Background
There is a growing consensus that decreasing the environmental impact from food production and consumption are crucial to meet the Paris Climate Agreement and its goal to limit global warming [
1,
2]. This is not surprising considering that agriculture and food production contribute an estimated 25% of total greenhouse gas (GHG) emissions [
3]. Meat and dairy production is observed to be a disproportional contributor of emissions, attributing approximately half of food-derived GHG emissions, while only accounting for one-third of the dietary energy intake worldwide [
1,
2,
4,
5]. Our dietary pattern not only affects our environment, it also influences our health [
6]. A Western-type diet, characterized by a high red meat, processed meat, pre-packaged foods, fried foods, refined grains, and high-sugar drinks, has been strongly associated with non-communicable diseases (NCDs) such as cardiovascular diseases, diabetes, and cancer [
1,
7,
8]. Consumption of red and processed meat exceeds recommended levels in most high and middle-income countries and has been associated with both negative health and environmental impacts [
9,
10].
Recent modelling studies suggested that replacing meat with plant-based foods [
8,
11], or selecting foods with low carbon footprints [
11] reduce environmental impact and increase health. Nevertheless, the current Western diet is neither sustainable nor healthy [
12‐
15]. The link between individuals’ environmental concerns as citizens and their behaviour as consumers were found to be quite weak and did not appear to influence meat-buying habits [
16]. Consumers are not clear about healthy eating and everybody interprets it differently [
17,
18]. This might even more apply to what it means to eat sustainable and environmentally friendly.
Governments may implement policy measures to stimulate healthy and sustainable choices. Systematic reviews demonstrate that subsidies to increase consumption of healthy foods and taxes to decrease consumption of unhealthy foods might be effective interventions in improving dietary behaviours and health [
19,
20]. Similarly, modelling studies from various European countries predict taxes based on GHG emissions to be feasible to change dietary behaviours towards food groups with a lower environmental footprint [
21‐
24]. Springmann et al. (2016) have estimated that the worldwide impact of taxing diet-related GHG emissions could result in a 9.6% decrease in GHG emissions originating from food production, while avoiding 500,000 deaths annually [
25].
Price interventions may not only influence environment and health, but also have effects on other aspects of society, for example economic and distribution effects. A social cost-benefit analysis (SCBA) can incorporate these effects into a single analysis [
26]. The SCBA is an instrument that can provide an overview of the (dis)advantages of measures, if possible quantified in euros and presented as a balance [
26]. In this study, a SCBA was used to estimate and monetize the 30-year societal effects of a tax on meat and subsidization of fruit and vegetables (F&V) in the Netherlands.
Discussion
A price increase for meat through a tax could lead to a net societal benefit for the Netherlands of about €3100–7400 million or about €4100–12,300 million for respectively a 15 or 30% tax over a period of 30 years. A price decrease of F&V by means of a subsidy could lead to a net societal benefit of about €1800–3300 million. Important contributors to net welfare gains or losses are consumer surplus and policy revenues/costs. Several assumptions, such as estimated costs of environmental impact indicators and production efficiency gains over time, as well as selected discount rate or using the informed consumers’ assumption were shown to have a large impact on the estimated results. However, in almost all sensitivity analyses the total estimated effect was still a net welfare benefit for society.
A SCBA aims to take into account all types of costs and benefits of interventions, irrespective of which stakeholders win or lose from the policy scenarios. However, it is also important to show the distribution of these benefits and costs over the different parties, especially when the losses are financial while the gains are non-financial, such as a gain in QALYs. In the tax scenarios, consumers could be net payers because of the loss of consumer surplus, whereas the government could gain income from the tax revenues. For the subsidy scenario, this is the other way around.
Even though no study as of yet has estimated the total societal effects of a tax on meat or subsidies on F&V, various studies have assessed the effects on consumption, health or environment separately. Mhurchu et al. estimated a 2% decrease in all-cause mortality following a 20% subsidy on F&V in a modelling study in New Zealand [
50]. In Sweden, Sall and Gren estimated a variable environmental tax (9–30%) on meat and dairy to decrease GHG emissions by 12%, at a specific point in time [
21]. Briggs et al. applied a tax of £2.72/ton carbon dioxide equivalents/ 100 g product applied to all food and drink groups with above average GHG emissions in the United Kingdom [
51]. They estimated GHG emissions reductions up to 18,683,000 ton CO
2−eq per year compared to the current situation, while saving 7700 lives per year. In our current study, introduction of a 15% tax on meat could reduce GHG emissions by 3,600,000 ton CO
2-eq in 2048 if our population size were similar to the UK, which population is around four times larger. This might be an indication that taxing of all foods with above average GHG emissions is more effective than singling out only meat.
In a recent paper, Springmann et al. estimated the global and national health care costs related to meat consumption and calculated from this the price of meat if all health effects were incorporated in the price [
52]. In Western countries, such as the Netherlands, the increase in price should be 21.3% for red meat and 111.2% for processed meats.
In addition to integrating health in the price of foods, also environmental and social factor could be added to calculate the ‘true price’ of a food [
53]. Evidently, as Springmann et al. described that integrating health would already add 20% (red meat) and more than 100% (processed meat) to the price, integrating these other components would add to this even more. From a societal perspective, in the current analyses, it showed that the meat tax scenarios of 15% or 30% could result in net benefits for the society. Although it is clear, that not all health and environmental costs are then covered.
