This note is written in response to a request about the details of how to do a social cost study on alcohol (misuse). It is written by an economist who has long been involved in the area both theoretically and practically. 
The point of this note is that it is important to understand that there is a deep theory for social cost estimates as far as economists are concerned. That creates a framework for the measurement of social costs. Economists doing a social cost should have recourse to the theory to apply the framework.
At the heart of the deep theory is the notion of that an economic cost is an ‘opportunity cost’ that is the cost of anything is the opportunity forgone. For an economist, then, one is always looking at an alternative,
In social cost estimation, the alternative is called the ‘counterfactual’ to be compared with the ‘actual’ situation. A social cost measurement exercise should always state what the counterfactual is. Without one the exercise is meaningless. There is rarely a unique or even obvious one.
For instance, it might be that the counterfactual is that there never has been any alcohol consumption, or it that all alcohol consumption ceases on a certain date or that on a certain date that alcohol consumption is reduced to the point where there was no misuse of alcohol although moderate consumption continued, or that there has never been any alcohol misuse. (The latter is the one I used.) Each counterfactual will lead to a different magnitude for the social cost.
(There is much puzzlement over the necessity of considering ‘avoidable costs’. They arise from a counterfactual scenario where, say, alcohol consumption had existed in the past but it stopped overnight. Costs arising from the past consumption will continue to occur after the cut off day (for instance there will be still be treatment costs and deaths from cirrhosis of the liver long after the cutoff date.) The costs are considered unavoidable and the counterfactual focuses on avoidable costs those that do not continue.)
The actual and counterfactual scenarios will have different outcomes. In principle a list is made of all the items in the two outcomes with particular attention to the differences between them. (For instance as alcohol consumption is reduced or eliminated, cirrhosis of the liver will decline.) The items are then valued according to the standards of economic ‘value theory’ thereby giv8ing two gross magnitudes, the differences of which is the social cost revealed by the counterfactual.
In practice every item is not identified – some remain the same, others are compared in aggregate. So the exercise is not as complicated as the previous paragraph might suggest. But that is the underlying notion and should be referred to when there is a doubt.
I do not propose here to discuss the valuation method that could end up a very long note.
However the implicit assumption is that all people’s consumption is valued the same, and that distributional issues are irrelevant. That is almost standard in economic analysis of alcohol consumption, except for some hand-waving. It is not my choice but I have found it difficult to offer an alternative paradigm. The best I can do is to split the aggregate into sectors – the recommended practice rarely followed in any detail. I add that this issue explains why transfers (such as taxes and benefits) are not included in social costs, although those not using the deep analysis sometimes get confused by how to treat transfers.
The simple summary is this. There is a well-developed tool in economics called cost benefit analysis (CBA), based on deep theory, with manuals that explain the framework and practice. It is used for many purposes, ranging from project evaluation to pharmaceutical evaluation. The economists’ approach to social costs of alcohol (and other drugs) is based upon CBA. Indeed done properly, were the need for a CBA arose, the calculations could use the cost-of-illness material. In effect the cost-of-illness exercise is a CBA with a particular counterfactual scenario. The rules and interpretations are the same.
One further issue needs to be mentioned. When CBA was first developed, it was applied to situations where the purpose was to change market output – GDP. However once it became used for analyses in the social sector, the issue of consequences in the nonmarket economy needed to be considered. For instance, two different methods of treating an illness may involve either the person being kept in a hospital or sent home. The first option impacts on GDP but the second has a similar impact on the care provided by the household which is not included in GDP. To ignore the household impact would be to bias treatments toward cost-shifting. Nowadays one would want to take the nonmarket sector into account in a CBA and therefore in a cost of illness study.
The issue deepens when one is concerned with wellbeing which is hardly measured in market GDP. For instance the effect of a pharmaceutical treatment may be to add to a person’s quality of life or to extend that life. The issue was not unappreciated in early CBAs. (One of my earliest introductions to CBA was how to deal with deaths in road accidents.) If this wellbeing is not taken into consideration treatments would be based in favour of one which resulted in shorter lives and greater discomfort for the patient because it was so much cheaper. The same analysis applies to social costs of drug measurement. In my experience the difference in quality of life/wellbeing aggregates far exceed to GDP ones.
 e.g. International Guidelines for Estimating the Costs of Substance Abuse, 2ed (WHO 2003) (also Canadian Centre for Substance Abuse, 2002) with E. SINGLE, D. COLLINS, H. HARWOOD, P. KOPP, H. LAPSLEY & E. WILSON; and The Social Costs of Tobacco Use and Alcohol Misuse (Department of Public Health, Wellington Medical School, Research Paper 2)