Drugs policy: a role for economic research?

Drugs policy has a long and inglorious history and has often been driven less by reason and evidence than by prejudice and sentiment, moral panic and conviction, and political fear and opportunism

By Steve Pudney

Drug abuse gives rise to intractable and enduring social problems and its policy counterpart, the ‘war on drugs’ continues to be one of the most hotly debated areas of public policy. Economists play a surprisingly minor role in this debate. Illicit drugs are just market commodities, drug dealers and users are just market participants and – if economists know about anything at all – one would expect them to know about the working and regulation of markets. Economics also has a lot to say about the nature of behavioural responses to incentives created by the policy environment (some of which may be unintended) and about the need to compare social costs and benefits when considering policy options. Despite this potential contribution, the economic research literature on illicit drugs is small and the field remains mainly the province of medical and non-economics social science research.

The proper basis of public policy

I would argue that any policy on harmful substances should be based on reasoned argument and sound evidence – and that we should bear in mind the possibility that there may be some social harms that feasible public policy cannot remedy. Economists are used to the idea of policy ineffectiveness, but this idea is absent from much of the non-economic literature on drug abuse. Many commentators appear to believe that the demonstration of harmful effects (of cannabis on mental health, for instance) is sufficient in itself to justify a ‘tough’ drugs policy. Economics teaches us otherwise: that tough policy may be ineffective, may have damaging unintended consequences or may involve a level of cost that outweighs any likely social benefit.

The evidence on illicit drugs

The quality of the available evidence is an obvious concern. For most of the substances regarded as ‘drugs’, possession and market transactions constitute illegal activities and drug users therefore have a strong incentive to conceal their behaviour, making measurement highly problematic.

I classify evidence on illicit drugs into three categories: quasi-facts, pseudo-facts and fantasy-facts. Quasi-facts are possibly prone to serious error but they are at least based on a clear and direct measurement process, usually involving large-scale sample surveys. An example is the regular publication of general-population prevalence rates published in the UN World Drugs Report (Figure 1), which portrays Australia as a world leader in the consumption of cannabis, amphetamines and ecstasy. Such ‘facts’ are undoubtedly prone to varying degrees of under-estimation through under-reporting and to non-comparabilities across countries.

Figure 1: Quasi-facts: general population prevalence rates (UN World Drugs Report 2006, proportion of 15-64 year-olds reporting use in last year ) [view enlarged image]

Pseudo-facts involve some manipulation of directly observed data and further assumptions (or, less charitably, guesswork). An example is my own study of the size of the UK drugs market¹, which combined multiple data sources to generate an overall estimate of £5.27 billion, equivalent roughly to a third of the tobacco market. Despite the wide range of uncertainty associated with this estimate (at least ±£1.3 billion), this figure has been widely reported as accepted fact in the UK press.

Fantasy-facts are another order of magnitude more uncertain, involving more heroic assumptions. For example, the UK Drug Harm Index² combines data from many sources with large assumptions about imponderables like the value of a life cut short by drug-taking (Figure 2). These attempts at measurement are often very sophisticated and may provide a useful guide to the evolution of the drugs problem and the impact of policy – and they are certainly better than nothing – but it is important to avoid treating them as clearly established ‘fact’.

Figure 2: Fantasy-facts: an index of drug harms (Source: MacDonald et al., op. cit.)

History of drugs policy

Drugs policy has a long and inglorious history and has often been driven less by reason and evidence than by prejudice and sentiment, moral panic and conviction, and political fear and opportunism. Jessica Warner’s book Craze: Gin and Debauchery in an Age of Reason (2002) draws a clear parallel between the eighteenth century moral panic about gin and contemporary worries about illegal drugs. In both cases, we see policy strongly affected by public outrage, largely fuelled by the media and often based on little or no evidence. Compare these quotations from the London press 250 years apart. All the elements are the same, only the writing style and the physical substances are different.

