How Do State Gambling Monopolies Affect Levels of Gambling Harm?

This systematic literature review evaluates the potential of gambling monopolies to affect gambling harms. It compares the occurrence of gambling harms in jurisdictions with gambling monopolies to jurisdictions with license-based regimes. The review identified 21 publications concerning three gambling-related harm indicators: problem gambling prevalence, total consumption, and the appearance of conflicts of interest. Due to the dearth of literature, concept papers and older publications were also included. Results show that there is a paucity of empirical research on the effectiveness of different regulatory regimes in affecting gambling harms. Available research demonstrates that monopolistic regimes appear to perform somewhat better in terms of problem gambling prevalence and total consumption but may also be more prone to conflicts of interest than license-based regimes. Monopolistic configurations also differ between themselves, and issues such as availability, accessibility, product range, scope of preventive work, monitoring, as well as the recognition of the public health approach may better predict the levels of harm in society than the existence of a monopoly.


Introduction
The organisation of legalised gambling can take place through different regimes, including prohibitions, monopolies, and various forms of licensing. The most common forms of regulation are public monopolies and licensing to the private sector, but even these basic regulatory models can differ depending on other policy choices such as product ranges, availability, 'responsible gambling' policies, or ownership structures. The choice of regime ranges between jurisdictions and between different forms of gambling. Approximately 20% of the regimes globally are still public monopolies [1••]. Public monopolies are particularly prevalent in lottery gambling across jurisdictions, including the United States, but in many European as well as Canadian and Australian jurisdictions, also casino gambling may be subject to a local, often privately operated, monopoly. The monopoly model also appears to be gradually disappearing. Particularly in Europe, where gambling monopolies were still the prevalent form of gambling provision some 20 years ago [2], these have been replaced by licensing models under which former monopolies have become one of many licensed operators in many market segments. Currently only Finland and Norway operate fully monopolistic gambling regimes in Europe.
The choice of regime is partly constrained by the legal status of gambling. This issue is particularly pressing in the European Union (EU), where institutional constraints have impacted national legislations on gambling via legislative means. These include, inter alia, infringement proceedings initiated by the European Commission, and court cases brought to the Court of Justice of the European Union (CJEU). The rulings of the CJEU, in particular, have gained considerable importance in the choice of regulatory regimes in the EU. In practice, the provision of gambling from other EU countries can be restricted by means of a monopoly if this restriction is justified in acceptable terms, including prevention of gambling harms and promotion of public order [3, 4•].
This article is part of the Topical Collection on Gambling Furthermore, the CJEU has ruled that Member States need to show a 'causal link' between restrictions in their gambling legislation and their justifications [3]. Although the burden of proof is minimal, this does put pressure on Member States to show some connection between their gambling regime and its potential to reduce harms [3].
Monopolies have become increasingly burdensome to maintain particularly since the rise in popularity of online gambling. Channelling consumption to the monopoly holder is costly and may require heavy control measures such as IP blocking [5, 6••, 7]. Monopolies are also increasingly difficult to justify both legally and to consumers [6••]. European countries that have maintained their monopolies are increasingly focusing their gambling legislation on the prevention of gambling-related harms and the 'particular capability of monopoly-based systems to tackle these problems' [8]. In official justifications, monopolies are claimed to better achieve the task of both harm reduction and harm prevention. For example, in Finland, the Lotteries Act has shifted focus from revenue collection towards consumer protection and harm prevention to better justify the monopoly system after an official note from the European Commission in 2006 challenging its premises [9].
Despite these official justifications, previous research literature has not been conclusive on whether monopolies are actually better capable of preventing or reducing gamblingrelated harms than other systems (e.g. [10,11] show some support for this statement; [12][13][14] do not). It has been suggested that licensed regimes may be more effective in preventing harms despite increased availability, as licensing allows regulating otherwise unregulated offshore provision and a wider implementation of the so-called 'responsible gambling' tools [15].
In the current study, we conduct a systematic review of existing research evidence on whether gambling monopolies address harm levels better than licensing systems. Resulting from the limited space here, the focus of the paper is to compare the two most prevalent regimes, mostly governmentoperated monopolies and private licensed markets, although other regulatory options would also be available.
The definition of harms in the study is wide, acknowledging that gambling causes a range of harms not only to individuals but also to families, communities, and societies [1,[16][17][18]. Two previous review studies have addressed the issue [19,20] and one non-systematic summary of evidence [1••]. Planzer and Wardle [19] conducted a review of previous literature using the rapid evidence assessment method and concluded that there was no empirical evidence available that would directly address the effectiveness of regulatory regimes in gambling. The reviews by Williams et al. [20] and Sulkunen et al. [1••] both concluded that a less restricted approach is more likely to lead to gambling expansion, whereas government-run gambling monopolies tend to offer more prevention and treatment initiatives. At the same time, governments may be more prone to conflicts of interest (COI), as they run simultaneous roles as providers, regulators, and beneficiaries of gambling [1,21,22].
As the existing reviews are already somewhat dated or nonsystematic, there is a need for an update. In the following, we will first describe the methods and data, including the systematic search for evidence and inclusion criteria. The systematic search yielded results on three categories of harm indicators that we developed for the purpose of this review: the prevalence of problem gambling, total consumption (TC), and appearance of conflicts of interest. In the final section, we will discuss the results and identify the main gaps in extant research.

