Report prepared for the New Zealand Law Commission. Filed 30 June, 2009. Executive Summary Keywords: Health; : Health;1. The Policy Framework It is assumed that any review of the Sale of Liquor Act 1989 will continue the policy framework on which it was based. Previously, the implicit assumption was that almost all alcohol consumption was harmful, but there were some exceptions. Thus it was necessary to control all aspects of liquor supply as far as was practical. The Sale of Liquor Act was based on a quite different premise. It took the view that most consumption of liquor was benign or even beneficial. However some was harmful – even extremely harmful in comparison to most products. Its approach was that rather than control everything to do with liquor production, supply and consumption, the aim was to target consumption which was markedly harmful. It involved a marked liberalisation of the supply of liquor, moving from ‘quantity licensing’ (the number of outlets) to quality licencing (anyone could set up an outlet, providing they met certain quality standards). At the same time there were associated measures to reduce certain kinds of harm, including more vigorous pursuit of drink-driving. In my opinion, the Sale of Liquor Act was perhaps the most successful social reform of its times. It vastly improved access to liquor for moderate drinkers, transforming and enlivening inner cities with a plethora of small bars and restaurants (with a resulting lift in the standards of cuisine). At the same time there is no evidence that harm rose – indeed the downward trend of absolute consumption per adult continued across the reform period. A minimalist assumption is that the reform (which included other measures such as tougher drink-driver enforcement) had no effect on the level of harm from alcohol consumption. In recent years there has been some evidence of rising harm in some areas including – increasing teenage drinking. This is an international phenomenon, so it cannot be attributed to the New Zealand Sale of Liquor Act; – increased evidence of binge drinking, especially in the 17 to 27 age range (although it is less clear whether this is a rising phenomenon or whether it is just being better identified); – new forms of alcohol (including light spirits and ready-to drinks) which may, or may not, require specific policies; – evidence of the magnitude of the social harm from alcohol – discussed below – which is substantial. These are social problems which need to be addressed. The assumption in this paper is they should be addressed by targeted interventions in a spirit similar to the 1989 regime reform, rather than a return to the pre-1989 regime of widespread control of supply and consumption. Even so, there needs to be a review of the 1989 Sale of Liquor Act and subsequent amendments to see whether its performance can be improved. The concern here would be to ensure that the supply system is working with the maximum efficiency (the minimum of use of resources) while contributing – as far as it is able –to reducing harm. Because the Sale of Liquor Act is a part of the totality of harm-addressing programs a review of the act necessarily involves looking at the efficacy of those other programs. 2. The Harm From Alcohol Since the 1989 Sale of Liquor Act there have been two major attempts to provide estimates of the social costs of alcohol misuse (Easton 1995, BERL 2009). Their exact estimates may be challenged but they each bring together the existing available data on social harm and value them. While the valuations may be challenged, the underlying data has not thus far been criticised by the challengers. What it shows – whatever the valuations – is that there is substantial harm caused by the misuse of alcohol. This harm is so large relative to the overall consumption that it makes policy sense to address it by specific measures. (It is acknowledged that ordinary goods and services can cause also harm. Except in particular cases, there is generic legislation – perhaps with accompanying specific regulation – to minimise this harm. In the case of alcohol consumption – and a few other products – the unit harm can be so large that a specific policy framework is thought necessary.) The harm can be divided into three components. First, there is the additional costs to the public purse. Notable elements are costs to the public health system, to the police and justice system, and to the regulatory system. Second, there are the material (or ‘tangible’) costs which are borne by the private sector. For instance, car insurance costs are higher as the result of accidents from drunken driving. Third, there are the human (or ‘intangible’) costs covering the loss of quality of life (not only from poorer health but from the violence to others) and early mortality. (Note that the resources used to produce the beverage are not a ‘harm’, in this formal sense, since they are netted off against the benefit to the imbiber.) A major issue is the degree to which these costs are ‘internalised’, that is, taken into account by the person who purchases or imbibes. The government costs are certainly not taken into account, and it seems likely that neither are the tangible and intangible costs which fall upon others (including to the insurance industry). These external costs are a major justification for excise duty on alcohol, since insofar as the drinker fails to take them directly into account when they make the consumption decision. If they do not, the benefit of the consumption to the drinker is greater than the benefit to society as a whole; indeed the benefit to society as a whole may be negative if the ignored costs exceed the benefits to the drinker. 