Tsunami Or Ripple

Presented at the 30th Annual Alcohol Epidemiology Symposium of the Kettil Bruun Society for Social and Epidemiological Research on Alcohol in Helsinki, Finland, 31 May – 4 June, 2004

Tsunami or ripple? Studying the effects of current Nordic alcohol policy changes

Robin Room

Centre for Social Research on Alcohol and Drugs (SoRAD)

Stockholm University

Abstract

Under pressure from the single-market rules of the European Union, Nordic countries have embarked on one of the biggest changes in policy on alcohol availability in 50 years. The changes include a 45% reduction in Denmark and a 44% reduction in Finland in spirits taxes, along with reductions in taxes for other alcoholic beverages in Finland, and substantial increases in cross-border traveler’s allowances borders within the EU, which encourages cross-border purchases in low-tax countries such as Germany, Estonia and now Denmark. These changes are expected to have effects not only in Denmark and Finland, but also in southern Sweden, given its easy access to Denmark. On the other hand, northern Sweden is expected to be little affected, and thus can serve as a control site for studying the results of the changes. The paper described the design and research questions of a collaborative Nordic study to measure the short- and medium-term effects of the changes, both on drinking amounts and patterns and on alcohol-related problems. The design includes annual reinterviews of panels of respondents to examine the question: Whose drinking is affected by such changes, and by how much and under what circumstances? In particular, how is the behaviour of heavier drinkers affected? Hypotheses from three theories of patterns of change in drinking will also be tested: (1) Norström’s application of the gravity model of decrease in effects with the square of the increase in distance from the border; (2) Skog’s theory of the collectivity of drinking cultures; and (3) some predictions from the Becker theory of rational addiction.

Key words: alcohol availability, traveler’s allowance, alcohol taxes, gravity model, rational addiction, collectivity of drinking

Tsunami or ripple? Studying the effects of current Nordic alcohol policy changes[1]

Robin Room

Centre for Social Research on Alcohol and Drugs (SoRAD)

Stockholm University

General background

It is well recognized that the availability of alcohol substantially affects levels and patterns of drinking, and rates of alcohol-related problems (Edwards et al., 1994; Babor et al., 2003). A major dimension of alcohol availability is its price, which can be substantially affected by the level of taxation (Österberg, 2001), and a substantial economic literature considers the effects of taxes on alcohol consumption and problem levels (Cook & Moore, 1999; Chaloupka et al., 2002). High taxes, particularly on distilled spirits, have been a major element of alcohol policy in all Nordic countries since the first decades of the 20th century. All except Denmark have also substantially restricted alcohol availability in other ways, for instance by requiring bottles or cans of alcoholic beverages to be purchased in a limited number of government-owned stores.

The general effects of alcohol taxation and prices on alcohol consumption and problems have been widely studied at the population level, with a number of such studies carried out in the last half-century in the Nordic countries (see Mäkelä et al., 2002; Holder, 2000). Whether in the Nordic countries or elsewhere, there has been much less study of the differential effects on different segments of the population. Little direct information is available on whose drinking changes by how much and in what circumstances when there is a change in alcohol taxes or availability. The study described in this paper is designed to study precisely these questions, with individual-level data. In this, the study will complement the aggregate-level time-series and other analyses of the same historical experience which can also be expected to be conducted in the next few years.

Studying the effects of a general policy change on consumption in different segments of the population also offers a chance for direct testing of major theories of change in drinking, as described below, including the theory of the collectivity of drinking cultures (Skog, 1985). This theory is an important component of what is variously called the “new public health approach” or the “distribution of consumption” model of the relation between alcohol policies, alcohol consumption levels and patterns, and drinking-related problems (Room, 2002a).

The changes to be studied

In recent months, substantial changes have occurred in Nordic alcohol policies, both in terms of levels of taxation and in terms of the amounts of alcohol which travelers can legally bring across national borders (see Table 1).

This paper describes a coordinated study which aims to study the effects of these changes. The primary changes to be studied are: (1) the effects in Denmark and southern Sweden of a large reduction in Danish spirits taxes, and in Finland of a large reduction in alcohol taxes, (2) the effects of substantial increases in traveler’s allowances for alcohol imports in southern Sweden (primarily imports from Germany and Denmark) and in Finland (primarily imports from Estonia).

