EU on drugs: how do politics and institutions affect drug policies in the EU member states?

Researchers who studied the strong difference in approaches to DCRs in the UK (where there are none) and Germany (where there are many), noted two important points. On the one hand, the establishment of DCRs in Germany took place at the level of private initiatives emanating from the owners of establishments and shops in areas where, prior to the introduction of these rooms, there were the most prevalent drug-related crimes. Interest groups in different cities arranged negotiations with local authorities and through this achieved federal recognition, at first existing semi-legally, but with the support of local authorities and the police. Lloyd C., Sto?ver H., Zurhold H., Hunt N. Similar problems, divergent responses: drug consumption room policies in the UK and Germany, 67. On the other hand, the experience of attempts to introduce such an initiative in the UK has failed twice. The Parliament rejected this project, citing the lack of clear research and the unpopularity of measures among the media and the public. Attempts of such private initiatives failed as well. Even with financial support, these projects did not receive support from the local authorities. Ibid., 69. It is important to understand that levels of problematic drug use among their populations are similar. Ibid., 66.

Thus, in this study, drug policy will be understood as a set of specific features of each EU member state that distinguish them within a single framework. As described earlier, the main areas of differentiation are healthcare and possession and use. In line with the framework set out earlier in this chapter, it is possible to move on to specific issues that differ among the member states of the European Union and constitute their unique drug policies. The values that are the most varied among all states will be taken for further comparison.

Three main areas of differences in healthcare approaches will be considered:

- Heroin-assisted treatment (HAT). Substitution therapy in one form or another is present in every state of the European Union, but only a few countries have HAT at their disposal. Heroin-assisted treatment is the use of medical heroin in diamorphine substitution therapy. It is used when other opioid substitution treatment methods such as methadone are not effective. Studies have shown that this type of therapy is highly effective in treating heroin dependence, it reduces the use of street heroin and related criminal situations, and has a generally positive impact on treatment outcomes. The State of Harm Reduction in Western Europe 2018 (Harm Reduction International, 2019), 9 Only five states are able to benefit from such type of programmes.

- Drug consumption rooms. As noted earlier, few states have DCR's at their disposal for various reasons. Studies show that their use greatly improves the crime situation by moving problematic drug use from the streets to rooms designed for serious surveillance of drug users. Ibid., 6. At the moment, only seven states have DCR's.

- Take-home naloxone. Naloxone is an antidote that allows a person to quickly get out of an overdose and thus save a life. Most EU countries have naloxone at their disposal, but only a few countries allow it to be used outside medical facilities/DCR's. Currently, take-home naloxone is available in 9 European Union countries. The State of Harm Reduction in Western Europe 2018 (Harm Reduction International, 2019), 12.

It is important to note that in the established framework, it is not the number of certain decisions concerning healthcare that is important, but their very existence, since in this case each of the three features is a clear indicator of how the state treats drug policy.

As far as the possession and use are concerned, the approaches of the states are much more diverse. In terms of obtaining penalties for drug use, in 10 of 28 states criminal penalties can be avoided. At the same time, only 8 states can avoid punishment for possession (still, it is possible to get punishment for a certain type of crime). An alternative to criminal punishment for use is found in the legislation of 25 countries, and for possession in 23. 8 EU countries have legal delimitations on drugs with respect to drug use (i.e., 10 countries where it is possible to avoid criminal punishment for use are not included in this list). In terms of possession of drugs, 14 countries are delineated. Penalties for drug use can be avoided in 6 countries (+10 countries described above). With regard to possession, it is possible to avoid punishment for using drugs during recidivism for the same crime in 11 countries (+10 above). In addition to the above features, it is also important to differentiate between the penalties for possession and the weight of the drug seized. The penalty for possession is thus delineated in 10 countries. All the information provided above is taken from the legislative database provided by the EMCDDA. Penalties for drug law offences in Europe at a glance // EMCDDA URL: http://www.emcdda.europa.eu/publications/topic-overviews/content/drug-law-penalties-at-a-glance_en (accessed: 04.04.2020)

Thus, in view of the fact that the common drug policy of the European Union works in the framework of Multi-level governance, where the issues of law enforcement towards supply are fully harmonized, healthcare issues are partly harmonized, and the issues of possession and use are not harmonized, the second chapter of this study will present a cluster analysis, which will result in obtaining groups of states that are the most and least liberal in terms of drug policy. The results of this study will, on the one hand, allow for a more detailed assessment of the existing drug policies in the EU countries and will show similarities and differences among states, as well as use the findings in a further study of the impact of political/institutional conditions and their configurations, which will be described in the next section.

