Nordic Studies on Alcohol and Drugs

Transcription

1 Nordic Studies on Alcohol and Drugs vol. 18/2001 English Supplement C o n t e n t s INTRODUCTION Jussi Simpura Trends in alcohol consumption and drinking patterns: sociological and economic explanations and alcohol policies 3 RESEARCH REPORTS Håkan Leifman Homogenisation in alcohol consumption in the European Union 15 Jussi Simpura & Thomas Karlsson Trends in drinking patterns among adult population in 15 European countries, 1950 to 2000: a review 31 Håkan Leifman Estimations of unrecorded alcohol consumption levels and trends in 14 European countries 54 Björn Trolldal Alcohol sales figures in 15 European countries: corrected for consumption abroad and tax-free purchases 71 Thomas Karlsson & Jussi Simpura Changes in living conditions and their links to alcohol consumption and drinking patterns in 16 European countries, 1950 to 2000 _ 82 Kalervo Leppänen & Risto Sullström & Ilpo Suoniemi Effects of economic factors on alcohol consumption in 14 European countries 100 Thomas Karlsson & Esa Österberg A scale of formal alcohol control policy in 15 European countries 117 NORDISK ALKOHOL- & NARKOTIKATIDSKRIFT VOL. 18, 2001 (ENGLISH SUPPLEMENT) 1

2 NORDISK ALKOHOL- & NARKOTIKATIDSKRIFT Vol. 18, 2001, English Supplement, Helsingfors NORDIC STUDIES ON ALCOHOL AND DRUGS Vol. 18, 2001, English Supplement, Helsinki Nordisk alkohol- & narkotikatidskrift is published by STAKES, the National Research and Development Centre for Welfare and Health (Finland), in co-operation with the Nordic Council for Alcohol and Drug Research (NAD). The journal is supported by the National Directorate for Prevention of Alcohol and Drug Problems (Rusmiddeldirektoratet, Norway), the Norwegian Wine Monopoly (Vinmonopolet), the National Institute of Public Health (Folkhälsoinstitutet, Sweden), Alkoholpolitisk Kontakt udvalg, the Ministry of Health (Denmark), and Alko Inc. (Finland). Editorial board ASTRID SKRETTING (chair), National Institute for Alcohol and Drug Research, Oslo, Norway HILDIGUNNUR OLAFSDÓTTIR, The Reykjavik Academy, Reykjavík, Iceland PEKKA HAKKARAINEN, Alcohol and Drug Research Group, STAKES, Finland LENNART JOHANSSON, Department of History, University of Växjö, Sweden PIA ROSENQVIST, Nordic Council for Alcohol and Drug Research, Helsinki, Finland ANETTE SØGAARD NIELSEN, Outpatient Treatment Clinic, Odense, Denmark Editor-in-chief KERSTIN STENIUS phone: (0) Editorial secretary LEA HÄLLSTRÖM phone: (0) Editor THOMAS ROSENBERG phone: (0) English language revision PIRKKO HAUTAMÄKI Editorial office STAKES, P.O.B. 220, FIN Helsinki, Finland Telefax (0) nat@stakes.fi Graphic design ANDERS CARPELAN EKENÄS TRYCKERI AB, Ekenäs 2001 ISSN NORDISK ALKOHOL- & NARKOTIKATIDSKRIFT VOL. 18, 2001 (ENGLISH SUPPLEMENT)

3 JUSSI SIMPURA INTRODUCTION Trends in alcohol consumption and drinking patterns: sociological and economic explanations and alcohol policies An introduction to the themes of the European Comparative Alcohol Study When we aim to explain variation between countries in the prevalence of alcohol-related harm, what are the factors that matter? The favourite explanatory factor has been per capita alcohol consumption (e.g. Edwards et al. 1994), but more and more voices are heard to insist that drinking patterns could matter more than the mere level of consumption (e.g. Rehm 1999). Beyond this basic dispute, a number of others lurk behind and require attention. What can actually be known about changes in alcohol consumption and variation therein between countries? How exactly can we establish the level of alcohol consumption, given the fact that a significant proportion of consumption may go unrecorded and be left outside the official consumption statistics? What do we know about changes behind consumption curves, and about trends in drinking patterns in particular? Are we led by our imagination more than by research evidence in believing that drinking patterns have changed and can be changed? What about possible explanations for trends and differences in alcohol consumption and drinking patterns? What is the power of the economic explanations? Where and how do cultural factors and changes in living condition enter into the picture? And finally, what could be the role of alcohol policy measures in influencing alcohol consumption, drinking patterns and NORDISK ALKOHOL- & NARKOTIKATIDSKRIFT VOL. 18, 2001 (ENGLISH SUPPLEMENT) 3