Table 1. Major tax and availability changes studied

Date of implementation / Nature and place of change / Mainly affects:
Oct. 1, 2003 / Danish spirits tax reduction / Denmark,
southern Sweden
Jan. 1, 2004 / Finnish, Swedish, Danish increase in traveler’s allowances / Denmark, Finland, southern Sweden
May 1, 2004 / Finnish change in traveler’s allowance from Estonia / Finland
March 1, 2004 / Finnish tax reduction / Finland

The effects of these changes in different segments of the population will be studied in Denmark, Finland and southern Sweden with panel (longitudinal) as well as repeated cross-sectional population surveys, using northern Sweden as a control site.

The Danish spirits tax decrease and cross-border shopping. Danish taxes on spirits were lowered on October 1, 2003 by 45%, to 43.75 DKK per bottle (70 cl. of 40% ethanol). ( The price of cheaper brands of spirits in Denmark dropped by about 25%. Danish spirits taxes are already considerably lower than in other Nordic countries, and the reduction further increased the discrepancy.

With the Öresund bridge linking Copenhagen with Malmö, Sweden’s third-largest city, there had already been substantial cross-border importing of alcohol, particularly in southern Sweden. Only 30% of the alcohol consumed in the Skåne (southernmost) region of Sweden was purchased from the Swedish government alcohol stores, with much of the remainder accounted for by spirits purchased in Germany and Denmark (Leifman & Gustafsson, 2003, Table 5.5). A discussion has started in Sweden of whether and when Sweden will feel compelled to lower its spirits taxes, also. The Swedish Prime Minister has stated that the government will wait until it sees what happens after the Danish change before deciding on any decrease of its own. In January 2004 a member of Parliament was named to lead an investigation of the alcohol situation in southern Sweden lasting at least 6 months and to recommend policy responses. Håkan Leifman, on leave from his position at SoRAD, is serving as the secretary of this investigation. In this circumstance, it is unlikely that any Swedish alcohol tax change would take effect before the beginning of 2005.

While spirits taxes are lower in Denmark than in other Nordic countries, they are higher than elsewhere in the European Union (Karlsson & Österberg, 2001), and higher particularly than in Denmark’s only land neighbor, Germany. The Nordic countries in general, along with Britain and Ireland, have been under pressure from the rest of the European Union to harmonize their alcohol taxes downward. Formal negotiations over harmonization have largely failed to bring tax rates closer, due in part to resistance to imposing any taxes on wine in 6 EU countries. The most recent discussion of the harmonization of alcohol taxes, on 18 September, 2002, again failed to reach resolution of the issue ( In the absence of harmonization by intergovernmental agreement, the European Commission has relied on mechanisms such as large cross-border traveler’s allowances to create pressure to bring tax levels down in the high-tax jurisdictions (Nordlund & Österberg, 2000). The Danish reduction in spirits taxes, to a level much closer to the German tax level, and the Finnish reduction ahead of Estonia’s accession to the EU, may be seen as in part successes for this implicit policy.

There is no doubt that, for consumers living near the border, cross-border shopping becomes popular when there are substantial differences in alcohol taxes and a relatively open border. This has been documented for the Danish-German border (e.g., Bygvrå, 1990), and has become a substantial factor, as mentioned, in the part of Sweden nearest to Denmark. An analysis of Swedish data by Norström (2000b), however, suggests that willingness to travel to purchase alcohol started falling off at about 100 km. from the border, and disappeared entirely at between 250 and 300 km. (See the discussion of “The gravity model of distance effects” below.)

The Finnish tax decreases on spirits and other alcoholic beverages. Estonia enters the European Union in May, 2004, at which point within-EU traveler’s allowances on import of spirits come into effect for travelers to Finland. Taxes on alcoholic beverages, including spirits, are low in Estonia, and the trip by fast ferry between Helsinki and Tallinn takes only about an hour.

