Using a return on investment analysis to estimate the economic impact of potential changes to alcohol control policies in Estonia

Abstract

Estonia is planning an update of its national alcohol policies where an economic return on investment (ROI) analysis is needed to guide decisions against their monetary returns. Using mostly national data sources, the ROI analysis was based on direct healthcare costs and productivity losses due to premature mortality. The interventions compared comprised availability restrictions and taxation increases. For taxation increases, associated revenue increases to government were included. All analyses used a one-year time horizon and different sensitivity analyses. In 2023, all alcohol-attributable harms in Estonia totalled €510.00 million (1.3% of the Gross Domestic Product of Estonia) with €263.91 million direct costs and €246.08 million indirect costs. The proposed availability reductions are expected to yield a net benefit of €6.33 million, whereas a 15% increase in alcohol excise taxation could lower healthcare costs and productivity losses by €1.77 million in addition to increasing tax revenue by €32.27 million. Moreover, the interventions were estimated to lead to substantial reductions in mortality and hospitalizations. In terms of ROI, the availability interventions would result in €15 gained for each euro invested, and the taxation increase in €477 per euro invested, and without revenue in €25 per euro invested. Positive ROI was also shown in all sensitivity analyses. The proposed alcohol control policies for Estonia would not only reduce mortality and morbidity but also bring sizeable gains for each euro invested. Higher ROI for taxation increase compared to availability restrictions was mainly due to the added tax revenue.

Introduction

Alcohol use is a major public health concern due to high disease burden [1]. However, consequences of alcohol use extend beyond health, and cause a wide range of social harms [2]. Thus, alcohol use places a heavy economic strain on societies through healthcare expenditures, productivity losses, and social welfare impact [3]. To counteract these effects, all countries have implemented at least some interventions to reduce its harm and economic burden [4].

To select the best policy interventions in a given environment, international organizations such as the United Nations or the World Health Organization (WHO) suggest that investment case studies should be conducted [5, 6]. Investment cases have economic and political components, and a return on investment (ROI) analysis [7] represents its main quantitative economic component using a ‘business as usual’ scenario to evaluate the costs of inaction and of implementing country-specific interventions against its return in monetary terms. ROIs have increasingly been used in public health evaluations [8], and recently to evaluate alcohol interventions [9].

Estonia, the northernmost of the Baltic countries, regained independence in 1991. In the immediate years that followed, the adoption of a free-market ideology resulted in high availability of alcohol, insufficient control policies, and an increase in consumption and harm [10, 11]. In the first decade of the twenty-first century, despite the introduction of additional alcohol policy measures, consumption continued to increase until the global recession of 2008/2009 [10, 12]. In the second decade, the Green Paper on Alcohol Policy [11] consolidated and formalized a wide range of interventions and activities into a coherent and systematic national alcohol policy (for a comprehensive listing of actions and measures, see [10, 13]). While this strategy was successful overall [13], its effectiveness waned at the end of the second decade, partly due to decreases in alcohol excise taxation and the COVID-19 pandemic [13, 14]. Furthermore, Estonia experienced the highest increase in 100% alcohol-attributable deaths in the European Union during COVID-19 in part or fully due to polarization of drinking ([15]; for background [16]).

It took until 2024 to revive the efforts of a new integrated alcohol control policy ([17]; for a definition of integrated, see [18]), with current discussions in Estonia centreing around taxation increases and cuts on availability, the latter one of the key problem areas identified in the recent policy evaluation [13]. For availability, the Estonian Ministry of Social Affairs (MoSA) has proposed reducing the opening hours for off-premise sales by two hours on weekdays and Saturdays (to 10 a.m.–8 p.m.) and by 6 h on Sundays (to 10 a.m.–4 p.m.), banning the sale of alcoholic beverages at gas stations, and restricting delivery services by instituting identification checks and delays between receiving online orders and their delivery [17]. Online orders have been playing an increasing role for alcohol sales not only in Estonia, and this role is projected to further increase.