This wicked GIN of all Defence bereft
And guilty found of Whoredom, Murder, Theft
Of rank Sedition, Treason, Blasphemy
Should suffer Death, the Judges all agree
London Evening Post, March 1751

Drugs take lives and tear apart communities. They also undermine all our efforts to combat crime. The Government needs to get an urgent grip on this problem.
Daily Mail, August 2006

The gin and tonic drinking Daily Mail reader who deplores cannabis-smoking youth is to some extent the product of random swings in our views about the substances which merit concern, and recent research³ suggests that gin and tonic might in fact be a greater source of health concern than cannabis.

Figure 3: A UK drug surge? (British Crime Survey, self-reported use in last year, 16-59 year-olds)

The strong public pressure on policy makers to solve the drugs problem has generated a good deal of tough talk by politicians, but remarkably little evidence of any strong effect on drug use. For example, in his 1989 inaugural address, George Bush Snr. promised ‘there is much to be done and to be said but, take my word for it: this scourge will stop.’ Well, it did not. If we look at quasi-facts from the annual US Monitoring the Future (MTF) survey, the proportion of 12th grade students who reported using marijuana continued its long standing downward trend that began in 1978-9 to a low point of 22 per cent in 1992, before starting a sustained rise to a peak of 39 per cent in 1997. Throughout the period, the proportion of respondents reporting that marijuana was easy to obtain remained well above 80 per cent. Similarly, the UK government’s reclassification of cannabis from class B to class C status in January 2004 (which weakened the penalties for possession) has apparently had no impact on the downward trend in self-reported cannabis use which began in 2003 (Figure 3), despite the dire warnings of possible consequences which appeared in the press.

The apparent ineffectiveness of much drugs policy does not mean that drug consumption is unaffected by events – just that those events are often not under the control of policy-makers. An excellent example from the US is the dramatic fall in self-reported cocaine use (in the year preceding interview) by 12th grade MTF survey respondents. 1986 witnessed the start of a six-year decline in cocaine use from 13 per cent to three per cent, and an unprecedented 16-point single year increase in the percentage seeing ‘great risk’ in using cocaine just once or twice. Was this the outcome of a radical new policy initiative? No. It resulted largely from a media frenzy surrounding the death of a basketball player, Len Bias, reported (incorrectly, as it turned out) to have resulted from his first use of cocaine. Available data suggest that this is a permanent effect – cocaine use in the US school population has remained below half its previous level.

The gateway hypothesis

The most distinctive part of the economist’s contribution to policy analysis is the analysis of behavioural response to the incentives created by policy. A good example is the hotly-debated ‘gateway hypothesis’, which asserts that the use of cannabis increases the risk of subsequent use of hard drugs, in a direct causal sense. Superficially, the evidence seems to be consistent with this idea. For instance, survey evidence suggests that almost all hard drug users had used cannabis before moving on to harder drugs. However, precedence in time does not necessarily imply causation. (Consider: does the sending of Christmas cards cause the occurrence of Christmas?) We also find that hard drug users on average display higher rates of early family disruptions, psychological impairment, involvement in petty crime and truancy, and so on. It is quite likely that cannabis and hard drug use are frequently joint outcomes of a common underlying set of causal factors, with the earlier onset of cannabis merely a reflection of easier access and lower price.

A disappointing feature of the research literature on the gateway hypothesis is the absence of analysis of policy design. If we could demonstrate the existence of a gateway effect, what would be the implications for policy? To answer this question, we need to understand the behavioural basis of the gateway effect. There are many possibilities, but an interesting one is the supply-side or access gateway. Some people strongly believe in this theory: for example, in October 2007, a primary school teacher called Nicola Cooper, from the Suffolk town of Bury St Edmunds, was convicted of supplying her own children with cannabis. Kevin McCarthy, defending, told St Edmundsbury Magistrates’ Court ‘she had effectively become a drug supplier to keep her children away from other suppliers. The reason for the supply was to keep those cherished children away from the drug culture.’ Mrs Cooper’s belief in the dangers of the access gateway is shared by Dutch policy makers responsible for the tacit legalisation of retail cannabis supply through Amsterdam coffee shops, which can be interpreted as an attempt to segment the drugs market and separate cannabis users from supplies of other drugs.