3. Rational and Irrational Drinkers What costs are included in a drinker’s decision is an empirical matter. For instance, the drinker may think of others as when a man choses to limit his intake because his wife does not like him drunk. At another extreme the drinker may not take into consideration all the costs which fall on her or himself. What is taken into consideration is a heavily disputed issue in economics (although other disciplines seem to accept as a matter of common sense that drinkers do not take into consideration all social costs). The default position of economists is to assume ‘rational economic man’ (Homo economicus) who takes into consideration all the costs of a consumption which impact on him (invariably he is treated as a man), but none of the costs which impact on others. A variation is that the man may not have full information (or the required computational power) and is ‘boundedly rational’. (Most economists would have little difficultly accepting that certain groups of people – say the young – may be so information limited, that the concept does not usefully apply to them.) Some economists treat the notion of rational economic man as a useful analytic device for want of a better hypothesis; others treat the notion as a fundamental economic assumption which may not be challenged (at least in its bounded form). The second group would argue that, say, drinkers take into consideration all the effects of the drinking on them (but usually not on others) including assigning a probability of serious damage or death, but with diminishing significance because future events are discounted. The first group of economists might accept that an inebriated person may miscalculate the probabilities, underestimate the harm and too heavily discount the future (even ignoring it). The difference between economics leads to a major disagreement as to what is or is not included in social costs. This is not a dispute about the facts but one of which harm is or is not to be included. It hinges on how rigorously we should take the notion of rational economic man. There is hardly any direct evidence, other than from introspection, that rational economic man is a realistic account of how humans make decisions. (Drunks are hardly notorious for this sort of introspection.) It does, however, make some predictions which are consistent with observation (such as about how people make normal purchases). That is why many economists treat the notion as a working hypothesis, acknowledging that other hypotheses about economic behaviour may give similar predictions in the same circumstances, but different ones in less standard ones (or which drinking might be a prime example). That a theory predicts well in some circumstances does not mean it will predict well in all circumstances; inebriation seems to be a circumstance where it does not (and one with potentially substantial differences in outcomes.) In recent years an alternative framework has begun to evolve around ‘behavioural economics’. It would be misleading to argue that it has been adopted by all the economics profession, but it was important in justifying to some economists the ‘Kiwi saver’ scheme. Depending on the definition of the subject it has already received two prizes in economics in honour of Alfred Nobel (Daniel Kahneman: 2002, Vernon Smith: 2003) and another leading innovator, Matthew Rabin, is expected to win one in due course following being awarded the 2001 John Bates Clark Medal – the second most prestigious prize in economics. (Another key thinker is University of Chicago professor, Richard Thaler.) Behavioural economics is characterised by close attention to psychology’s research and theories (Kahneman is a psychologist). It takes much less an a priori approach than that upon which rational economic man is based. The particular element of the approach relevant to this report is ‘time inconsistent’ decision making, which occurs when without any new information a person may regret a decision which earlier had been made rationally. A simple example is when someone plans to have a couple of drinks in a bar but, in the event, over imbibes and yet the following day regrets the decision. Each decision is taken rationally and yet collectively there is an inconsistency between the three ‘rational decisions’. The time inconsistency arises because the discounting of decisions through time differs from that which is assumed for rational economic man. It turns out that time inconsistent drinkers will welcome a tax on their consumption since it limits the excessive drinking which subsequently they will regret (see Rabin and Ted O’Donoghue on Optimal Sin Taxes). That means some consumption is not (subsequently) valued by the consumer, and therefore is an externality and a contributor to social costs. 