To help ground discussions of alcohol policy options on a broader evidence base, we conducted a ROI analysis.

Methods

Overall modelling strategy

While all components of cost of illness studies on alcohol can be used in a ROI analysis from a societal perspective [19], our analysis included only (1) cost to the healthcare system, which is usually the largest contributor to alcohol-attributable direct costs, and (2) productivity losses due to premature mortality before the pension age of 64.5, the largest contributor to indirect costs [3]. These components were chosen to avoid conducting a lengthy full cost of illness study.

In addition, we indirectly estimated all alcohol-attributable costs based on these two components, and the proportional relationships between healthcare costs (first main component) and the other major component of direct costs, law enforcement costs [3]; and between productivity losses due to premature mortality and all indirect costs based on a prior alcohol cost study conducted on Estonia [20]. As the estimation was only indirect, we did not include it in the ROI analysis.

For the estimation of both main components, we used the alcohol-attributable fraction (AAF) methodology for all disease and injury categories causally related to alcohol consumption [21]. AAFs are defined as the proportion disease and injury categories that would not have occurred in the absence of alcohol use. They sum up exposures to alcohol for all members of the population combined with their respective sex-, age- (where applicable), and disease-specific risks ([22]; for details, including formulae, see Supplementary Materials 1).

In addition to these components for the taxation scenarios, we included potential increases in government revenues from alcohol excise taxation. While revenue changes are only relevant if the ROI is from a fiscal perspective, for decision makers these changes constitute a key element in decision-making, and are often central to discussions in parliament.

The year 2023 was selected for the ROI analysis since it was the last year in which all necessary data were available. AAFs were first estimated based on the actual exposure to alcohol in 2023, and then based on counterfactual scenarios wherein all new alcohol policy measures were assumed to have been implemented in the beginning of 2023, with immediate effects (for a theoretical underpinning of temporality, see [23]). The flowchart providing a schematic overview can be found in Supplementary Materials 2. To be conservative and of more relevance to decision makers, a timeframe of 1 year for direct costs was used. Such a timeframe avoids predicting trends of future developments, and allows for quick verification of whether the predictions for monetary benefits will actually be reached (the exact formula can be found in Supplementary Materials 2).

Modelling interventions

The interventions compared were taken from suggestions made by the MoSA [21] and included: (i) the suggested availability reduction policies implemented simultaneously (reduced hours of off-premise sales, ban on sales at gas stations, and the delayed delivery and identification check for online and telephone deliveries), and (ii) an increase of alcohol excise taxation.

The three availability restrictions were modelled together, as control of online and telephone deliveries is crucial for ensuring the enforcement and effectiveness of permitted times and places of sale [24]. The estimates on the impact of the availability interventions were based on effects reported in Baltic countries [25], and the impact of taxation interventions including revenue was calculated based on beverage-specific price elasticities [26]. A detailed description of the interventions and underlying assumptions can be found in Supplementary Materials 3.

Estimating revenue

Tax revenues were calculated based on volume of recorded consumption by beverage type, and the tax rates currently employed (taken as baseline) vs. the tax rates associated with the various policy scenarios to increase taxes. The calculation methodology used was validated against reported revues in all EU countries, including past and current revenues reported by the Estonian government, and the agreement was found to be very high [27].

The increases in excise taxation were assumed to result in increases in price (full pass-through assumed), and the amount of alcohol purchased and consumed after the taxation increase was estimated based on own-price elasticities of alcohol demand, taken from the literature. The whole procedure is explained in more detail in the Supplementary Materials 3.

Costs of the policy interventions

The costs of alcohol policy interventions (i.e. availability reductions and taxation increases) were estimated using the WHO’s costing tool as a guide [28]. This tool provides a comprehensive list of cost components to consider when estimating the costs of new policy interventions, such as staff involved in planning, implementation, enforcement, and evaluation, and covers the staff’s training, meetings, and supervision. Other costs include media communications necessary to introduce the new interventions, supplies, and equipment. The costs for the three availability reductions were calculated separately and then combined. For taxation, the same costs were taken into consideration, independent of the tax rate introduced.