4 alcohol-related harm? Many would want to believe in them, but there is wide disagreement on which measures are feasible. All these questions lie in the background of the European Comparative Alcohol Study (ECAS). This collection of articles, a thematic issue of the NAT, will deal with trends in alcohol consumption and drinking patterns, with underlying economic factors and changes in living conditions, and the alcohol policies in different countries. Another issue (Addiction, 2001, edited by Thor Norström) will gather material from a set of studies about the link between the variation of alcohol consumption and changes in the prevalence of alcoholrelated harm. An official final report of the ECAS project will bring these two flanks of the study together in summer This introduction depicts the structure of the ECAS project and the relationships between the different sub-studies. It also discusses some of the methodological issues shared by the articles in this collection, drawing our attention to the limits of empirical knowledge on long-term trends and to the tension between beliefs and evidence. Drinking, alcohol consumption, and measures to prevent alcohol-related harm are all culturally and politically sensitive issues with large underlying economic and health interests. What matters in this minefield is not always scientific evidence only. Beliefs and aspirations count as well. An important aim of this collection of articles is to provide a basis for seeing the limits of knowledge but also the limits of believing. The degree of inaccuracy and uncertainty about trends in alcohol consumption and related processes may be higher than many would believe. But still, the main aim of the articles is to extend our understanding of the dynamics of changes in alcohol consumption and drinking patterns. Two substantial issues stand out. The first is the question of homogenisation of consumption and patterns, and also of underlying processes in economy, living conditions and policies. The issue here is to be aware of how far the homogenisation has come and, at the same time, how large the differences still are between the 15 European countries. The second question is about the natural time scales of changes in alcohol consumption and drinking patterns. Better understanding of the time structure may help in assessing the realistic scope of alcohol policy measures. But before entering these general questions, what needs to be introduced is the overall structure of the ECAS project and the main findings of the articles in this collection. The European Comparative Alcohol Study, ECAS The ECAS project was initiated in negotiations between the EU DG V (Employment, Health and Social Affairs) and a number of governments of the member countries after Funded by a grant from EU DG V/F since August 1998, the project 1 will be completed by summer The project deals with alcohol policies, alcohol consumption, and alcohol-related harm within a comparative and longitudinal approach. The focus is on the time period in the member states of the European Union as of However, Luxembourg was not included, while Norway has been added to the set of countries for investigation. The project is structured into four interrelated yet distinct areas and a fifth part that will be a general conclusion: 1. Analyses of alcohol control policies. These analyses comprise a systematic description of a) different methods used to prevent alcoholrelated harm, including price and tax policy, physical availability, licensing policy, advertisement regulations, and education and information; b) regulations and enforcement 4 NORDISK ALKOHOL- & NARKOTIKATIDSKRIFT VOL. 18, 2001 (ENGLISH SUPPLEMENT)

5 practices of public drinking, including drunk driving, on-licence outlets, and custody of drunk persons; c) the administrative structure of preventive policies and treatment; and d) a study of the feasibility of preventive policies, including an analysis of public opinion, pressure groups, and the effects of internationalisation and cross-border trade. 2. Analyses of trends in overall consumption. This work is based on sales data for alcohol beverages. Econometric analyses of price and income elasticities are performed, and a natural continuation is to explore the extent to which variations in economic factors across time and countries can account for the variation in consumption. Also addressed is the issue of unregistered consumption. 3. Analyses of drinking patterns. Consumption and drinking pattern surveys have been conducted in a large number of countries. However, few systematic analyses have addressed the variation across countries. Problems of comparability between countries and over time have been a major obstacle. Improved comparability has here been sought through expert collaboration, which has made it possible to approach issues such as the prevalence of heavy drinking in a comparative setting. 4. Analyses of alcohol-related harm. The only indicators of alcohol-related harm that meet reasonable standards of comparability are mortality data, although these are far from unproblematic. The core list includes causes of death where alcohol is an established risk factor: alcohol poisoning, alcoholism, alcohol psychosis, liver cirrhosis, pancreatitis, accidents, suicide, and homicide. A first basic task in this work is to aim at a description of harmrate profiles (e.g. acute vs. chronic harm) in the study countries. Further, the relationship between per capita alcohol consumption and the various alcohol-related harm rates are estimated by means of time-series analysis. These analyses are country-specific; i.e. the time series are analysed within each country, and, for some countries, reported in the literature. The project will pursue this kind of work but on a larger scale by including a more complete set of harm rates and countries. 5. Integration of the findings. In this final stage the aim is to interrelate the findings from the areas above. Issues to be addressed include: Is the pattern of harm-rate profiles related to alcohol policy regimes and to drinking patterns in the study countries? Is the strength of the relationship between alcohol and harm contingent upon drinking patterns and alcohol policy regimes? What are the implications of the findings for co-ordinated and efficacious alcohol policies within the European Union? The articles in this issue of the Nordic Studies on Alcohol and Drugs deal with questions concerning the first three points of the fivepoint programme above: first with alcoholspecific issues of consumption and drinking patterns, then with the background issues of economy and living conditions (cultural changes being discussed as an aspect of living conditions), and, briefly, with issues of alcohol policy. Putting the ECAS project into a schematic format, the role of the drinking patterns can be illustrated as in Figure 1, as an intervening factor between alcohol consumption and alcohol-related harm, and also as contributing to the understanding of dynamics in alcohol consumption: Each element in this scheme, indicated by small letters from a to g, pertains to a certain part of the ECAS project as follows: a. Alcohol policy b. Changes in alcohol consumption, registered and unregistered NORDISK ALKOHOL- & NARKOTIKATIDSKRIFT VOL. 18, 2001 (ENGLISH SUPPLEMENT) 5