In this circumstance, the Finnish government decided to reduce the tax on spirits by 44%, on fortified wine by 40%, on table wine by 10%, and on beer by 32%, effective 1 March, 2004. (Helsingin Sanomat, international edition, 21 Aug. 2003)

Increases in traveler’s allowances. On 1 January, 2004, the traveler’s allowances for those entering Denmark, Finland and Sweden from other EU countries increased to the general EU levels, and in May 2004 these levels will come into effect, also, between Estonia and Finland. Table 2 shows the relevant allowances for tax-free imports for Denmark, Sweden and Finland prior to the changes, and the general EU “guidance” on

Table 2. Traveler’s allowances for alcohol imports, in liters of each beverage

Within-EU to Denmark
<1/2004 / Within-EU to Sweden <1/2004 / Within-EU to Finland
<1/2004 / Estonia to Finland <5/2004 / General within-EU “guidance”: imports for own use
Spirits & / 1.5* / 5 / 1 / 1* / 10
Fortified wine & / 20* / 6 / 3 / 2* / 20
Table wine & / 90 / 52 / 5 / 2 / 90
Beer / no limit / 64 / 64 / 16 / 110

*either spirits or fortified wine.

Source:

imports that are assumed to be for the traveler’s own use and thus not subject to duty on entry. Except as noted, the amounts are cumulative, not alternatives to each other. These general EU guidelines came into effect for the three countries at the beginning of 2004. It should be noted that in April, 2004 the EU Commission has proposed an amendment to the directive on General arrangements for products subject to excise duty, which would end any “guidelines” on traveler’s allowances altogether and require cross-border sales of alcoholic beverages and through carriers be allowed ( The British and Swedish finance ministers have announced their opposition to this amendment.

Previous experience of the effects of changes in taxes and traveler’s allowances

Studies of the aggregate effects of tax changes. The relatively high spirits-tax regime in Denmark has been in effect since 1917, when, in wartime conditions of food shortages, the spirits taxes were raised to over 12 times their former value. The results were large decreases in per-capita alcohol consumption and in acute and chronic health consequences of drinking, and an overnight change from Denmark being a mainly spirits-drinking to being a mainly beer-drinking culture (Bruun et al., 1975:73). With differential tax rates and availability disfavoring spirits in all the Nordic countries, the other Nordic countries have gradually also turned from “spirits-drinking” to “beer-drinking” societies, although the predominance of beer has come to them much later than to Denmark (Leifman, 2001b).

As the Danish experience in 1917 vividly demonstrated, tax rates affect the level of alcohol consumption and the rates of alcohol-related problems in a population. High taxes on alcoholic beverages are viewed in the alcohol policy literature as one of the most effective measures to hold down levels of alcohol consumption and drinking problems (e.g., Edwards et al., 1994). These conclusions are in considerable part based on studies of the effects on the level of alcohol consumption of alcohol taxes or prices, which are common in the literature, including in the Nordic countries (Österberg, 1995).

There have been a number of studies of price elasticity in Nordic countries (Österberg, 1995). For Sweden, the price elasticity estimates commonly given are –1.3 for beer, -0.9 for wine and -0.9 for spirits (Assarsson, 1991), while for Finland they are –0.6, -1.3 and –1.0, respectively (Salo, 1990; Österberg, 1995). The estimates by Leppänen et al. (2001) of price elasticity for alcoholic beverages as a whole in EU countries find slightly less elasticity in Denmark (-0.57) and Finland (-0.68) than in Sweden (-0.86). These price elasticity estimates for Nordic countries do not differ greatly, for instance, from estimates of elasticities in the United States, although the Nordic demand for beer seems to be somewhat more price elastic than the U.S. demand. (For the U.S., the approximate values usually given are -0.3 for beer, -1.0 for wine and –1.5 for spirits (Chaloupka et al., 2002), values that derive originally from estimates by Ornstein and Levy (1983).)

Drawing on the literature on price elasticity, an international working group in 1995 estimated the effects on consumption in Sweden, Norway and Finland if prices there fell to the existing Danish level and to the existing German level (Holder et al., 1995). Their estimates were that the Finnish consumption would rise by 23% and 46%, respectively. The forthcoming changes in Finland will offer a direct test of the predictive model used in those estimates.