Although we have no relevant evidence beyond a few fantasy-facts, it is possible to construct a story that illustrates the way incentives might work to produce an access gateway. Comparing the US and UK, the US appears to have a more uniformly tough policy on the supply of all drugs, whereas the UK has a more graduated scale of penalties. In UK courts, the mean sentence for heroin/cocaine trafficking relative to cannabis trafficking is 5.5:1, while in US Federal courts it is only 2.8:1. There is consequently a stronger incentive for UK cannabis dealers to avoid co-supplying hard drugs – which appears to be reflected in a rate of co-supply higher in the US (37 per cent) than the UK (28 per cent). This in turn is reflected in higher proportions of US cannabis users reporting easy access to hard drugs (forty per cent compared with 22 per cent)⁴. These figures are all profoundly uncertain, but they do illustrate the potential value of insights into incentive effects that an economic approach to policy design provides.

The need to use evidence intelligently

Returning to the critical issue of the quality of evidence on drug use, we have to accept that the evidence we have to work with is potentially unreliable, because of the incentive that exists for drug users to conceal their activity. Nevertheless, it would be a mistake to disregard the evidence just because it is imperfect. We need to find ways of using evidence intelligently, making allowance for its imperfections. There is scope for progress here. An example is the case of children’s illicit smoking. The British Household Panel Survey involves annual interviews with a large panel of children aged 11-15. Each year, the children are asked the question: ‘Have you ever tried a cigarette, even if it was only a single puff?’ Examination of the sequence of responses for each child over five consecutive years reveals many inconsistencies, which gives us an insight into the nature and effects of misreporting. A statistical analysis of reporting error suggests that, in a household where the child’s mother smokes, there is a four per cent probability of over-reporting for a child who has not smoked and a 13 per cent probability of under-reporting for a child who has smoked. In a non-smoking household, the over-reporting rate falls to one per cent and the under-reporting rate rises to 22 per cent. In other words, there appears to be a tendency for children to misreport their behaviour in a way that brings them more in line with family norms.

This kind of misreporting can have a big impact on research findings. For example, if we treat the data as completely accurate and analyse the link between parental smoking and children’s smoking, we find the child’s annual risk of starting smoking to be more than doubled by having a mother who smokes (4.4 per cent rather than two per cent). On the other hand, if we allow for the possibility of misreporting by children, we find evidence of a much larger risk, which is little affected by parental smoking (8.4 per cent rather than seven per cent)⁵. Although we cannot trust our data completely, this doesn’t mean we can’t find effective ways of using it.

1 Pudney, Badillo, Bryan, Burton, Conti and Iacovou, 2006, Estimating the Size of the UK Illicit Drugs Market, in Measuring Different Aspects of Problem Drug Use: Methodological Developments, Home Office Online Report 16/06, pp. 46-120
2 MacDonald, Tinsley, Collingwood, Jamieson and Pudney, 2005, Measuring the Harm from Illegal Drugs using the Drug Harm Index, London: Home Office Online Report 24/05
3 Nutt, King, Saulsbury and Blakemore, 2007, Development of a rational scale to assess the harm of drugs of potential misuse, The Lancet
4 See Pudney ‘Cannabis and hard drugs: is there a supply-side gateway effect?’, Working Paper, ISER
5 Figures adapted from results presented in Pudney, 2007, ‘Rarely pure and never simple: extracting the truth from self-reported data on substance use’. Institute for Fiscal Studies: CeMMaP Working Paper CWP11/07

Top ^

A condensed version of the 20th Richard Downing Lecture delivered at the University of Melbourne on 6 December 2007.

Professor Steve Pudney is Director of the Research Centre on Micro-social Change at the Institute for Social & Economic Research, University of Essex.