Market-failure reasons for policy intervention suggest various policy options.

1. Information-related policies: The public sector may need to intervene to provide information about negative consequences of consuming alcohol and to restrict advertising that falsely presents optimistic consequences of consumption.

Accurate information on negative aspects of consumption includes information on self-control issues and on genetic information that might suggest future dependency problems. Goldstein (2001: 113) argues that ‘…children of alcoholics should be advised to never touch alcohol; certainly they should be taught the special hazards that alcohol holds for them — hazards not shared by their peers’.

A difficulty stems from disagreement over precise health consequences of consuming alcohol among those without genetic predispositions to alcoholism. The debate over possible health benefits from reducing heart disease is an instance. Health benefits claimed to arise from consuming alcohol appear to be a spurious consequence of including among non-drinkers those who have ceased drinking because of poor health (Fillmore 2006). Apart from providing known information there is a case for investing in improved information.

As mentioned, the case for health-risk warnings is weakened if people already exaggerate risks. If particular at-risk groups or particular health concerns arise, these should be targeted rather than providing general health warnings to groups who, on average, assess risks adequately. Information targeting youth should address risks of consumption, while information targeting problem drinkers needs to address risks, problem-recognition, denial and dependency treatment.

2. Self-control policies: Problem drinkers are a sizeable sub-population whose excessive drinking is hazardous to health.

For many heavy drinkers, alcohol is not an ordinary consumer good whose consumption can be analysed using static, rational choice models. People become addicted to alcohol in an unconscious process that eventually takes over their lives. Unless consumers set out initially to become dependent, such heavy drinkers are not ‘rational addicts’ (Becker and Murphy 1988). Heavy drinkers typically start drinking as adolescents when excessive consumption is linked to incomplete development of brain regions involved with executive control and motivation. The vulnerability to alcoholism is greatest among individuals who start consuming early in life but those with difficulties making rational choices at any age tend to be alcohol abusers. Abusers have a higher incidence of mental disorders (depression, anxiety, ADHD and schizophrenia) than the general population and apparently use alcohol and other drugs to self-medicate (Koob and Moal 2006).

Genetic influences on propensities to consume alcohol may underlie personality traits such as impulsivity, risk-taking and stress responsivity that drive excessive consumption. Family and twin epidemiological studies suggest heritability of vulnerability to addictive diseases of 30–60 per cent (Kreek et al. 2005).

Like other addictive drugs (opioids, stimulants, nicotine, marijuana) and natural rewards (food, sex, water) alcohol produces euphoria by activating pleasure centres in the brain. Like other drugs, alcohol releases dopamine in the brain where pleasure centres that have evolved to ensure survival get ‘hijacked’. The euphoria induced by alcohol, particularly if enhanced by a genetic predisposition, encourages repeated use. Over time, alcohol disrupts brain reward circuits and can produce withdrawal and craving if consumption ceases. Such negative reinforcements alternate with positive reinforcement to drive a cycle of addiction that becomes etched into brain structures. This etching reinforces pursuit of alcohol consumption, as a surrogate for survival-related behaviour, by dominating attention and decision-making.

Problem drinkers come to drink too much and face difficulties limiting consumption during particular episodes, with the first drink leading to uncontrolled drinking. Moreover, these compulsions can be long-term. Environmental cues associated with alcohol (people, places, advertisements) can trigger intense cravings among those addicted which cause relapse into use even after protracted abstinence.

Consumers in this cue-driven environment make consumption decisions with limited rationality and foresight. Policies for improving self-control include helping to demonstrate that alcoholism can be a consequence of recurrent drinking and desensitising and limiting exposure to cues. There can also be attempts to improve self-control by promoting ‘personal rules’ or heuristics to control behaviour. Since problems with alcohol consumption primarily stem from excessive consumption, ‘personal rules’ relating to the number of ‘standard drinks’ consumed per week, numbers of alcohol-free days or, in some cases, pursuit of total abstinence are useful information policies.

Self-control can also be improved by supply restrictions such as limiting the availability of alcohol outlets, trading hours and promotion of advertising that might trigger cues to drink. Unfortunately, such policies increase ‘user costs’ of consumption that impact on all consumers.

Supply-control measures on alcohol outlets and on opening hours will limit external costs of alcohol consumption (Donnelly et al. 2006) and, by limiting cues, promote individual abilities to control drinking. By keeping away from cues and knowing that, beyond a certain time, drinking is impossible, people with self-control problems can be helped to formulate personal rules which limit drinking.

Restrictions on advertising help curb drinking among abstaining problem drinkers subject to cues. Indeed, as with smoking, there is a case for limiting cues in the media that elicit consumption.

For age and ethnic groups prone to excessive consumption because of self-control problems, there can be a case for consumption bans. Minimum-age laws and restrictions of alcohol availability to particular ethnicities, such as Australian aborigines, are controversial policies. Anderson and Wild (2007), however, supported such interventions resulting in the Northern Territory Emergency Response by the Howard Government in 2007. Aboriginal people are often alcohol-abstinent; but, those who do drink, often consume more than 13 standard drinks per episode (ABS 1994) and are therefore an ‘at-risk’ group.