6 Figure 1. The relationships between the various components of the ECAS project - c. Changes in drinking patterns d. Changes in living conditions e. Econometric analysis of changes in alcohol consumption f. Changes in the prevalence of various consequences of drinking g. Summary: the role of alcohol policy in influencing alcohol-related harm. The findings of this volume The first article (by Håkan Leifman) in this volume analyses trends in alcohol consumption, followed by two more articles that deal with the inaccuracies of semi-official consumption statistics, caused by unrecorded alcohol consumption (Håkan Leifman), and by tourism and tax-free purchases (by Björn Trolldal). The second issue to be discussed are trends in drinking patterns (by Jussi Simpura & Thomas Karlsson). Thirdly, two important issues that influence alcohol consumption and drinking patterns are analysed next, one (by Thomas Karlsson & Jussi Simpura) dealing with changes in living conditions, the other (by Kalervo Leppänen, Risto Sullström & Ilpo Suoniemi) examining the demand for alcoholic beverages, and clarifying the role of economic factors in alcohol consumption trends. Fourthly, a brief summary of trends in alcohol policies is given in an article (by Esa Österberg & Thomas Karlsson) that describes an attempt to develop a scale for the scope and strictness of alcohol policies. The main body of the findings from the alcohol policy section of the ECAS project will be published in a collection of country reports (Österberg & Karlsson, forthcoming 2001). More detailed accounts will also be available on alcohol con- 6 NORDISK ALKOHOL- & NARKOTIKATIDSKRIFT VOL. 18, 2001 (ENGLISH SUPPLEMENT)

7 sumption (Leifman, forthcoming 2001), drinking patterns (Simpura & Karlsson, forthcoming 2001), and the demand for alcoholic beverages (Leppänen et al., forthcoming 2001). All articles will examine the 15 EU countries, with the exception of Luxembourg, but including Norway. In some cases where the database is insufficient or otherwise problematic for some countries, these will have been excluded. Alcohol consumption: registered and unregistered, with correction for tourism and tax-free sales. Homogenisation in alcohol consumption is the main topic in Håkan Leifman s article about registered alcohol consumption. In the period from 1950 until the mid-1970s, the homogenisation of beverage preferences was visible in all country groups (Mediterranean wine-drinking countries, Central European beer-drinking countries and the former spirits-drinking countries of Northern Europe). In each group of countries, the dominating beverage lost some of its popularity. After the mid-1970s, the process of homogenisation has slowed down, only sustained by the drastic reduction of wine drinking in the Mediterranean countries. Homogenisation of per capita alcohol consumption has also followed a similar two-step process, with more dramatic changes in the first half of the 50- year period. In his second article, Leifman continues to analyse the levels and trends of per capita alcohol consumption by compiling the existing information on unrecorded alcohol consumption. The share of unrecorded consumption is significant and varies between countries, but also over time. In general, low-consuming countries in the north of Europe have relatively higher unrecorded consumption than do the other countries. To take the unrecorded consumption into account obviously evens out some of the differences between countries, but it does not change the ranking of the countries by per capita consumption in any significant way. The third article in this set, by Björn Trolldal, studies in detail the effects of one source of error in estimating the actual per capita alcohol consumption, that of tourism and tax-free sales in The correction needed because of these factors was in most countries below 10 per cent (with the exception of Norway), and did not change the ranking of the countries. In three countries (France, Austria, Greece) the correction was negative but very small, in another three (Spain, Portugal, Italy) it was virtually zero, and in the remaining nine countries the effect was clearly positive. Drinking patterns. Abstinence rates, frequency of drinking, the distribution of alcohol consumption by gender and age as well as the shape of population distribution of alcohol consumption, the prevalence of binge drinking, and drinking in certain specific contexts were the set of indicators used to identify trends in drinking patterns, as analysed by Jussi Simpura and Thomas Karlsson. Information on drinking patterns was compiled through a network of contact persons in all the 15 countries. It turned out that only a few countries could provide data from the 1950s and 1960s, and in most countries sufficient data was not available until the 1980s and 1990s. A decline in abstinence rates took place in the northern European countries in the 1970s and 1980s. The data does not show any common European trends for the women s share of alcohol consumption or in the distribution of alcohol consumption between age groups. Women contributed around 30 per cent to the total alcohol consumption, and the heaviest drinking age groups were in most cases between 30 and 50 years of age. Binge drinking, or the prevalence of occasion with high alcohol intake, could only be studied in a few countries, and even then, no clear trends would emerge. Data on drinking con- NORDISK ALKOHOL- & NARKOTIKATIDSKRIFT VOL. 18, 2001 (ENGLISH SUPPLEMENT) 7