4. The Case For Taxing Alcohol It is clear from the evidence that alcohol consumption causes considerable harm which is not always taken into consideration when individuals make decisions to imbibe. As a consequence public policy has introduced a range of interventions which aim, one way or another, to internalise the decision, so that the drinker takes into consideration more of the harm which the drinking causes. In practice it has not been possible to eliminate all the social harm by education, private arrangements and statutory and regulatory interventions. Drinking behaviour is too complicated; even heavy interventions, which impact onerously on moderate drinkers, are likely to be circumvented to some extent – as the prohibition era illustrates. It has therefore been a standard practice to use specific taxes on alcohol to attempt to deal with the remaining social harm. There are two channels by which this may work – modification of drinking patterns and compensation for social harm; one leads to an efficiency gain, the other to an equity gain. They are mediated by the distributional impact of taxes. Modification of Drinking Patterns It is generally assumed that the demand for alcohol is largely price inelastic, that is a hike in the price arising from, say, higher taxation on alcohol does not reduce the demand for alcohol greatly. However careful econometric studies shows higher prices do reduce some consumption. It is believed (on the basis of fragmentary empirical evidence – it being hard to obtain systematic evidence on individual price responsiveness to price changes – and on a priori grounds) that the main groups whose consumption is sensitive to changes in prices are – the young; – binge drinkers (insofar it will reduce the number of purchases during a bing); – heavy drinkers. Since each of these represent a known source of social harm, any reduction in the quantities they drink will reduce social harm to some extent. Thus the tax will increase the efficiency of the system. It is to be noted that the Rabin-O’Donoghue paper suggests that where there is time inconsistent decision-making the tax will make such drinkers judge themselves retrospectively better off in the long run, and that they will welcome it from this perspective. However, strictly the Rabin-O’Donoghue paper assumes that the additional tax revenue from the drinker is recycled back to the drinker (say in a lump sum income grant). In practice this does not happen; the recycling is not that neutral. This analysis also assumes that the impact on the quantities drunk by moderate (time consistent) drinkers is zero. Yet they pay more tax. Insofar as it is not recycled back to them, they may be worse off in real income terms (although they may be better off from lower social harm). The Distributional Impact of Taxes on Alcohol The previous two paragraphs show that a second role of specific taxation on alcohol is that it raises revenue. What exactly happens to this revenue depends upon fiscal policy, but since it is on the margin of fiscal revenue, we can simplify analysis with little loss of generality by assuming that the additional revenue is used to reduce income tax and/or to increase social transfers. (This assumes that the degree of other interventions, and hence the resources required for them, are set independently of available tax revenues.) It is unlikely that the income increases can be targeted on drinkers or on those who suffer social harm from drinking, and in any case the government has considerable discretion as to whom it recycles the additional tax revenue. Thus it is not possible to state with any precision the exact winners and losers of a specific tax on alcohol, except non-drinkers will be beneficiaries since irrespective of whether they receive any of the recycled tax revenue, they will pay none of it, and they will (probably) benefit from reductions in personal harm. By extension, low alcohol consumers will also benefit, but the threshold below which they benefit cannot be predicted a priori. It is also likely that despite some gains from reductions in social harm heavy drinkers may be worse off after they pay the higher alcohol specific taxes. More generally, since there will be overall reductions in social costs from higher specific taxes, the aggregate gains (including the reductions in social costs) of the winners will be greater than the losses of the losers. That is, there will be a net social gain from this efficiency gain. However it will not be distributed evenly through the community. Compensating for Harm The previous subsection on the distributional impacts of a specific tax (or an increase in the specific tax) on alcohol treats the status quo as the reference point. That perspective ignores that the situation is one where there is social harm from the drinking, some of which is impacting on non-drinkers (and socially responsible drinkers). This includes the fiscal costs of the interventions and the private costs which cannot be prevented by the interventions. It is surely unjust when non-drinkers and moderate drinkers pay for the harm of other drinkers. It follows that it could be argued there is justice in specific alcohol tax transferring some of the burden of these costs from the non-drinkers and moderate drinkers to the drinkers who are generating harm. Such compensation improves the equity of the system. 