Main assumptions used in estimating costs of alcohol use

For healthcare costs, we applied the above-described AAFs to the primary diagnosis for attributable diseases and to injury codes, while including full costs for diseases 100% attributable to alcohol.

For productivity losses due to premature mortality, we followed a human capital approach [29]. The pension age in Estonia in 2023 was 64.5 years, and we estimated the years of life lost due to mortality before reaching pension age, taking into consideration employment rate, defined as the share of employed persons in the working-age population (for the exact definition, see Supplementary Materials 4). All calculations were sex-specific, as average salaries, employment rates, and AAFs differed by sex in Estonia; however, we did not consider age variations in salaries due to lack of empirical data.

Sensitivity analyses

Regarding the costs for the interventions, we varied assumptions about the personnel necessary for enforcement of the availability restrictions, and made assumptions regarding whether more or less personnel would be needed in the Estonian Tax and Customs Board to effect any change in the taxation rate.

Different sensitivity analyses were conducted on the impact of the availability restrictions (assuming less impact) and taxation increases (modelling alcohol excise taxation increases of 10%, 15%, 20%, and 40%). Moreover, we modelled two alternative ways to calculate costs of alcohol use disorder (AUD) and other 100% alcohol-attributable disease (see Supplementary Materials 1), a 2.5% discount rate for productivity losses, and excluding only unemployed people instead of using the labour participation rate (see Supplementary Materials 4).

Data sources

The list of alcohol-attributable disease and injury categories and their relative risk functions were taken from the latest WHO comparative risk analyses [4]. All data were obtained from official Estonian sources (for a full list, see Supplementary Materials 4). As per standard procedure for comparative risk analyses on alcohol, exposure was triangulated between aggregate and survey data [30].

Data on costs of medical services were obtained from the Estonian Health Insurance Fund (EHIF), which covered 96% of the population in 2023. The EHIF is the national mandatory health insurance fund in Estonia, and covered 72% of direct healthcare costs in Estonia in 2023, with most of the other costs covered by the patients [31]. Cost data were based on anonymized individual invoices submitted to the EHIF by healthcare service providers and included primary and specialized healthcare services, reimbursable prescriptions, and sickness benefits for 2023. These invoices also included the diagnosis information used to derive AAFs.

Data on productivity losses were based on official cause-of-death statistics, and we calculated the years of life lost until the average pension age of 64.5. And, once again, the sex-, age-, and disease-specific AAFs were applied.

The relevant cost input data for the selected interventions were provided by the MoSA.

Results

In 2023, 1932 (95% CI: 1707–2176) or 12.3% (95% CI: 10.8%–13.8%) of all adult deaths (defined as 15 years and older) in Estonia were attributable to alcohol use. In adults aged ≤64.5 years, there were 781 (95% CI: 749–810) deaths representing 26.9% (95% CI: 25.8%–27.9%) of all deaths. A total of more than 10 000 life years were lost prematurely, the overwhelming majority in males. An overview of the underlying cause-of-death categories are given in Fig. 1, in total and by sex. Large sex differences are not only seen in the total number of deaths but also in the relative distribution across cause-of-death categories.

Figure 1.

Distribution of premature (<64.5 years) years of life lost in Estonia by causes of death, 2023.

Costs of the potential alcohol control policy interventions

The total cost of implementing, enforcing, and evaluating the three availability restrictions in Estonia was €428 208 (with an estimated range of €362 040 to €494 376). Within availability restrictions, decreasing off-premise hours of sale had the largest cost (€208 904) followed by a ban on alcohol sale in gas stations (€149 353) and delayed delivery and identification checks for online and telephone purchases (€69 951; Supplementary Tables S3a and b). The cost of implementing the taxation policy was lower than the costs for implementing the availability restrictions (€71 368; range between €49 312 and €93 424; Supplementary Tables S4a and b).

