week versus 26 - it is clear that those on higher incomes still buy cheap alcohol. It is not just low-income people who will be affected.

The Sheffield alcohol policy model, SAPM, takes account of these different purchasing patterns, and the estimates of annual reductions in consumption for each of these groups reflect this. We estimate that the annual reduction in consumption for low-risk drinkers would be very small; they would be largely unaffected. It is estimated that low-income low-risk drinkers would reduce their consumption by just 25 standard drinks per year, which is equivalent to drinking three fewer bottles of wine per year. That is a small reduction. Compare this to high-risk drinkers, who, it is estimated, would reduce their consumption by well over 500 standard drinks per year for those in poverty and 480 standard drinks per year for those not in poverty. This is equivalent to approximately 60 to 70 bottles of wine per year. That shows the much greater impact on high-risk drinkers.

A concern for some parties has been the potential financial impact on low-income drinkers. Our modelling does not support these concerns. We estimate that spending would fall in most groups. Rather than spending more to maintain their level of consumption, most drinkers tend to reduce their consumption and spend less on alcohol as a result. Spending is only estimated to increase in the higher-income groups. It is worth noting from the graph that these spending changes are fairly small. The largest reduction is €159, but most reductions are in the tens of euros. When one compares that against how much high-risk drinkers spend on alcohol per year - an average of over €5,000 - we are not making much of a dent in people's alcohol spending. Spending is not the significant factor; it is the pattern of consumption, and who is affected by those consumption patterns.

All research has limitations. Our aim, which we hope to talk about in greater detail, is to be transparent about these and help policy makers understand their implications for our estimates. Some of the key limitations of the SAPM include our assumption that prices of products above the minimum unit price threshold will be totally unaffected. We assume that prices below the threshold come up to the threshold and everything above it is unaffected. This is unlikely to be the case in reality because, for instance, Diageo is unlikely to want Smirnoff to become the cheapest vodka on the market. We are likely to see what we call premiumisation - products being pushed up to assume a certain place in the market. What that means is that we are probably underestimating the impact of the policy. Our estimates are probably a little conservative.

There is a lack of clear evidence for the impact of such policies on the use of illicit alcohol or alcohol substitutes such as illicit drugs. We know from elsewhere that when prices increase we see falling levels of harm, and this suggests that any negative side effects are not sufficient to outweigh the benefits. We do not ignore these limitations and, where appropriate evidence is available, we test the sensitivity of our results to a range of assumptions, data sets and analytical methods. What those sensitivity analyses have shown is that although the numerical results change, the broad conclusions stay the same. There is strong and consistent evidence that price increases reduce alcohol consumption and related harm. Minimum pricing is a targeted form of price increase, as it tackles disproportionately cheap alcohol purchased by heavier drinkers. We estimate that it would not penalise low-risk drinkers, irrespective of income, because they buy little of this cheap alcohol. In contrast, high-risk drinkers buy large quantities of cheap alcohol and would be affected. These conclusions have been found to be robust to a range of alternative assumptions, data and analytical methods.