8 texts was too scattered for any generalization. We can only conclude that there are still many differences between the countries, and it is very difficult to say whether a homogenisation is taking place here, too, or not. Many qualitative features of drinking patterns seem very persistent, and immune to change, even over decades. Underlying factors: living conditions and economic factors. The research question in Thomas Karlsson s and Jussi Simpura s study of changes in living conditions is again related to the general issue of homogenisation. It could be assumed that similarities in living conditions would, in the long run, contribute to a uniformity of alcohol consumption and drinking patterns, too. Indicators of living conditions in 15 European countries point to increasing urbanization, internationalisation and even globalisation. There is, indeed, much more homogenisation in living conditions than in alcohol consumption and drinking patterns. Modernization, or abandoning traditional ways of life, may evidently work into different directions in different countries, when it comes to alcohol consumption. In many countries in Northern Europe, increasing alcohol consumption has been linked with modernized life styles, whereas in many Mediterranean countries, the modernization has been accompanied by decreasing winedrinking in particular. There is no direct functional or technical link between living conditions and drinking habits; rather, the links are collective and cultural by nature. The econometric part of the ECAS project by Kalervo Leppänen, Risto Sullström and Ilpo Suoniemi covers all the 15 EU member countries (excluding Luxembourg and Germany) and Norway. Their research question is about the effects of alcohol prices and of total expenditure (i.e., purchasing power) on the demand for alcoholic beverages. Their timeseries analysis also allows for an estimation of the effects of non-economic factors, after having taken the economic factors into account. The analysis shows that country-specific level parameters (e.g. culture and traditions, alcohol policies and non-economic features of living conditions) seemed to account for the major part in explaining the differences of alcohol consumption among the countries. The economic factors, i.e. the relative prices of alcohol and real expenditure, played a subsidiary but important role in explaining the differences between the countries. The results for price elasticities indicate that demand for alcoholic beverages is more easily controllable by excise taxes in the northern European monopoly countries than elsewhere. However, the values of price elasticities indicate that the taxes in these countries have not been set at their revenue-maximizing levels. In the wineproducing countries the demand is relatively inelastic with respect to price. Such differences were not found in the effects of purchasing power (expenditure elasticities), but common expenditure elasticity could be estimated for all countries. The elasticity degree suggests that alcoholic beverages are considered to be normal commodities rather than luxuries. A scale of formal alcohol control policy. The major part of the results on alcohol policy will be published in detail in a separate report (Österberg & Karlsson, forthcoming 2001). In their article in this volume, Esa Österberg and Thomas Karlsson give a critical review of the earlier attempts to develop a scale for the strictness of formal alcohol control policy. They make an attempt of their own, too, although with important cautions against the risks of misuse of the scale. Their analysis is the first application of this approach on all EU member countries and Norway. Importantly, taxation of alcoholic beverages was not included in the scale. Nor can the scale account for informal social control of drinking that is regarded as an important factor in many countries. On 8 NORDISK ALKOHOL- & NARKOTIKATIDSKRIFT VOL. 18, 2001 (ENGLISH SUPPLEMENT)

9 the whole, the scales show a homogenising tendency in formal alcohol control policies. A closer inspection reveals, however, significant differences between the countries in how formal control is enforced and interpreted in practice. The limits of knowledge Much of current and previous research on the link between alcohol consumption, drinking patterns and consequences of drinking builds on the assumption that sufficiently reliable and accurate data will be available, preferably on a long-term basis in comparable time series. This is also the case with the other flank of the ECAS project, the time-series analysis on alcohol consumption and alcohol-related harm (see the special ECAS issue of Addiction, 2001). In existing research on drinking patterns, the access to data turns out to be a major obstacle, at least when any long-term developments are considered (see Simpura & Karlsson in this volume). Few countries provide a basis for long-term descriptions of changes on drinking patterns covering most of the ECAS investigation period (Finland, the Netherlands, and Norway). A few others can be studied in detail from the 1980s onwards (Sweden, Britain, Denmark). Even more recently, data is available from Austria, Germany and Italy, and in a more scattered fashion, from France and Spain. Finally, a number of countries (Belgium, Greece, Ireland, and Portugal) provide only a few rare studies on drinking patterns in adult population, based on general population surveys. Therefore, the basic question of homogenisation and the possible effect of drinking patterns as an intervening factor between alcohol consumption and alcohol-related harm remains largely unanswered. Similarly, little evidence is available on long-term changes in the more simple aspects of drinking patterns such as the share of women and men in alcohol consumption, or the more subtle aspects such as binge drinking. Somewhat similar, although less dramatic difficulties were met when establishing time series on per capita alcohol consumption (see Leifman s first article in this volume). The problems begin already with the official sales statistics where inclusion and exclusion of various types of beverages has varied between countries and over time. The alcohol contents of beverages vary, too. The problems can, however, be managed to some extent by a detailed inspection of the various sources. More problematic is the question on unrecorded consumption (home-made alcoholic beverages, tourist imports and consumption, smuggling, use of non-beverage alcohol as a potion). All estimates of the volume of unrecorded consumption are inaccurate, and again, the degree of inaccuracy varies over time and differently in different countries. In some cases, the accuracy can be improved by a detailed inspection and use of different parallel statistical materials, as has been done in the article on correction for tourist use and taxfree sales of alcohol (Trolldal in this volume). The article shows that a major effort was required to produce the correction even for a couple of years, and the effort needed for correcting the series over 50 years would be formidable indeed. From the point of view of studies on the link between alcohol consumption and alcohol-related harm, the consumption figures should reflect the actual consumption (recorded and unrecorded together) and not the official sales statistics only. Unfortunately, the studies using per capita alcohol consumption as an explanatory variable seldom can use anything more than the official sales statistics, with all their inaccuracies. This was also the case with the ECAS time-series analysis on the link between alcohol consumption and NORDISK ALKOHOL- & NARKOTIKATIDSKRIFT VOL. 18, 2001 (ENGLISH SUPPLEMENT) 9