Methods and Data
We conducted a systematic review into the potential of gambling monopolies to affect harms. Following the PICO mnemonic [23], the intervention that we reviewed was therefore to compare monopoly provision to a licensing system to determine whether a monopoly system performs better. The choice to compare monopolies to licensing is motivated by the fact that this is the most plausible alternative to monopolies, particularly in online environments [1••]. Studies that include such a comparative setting fell into three categories: (a) those comparing jurisdictions with a monopoly to jurisdictions with licensing, (b) those comparing operators within a monopolistic and license-based market segments in one jurisdiction, and (c) longitudinal studies that examine the change of regime within one jurisdiction. In order not to focus on predetermined harms, we conducted a generic literature search on monopolies and licensing systems. The utilised search terms were 'gambling & monopoly', 'gambling & license/concession', and 'gambling & regulation'. In addition, we alternated the term 'gambling' with different alternative forms: 'lottery', 'casino', and 'betting' (i.e. 'lottery & monopoly' and 'lottery & license/concession). In total, 16 search term pairs were used in the following databases, using default search settings in the first half of the year 2020: If a search yielded more than 100 hits, only the first 100, sorted by relevance, were considered for each search pair. The included databases also included grey literature such as newspaper articles and reports. Figure 1 presents the PRISMA chart of the search [24]. The 4342 identified records were first cleaned from unrelated hits and duplicates, yielding a total of 911 records. These articles were screened for their relevance based on their abstracts, resulting in the exclusion of further 770 papers. The inclusion criteria at this stage of abstract reading were that the paper (1) appeared to concern the regulation of gambling via a monopoly or license system; (2) reported results on some outcome measures related to some outcome measure related to individual, community, or society level harms; (3) was available in a language that the authors could read (English, German, French). At this stage, we also decided to exclude papers regarding the levels of crime in different regulatory systems.
The remaining samples of 141 articles were carefully read. The final inclusion criteria were that the paper (1) compared monopolistic and licensed systems in terms of their potential to prevent harm and (2) gave either an empirical or an argumentative result regarding how well monopolies affect harms. Initially, we only planned to include papers that presented empirical evidence to this effect, but as their number was very limited, we followed the example of Planzer and Wardle's previous review study [19] and also included concept papers.
The final sample consisted of 21 papers, described in Table 1. Of these, 19 were in English, one was in German, and one was in French. Fifteen papers dealt with European jurisdictions at least partly. This was presumably due to the pressing nature of the monopoly topic in the European Union following the contradictory status of monopolies in light of the principle of the free movement of goods and services in the internal market. The remainder of the sample focused on the United States, Canada, Australia, and/or New Zealand.
The types of harm indicators addressed in these papers fell into three categories: prevalence of problem gambling, total consumption, and appearance of conflicts of interest. We included each in this review, as they offer an encompassing view of gambling harms at individual, community, and societal levels [1••]. In total, the 21 studies included 24 observations regarding these three topics. Problem gambling prevalence was considered an appropriate harm indicator, as it is widely used in studies to describe individual-level harms. Although problem gambling symptoms are not the same as harms, they can be used as an indicator of them. Total consumption has also been used as an indicator of a variety of Records identified through database searching (n =4342)

Eligibility Identification
Records after duplicates and unrelated hits removed (n =911) Records screened (n =911)

Records excluded (n =770)
Full-text articles assessed for eligibility (n =141) Full-text articles excluded, with reasons (n=120) Studies included in qualitative synthesis (n =21) population level harms, such as exposure and availability in related fields, like in alcohol studies [25], but also increasingly in gambling [1••, 26, 27]. The appearance of conflicts of interest was also considered in the previous reviews by Williams et al. [20] and Planzer and Wardle [19] and was considered a justifiable measure of systemic-level harms caused by gambling: Gambling is lucrative business for governments and private operators, which has been argued to have an impact on regulatory choices [28][29][30][31][32]. The three harm indicators differ profoundly as outcome measures. As recommended for a realist systematic review approach [33], working with heterogeneous methods and under various research paradigms, we did not assess the quality of the research in a hierarchical way. However, we did note whether the research included empirical data and how large the dataset was to better assess the results.