Costs to the healthcare system before and after alcohol policy interventions

Table 1 summarizes the costs of alcohol use to the healthcare system before and after the suggested interventions. Alcohol use in 2023 caused more than €75 million in costs (see Table 1). The suggested alcohol policies would have reduced these costs by €2.57 million in the case of implementing the availability restrictions (3.4% of the overall costs), and by €0.65 million in the case of implementing a 15% taxation increase (0.9% of the overall costs).

Table 1.

Alcohol-attributable costs to the healthcare system in Estonia before and after the hypothetical alcohol policy interventions in million Euros in the year 2023^a

	Alcohol-attributable costs to the healthcare system 2023	Alcohol-attributable healthcare costs after hypothetical policy interventions
		Availability reduction	15% alcohol excise taxation increase
Females	15.78 (12.81–20.62)	14.27 (11.58–18.97)	15.56 (12.64–20.32)
Males	59.85 (55.16–63.87)	58.79 (53.57–63.08)	59.42 (54.75–63.48)
Total	75.63 (69.70–81.89)	73.07 (67.03–79.32)	74.98 (69.11–81.22)
Net benefits of intervention	Females	1.51 (0.69–2.42)	0.22 (0.14–0.39)
	Males	1.06 (0.48–1.82)	0.44 (0.34–0.50)
	Total	2.57 (1.49–3.73)	0.65 (0.52–0.80)

^aAll costs are in million €, with 95% CIs in brackets.

Productivity losses before and after implementation of alcohol policy interventions

An overview of alcohol-attributable productivity losses is given in Table 2. First, amounting to a total of €206.60 million, alcohol-attributable productivity losses in 2023 were estimated to be much higher than alcohol-attributable healthcare costs, by almost a factor of three. Second, the proposed interventions would lead to sizeable reductions in productivity losses amounting to €3.76 and €1.12 million for availability reductions and a taxation increase of 15%, respectively—albeit only affecting 1.8% and 0.5% of the overall productivity losses for 2023 in Estonia, respectively.

Table 2.

Alcohol-attributable productivity losses in Estonia before and after the hypothetical alcohol policy interventions in million Euros in the year 2023^a

	Alcohol-attributable productivity losses 2023	Alcohol-attributable productivity losses after hypothetical policy interventions
		Availability reduction	15% alcohol excise taxation increase
Females	29.21 (26.58–32.28)	27.56 (24.65–30.63)	28.99 (26.35–32.05)
Males	177.38 (169.16–184.52)	175.28 (166.53–182.77)	176.49 (168.10–183.73)
Total	206.60 (197.96–214.22)	202.84 (193.70–210.89)	205.48 (196.82–213.23)
Net benefits of intervention	Females	1.66 (0.75–2.79)	0.22 (0.17–0.29)
	Males	2.10 (0.95–3.52)	0.90 (0.69–1.01)
	Total	3.76 (2.18–5.48)	1.12 (0.90–1.24)

^aAll costs are in million €, with 95% CIs in brackets.

Sensitivity analyses for components of alcohol-attributable healthcare costs and productivity losses

The results for the various sensitivity analyses for net benefits with respect to healthcare and productivity losses are summarized in Fig. 2a and b (the exact definition of each scenario can be found in Supplementary Tables). It is clear that all interventions will lead to sizeable net monetary benefits from savings in healthcare costs:

Figure 2.

Sensitivity analyses on net benefits of policy interventions for healthcare (a) and productivity losses (b).