10 harm. The problem of inaccuracy may be quite significant, as the share of unrecorded consumption varies between 10 and 30 per cent of recorded consumption between countries, and also varies over time. The econometric analysis also requires struggle with data compilation and comparability. The authors of the ECAS econometric sub-study (Leppänen et al. in this volume) had to exclude Germany from their analysis because of the lack of sufficient data on alcohol consumption. This is, of course, a major backlash for an all-european analysis. In other countries, comparable data was available for periods of variable length, not always overlapping. Fortunately for the econometric analysis, there are well-established theoretical grounds for selecting variables and checking the usefulness of the data. This is not the case with the analysis of changing living conditions and their effects on alcohol consumption (Karlsson & Simpura in this volume). In the absence of theoretical guidelines, the selection of factors necessarily becomes somewhat arbitrary and serves as an illustration of what was happening parallel to changes in alcohol consumption rather than as an analysis of the influence of the various factors. Finally, the article on constructing an index for the strictness of alcohol policy (Karlsson & Österberg in this volume) is essentially a story about difficulties in the quantification of unquantifiables. The numerous and rather arbitrary choices needed in the construction of such indices leave much space for discussion. A major problem that could not be solved in the construction of the indices concerned the intensity or depth of alcohol policy measures and the degree of their enforcement and implementation. These difficulties seemed largely insoluble. It should be remembered that similar problems are met when working with data on the consequences of drinking. Even the most routinely used indicators, including liver cirrhosis mortality, hide numerous problems for international comparisons. And, of course, this is not a difficulty that would be met in alcohol research only. In all fields of comparative epidemiology, a major effort is required even to know what the differences are between national statistical records. The problem of filtering off the effect of these differences is a daunting task everywhere. What can be done to remedy these insufficiencies and inaccuracies? Some are beyond the scope of any possible efforts, as in the case of a lack of data on drinking patterns. It is practically impossible to construct quantitative indicators on drinking patterns in most of the countries in the 1950s or 1960s in retrospect. A scale of inaccuracy could be built to measure and analyse consumption estimates by using certain upper and lower limits. Economic indicators can be improved with tedious work, except for the incurable lack of data. Problems with indicators of living conditions are more conceptual by nature, as there is a wide selection of possible indicators and only a vague theoretical basis for selecting any of them to explain changes in alcohol consumption and drinking patterns. Finally, quantitative presentations of the strictness of alcohol policies will probably always be problematic, and no one golden rule will become available for making the indices more useful. The lesson, then, is complicated. Why should such problematic findings be reported at all? There is a strong in-built tendency in science to report only positive results or significant differences. Results that are conditional on a number of reservations concerning reliability and comparability will not easily be published. This creates a cult of trustworthiness, and the limits of our knowledge on, say, drinking patterns, will not be openly debated. Similarly, there is a cult of difference and change. Non-existent differences go unnoticed, and no change is nowhere near as inspiring as dramatic turns. This bias for signif- 10 NORDISK ALKOHOL- & NARKOTIKATIDSKRIFT VOL. 18, 2001 (ENGLISH SUPPLEMENT)

11 icant differences and changes may easily lead the audience to believe that the world is essentially in a state of almost turbulent change. Our findings suggest that this is seldom the case in alcohol-related issues, where the changes are typically very slow. The nature of long-term changes in alcohol consumption, drinking patterns and alcohol policies Still, there are changes in alcohol consumption and drinking patterns, in conditions that influence drinking, and in policy measures. From the point of view of broad preventive policies, a crucial question is the pace of change: how quickly do changes occur? Implicitly, this is also a question about how quickly changes can be produced by policy measures in any direction. Policies that become effective in a very long run instead of the normal time frames of political operations, that is, from 5 to 10 years, are much more difficult to introduce than instant tricks leading to results next year. The findings of the articles in this collection suggest that in all respects, alcohol-related phenomena are prone to change slowly, except for very radically exceptional conditions. Among the 15 countries studied here, there are perhaps only two examples of such exceptional change since 1950: the Swedish alcohol reform in 1955 and the Finnish alcohol reform in 1969 (see Karlsson & Österberg in this volume). Three patterns of long-term processes deserve particular attention: homogenisation, natural time frame of changes in drinking and long waves of alcohol consumption. Homogenisation refers to a set of processes such as diminishing differences in per capita consumption levels, increasingly similar beverage preferences and perhaps even an increasing similarity in some aspects of drinking patterns. Also, despite the problems of constructing indices for alcohol policy, there are certainly signs of homogenisation here, too. And of course, economic, social and cultural conditions that influence alcohol consumption and drinking patterns also show a kind of homogenisation. Despite all this, the differences between countries remain, and there is much scope yet for further homogenisation. But from the policy point of view, the trends of homogenisation become visible very slowly, in a decade or two, and they are likely to proceed very slowly in the future, too. It will therefore be difficult to build short-term or medium-term policy programmes on the idea that homogenisation will solve or create some of the alcohol-related problems. The idea of natural time frames for changes in alcohol issues emerges when considering how the qualitative features of drinking may change. Again, there are certainly significant changes in these features (drinking wine regularly at meals in the South, the high esteem of intoxication in parts of the North etc.). But again, they take decades and even longer to become visible. Some traditional qualitative features of drinking seem very persistent to change, even in the midst of major quantitative changes in consumption levels etc. Therefore, the analysis of this report suggests that the natural time frame for changes in drinking patterns is a generation, rather than a decade or any shorter period. If this is accepted, it implies that efforts to prevent alcohol-related harm by measures targeted at drinking patterns will produce gains only in the very long run, if ever. Finally, the post-world War II history of alcohol consumption and drinking patterns in 15 European countries is also a story about long waves of alcohol consumption. These waves seem to have operated earlier in many parts of the European cultural sphere, with a wavelength of some 50 to 70 years (see Skog 1986, Room 1991 for a discussion). A peak in alcohol consumption occurred in many countries NORDISK ALKOHOL- & NARKOTIKATIDSKRIFT VOL. 18, 2001 (ENGLISH SUPPLEMENT) 11