1.3 Factors affecting drug policy

Earlier, I have noted that the existing drug policy and its results can be viewed from completely different points of view: cultural, economic, social and political. In this part, I will try to bring more details regarding the approaches of understanding the differences in drug policies in order to convert this information into concrete conditions and outcomes in my future work. In addition, I will separately highlight the political factors that can be identified while understanding certain phenomena in drug policy.

The most common approach to drug policy is cultural. Basically it is about historically formed circumstances of the attitude of certain states towards the abnormal "states of mind". Duyne Petrus C., Levi M. Drugs and Money: Managing the drug trade and crime-money in Europe (London: Routledge, 2005), 4-5. There are both states in which any altered state of mind is unacceptable (Qatar, Saudi Arabia and other Islamic countries), and those in which the attitude is considered more comprehensively in different directions. Ibid., 9.

In addition, the cultural factor is about the historical development of the culture of drug use, which, again, varies greatly from state to state. In recent years, the experience of creating a culture of drug use has begun to become something of a product of export, since the actively developing online trade of illicit substances in the European Union creates the necessary information exchange for this area. Aldridge J., Stevens A., Barratt M.J. Will growth in cryptomarket drug buying increase the harms of illicit drugs? (Addiction, 113, 2011), 789. In general, it is noted that the personal practice of harm reduction, on which the culture of drug use is based, is built around a broad concept of risk. Bancroft A. Responsible use to responsible harm: illicit drug use and peer harm reduction in a darknet cryptomarket // (Health, Risk & Society, Vol. 19, Nos. 7-8, 2017), 337. The risk can be understood as the specific effect of certain substances on the body or knowledge of the laws of various states regarding the punishment of certain types of drug-related activities. Ibid., 343. Drug use trends are changing, and this is largely due to the exchange of information and the development of awareness and the so-called controlled risk, Ibid., 344. when consumers understand all the dangers of certain actions. An interesting example of this attitude is the use of two or more drugs together. Researchers call this phenomenon polydrug use. OLSZEWSKI D., MATIAS J., MONSHOUWER K., KOKKEVI A. Polydrug use among 15- to 16-year olds: Similarities and differences in Europe (Informa, 17(4), 2010), 288. Drug users largely do this for the sake of experiment, and then share the results in specialized forums.

When considering the social factors of drug policies, the demographic characteristics of their use are most often considered. So, it is believed that the main group of drug users are minors or young adults. COBO B., RUEDA Ma.M., LOмPEZ -TORRECILLAS F. Application of randomized response techniques for investigating cannabis use by Spanish university students (International Journal of Methods in Psychiatric Research, 26:e1517, 2017), 2-3. They are also called risk groups, because long-term dependencies are usually formed at this particular age. Schmits E., Mathys C., Quertemont E. Is Social Anxiety Associated With Cannabis Use? The Role of Cannabis Use Effect Expectancies in Middle Adolescence (JOURNAL OF CHILD & ADOLESCENT SUBSTANCE ABUSE. VOL. 25, NO. 4, 2016), 350. That is why many studies and solutions focus precisely on this social group as one of the main ones in the case of evaluating the results of drug policies. Reducing drug use in this group is treated as a clear success. This social group is the subject of programs for the prevention of drug use, which are further analyzed in terms of short and long-term effectiveness, Vigna-Taglianti F.D., Galanti M.R., Burkhart G., Caria M.P., Vadrucci S., Faggiano F. "Unplugged," a European school-based program for substance use prevention among adolescents: Overview of results from the EU-Dap trial (NEW DIRECTIONS FOR YOUTH DEVELOPMENT, NO. 141., 2014), 60, 76. as well as the impact of drug use on self-determination of the role of adolescents in the society. Schmits E., Mathys C., Quertemont E. Is Social Anxiety Associated With Cannabis Use? The Role of Cannabis Use Effect Expectancies in Middle Adolescence, 354. A special case is the fact that it is juveniles who are most likely to use new designer drugs, which are often very dangerous due to the lack of research and the artisanal mode of production. OLSZEWSKI D., MATIAS J., MONSHOUWER K., KOKKEVI A. Polydrug use among 15- to 16-year olds: Similarities and differences in Europe, 287.