Results
In the following, we will consider the results thematically based on the three harm indicators found in the systematic review: problem gambling prevalence, total consumption, and appearance of conflicts of interest. Table 2 presents the results of the included studies comparing the prevalence of problem gambling (PG) in monopolistic versus license-based regimes. In total, six observations in the data addressed this topic.

Problem Gambling Prevalence
Two studies comparing population prevalence surveys across countries [4•, 34] find that there is either no observed difference between the two systems or that differences are not related to systems but possibly prevention. Another paper comparing case studies from Europe found that there also appears to be no observable difference between how well jurisdictions with licensed versus monopolistic regimes prevent problem gambling [35]. Cross-jurisdictional comparisons of prevalence rates are complicated due to the imprecise nature of population studies, including differences in screening instruments used, variance in samples, and heterogeneity in outcome measures [36,37]. These differences have been argued to make it difficult to evaluate any cross-jurisdictional differences, including regime efficiency, using prevalence rates [37]. Kairouz et al. [34] also note that the levels of problem gambling within a population are probably not the most appropriate measure for evaluating to comparative advantages of gambling policies or regimes.
The remaining studies focused only in differences within one jurisdiction (cross-sectional comparisons of market segments or longitudinal policy change studies). One study was a concept paper, arguing rather a lack of evidence than any difference between regimes in the case of the UK [38]. Another study compared levels of PG across two market segments in Denmark [39] and found that while PG prevalence was higher for the products operated in the licensed market than in the monopolistic market, this difference was likely to result from the characteristics of these products rather than the regime. Another paper compared prevalence of PG in Norway before and after the monopolisation of the country's EGM market [40]. The monopolisation process appears to have reduced overall PG prevalence because it also influenced availability and product characteristics of EGMs.
Overall, the results indicate that monopolistic and licensed regimes have some differences in terms of PG prevalence. Monopolistic regimes or market segments appear to have lower or similar levels of problem gambling than licensing, suggesting that monopolies may be better at limiting this type of harm because of other policy-choices often connected to monopolies, such as limited availability or prevention programmes.

Total Consumption
Eight study observations compared monopolistic and licensed gambling regimes in terms of their potential to either increase or curb TC. The studies and their results are described in Table 3.
Three studies address the TC on gambling across jurisdictions [2,41,42]. Two of these [2,41] found little difference between regimes regarding TC, but the study by Gu [42]  argued that at least in the casino sector, monopolistic European providers in France and the Netherlands appeared to operate less expansionist business models resulting in lesser TC than the competitive US casino sector. However, this observation is only based on one type of game product (casino games) and dates back over 20 years, to a time when many European markets were still considered to have a less 'commercialised' attitudes than today [e.g. 43,44]. More recently, one study [45] compared TC on gambling in Switzerland between the licensed casino market and the monopolistic lottery market. Despite these differing products, the study found that both monopolistic and licensed operators operated in an expansionist manner looking to expand markets and TC. Similar results have been argued in a number of concept papers [43,[46][47][48], but two of these papers ( [46] comparing licensed and monopolistic offer in Germany and [48] describing the Australian experience) argue that licensed providers operating in a market environment and looking for financial returns may be more expansionist than monopolistic providers.
Overall, the results indicate that while both monopolistic and licensed offer is connected to applicability of high TC and expansionist attitude, this appears to hold more for licensed operators than for monopolies.