With respect to drink-driving issues, the installation of ignition interlock devices can improve the judgment and self-control of drinkers in their ability to safely drive after consuming alcohol (ICADTS 2002).

Finally, treatment options emerge as a way of addressing excessive use. A substantial literature shows that local GPs can be effective in providing early interventions to caution people concerning alcohol problems (Johansson 2002). Promoting a variety of behavioural and cognitive therapies makes sense, as do traditional routes to control such as Alcoholics Anonymous. Given the social costs of alcoholism, there is a case for subsidising such services on the basis of their cost-effectiveness.

For problem drinkers, treatment options include pharmacotherapies using drugs such as naltrexone (Volpicelli et al. 1992). Naltrexone reduces recurrent daily drinking, diminishes alcohol-induced cravings and even reduces cravings after a ‘priming’ drink. It therefore limits the tendencies to lose control. The drug acomprosate suppresses cravings and relapse problems and can be used with naltrexone (Rösner 2008). With external costs, there is again a case for subsidising pharmacotherapies.

3. Externality policies: Economists focus on externalities as the primary source of social costs in relation to substance abuse. The standard prescription is to levy a tax internalising the external costs.

Externalities such as drink-driving-induced accidents cannot be dealt with by a consumption tax since this externality stems from the combined activities of drinking and driving. Moreover, the probability of an accident, given a certain level of alcohol consumption, falls with age (Phelps 1997: 516). A driver aged 20+ who has consumed six standard drinks and then drives has 12-times the chance of a fatal crash than a similarly-aged sober driver. For a driver aged 16–19, the risk increases to 100-times that of a similarly-aged driver. Inevitably, highway patrols and booze buses must detect drink-driving, with the driver then being penalised, rather than employing taxes on consumption. It makes sense to impose stringent restrictions on alcohol consumption by young drivers.

It is only if certain alcohol-related externalities are related to overall population alcohol consumption — the ‘population health’ view (Young 1998) — that simple uniform tax policies make sense.

In Figure 1 the market demand for alcohol (q litres) is illustrated with the marginal production cost c(q). The area under the demand curve measures private benefits from drinking. Social marginal costs of consumption are also illustrated. These comprise the private marginal costs borne by consumers and the external social costs generated, including health and traffic accident costs borne by the community as well as dollar costs of violence and anti-social behaviour to others. As drawn, there are no external benefits associated with alcohol consumption at low consumption levels and external marginal costs are low. As market consumption increases, external costs rise at an increasing rate. Without taxes, consumers operate where private marginal benefits equal private marginal costs — they consume q₂ and pay price p₂. Because social costs exceed marginal benefits over the range q₁ to q_2, this consumption imposes net social costs or deadweight losses (DWLs) equal to area ABC.^[4]

The standard economic prescription to remove these costs is to levy a Pigovian tax t which, assuming competition, raises the alcohol price to p_1, leaving consumption where marginal private benefits equal marginal social costs at q₁.

Determining this tax requires assessing the scale of the unpaid-for social costs, how these are linked to consumption and the elasticity of demand for alcohol. The more responsive demand is, the smaller the tax can be. Selvanathan et al. (2004) provide elasticity estimates of –0.3, –0.4 and –1.3 for Australian beer, wine and spirit consumption, respectively. For alcoholic beverages as a whole, the estimated elasticity is –0.6, suggesting that a 10 per cent tax induces 6 per cent less consumption.^[5]

If social costs are related to the alcohol content of drinks, this tax should reflect this by being a volumetric tax related to the alcohol content of particular types of alcoholic beverages, not an ad valorem excise levied on the value of the product sold. Alcohol products in Australia are taxed approximately volumetrically, although spirits are subject to twice the charge on alcohol content than beer, and wine is subject to ad valorem duty (Commonwealth of Australia 2005: chapter 5). In 2004–05, $5.1 billion was collected as excise and customs duty (AIHW 2007). For years where comparable data is available, these revenues greatly exceed gross health and even costs attributable to traffic accidents (Collins & Lapsley 2002: 62–65).

This economic approach to taxation does not focus on the gross costs of consumption, GC — the total medical and other costs attributable to alcohol. In the figure, these are the area 0q₂BE. Nor does it focus on non-internalised net costs less benefits (NC) of alcohol — the medical and other costs not borne by alcohol consumers less consumption benefits, given by area ECB. Instead, the economic approach recognises that alcohol consumption yields benefits to consumers, given by the area under the demand curve. Thus a tax-inclusive price is sought that maximises the net social advantage — benefits derived less all costs, whether internalised or not.

Indeed DWLs << NC << GC where << means ‘is much less than’. Thus setting tax t eliminates society’s deadweight losses and reduces, but does not eliminate, net and gross costs of consumption.

Note that optimal tax revenues, tq₁, exceed non-internalised social costs EAF because social marginal costs increase strongly as consumption levels increase. Therefore, evidence showing that alcohol consumers pay more taxes than are imposed in external costs does not support the conclusion that taxes are ‘too high’ as claimed by, for example, Heien (1995) using ‘reparationist logic’. It is only if non-internalised, marginal external costs were constant, so private and social marginal cost curves are parallel, that tax revenues should coincide with unpaid external costs to achieve social optimality. But social damages are, most plausibly, strongly increasing in alcohol consumption. Thus taxes ideally should exceed damages, rather than merely compensating society for damages done.