5. The Taxation Implications for New Zealand There is a tendency to argue that a hike in taxation is necessarily a good thing – perhaps using arguments analogous to those in the previous section to justify the policy. Practically policy needs to be more refined in order to determine the way alcohol is to be taxed, and the level at which it is taxed. But first the price elasticity analysis of the previous section needs to be extended. Each alcohol-containing drink involves at least two characteristics. One is the amount of absolute alcohol in the drink, the other – which actually summarises a whole lot of characteristics – we shall call ‘quality’. Absolute alcohol can be measured directly while the quality characteristic(s) is usually evaluated by the price a consumer is willing to pay. Thus two bottles of wine may have the same quantity of absolute alcohol but the higher priced one is judged to be of better quality. (More subtly we pay more for the same bottle of wine in a restaurant than a supermarket because we are purchasing ‘quality’ as a superior venue.) Now suppose the price of all alcohol goes up (proportionally to avoid the complexities of measuring the quality, as distinct from ranking it as applied in the previous paragraph). The purchaser-consumer is likely to adjust their consumption in response to that price increase. At the extremes they may reduce the consumption of absolute alcohol and maintain the quality, or they may maintain the quality and reduce the consumption of absolute alcohol. Only the second reduces harm. One of the reasons why the demand for alcohol is price inelastic (insensitive to a rise in price). Is that when the price goes up many people have the option of maintaining their absolute alcohol for the same cost, by reducing the quality of what they drink. (Because beverages come in fixed volume containers it is likely that in many situations – especially moderate – drinkers cannot easily adjust their absolute alcohol consumption and the adjustment is entirely via quality. This suggests more attention should be paid to container size in alcohol control policy.) There is however one group of drinkers who do not have the option of reducing the quality of their consumption when there is a price hike. There may be no cheaper form of absolute alcohol. In effect they are drinking entirely for the characteristic of absolute alcohol disregarding the quality characteristic. This group includes alcoholics but it also includes many immature drinkers, binge drinkers and heavy drinkers. (This is the reason why these groups are likely to be more price sensitive.) Some of the most harmful drinking occurs in these circumstances. The conclusion which follows from this analysis is that insofar as the aim is to use taxation to regulate absolute alcohol consumption with the objective of reducing harm, most attention should be paid to the price of absolute alcohol levels, particularly where they are cheapest. The New Zealand alcohol taxation regime is particularly suitable for this purpose since it is levied on the basis of absolute alcohol content (details are below). Thus it raises the price on the cheapest alcohol (measured in absolute alcohol terms), where drinkers cannot avoid the price increase, relative to the more expensive forms where drinkers can avoid the increase by reducing the quality of their purchases. Apparently the regime was introduce in the 1980s, because it was thought that absolute alcohol was the best indicator of social harm and so taxing it was the most appropriate way of implementing the compensation principle. It seems unlikely that the efficiency gain from targeting the cheapest drinks was fully appreciated. Although one can think of more complicated regimes which might have some advantages (such as a duty which is reduced for higher quality drinks so that for the same overall revenue the impact on the cheapest was even greater) they would not be easy to implement. It seems likely that in current circumstances the best duty is in proportion to the absolute alcohol content. What should be the level of the duty? There is a general acceptance that the aggregate revenue from the excise duty should at least cover the fiscal costs addressing alcohol harm. The compensation principle suggests it might also cover the social cost to the non-drinkers. However recall that some measures of social cost include the costs to themselves which drinkers do not take into consideration when making the purchase decision. Thus it could be inappropriate to equate the full revenue recovery to the total social costs of alcohol, even if there were an universally agreed measure. In practice it is likely that full cost recovery would not be complete and that some harm would not be compensated – even crudely – through the tax system. An alternative approach arises from the focus on modifying drinking patterns. In this case the excise duty level would be chosen to target some minimum price of absolute alcohol. This was discussed in Taxing Harm (Easton 2002) and it and subsequent developments illustrate the principle. At that time, seven years ago, there was a product, light spirits, a bottle of which contained 21 standard drinks which could be purchased for a price equivalent to the