Cannabis and asbestos: The cost-effectiveness of prohibition as health policy

The aim of a health intervention is generally to cause a cost-effective increase to some normative outcome such as by reducing mortality, increasing productivity and so forth. Often policy makers want to encourage or discourage certain behaviours by using incentives or disincentives. One such way the government may intervene is through prohibition. Often drug prohibition is touted as a health policy intervention to prevent harms both clinical and social. However, it is probable that this form of intervention is not cost-effective.

Let’s distinguish prohibition from other forms of disincentivisation by saying that prohibition does nothing to internalise the externality that harmful behaviours create. A tax on cigarettes disincentivises the consumer into smoking less but the revenue can be used against the negative effects of smoking, it is a Pigouvian tax; whereas putting someone in prison compensates no-one. Admittedly such laws are being reformed to help addicts rehabilitate rather than punish them but these programs are not funded by drug use itself.

One of the most widely discussed prohibitions is against cannabis. (I don’t want to wade into any other arguments than whether it is a successful health policy intervention.) A recently, well-publicised study showed that cannabis use in under 18s was associated with a reduction in IQ of less than 1 point. Firstly, I would argue that this paper does not demonstrate causal effects. The authors used OLS with the outcome as adult IQ minus childhood IQ, and the independent variables were cannabis and other drug use. Without delving too deeply into the discussion, there are potential standard error and endogeneity issues.  Second, this effect is quite small. But, other studies show that cannabis is also associated with an approximate 40% increase in the risk of psychosis. Prohibiting it may be a good way of intervening to prevent these negative effects. But what are the costs?

A report suggests that the cost of policing cannabis in the UK is about £500m annually. On top of this there is a possible opportunity cost in not legalising it of about £6.4b from lost revenue. So what is the cost-effectiveness of prohibition in this case? The current prevalence of psychotic disorders is about 5 in 1,000 in the UK. If everyone used cannabis then this could increase to 6 in 1,000 in the worst case that would be around 60,000 cases of psychosis which may equate to a loss of 30,000 QALYs (based on a QALY of 0.475).  This incredibly crude calculation yields a cost per QALY of about £250,000/QALY. There may be other benefits, such as modestly increasing the IQ of a small number of people, but that is certainly not going to be enough to justify the costs. And since only a small proportion of the population uses cannabis regularly this value is going to be many times larger, and certainly over £1m/QALY. Clearly it is not cost-effective.

This may be true of other bans as well. An older report by the World Bank in 1994 suggested that the ban in asbestos in the US valued the cost per statistical life saved at about $49m, much more that the standard compensation workers receive for risk, about $5m per statistical life.

I would think that there would be much more effective ways of spending such sums of money in increasing health and safety and protecting the worker or consumer. This could be through taxes or subsidising substitute products or compensating workers better and internalising the externality. This is not an argument for a free market, just a well regulated one that uses cost-effectiveness policies more effectively.


  • Health economics, statistics, and health services research at the University of Warwick. Also like rock climbing and making noise on the guitar.

Join the discussion

This site uses Akismet to reduce spam. Learn how your comment data is processed.