• availability restrictions from €1.01 to €2.63 million

• taxation increases from €0.43 to €1.77 million

and in productivity losses:

• availability restrictions from €1.46 to €5.97 million

• taxation increases from €0.74 to €3.07 million

Revenue gains

When conducting an ROI analysis, the resulting increases in revenue from changes in taxation must be taken into consideration for all the taxation scenarios. These increases occur because most alcoholic beverages are a relatively inelastic commodity and, thus, the decrease in sales is proportionally less than the increase in price. Using the usual beverage-specific elasticities, we estimated gains of €21.73, €32.27, €42.61, and €81.85 million for increases in excise taxation for alcoholic beverages by 10%, 15%, 20%, and 40%, respectively. This means that the monetary revenue gains derived from taxation increases clearly outweigh the other benefits realized from the alcohol policies considered.

Return on investment

All scenarios showed a positive ROI. The main scenarios for availability reductions resulted in about €15 gained for each euro invested, and about €477 for each euro invested for taxation increases of 15%, and €25 for each euro without revenue. While ROI values varied based on different assumptions made for the sensitivity analyses, it was substantial in all scenarios. ROIs for availability reductions varied between a 5:1 ratio for the lowest net gain and the highest cost of intervention and 24:1 ratio for the highest net gain and the lowest cost of intervention. The corresponding variations in ROI ratios for taxation increases were between 245:1 and 1758:1. The higher ROIs for the taxation increases are mainly due to the added revenue, in addition to lower costs to plan, implement, and enforce them.

Overall costs of alcohol use to Estonian society

So far, we have restricted ourselves to the main alcohol-attributable cost components: healthcare and productivity losses from premature mortality. To consider all cost components, we would need to include, minimally, law enforcement costs as a major direct cost, and presenteeism/absenteeism and disability as major indirect costs [3]. For the former, we can make our estimate based on the cost study of Saar [20], which would result in €188.28 million (95% CI: €173.51–€203.85 million) for law enforcement costs in 2023, and €263.91 million (95% CI: €243.21–€285.73 million) for total direct costs. Together with an estimate of €246.08 million (95% CI: €235.80–€255.17 million) for total indirect costs, the total economic cost caused by alcohol use in Estonia was estimated to be €510.00 million (95% CI: €484.07–€538.06 million), corresponding to 1.34% (95% CI: 1.27%–1.41%) of the national gross domestic product (GDP).

Discussion

Alcohol use causes considerable health burden in Estonia with 26.9% of deaths between ages 15–64.5 years being attributable to this risk factor. All alcohol-attributable harms were estimated to result in a total of €263.91 million in direct costs and a total of €246.08 million in indirect costs, resulting in a total of €510.00 million in economic costs, 1.3% of GDP. The proposed reductions of off-premise sales and other availability reductions are projected to result in a net benefit of €6.33 million in reductions in healthcare costs and lowered productivity losses, and an alcohol excise taxation increase of 15% was estimated to lower the costs by €1.77 million in the same cost categories, plus an expected increase in tax revenue of €32.27 million. According to the results of the ROI, within 1 year after policy implementation the availability interventions would result in about €15 gained for each euro invested, and a taxation increase of 15% is about €477 for each euro invested, or €25 gained for each euro invested without tax revenue. Thus, the higher ROI for the taxation increases was mainly due to the additional tax revenue generated, in addition to lower costs required to implement and enforce the intervention. In addition, all sensitivity analyses showed sizeable positive ROIs.