Geographical factors of drug policy. Scholars that research the subject of drugs note that drug consumption in Western Europe is much higher than in all other parts. Spatial differences and temporal changes in illicit drug use in Europe quantified by wastewater analysis (Addiction. 109, 2014), 1338. However, even here there are conditions, since geographical differences exist not only among states, but also within them. Thus, drug use varies greatly between the inside of two parts of Germany - the former GDR and the former FRG, where the western part consumes more than the eastern. Ibid., 1349 Geographical factors are also largely dependent on the existing drug transition routes. European Drug Report Trends and Developments, 11. However, it is not possible to draw any specific conclusions regarding these factors, since the path of drugs and their "stopping" intersects with all other types of factors. For this reason, geographical factors of drug prevalence can be considered as a necessary background, but not as the basis for explanation.

Economic aspects mainly consist of phenomena that affect the cost of drugs in a given region, and, accordingly, their role in the state economy. Drug prices in each country also vary due to very different circumstances. In general, the cost of a substance depends on how restrictive the legislation of countries is. Researchers point out that more restrictive laws inflate the cost of drugs, as manufacturers and distributors at different levels charge more for risk. COBO B., RUEDA Ma.M., LOмPEZ -TORRECILLAS F. Further insights into aspects of the EU illicit drugs market: summaries and key findings, 12. In general, the economy of illegal drugs is a topic of many debates, both in terms of real production, the supply and demand of drugs, and the role of money laundering in the regional and world economy. Ibid., 2-4. The drug economy is the most difficult, closed, and confusing realm, which is why research usually concentrates on more understandable and measurable factors.

As mentioned earlier, there are practically no direct studies on the political factors that influence the course of drug policy. This can be explained by the fact that the field of drug policy in the European Union (as well as in the world) has developed not so long ago. Therefore in this study I want to give a first look at what possible configurations of political/institutional structures of the EU states might be relevant from the point of view of a particular drug policy. In order for this study to be relevant enough, I will collect a set of different conditions in which EU states exist in relation to averages of indices with known methodology. In this case it may be noted that the indices are best suited due to the peculiarities of the QCA method of this study and the given framework. The point is that for a full comparison of states it was necessary to find (1) comparable values - indicators of certain political/institutional conditions in which all 28 states (including the UK) in question exist. In addition, (2) the measurement data should be suitable for asymmetric encoding, which is implied by the QCA methodology. Categorical data are not appropriate, and numbers of statistics or other simple numerical measurements cannot be fully appropriate because QCA requires weighted conditions. Thus, data on the one hand, should fit into a model where "1" is the full presence of a given parameter and "0" is the absence of presence rather than the usual "absence" as in statistical methods. On the other hand, the data should be methodologically and theoretically homogeneous in order to make the assumption of a configuration leading to a certain outcome relevant, given the lack of such literature. That is why it was decided to select some World Bank indexes and indicators as the initial data, World Bank Open Data // THE WORLD BANK URL: data.worldbank.org (accessed: 04.04.2020) as in the majority of cases they are composed according to the unified measurement system (-2.5 -> 2.5), except for the indicator of the part of the budget expenditures on health care relative to GDP. Since in general the results of the European Union countries fit into the range from 0 to 2.5, it was decided to encode such measurements through the calculation of the average factor value for the most relevant comparison. Thus, "1" would mean that the value of a certain condition is above the European Union average.

Six conditions will be taken (more details about each of them and the features of the encoding will be explained further):

1) Indicator of political stability, non-violence and terrorism. Political stability is an important dimension to assess the overall situation in a country, as the presence or absence of political stability can directly influence which drug policy course a particular state chooses. In addition, the EMCDDA largely links the global drug trade problem with terrorism and general political instability. European Monitoring Centre for Drugs and Drug Addiction and Europol, EU Drug Markets Report 2019, (Publications Office of the European Union, Luxembourg, 2019), 32.

2) State regulation quality indicator. How the state copes with the regulation of certain areas of society is an important condition for building the right course of different policies.

3) The rule of Law. The problems of drug policy are largely related to the ability of the state to control the implementation and operation of legislation.