Empirically the social costs of alcohol consumption must be computed net of external benefits. People who die from alcohol consumption create a social benefit in terms of reduced costs of aged care and reduced costs of treating other diseases that would have impacted on mortality if alcohol consumption was curtailed. If there are external benefits from alcohol’s ability to catalyse social interactions which help create social capital, these too should be netted out.

There are several reservations concerning this standard tax prescription.

(i) The approach optimises the overall social advantage but has distinct effects on different community groups. Taxes levied are inevitably uniform — they do not vary by consumer. If a minority of alcohol consumers has very inelastic demands because of self-control issues then their consumption would need to be constrained by high taxes if it is to be curtailed. Such taxes would impose heavy DWLs on those consuming at moderate levels, who impose low social costs. Some 74 per cent of Australians aged 14 years and above consumed alcohol in quantities considered ‘low risk’ to health but paid hefty alcohol taxes (AIHW 2005). Setting lower taxes, however, to cater for those imposing low social costs, leaves the behaviour of those with inelastic demands who impose large social costs largely unchanged while significantly reducing their real incomes. If heavy consumers have low incomes, this raises equity issues. While gainers from efficient taxes can, in principle, compensate losers, this compensation is difficult to engineer in practice.

Such distributional issues constrain policy. Indeed, some argue that ‘a tax on alcohol would reduce consumption indiscriminately…and therefore reduce the satisfaction experienced by millions of sensible drinkers without necessarily reducing the harm caused by a few excessive drinkers’ (Littlefield 1986: 274).

High taxes may encourage the decision to quit entirely but, given an alcoholic’s compulsion to not stop at one drink, will not markedly reduce the intensity of drinking during particular consumption episodes. In their favour, high taxes have major effects in restricting alcohol consumption among adolescents (Saffer and Dave 2003).

The question of whether low levels of drinking are harmful — or even beneficial — is controversial. In assessing workforce costs, Pidd et al. (2006) used self-reported measures of alcohol-related absenteeism to estimate 2.7 million workdays were lost due to alcohol use in 2001, costing $437 million. Self-reported measures of illness or injury absenteeism to determine absenteeism attributable to alcohol use resulted in an estimate of 7.4 million workdays lost, costing $1.2 billion. Low-risk drinkers and infrequent or occasional risky and high-risk drinkers accounted for 49–66 per cent of this absenteeism. This estimated cost of alcohol-related absenteeism is far greater than previously reported and the high incidence of costs by low-risk drinkers and those who infrequently drink heavily was unexpected.

One approach to addressing concerns that external costs are concentrated among a narrow group of high-consumption users is to levy moderate taxes but to penalise intensively socially costly actions associated with high alcohol consumption. Drink-driving and alcohol-related assault should be intensively penalised as activities in themselves. Thus, the intoxicated person who walks home from the pub and ‘sleeps it off’ is not taxed prohibitively; but if the same person drives home or assaults a family member, then stringent penalties obtain. Alcohol taxes in this event capture non-extreme social costs while particular laws seek to capture externality costs.

(ii) Using volumetric taxes to penalise low-value, high-alcohol products creates incentives for unfavourable substitutions towards other intoxicants. A widespread intoxicant in Aboriginal communities is sniffed petrol, so there are incentives to shift towards this if low-cost high-alcohol beverages become more expensive. This is a ‘second-best’ constraint on alcohol taxes that reflects the existence of untaxed substitute intoxicants.

Whether alcohol taxes will induce substitution towards alternative drugs depends on whether alcohol and the drugs are substitutes — Cameron et al. (2001) suggest cannabis and alcohol are substitutes — or complements. Williams et al. (2004) suggest alcohol and cannabis are complements, so increasing the price of alcohol will reduce cannabis consumption.

Attempts to tax alcohol-induced social costs efficiently require that close-substitute intoxicants also be taxed or regulated efficiently. Policy design must account for such cross-market interactions.

(iii) Measures of consumer gain from alcohol consumption given by the area under a demand curve are exaggerated if there are self-control problems. These can be dealt with by adjusting downward benefit measures to account for compulsive consumption: the technique is used for gambling addictions (Productivity Commission 1999).

Finally, apart from implementation difficulties some have argued that there is a case for hypothecating a proportion of alcohol taxes to targeted alcohol treatment and information programs (RACP 2005). The motivation is that such revenues reflect social costs that alcohol creates. This is not true with the social costs we have envisaged, which increase more than proportionately with total consumption so optimal tax revenues exceed non-internalised costs. Hypothecation is unsound anyway since desired investment depends on program effectiveness, not only the damages alcohol inflicts. To take an extreme case, if such programs had low effectiveness then little should be allocated even if social costs are large. The function of taxes on alcohol is primarily to signal the social costs consumers are imposing at the margin. Their revenue-yielding function is secondary.