Studies of differential effects of tax changes on different population subgroups. Most studies of price elasticity are based on analyses at aggregate, population levels. They therefore contribute no direct evidence on whose drinking changes or by how much when taxes and prices change, a question which holds important implications for alcohol policy. One possibility would be that the change is concentrated among lighter drinkers. Arguments along this line are commonly backed up by the idea that addicted drinkers would not be influenced by such contextual influences as the price of a drink. A 1995 study (Manning et al., 1995) found evidence that heavy drinkers were less influenced by price than moderate drinkers. This study has been influential in the literature, although it is based on an analysis of cross-sectional data, rather than directly measuring the effects of change. A more recent cross-sectional study in which Manning was also involved (Farrell et al., 2003) found evidence in the opposite direction, concluding that there was substantial price elasticity for heavy drinkers with symptoms of alcohol abuse or dependence. This conclusion is in accord with other findings in the literature, such as the conclusion of Laixuthai and Chaloupka (1993) that beer tax rates had a greater effect among American youth on frequent or heavy drinking than on less frequent or lighter drinking.

Reviewing the current econometric literature on the effects of price on alcohol consumption and problems, Chaloupka et al. (2002) note critically that the literature is “not based on natural experiments. For example, no data are available comparing the amounts of alcohol consumed by individuals or groups at different prices, with all other variables held constant. Instead, researchers use cross-sectional data … or time series of such cross-sectional analyses”. The present study is an answer to this criticism.

Indirect evidence on the effects of price or tax changes on different groups of drinkers can be gained from studies of the effects of the changes on alcohol-related problems. If alcohol tax increases affect rates of cirrhosis deaths, of drinking-driving casualties, or of violent crimes, this is not only an important policy finding in its own right, but also implies that the drinking patterns of those prone to these outcomes are affected. Such effects have been found in U.S. studies (Cook, 1981; Cook & Moore, 1993).

The most direct way to measure whose drinking changes and by how much when there is a change in policy is with cohort or follow-up data from individual drinkers, interviewed before and after a significant policy change. A quasi-experimental design with interviews also in a control population increases confidence that the changes in drinking and alcohol-related problems are due to the change in policy.

There have been relatively few such studies of the effects of changes in alcohol taxes, and none that has included a control population. Citing three literature reviews, Kuo et al. (2003) remark that “the critical lack of recent research on price effects at the individual level has been widely observed”. Kendell et al. (1983) used longitudinal data to study the effects of a rise in alcohol taxes in Scotland, finding that heavy drinkers reduced their drinking at least as much as light drinkers. However, with the exclusion of abstainers from the follow-up, the study’s design was not optimal.

A study of the effects of a substantial reduction in the tax on foreign spirits in Switzerland has been carried out by Gerhard Gmel, Jürgen Rehm and colleagues (Kuo et al., 2003). Hypotheses to be tested in the study included: that the proportion of spirits in total consumption would increase after the tax change; that heavy drinkers would be more responsive than others to the change; that alcohol-related harm indicators would increase; that the effect of the change would be greater on respondents living nearer the border, and lesser on those in regions with more socially integrated drinking of spirits; and that the change would result in an increase in alcohol consumption in risky situations. An initial sample of 4007 completed interviews interviewed in spring 1999 was followed up, with 2923 (73%) reinterviewed 2½ years later in autumn 2001. Between these two main datapoints, two follow-ups of a subsample of the original sample were also conducted in autumn 1999 and spring 2000. Three reports on the study have been published or are in press (Kuo et al., 2003; in press; Heeb et al., 2003), and analysis and publication on the study continues.

Foreign spirits, which accounted for 53% of the Swiss spirits market before the change, dropped in price by 30-50%, while prices of domestic spirits stayed much the same. An analysis comparing the “before” data with follow-up data at 28 months (Kuo et al., 2003) found an increased consumption of spirits in all subgroups studied: in men and women, in different age groups, and at different levels of consumption. The study found that the increase in consumption was relatively higher in moderate than in heavier drinkers, though regression to the mean effects may not have been sufficiently controlled for.