4) Indicator of public administration efficiency. The extent to which the state performs its duties affects all areas of society, including drug policy.

5) Corruption Control Indicator. One of the big and well-known problems of drug policy is corruption at various levels, including among doctors. Vokinger K.N. Opioid Crisis in the US - Lessons from Western Europe (The Journal of Law, Medicine & Ethics. №46, 2018), 189. Corruption allows drug trafficking to pass through ports and borders. Corruption on the ground can allow criminals to escape accountability. Given the large amount of money involved in drug trafficking, I believe that the extent to which the state is able to control corruption is an important part of understanding the bigger picture.

6) Percentage of state budget spending on health care relative to GDP. The introduction of this condition is one of the assumptions of this study, because, like the statistics on the results of drug policy, spending on healthcare on specific measures to address drug-related problems is heterogeneous and impossible to compare. However, the introduction of general spending on healthcare fits within a system of state configurations and can also provide a general understanding of how important general healthcare problems are to a state.

As an outcome, two provisions of drug policies will be considered for two QCA models. I will analyze which configurations lead both to the softness of drug policy above the EU average and the negate value of this outcome for a full understanding of the whole picture.

It is important to understand that the indices may not have full explanatory power, but, as noted earlier, they are good for comparing and analyzing countries beforehand, seeing certain patterns that will be interpreted in relation to each individual case. In the absence of adequate statistics and theoretical underpinnings, the indices can provide a new perspective on this issue.

This study avoids forming hypotheses not only because of the use of QCA but also because of the lack of a large amount of literature on the topic under consideration. Therefore, no hypotheses will be formed in this study, since the analysis itself is conceived as the first glimpse of the problem stated. Thus, by creating an important basis for the study, it is possible to proceed to the first stage - the formation of groups with different drug policies in the EU countries.

2. Drug policies in the EU Member States

This section will present detailed rules for coding cluster analysis variables. The first part will be devoted to healthcare and the second - for possession and use.

Healthcare

Variables for healthcare are:

- Drug consumption rooms (=DCR_s);

- Heroin-assisted treatment (=HAT);

- Take-home naloxone (=THN).

The value of the variable becomes "1" if there is a considered feature in the State, respectively, "0", where this feature is absent. Detailed description of the variables is presented in Table 1.

Table 1. Description of variables for healthcare

Possession and use

As noted earlier, the squeezers from modern drug-related legislation are provided as answers to certain questions on the EMCDDA website. Penalties for drug law offences in Europe at a glance // EMCDDA URL: http://www.emcdda.europa.eu/publications/topic-overviews/content/drug-law-penalties-at-a-glance_en (accessed: 04.04.2020) Each of the types of legislative regulation of drug issues has the same set of questions, 6 for crimes related to drug supply and possession, and 5 for use (there is no question "Penalty varies by quantity?", as it is related to the seizure of a certain amount of drugs, which is logically not appropriate in the case of drug use). Four of the six typical questions initially have a yes or no answer, which allows them to be coded as binary variables. These questions relate to the differentiation of drugs by types in the legislation, the differentiation of the threshold mass of the substance, the attitude to punishment considering the addiction problems of the detainee, and the consideration of recidivism for such an act in the past to choose responsibility. The other two questions are as follows: "what is the punishment for the offence" and "what are the alternatives to punishment for the offence?" do not imply the answer "yes" or "no", but can be redrafted according to the logic of constructing laws regarding use and possession separately.

Table 2. Questions on specifics of legislation and name of variables

In the first case, instead of asking about the type of punishment, the wording "Is there a punishment for the offence?" could be used, since there are 10 States for which there is no liability for this type of offence ("N/A" for some variables in the Table №3 because of full absence of punishment for use). When the question for possession was "Is it possible to avoid criminal punishment?". In all EU states, it is possible to obtain penalties for possession, but this depends on different conditions. That is why we can reformulate it for whether it is possible to avoid punishment in general.

In the case of the question "What are the alternatives to punishment for the offence?", it is possible to use the wording "Is there an alternative to punishment for the offence?" because of the possibility of an alternative to punishment that does not exist in all EU countries.

It is important to mention that the use of these questions is relevant for this study, as each of them focuses on a specific feature that has no intersection with the others. The whole set of questions is sufficient for an overall assessment of drug policy, as it considers each important feature of the legislation. For a description of the rules for variable naming, see Table 2, and Table 3 for a detailed description of each variable.