12 (but not in the Mediterranean wine-drinking countries) around the mid-19th century, and the driest years coincided with the early decades of the 20th century. After WWII, there was a new rise in many countries until the 1980s and even later. But this is not the whole story: a parallel but opposite long wave of declining alcohol consumption started to operate first in France already in the 1950s, and later in other wine-drinking countries. It remains an open question how far the homogenisation is a mixture of two somewhat independent long waves. But again, from the policy perspective, long waves may be as real background factors as are changes in social, cultural and economic conditions. And once more, long waves seem to operate largely outside the realm of policy operations. the existence of the basic link. The insufficiency and inaccuracy of the data are major problems in all our analyses and leave space for speculation and beliefs. An important lesson drawn from this part of the ECAS project is that even in the worst cases of insufficient and inaccurate data, it is possible at least to identify such issues where evidence simply is unavailable. These issues, although often popular in political debate, could then be clearly demarcated outside the realm of Realpolitik around alcohol. But politics is not about facts and evidence only: for many, beliefs are as real as any evidence, and it may be beyond the capacity of scientific effort to change this world of ideas. Beliefs vs. evidence: concluding remarks According to the findings of the ECAS analysis on alcohol-related harm (see the ECAS issue of Addiction, 2001), the link between per capita alcohol consumption and prevalence of alcohol-related harm is not disappearing, although it is much more nuanced and far less mechanical than suggested in the most simplistic interpretations. The findings reported in this collection would suggest some additional reservations, but they do not question NOTE 1 Project co-ordination has been in the hands of a group of Swedish and Finnish researchers, with Dr. Thor Norström (University of Stockholm, Institute for Social Research (SOFI), Sweden), as scientific director and Dr. Jussi Simpura (Alcohol and Drug Research Group, National Research and Development Centre for Welfare and Health (Stakes), Finland), as deputy director. The administrative manager has been Dr. Ola Arvidsson at the Public Health Institute, Stockholm, Sweden. The project created a network of contact persons in all the 15 countries involved (EU member countries, Luxembourg excluded, and Norway). 12 NORDISK ALKOHOL- & NARKOTIKATIDSKRIFT VOL. 18, 2001 (ENGLISH SUPPLEMENT)

13 REFERENCES Edwards, Griffith et al. (1994) : Alcohol Policy and the Public Good. Oxford: Oxford University Press Karlsson, Thomas & Simpura, Jussi (2001): Changes in living conditions and their links to alcohol consumption and drinking patterns in 16 European countries, 1950 to Nordisk alkohol- & narkotikatidskrift 18 (English supplement) (in this issue) Karlsson, Thomas & Österberg, Esa (2001): A scale of formal alcohol control policy in 15 European countries. Nordisk alkohol- & narkotikatidskrift 18 (English supplement) (in this issue) Leifman, Håkan (2001): Homogenisation in alcohol consumption in the European Union. Nordisk alkohol- & narkotikatidskrift 18 (English supplement) (in this issue) Leifman, Håkan (2001): Estimations of unrecorded alcohol consumption levels and trends in 14 European countries. Nordisk alkohol- & narkotikatidskrift 18 (English supplement) (in this issue) Leppänen, Kalervo & Sullström, Risto & Suoniemi, Ilpo (2001): Effects of economic factors on alcohol consumption in 14 European countries. Nordisk alkohol- & narkotikatidskrift 18 (English supplement) (in this issue) Österberg, Esa & Karlsson, Thomas (eds.) (2001): Alcohol Control Policies in EU Countries and in Norway. Helsinki: Stakes, Reports, forthcoming 2001 Rehm, Jürgen (1999): Draining the ocean to prevent the shark attacks? Nordisk alkohol- & narkotikatidskrift 16 (English supplement): Room, Robin (1991): Cultural changes in drinking and trends in alcohol problem indicators: Recent U.S. experience. Pp in: Clark, W. & Hilton, M. (eds.): Alcohol in America. Albany: State University of New York Press Simpura, Jussi & Karlsson, Thomas (2001): Trends in drinking patterns among adult population in 15 European countries, 1950 to 2000: a review. Nordisk alkohol- & narkotikatidskrift 18 (English supplement) (in this issue) Skog, Ole-Jørgen (1986): Long waves of alcohol consumption: a social network perspective on cultural change. Social Networks 8:1-32 Trolldal, Björn (2001): Alcohol sales figures in 15 European countries: corrected for consumption abroad and tax-free purchases. Nordisk alkohol- & narkotikatidskrift 18 (English supplement) (in this issue). NORDISK ALKOHOL- & NARKOTIKATIDSKRIFT VOL. 18, 2001 (ENGLISH SUPPLEMENT) 13

14 14 NORDISK ALKOHOL- & NARKOTIKATIDSKRIFT VOL. 18, 2001 (ENGLISH SUPPLEMENT)