Conflicts of Interest
Ten study observations concerned the potential of monopolies versus licensing models to be subject to COI. The studies are listed and described in Table 4.
Six of the included studies were concept papers. Of these, five papers [39,[49][50][51][52] connected monopolies, particularly when state-run, to a stronger COI in comparison to a licensed system. This difference was mainly attributed to the same entity (a state) taking several contradictory roles, including operator, regulator, and beneficiary of gambling. One concept paper [53] did argue that the state also has a similarly strong interest in revenue generated by licensed operators, and that any COI may be more strongly connected to total revenue from gambling than the regime. Private companies can also gain significant lobbying power and have been argued to foster 'poisonous' relationships with governments [54-56, 57•].
Only four studies used empirical evidence to investigate the scope of the COI. As this type of harm is difficult to operationalise, the studies are a mixture of focus groups and financial records, complicating comparisons [35, 58, 59, 60•]. The results were nevertheless similar to those argued in the concept papers: state-operated monopolies appeared to be related to a stronger COI than licensed regimes. At the same time, a licensing system does not protect from COIs if industry actors gain influence. As shown in the case studies analysed by Marionneau et al. [35], the financial stakes of states, other beneficiaries, as well as gambling industries create a variety of interests that may have important influence on national gambling legislations.
The results of this section indicate that although the empirical basis is weak, government-operated monopolies appear to be more connected to the presence of COI.

Conclusion
The current review draws three main conclusions. First, and as was already established in the two previous reviews [19,20], there remains a paucity particularly regarding empirical research on the effectiveness of different regulatory regimes in preventing and addressing gambling harms.
Second, based on the evidence that is available, different regulatory regimes appear to be connected to different types of gambling harms. In this paper, we have reviewed existing evidence on how regulatory models impact problem gambling prevalence, total consumption, and the appearance of conflicts of interest. In terms of problem gambling, monopolistic configurations appear to be connected to lesser or at least similar levels of harms in comparison to licensing systems. As for total consumption, licensed markets may lead to increased consumption in comparison to monopolistic providers, but this also depends on what kind of business model operators have. Regarding conflicts of interest, state monopolies have been connected to stronger COIs due to their several roles, but it has also been suggested that a similar situation may arise if private operators gain significant political influence via lobbying or other connections with governments. This indicated that fiscal motivations have had priority over advancing the good of the society across regimes [61].
As the evidence base on the relative effectiveness of monopoly regimes to reduce harms is slim (or even contradictory), the third conclusion we draw is that instead of regulatory models, it would be more pressing to compare how these are organised. Even seemingly similar regulatory approaches can differ considerably, as also noted by Planzer and Wardle [19]. Comparison of European country cases presented in the edited book shows that both licensed and monopolistic systems create COI, but COI appears particularly strong when the government takes several roles Miers (1996) [51] UK Concept paper Argues that when the lottery is operated by a private operator, the COI of the state is less pressing than in a case where the state is also the operator via public monopoly.
Paldam (2008) [39] Denmark Concept paper The regulation of state-operated monopoly gambling is argued to be 'softer and more informal' than that of market-based operation, creating a stronger COI.
Valleur (2015) [52] France Concept paper The strong COI related to the dual role of the state as an operator and regulator connected to opening the French online markets to licensing.
Even though gambling monopolies, particularly in the European Union, are justified in terms of consumer protection and harm prevention, monopolies may also operate in an expansionist manner. For example, despite being one of the last remaining fully monopolistic markets in Europe alongside with Norway, Finland also has one of the highest per capita gambling expenditure rates in the continent [62]. Similarly, in Sweden, Canada, or the Netherlands, state-owned or statecontrolled monopolistic companies have been argued to operate increasingly commercial attitudes [43,44,63]. In contrast, monopolistic offer in Norway has been accompanied by comparatively effective harm prevention [40]. While the mostly license-based regime in New Zealand has recognised gambling as a public health issue already in early 2000s [53]. Based on these results, it appears that rather than the regime, issues such as availability, accessibility, scope of preventive work, responsible gambling policies, the existence of a sufficiently resourced independent monitoring body, as well as the implementation of a public health approach to gambling may better predict the levels of harm in society. For example, Molinaro et al. [64] have found a negative association between the expenditure on public health and levels of PG prevalence in European countries, whereas Baxter et al. [65] have found that rather than regimes, the focus on public health research appears to be a dividing line between jurisdictions.
The current study has been limited to reviewing existing research on the potential of gambling monopolies to affect harms. We have only compared two models: the typically state-run monopoly model and the market-based licensed model within and between certain jurisdictions. Further research should also take into consideration the variations between these ideal types, including a comparison of public and private monopolies, and different types of licensing. The review has also not addressed the causality between regime and harm. It may be that a specific regime is more prone to a certain type of harm, but it may also be that the harms precede the choice of regime [cf. 66]. Nevertheless, the paper has given an overview on existing evidence regarding the incidence of different types of harms under monopolistic versus licensed gambling systems.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.