Table 3. Description of variables for possession and use

2.3 Results and interpretation

As a result of 10 manual iterations of K-modes clustering performed through the R environment using the "klaR" library, a set of countries was obtained, divided into four groups. To determine whether a group belongs to a drug policy characteristic, the brightest representatives of each of the resulting clusters were selected. Thus, based on the literature studied, Sweden and Finland are countries with completely prohibitionist drug policies, Chatwin C. Drug Policy Harmonization and the European Union (London: Palgrave Macmillan, 2011), 33. and the Netherlands and Portugal with a fully liberal drug policy, the results of different iterations of the cluster analysis identified four relevant groups of countries:

The first group, five countries are Croatia, Estonia, Finland, Greece, Sweden. Each country is characterized by a prohibitionist drug policy. In the case of Estonia, there is a difference - it is the only country on this list that has at least one of the points of healthcare, namely, the take-home naloxone. Interestingly, it is Estonia that the press calls "drug overdose capital of Europe" because of the major fentanyl epidemic of recent years. How Estonia became the drug-overdose capital of Europe // The Economist URL: https://www.economist.com/graphic-detail/2019/11/12/how-estonia-became-the-drug-overdose-capital-of-europe (accessed 04.04.2020)

The second group, five countries - Bulgaria, Republic of Cyprus, Hungary, Latvia, Lithuania. More prohibitive than liberal drug policy. The countries on this list are either not interested in active drug policies or have a historical negative attitude towards drug use.

The third group, 11 countries - Austria, Czech Republic, Ireland, Italy, Luxembourg, Poland, Romania, Slovakia, Slovenia, Spain, UK. They are more liberal than prohibitive drug policies. The largest of the groups. Most of these countries are either in transit to or already have liberal drug policies, but domestic measures are not sufficient to consider these countries completely liberal.

Fourth group, seven countries - Belgium, Denmark, France, Germany, Malta, Netherlands, Portugal. A group of countries with the most liberal drug policies. Interestingly, each of these countries is a transit country in drug trafficking, because each of them has ports where cargoes of cocaine and other imported drugs arrive, which are further diverged throughout the European Union and beyond. PERSPECTIVES ON DRUGS: Cocaine trafficking to Europe (EMCDDA, 2016), 2. This is a rather interesting pattern, which can be highlighted in further research.

When putting all states on the map (see Figure 1), we can see an even more interesting pattern: the liberal countries are those that are closer to the sea and to world drug supplies from Morocco, Colombia and Puerto Rico. However, the further a country is from the main drug supply chains, the less liberal its policy becomes, except for a few states. This detail may be a good start for a more detailed study.

Figure 1. Types of drug policies in the EU member states, May 2019.

Thus, having obtained four main groups on the types of drug policies in the European Union, we can proceed to the third and last part of this study - QCA. The results obtained in this chapter will be used as outcomes in two QCA models.

3. How political/institutional factors lead to certain drug policies in the EU Member States

Qualitative comparative analysis (QCA) is a research method based on the study of how certain causes (conditions) lead to a certain effect (outcome) among sets (groups) of cases. QCA is an alternative to the case-oriented approach (qualitative methodology) and the variable oriented approach (quantitative methodology) which investigates complex causality. Schneider M.R. Eggert A. Embracing Complex Causality with the QCA Method: An Invitation (J Bus Mark Manag Vol 7, No 1, 2014), 314. Complex causality is characterized by four main features: Rihoux, Benoоt & Ragin, Charles. Configurational Comparative Methods: Qualitative Comparative Analysis (QCA) and Related Techniques. Applied Social Research Series. 2009), 6-11

- conjunctural (variable-oriented + case-oriented);

- equifinality (how conditions lead to outcome);

- context-specific (важно хорошее теоретическое знание случаев);

- asymmetric values (e.g. negate of rich country is not poor country, but not rich).

Using QCA together with a solid theoretical base allows the researcher to look at the internal processes between specific cases on the one hand, and to make a comprehensive comparison of up to 100 cases on the other hand. This study will use crisp-set QCA, where conditions and outcomes contain binary asymmetrical values, which will be described and calibrated below. In this chapter I will analyze two QCA models, where the cases will be EU countries, and the conditions will be the Worldwide Governance Indicators provided by the World Bank, Worldwide Governance Indicators // THE WORLD BANK URL: https://datacatalog.worldbank.org/dataset/worldwide-governance-indicators (accessed: 03.24.2020). and the outcome: (1) the liberal drug policy (derived from the previous chapter) and (2) the negate value of liberal drug policy. Each solution provided by QCA will be interpreted, and rules for value calibration, truth tables, and the calibrated dataset used will be presented.