15 HÅKAN LEIFMAN RESEARCH REPORT Homogenisation in alcohol consumption in the European Union This study examines the developments in alcohol consumption in 15 European countries (14 European Union or EU member states and Norway) from 1950 to This very long time period is one of several dramatic transformations, both socially and economically, beginning with the reconstruction period after the Second World War, followed by the Golden Age of economic growth, increased purchasing power and the extension of the welfare systems in the 1960s. Optimism would give way to the harsh economic realities in the 1970s, which saw the rise of mass unemployment in most Western European countries. In the booming economies of the 1980s, too, mass unemployment was a constant and striking feature, while it is perhaps the ever-deepening globalisation of the economy and the ongoing economic and political integration that best characterize the 1990s. The drinking habits, too, have changed substantially over the years. One of the most frequently used labels to denote the trends in alcohol consumption in the industrialised world is that of homogenisation. What this usually refers to is that national differences in consumption levels are growing smaller and/ or that the countries traditionally dominant beverage type is losing ground in relative terms (e.g. Sulkunen 1976; Simpura 1995). Previous studies have addressed only parts of the time period between and have not fully discussed the importance of homogenisation and globalisation for national policies. A new appraisal of the evidence is therefore in order. It is not really clear whether we can detect a process of homogenisation for the whole study period, in particular in absolute levels (litres 100% alc. per capita) for the three main alcoholic beverages, and for overall consumption. These questions are im- This article is part of the European Comparative Alcohol Study (ECAS), a European Union (EU) financed study carried out by the Swedish National Institute of Public Health (NIPH) and the Alcohol and Drug Research group at the National Research and Development Centre for Welfare and Health (STAKES), Finland. NORDISK ALKOHOL- & NARKOTIKATIDSKRIFT VOL. 18, 2001 (ENGLISH SUPPLEMENT) 15

16 portant, especially from an alcohol policy point of view. If there is evidence of long-term reduced differences between the countries in levels of alcohol consumption, this means that national characteristics in culture, economy, and politics do not affect the development of alcohol consumption to the extent they used to. The question could then be raised whether there is any room for, and any reason to uphold, national alcohol policies. Alternatively, the impact of these global factors may differ in different countries and may be overshadowed by the specific national characteristics. Most analyses have focused on changes in beverage preferences, rather than on the absolute consumption levels, and have often been interpreted in the light of two main hypotheses. The substitution hypothesis assumes that new patterns, or new beverages, substitute for older ones, while in the hypothesis geared around addition, new beverages or patterns are added to the old ones that remain intact. By means of recorded alcohol consumption, Sulkunen (1976) found significant structural changes of drinking between (mostly) Western countries in , which is when the consumption grew in most of the countries. The proportion of non-traditional beverages in total consumption had risen. According to Sulkunen (1976), this levelling out of beverage preferences meant that inter-cultural diffusion had taken place in drinking patterns. He labelled this form of convergence as qualitative homogenisation. Simpura (1995) found evidence for a continuing convergence in Western Europe in beverage preferences in the 1980s, and Holder et al. (1998) showed that this trend of equalisation in the relative share of each alcoholic beverage between the EU countries has continued (from 1970) to Several studies have shown that, according to recorded consumption data, wine has increased its market shares in countries where it previously was rather rare and has lost its market shares in the typical wine-drinking countries in the Mediterranean. Spirits have lost in popularity especially in the former typically spirits countries in Northern Europe whereas beer has increased its market share both in Southern and Northern Europe (e.g. Pyörälä 1990; Simpura 1995). These are examples of harmonisation processes in beverage preferences, while not in the absolute levels of per capita alcohol consumption, although evidence of this, too, was found in Sulkunen s study (1976) of the period It was shown that the level of recorded consumption had risen more rapidly than average in the countries where the level of consumption was low to begin with, i.e. in the (at that time) spirits-drinking countries. This had the effect of bringing the countries closer to one another also as regards the volume of alcohol consumption. Simpura (1995) notes, too, that the overall consumption continued to converge in the 1980s. Holder et al. (1998) state that the countries with the lowest consumption increased their consumption between 1970 and 1996, whereas consumption decreased in the countries with the highest levels. Homogenisation is often discussed in relation to another frequently used term, that is, globalisation. Other related terms in use include modernisation (see, for example, Simpura et al. 1995), internationalisation and even McDonaldization (the increasingly global spread of rationalisation, Ritzer 1996). The terms are not completely synonymous but they embrace the basic idea that the world is shrinking, and that one likely consequence of this process (though not automatically) is convergence, standardisation or harmonisation between countries or regions in different spheres of life. The hypothesis of globalisation, and of homogenisation, is not a new idea but has been used for many decades in different disciplines. Generally speaking it is discussed in three 16 NORDISK ALKOHOL- & NARKOTIKATIDSKRIFT VOL. 18, 2001 (ENGLISH SUPPLEMENT)