Before we discuss implications, we would like to list the limitations of study. First, we examined only two alcohol-attributable cost components in detail: healthcare costs and productivity losses. However, these cost components are proportionally the largest contributors for direct and indirect costs of alcohol use. Moreover, since the chosen interventions have been shown to reduce overall level of alcohol consumption [2, 18, 32], there is no reason to believe that the other cost components would not be similarly affected. Thus, the chosen method is overall conservative. Second, the timeframe of 1 year also underestimates the true impact of the interventions to reduce alcohol-attributable burden and costs. Clearly, while the impact of interventions may weaken over time—because people may get accustomed to availability restrictions and obtain their alcohol within these restrictions [33], and due to inflation and increased disposable income [2]—these effects last longer than a year. Third, the chosen AAF methodology, combined with cross-elasticities, seems to underestimate the true effects of interventions such as taxation increases on mortality and other health outcomes, as a recent methodological study for the neighbouring country of Lithuania showed [34]. Part of this underestimate seems to be due to the impact of availability on heavy drinking occasions, which are linked not only to injury, but also to cardiovascular events and 100% alcohol-attributable mortality [35, 36], over and above the effects of changes in level of drinking modelled here. Fourth, the chosen human capital methodology is standard in guidelines for economic cost studies, but may overestimate productivity losses [37]. Fifth, it may be argued that fiscal effects are often converted into a reduction in the efficiency costs of taxation, meaning that only a certain percentage of additional tax revenues can truly be considered a ‘return’ to society. We chose to list the full expected increase in revenue from alcohol excise taxation, as this best captured the fiscal perspective of the decision makers. Finally, as is customary for comparative risk assessments, we assumed that all impacts took place within the first year, which is an overestimate for certain chronic disease outcomes, even though most of the effects for chronic disease tend to materialize within 1 year [38].

While potential biases go into both directions, most of the large cost factors point to an overall conservative approach. In addition, we would also like to mention the main strengths of our approach: all data were available on the individual level and fully anonymized for the same year, restricted to a 1-year frame which is politically relevant, and relied on standard WHO methodology.

The ROI values obtained are substantive and reflect that alcohol control policies not only reduce the population level of alcohol use and attributable harm but also provide monetary gains to the society. As for increases in excise taxation, our study reiterates that the large ROI is mainly due to increased taxation revenue. The theoretical and practical considerations of including taxation revenue into a ROI in general and for Estonia in particular are further discussed in Supplementary Materials 5.

In conclusion, the proposed set of availability restrictions and increases in alcohol excise taxation for Estonia would not only reduce mortality and morbidity, but would also bring sizeable gains for each euro invested. These interventions seem timely for renewing efforts to reduce alcohol-attributable harm in a country that still ranks among those with the highest such burden in Europe, in part because no systematic efforts have been made since the fading of the effects of the last national alcohol strategy [11].

Supplementary data

Supplementary data are available at EURPUB online.

Conflict of interest: The authors declare no competing interests.

Funding

This evaluation was financially supported by the National Institute on Alcohol Abuse and Alcoholism (NIAAA) project “Evaluation of the impact of alcohol control policies on morbidity and mortality in Lithuania and other Baltic states” (Co-PIs: M. Štelemėkas & J. Rehm; R01AA028224). J.R. and A.S.H. were additionally supported by a grant from the Canadian Institutes of Health Research via the Ontario Canadian Research Initiative Node Team (OCRINT) CRISM Phase II (CIHR FRN-181677).

Data availability

The original data used in this study must be requested from the Estonian authorities, who are the sole providers of the data required to fully replicate and recalculate the analyses. Due to data access restrictions, these data are not publicly available. The R scripts used for the analyses can be obtained from the first author upon reasonable request.

Key points.

• Alcohol use has been estimated to account for 26.9% of deaths among people aged 15–64.5 in Estonia in 2023.

• The total economic cost of alcohol-attributable harms in the same year was estimated to amount to €510 million (1.3% of GDP), including €263.9 million in direct and €246.1 million in indirect costs.

• Availability restrictions (including a reduction of off-premise sales hours, a ban on advertisement and delayed delivery for online sales) were projected to yield a net benefit of €6.33 million through reduced healthcare and productivity losses (ROI ≈ €15 per €1 invested).

• A 15% alcohol excise tax increase was projected to yield a €1.77 million reduction in costs plus €32.3 million in additional tax revenue (ROI ≈ €477 per €1 invested).

• Both availability restrictions and tax increases would reduce mortality and morbidity and generate large economic returns, with tax increases yielding higher ROIs, mainly through additional revenue. Estonia still has room for higher excise taxation on alcohol than a 15% increase, given the recent increases in disposable income in the region.