3.1 Setting up the QCA

Any QCA study consists of the following steps:

1) Theoretical background and QCA-oriented research design. This step is presented in the introduction and the first chapter of this study.

2) A description of the conditions and outcomes. Conditions were presented in section 1.3 of this study. In this section, their description will be added.

3) Calibration. Scoring of each case on the terms and conditions. Calibration will be fully presented in section 3.2. with description of the crossover point and participation of each condition and outcome in modeling.

4) Formation of the Truth table. It is presented in the analytical section 3.3.

5) Logical minimization. Obtaining different types of solutions. Both Complex and Parsimonious solutions will be presented in this study, since QCA involves parsing both full configurations and their shortened views.

6) Interpretation. In the last part of this chapter, I will move on to describe the resulting solutions and discuss the results relative to the available knowledge.

The research question of this paper was formulated in this way: "How configurations of political/institutional conditions affect current drug policy in the European Union member states?". Breaking it down in parts, you can see each of the stages of this research. First, the concept of drug policy in the European Union member states was introduced, meaning the different decisions of each EU member state in the context of harmonization of a large part of drug policy at the supranational level. After that, the main distinguishing elements regarding the use and possession as well as healthcare were highlighted. As a result, 4 groups of countries on the level of liberality of their drug policy were received. Based on the logic of building a research question for the QCA, I can use the results of the cluster analysis on the liberality of drug policy of EU member states as an outcome. The Worldwide Governance Indicators, derived from the World Bank's common methodology, will be used as a policy/institutional context, as mentioned earlier. Five of the six main indicators will be used, namely:

- Political Stability and Absence of Violence/Terrorism; Political Stability and Absence of Violence/Terrorism THE WORLD BANK URL: https://datacatalog.worldbank.org/political-stability-and-absence-violenceterrorism-estimate (accessed: 03.24.2020).

- Government Effectiveness;Government Effectiveness THE WORLD BANK URL: https://datacatalog.worldbank.org/government-effectiveness-estimate-0 (accessed: 03.24.2020).

- Regulatory Quality; Regulatory Quality THE WORLD BANK URL: https://datacatalog.worldbank.org/regulatory-quality-estimate-0 (accessed: 03.24.2020).

- Rule of Law; Rule of Law THE WORLD BANK URL: https://datacatalog.worldbank.org/rule-law-estimate-0 (accessed: 03.24.2020).

- Control of Corruption; Control of Corruption THE WORLD BANK URL: https://datacatalog.worldbank.org/control-corruption-estimate-0 (accessed: 03.24.2020).

It is important to note that the Voice and Accountability Index on Media Freedom and Freedom of Association Voice and Accountability THE WORLD BANK URL: https://datacatalog.worldbank.org/voice-and-accountability-estimate-0 (accessed: 03.24.2020). was also tested in the model, but did not perform as well as it did when the condition of percentage of health care spending in the country relative to GDP was introduced instead. It should be noted that there are no precise data on healthcare spending on drug policy, so a general control condition was chosen, which may indicate an interest in state regulation of healthcare and development of social programs in this case. The QCA can accommodate a limited number of conditions and the model does not work properly when all 7 are launched at once. In case of 6 conditions, the models turn out to be quite relevant.

This set of indicators explains many of the institutional/political conditions in which a state exists. In relation to the liberal (or non-liberal) outcome of drug policy, it is possible to assess which configurations of these indicators may be sufficient in explaining existing drug policies. A complete list of the names of conditions and outcomes according to the subject matter can be found in Table 4. I will then move on to calibrate them to introduce QCA models.

Table 4. Names of the variables.

3.2 Calibration

Five specified indicators from The Worldwide Governance Indicators list have normal distribution in the range from -2.5 to 2.5. At the same time, all the indicators of the EU member states are in the interval from -0.25 to 2.21. Therefore, as part of the calibration of the conditions of this study it was decided to take the intervals limited by the values of the EU countries for each individual condition. This is necessary because due to the common framework, the EU states in general are better able to cope with their responsibilities (with some exceptions like Greece and Hungary) than the world average, so due to the common field of rules of the game it is possible to consider the differences between countries within the EU itself.