17 fields. One is in the field of economics, where the capitalist world economy today knows no national boundaries. Deregulation of national markets and adaptations to the single global market economy occur everywhere. Multinational companies become even more multinational when they merge with each other. The European Monetary Union (EMU) is one example of the increasing economic integration in Western Europe. One likely outcome of this increasing economic integration is a gradual tax harmonisation within the EU. This may also be true of taxes on the different alcohol beverages. Analyses on alcohol prices and income (expenditure for private consumption) from the early 1960s to the mid-1990s have been conducted within the frame of the ECAS project, showing that both price and income changes have an effect on the overall consumption. The income elasticities were rather similar across all study countries, whereas the price elasticities proved stronger in the North and weakest in the South of Europe (Leppänen et al. 2001). Secondly, real incomes showed some tendencies towards increased harmonisation between the study countries. No such pattern was found for alcohol prices (Trolldal, forthcoming). Globalisation is also a force to be reckoned with in politics. At the European level, the EU, as a form of supra-national organisation, can be seen as a new type of political entity which is created not only to enhance the functions of the market, but in order to harmonise policies between the member states (see e.g. Montanari 1996). Thus, by entering the community, the EU member states have already given up part of their political autonomy. Hitherto, the dominating rationale of the EU (and especially of its forerunner the EC) has been the objective of creating an integrated and common market and, by degrees, an integrated economy. It has first and foremost been an economic-political (market economy) project. However, the Union also deals with other forms of political integration and perhaps even more so in the future with the Maastricht treaty (which came into force in 1991) and the treaty of Amsterdam (effective as of 1998). An analysis of the EFTA and EC countries until 1990 found no evidence of harmonisation of social policies with regard to social insurance programmes (Montanari 1995). Each EU country has its own national social policy. However, even here a future harmonisation is plausible. This may not necessarily be a direct consequence of EU regulations. It could be an indirect effect of common economic policies or the internationalisation of the economy which will make it increasingly difficult to uphold distinct national policies. The final shape of the political union of the EU remains to be seen, but that it will be a more politically integrated community than previously seems likely. Also, the ECAS project has conducted analyses of the extent of alcohol policy measures implemented in the countries involved (Karlsson & Österberg 2001; Trolldal 2001). In contrast to social insurance programmes, there are clear signs of homogenisation in the number of measures implemented from 1950 to The third field in which globalisation and homogenisation are discussed is culture. The last decades have witnessed uniform life-styles across countries. Perhaps the most obvious example is youth culture which recognizes few national boundaries. Increased cultural similarities between the countries must not be seen as a direct cause-and-effect relationship. Modernisation, or the abandoning of old lifestyles, rather implies that there are similar underlying structural conditions in most Western countries that make room for similar cultural patterns to emerge. The development of communication technology is one of these conditions, which has obviously induced more uniform life-styles across the countries. NORDISK ALKOHOL- & NARKOTIKATIDSKRIFT VOL. 18, 2001 (ENGLISH SUPPLEMENT) 17

18 All these changes have contributed, and probably will continue to contribute, to a progressive homogenisation of life-styles. The four freedoms of the internal EU market free movement of goods, services, capital and, not the least, of people will most likely speed up these processes even further. One aim of this study is to outline the general trends in beverage preferences and especially in aggregate alcohol consumption (in litres 100% alcohol). The emphasis is on the dispersion over time between all the countries. This will give an overview of all the study countries but will not yet highlight the individual countries nor specific trends. A second aim is therefore to look more closely behind these trends and to analyse the development for the different groups of countries. The results will be discussed in relation to the homogenisation processes in economy, politics and culture. Data and method The analyses in this article are based on aggregate recorded alcohol consumption on country levels and focus on aggregate changes in alcohol consumption and in beverage preferences. The data is published by the Brewers Association of Canada (1997). For a few countries lacking complete data, other data sources have been used as a complement or substitution. The statistics cover both the total per capita consumption of alcohol and the per capita consumption of beer, spirits and wine, expressed in litres 100% alcohol. All consumption data has been recalculated to per capita consumption per inhabitants aged 15 or older (per capita, 15+). In the calculations of the dispersion measures over time, each country s mean consumption (or proportions) counts for one unit, regardless of the population. The EU mean values presented below express the mean consumption of each country s consumption per capita, 15+. Each country therefore contributes to the same extent to the all-european estimates of beverage preferences (for example, the proportion of wine consumption of all consumption per capita 15+) and consumption levels (litres of total consumption or beer/spirits/wine consumption per capita, 15+). For most of the countries, the time period covered is from 1950 to All data suffers from validity problems, recorded consumption data included (see e.g. Simpura 1995; Leifman 2001). It should be mentioned that the recorded consumption is not equivalent to the total consumption and that the proportion recorded to total varies between the countries. Since this report focuses on cross-country comparisons in alcohol consumption, it is important to notice the possible changes that the inclusion of unrecorded alcohol consumption would have on the results obtained by analyses on recorded data. The size and development of unrecorded consumption in the study countries have been analysed in another ECAS paper and will not be repeated here. Some of the results will, however, be referred to. Two dispersion measures are used in order to calculate the dispersion between the study countries over time in alcohol consumption and beverage preferences. One is the coefficient of variation (CV), a relative dispersion measure expressing the standard deviation in relation to the mean. Since the CV tends to decrease when the mean value increases and the absolute variation is relative constant, the standard deviation (SD) will also be presented, referring to the absolute dispersion measure. For instance, if the countries involved show upward trends in per capita consumption but the absolute differences remain intact, the CV will decrease but the SD will remain the same. If the upward trend is combined with less absolute differences, the SD is also reduced. 18 NORDISK ALKOHOL- & NARKOTIKATIDSKRIFT VOL. 18, 2001 (ENGLISH SUPPLEMENT)