minimum hourly wage. The contents were enough to kill a person – as a couple of coroners found. The cheapness of the bottle was partly because the absolute alcohol came from very cheap industrial sources, but it also reflected anomalies in the tax system. When the anomalies were eliminated in May 2003, the price of a bottle of light spirits doubled and the product disappeared from the shelves. Taxing Harm　recommended that the excise duty on alcohol should be set to target some minimum price of absolute alcohol. This would mean that while competition would cause producers to seek to minimise price by minimising the resources they use, any efficiency gains at the bottom end would accrue to the public revenue. An alternative proposal is to set a minimum price for absolute alcohol but how that would be implemented is not explained, nor how it would avoid generating substantial profits for the alcohol production industry. They might be used for, among other things, promoting alcohol consumption. 　recommended that the excise duty on alcohol should be set to target some minimum price of absolute alcohol. This would mean that while competition would cause producers to seek to minimise price by minimising the resources they use, any efficiency gains at the bottom end would accrue to the public revenue. An alternative proposal is to set a minimum price for absolute alcohol but how that would be implemented is not explained, nor how it would avoid generating substantial profits for the alcohol production industry. They might be used for, among other things, promoting alcohol consumption.In summary the two gains from a specific tax on alcohol – the efficiency gain from moderating harmful drinking, and the equity gain from compensating for the harm – give slightly different recommendations for the level of an excise duty on absolute alcohol. It seems likely however that, whether the purpose is to ensure the minimum price of absolute alcohol discourages harmful drinking or whether the purpose is to adequately compensate for the harm, that the current excise duty is too low. One complication is that the excise duty on absolute alcohol where the beverage is above 14 percent proof is 82 percent higher than the duty for alcohol below 14 percent. (Wine, typically does not exceed 14 percent, so the higher rate is on fortified wines, liqueurs and spirits.) It should also be noted that in practice wine is levied as if its proof is 10 percent, although it is more typically near 14 percent which represents a 30 percent discount relative to the levy on the absolute alcohol in beer. The reasons for these differences seem to arise from history and political compromise. It is sometimes argued that some forms of alcohol are less harmful than others. While that may be true, the differences are not quantified and it seems unlikely that they are as great as the excise differentials. The harm difference case for excise duty differentials needs to be quantified before it can be incorporated systematically into policy. The only other justification I can think of (other than the fortunes of history and political compromise) is that the excise duty differentials may result in a rough equality in the minimum purchase price of absolute alcohol. Some evidence to this effect was found when collecting data for Taxing Harm. (If there is little substitution between types of beverages, there is little efficiency loss from this approach. But if drinkers switch between drinks according to price. They might, for instance, choose wine because of lower prices due to lower excise duty when they could get the same value from spirits for a cheaper resource cost to the economy.) There is a case for a zero excise duty rate on alcoholic drinks of very low strength, which do not cause harm. Tax rates need always to be set with the possibility of avoidance. This includes home brewing and similar activities for other beverage forms. It also includes duty-frees. Any review of the tax regime on alcohol must take account that New Zealand has various international obligations which affect its policy freedom. (Many would argue that it should vigorously pursue international reform to minimise the harm to which duty-frees contribute.) Another consideration is that much alcohol consumption is not paid for by the consumer – as when the employer pays for it. This is not only another reason for the price inelasticity of absolute alcohol, but indicates the need for other policies to reduce harm. The host responsibility program is an example. 　 6. Conclusion In summary – on the whole much alcohol consumption is benign or even socially beneficial, but some generates very great social harm; – this harm may be reduced by various interventions, but their effectiveness is limited because of the need to allow consumption which is benign and socially beneficial; – a specific tax on alcohol is a means of reducing the harm through internalisation of an external cost ( (the efficiency gain), and compensating those who suffer harm from others’ drinking (the equity gain). However neither objective can be precisely attained. – the New Zealand system on an excise duty on absolute alcohol has much to commend it. However more attention could be given to the minimum purchase price of absolute alcohol. Go to top