This study uses a direct calibration method: the average of each of the indicators for 2018 (last available data) was taken as a crossover point. Data only for 2018 on political indices is a kind of assumption for this study, but the studied data on legislation is presented as of May 2019, and the distinguishing features are taken from the EMCDDA report for 2019, the difference from 2018 is minor. In addition, the values of presented indicators in most countries also did not change much in previous few years, and where the changes were significant, a common vector was preserved. In addition, due to the fact that countries maintain a common vector relative to their indicators, the problems in the dynamic changes over the years are levelled by asymmetric binary calibration, which is required for crisp-set QCA models. Table 4 shows that the value "above the EU average" was taken from each of the indicators. This is one of the requirements for the conditions based on indicators used in QCA due to the asymmetric logic of the input values. The rules of variable calibration are presented in Table 5.

Table 5. Variable calibration rules.

Please refer to Appendix No. 1 for a calibrated dataset. In the next section, a direct QCA analysis and interpretation of the results will be presented.

3.3 Analysis and discussion

For correct and transparent QCA analysis, it is necessary to assign suitable parameters of fit. First, it's about Consistency threshold. Consistency is about how many cases there are in the truth table display the outcome. The standard for crisp-set QCA is the Consistency threshold = 0.75. Some modern researchers take a higher value for this parameter (e.g. 0.8 or 0.85), but this study has an interesting feature. In a model where LIBERAL_DP has a positive value, the Consistency threshold value can be set to 1 because lower values do not even fall under threshold = 0.75 (see Table 6). In the case of negate model LIBERAL_DP (~LIBERAL_DP) of integer set of 4 cases has Consistency = 0.75, so I decided to include them into the study.

Frequency threshold of both QCA models =1 due to small number of cases. This value is due to the small number of cases included in the further logical minimization. In other words, one configuration will explain only one case if it falls into the parameters of fit models. In this study it is suitable because it is interesting to look at all possible configurations for any country, even if the configuration does not form a group.

Among other things, it is worth saying a few words about software used and necessity testing. To create models in this work, the program fsQCA was used, which uses the Quine-McCluskey algorithm. The necessary conditions in both models were not revealed, because the cases are heterogeneous. Necessary condition is a condition that is necessary in all cases. For example, when all "1" or all "0" are placed in the truth table under a certain condition included in the consistency parameters.

As noted earlier, this study will present two models - for LIBERAL_DP = "1" and LIBERAL_DP = "0", their detailed analysis will be presented below.

Liberal drug policy

Model: LIBERAL_DP = f(HSPGDP, POLSTAB, REGQUAL, ROL, GOVEF, CONCOR)

Table No. 6 shows the results of the Truth table formation algorithm, where you can see the configurations included in the logical minimization algorithm according to the specified value of the consistency threshold.

Table 6. Truth table for model: LIBERAL_DP = f(HSPGDP, POLSTAB, REGQUAL, ROL, GOVEF, CONCOR)

As a result of logical minimization, results were obtained with coverage =0.7, which indicates that the model explains 70% of cases. Consistency of model =1, it means that the configurations included in the model fully explain the 70% of cases with coverage =1. The lists of solutions are presented below. NOTE: Only Complex and Parsimonious solutions are presented, but not Intermediate solutions, because there were no prime implicants. Symbol descriptions: "~" is for "NOT", "*" is for "AND", "+" is for "OR". The countries selected in italics have an outcome =0 and, entering the set, are not included in the consistency parameter =0.75 for this model.

Complex solution:

POLSTAB*REGQUAL*~ROL*~GOVEF*~CONCOR: Czech Republic, Malta

HSPGDP*~POLSTAB*ROL*GOVEF*CONCOR: Belgium, France, Germany, UK

HSPGDP*~POLSTAB*~REGQUAL*~ROL*~GOVEF*~CONCOR: Italy, Spain

HSPGDP*POLSTAB*~REGQUAL*ROL*GOVEF*~CONCOR: Portugal, Slovenia

~HSPGDP*POLSTAB*REGQUAL*ROL*GOVEF*CONCOR: Ireland, Luxembourg

Here we see an interesting pattern in which in three out of five configurations the presence of spending on health care is more important than the average in the European Union. The second configuration stands out because it lacks the value of regulatory quality, but there is political stability below average in all conditions above the EU average. Also interesting is the latter configuration, where among the group of states from Ireland and Luxembourg the level of spending on health is lower than the EU average, while all other conditions are above the EU average. Both states are representatives of less liberal but not prohibitive policies. All states with fully liberal drug policies have healthcare expenditures above the EU average, but Ireland and Luxembourg have lower healthcare expenditures than the EU average under all other conditions staying the same.

Parsimonious solution:

HSPGDP*~CONCOR: Italy, Malta, Portugal, Slovenia, Spain

HSPGDP*~POLSTAB: Belgium, France, Germany, Italy, Spain, UK

~HSPGDP*POLSTAB*REGQUAL: Czech Republic, Ireland, Luxembourg

This type of solutions in QCA models does not have full explanatory power, but can highlight some interesting points thanks to clarifications. Thus, we can see that in two configurations out of three spending for healthcare is higher than the average for the European Union. They describe five out of seven countries with the most liberal drug policy (in the first case: Malta and Portugal, in the second case: Belgium, France and Germany). The third configuration, where only countries with less liberal drug policies spend less than the EU average on healthcare. The difference between the first two solutions lies in the different second conditions that make up the sufficient configuration. In the first case, it is control over corruption, which is lower than the EU average. In the second case, political stability plays a role below the EU average. The third solution shows a set of countries where the spending on health care is lower than average, political stability and the quality of state regulation is higher than average.

In addition, it is important to look at configurations that were below threshold consistency, but more than 0, because there you can find the remaining solutions to understand the full picture. Remaining solutions:

HSPGDP*POLSTAB*REGQUAL*ROL*GOVEF*CONCOR: Austria, Denmark, Finland, Netherlands, Sweden

~HSPGDP*POLSTAB*~REGQUAL*~ROL*~GOVEF*~CONCOR*: Croatia, Hungary, Lithuania, Slovakia

~HSPGDP*~POLSTAB*~REGQUAL*~ROL*~GOVEF*~CONCOR: Bulgaria, Republic of Cyprus, Greece, Poland, Romania

In the first and third case two completely opposite pictures can be seen: the first one presents the configuration of countries where each of the conditions is above the EU average. These are rich and stable states, two of which (Denmark and the Netherlands) represent the most liberal drug policy. In the third solution, each condition is not above the EU average. Among them there are two countries (Poland and Romania) where drug policy is less liberal but not prohibitive. These cases require a separate study on how countries can maintain liberal drug policies in this configuration and why this is the case.

In the second case, the situation is such that all countries in the set have all conditions below the EU average, except for political stability. Accordingly, in the case of a less liberal but not prohibitive Slovakia, political stability may play a role among all other conditions.

Non-liberal drug policy

Model: ~LIBERAL_DP = f(HSPGDP, POLSTAB, REGQUAL, ROL, GOVEF, CONCOR)

There is no point in attaching the Truth table to this model, because due to the negate value of LIBERAL_DP, it is inverted. Where Table No. 6 consistency equals 1, in this case it equals 0; where =0.6, here =0.4; where =0.4, here =0.6; where =0.25, here =0.75; where =0, here =1.

Logical minimization led to a model with coverage = 0.5 (explains 50% of cases), and consistency = 0.8 due to the fact that one of the configurations has raw consistency = 0.75. The countries in italics have an outcome =1 and are not included in the consistency = 0.75 parameter for this model.

Complex solution:

POLSTAB*~REGQUAL*~ROL*~GOVEF*~CONCOR*~HSPGDP: Croatia, Hungary, Lithuania, Slovakia

~POLSTAB*REGQUAL*~ROL*~GOVEF*~CONCOR*~HSPGDP: Latvia

~POLSTAB*REGQUAL*ROL*GOVEF*CONCOR*~HSPGDP: Estonia

The first case was presented in the last section, only from the perspective of Slovakia, which in this case falls out. Interestingly, in each of the presented decisions, states spend less than the EU average on GDP. Latvia and Estonia are opposed under three conditions: Rule of Law, Regulatory Quality and Corruption Control. The former has lower than the EU average, while the latter has higher than the EU average. They are similar under the other three conditions.