The analyses focus on the Dutch consumption, but the global perspective used in the article of Springmann et al. indeed is important to take into account. When price policies are applied only nationally, there will be trading effects with economic consequences that are yet not included in the current analysis. In addition, foods and feed are imported from other regions. International or European agreements will be important to prevent the possibility of carbon leakage, in which the emission reduction by one country is followed by an increase in another country, as environmental effects are not confined to national borders [
23,
54]. The used environmental impact is based on average Dutch market (including import) food consumption data [
40] and their monetized values are not specific for the agro sector [
41]. For monetized local environmental effects of Dutch foods this might be a good estimation, but some of our consumed foods and feed are imported and the associated environmental impact and associated costs are located in other countries [
55].
Changing prices of food products will likely have diverse effects on groups within society, in particular regarding socio-economic status (SES). While the environmental impact of food consumption is similar between SES groups, low SES groups generally have an unhealthier diet and tend to eat less F&V and fish and more meat and fats compared to higher SES groups [
28,
55]. A recent Dutch study indicates that the socioeconomic differences in healthiness of the diet are likely to further increase [
55]. As low SES groups have been observed to be more responsive to the price of food, price intervention policies could be an effective approach to increase healthiness and environmental friendliness of diets [
29]. A combined taxation on meat and a subsidy on F&V could compensate some of the losses felt especially in the low SES groups. The current model did not allow a combined calculation of the subsidies and taxes and for stratification by SES, which is predicted to be more effective and result in higher societal benefits than single interventions. In addition, more research is needed to estimate cross-price elasticities by SES as well: which foods items will be consumed more often to replace the more expensive meats.
Discouraging meat consumption is highly related to current consumption levels and disease profiles of countries [
10]. Although linked to adverse health effects on noncommunicable diseases, such as colorectal cancer and cardiovascular diseases at high consumption levels, meat is a nutrient-dense food in itself. In developing countries, because of the very low meat consumption, an small increase of meat consumption could help to alleviate some of the micronutrient deficiencies [
10,
52]. It should also be noted that sustainable diets are only a part of the solution towards a more sustainable consumption pattern [
3]. Using renewable energy sources, recycling waste, using trains instead of airplanes are, among others, additional steps for a sustainable future [
56].
This study has a number of strengths. The consumption data was from a representative sample of the Dutch population [
28]. Estimated total Dutch consumption of meat at baseline is in line with other reports on Dutch consumption of 77 kg meat (excluding bones 38 kg [
57]). Relative risks related to food consumption are supported by consistent and strong evidence derived from a systematic literature review [
30]. The DYNAMO-HIA model allows dynamic simulations and used Dutch data as input with an integration of the range of the estimates. Additionally, the used environmental data were obtained from 2018 LCA analyses tailored for the Dutch food consumption context. The SCBA framework allows for comparison between different fields such as health, environmental and other societal effects by monetizing the effects allowing us to estimate the societal impact of taxes on meat or subsidies on F&V. The Dutch guidelines on SCBA analyses in the health and environmental domains were followed [
26,
31,
49]. Finally, several multiple sensitivity analyses were used to demonstrate the effect of different input parameters and methods on the net societal benefit.
Like any modelling study, the current study also has limitations to consider. Firstly, in the model, cross-elasticities could not be used to consider replacements following a decrease in meat consumption. Replacement foods could be healthy and sustainable, such as F&V, but also have high environmental impacts or negative health effects, such as cheese. In addition, the DYNAMO-HIA model is limited to one risk factor, and thus does not allow for analysis of a combination scenario of a meat tax and F&V subsidy. A replacement food may also partly compensate for the lost consumer surplus experienced in the meat tax scenario. Secondly, when considering the consumption in the reference scenario, meat and F&V consumption were assumed similar over time per age and sex category. An autonomous trend in food consumption patterns was not included, as there was no conclusive evidence available about the changes in meat and F&V consumption in the Netherlands [
28,
55,
57]. In addition, whether the number of vegetarians/vegans would increase following an increase in price of meat-based products or a decrease in price of F&V could not be predicted. Thirdly, the health modelling was performed using categorization of food consumption. As the observed effects in food consumption were relatively small, the categorization might lead to an underestimation of the true effect on health when the change is not large enough to be placed in another consumption category, while any change in food patterns, even small ones, may have positive effects for health and environment at population level. Fourthly, the used SCBA framework uses monetization of effects in order to estimate societal benefits or costs, comparing a scenario with a reference scenario. In the SCBA, consumer surplus was one of the largest contributors to the balance in both the tax (as costs) and subsidy (as benefits) scenarios. Although utility euro’s and not financial euros, consumer surplus is to be considered in a SCBA according to the Dutch guidelines [
26]. Another weakness is that this study has only estimated effects on the food consumption side, and not the food production side. For the tax on meat-based products, increases in export following a lower domestic demand for meat products could offset some environmental gains in terms of national GHG emissions and may worsen the financial situation for some livestock farmers. If not implemented in Europe, Dutch consumers might also increase spending across border to avoid the higher prices. With the Danish fat tax, however, this effect was observed to be relatively small [
58].
The attention for healthiness and sustainability of food consumption grows [
59]. Information like derived in this SCBA could prove crucial to accurately estimate long-term effects of new policies. To improve assessment of societal effects of price interventions, real-life assessment of price changes could prove highly informative in addition to price elasticities of demand. Virtual supermarkets such as those studied by Waterlander et al. [
60,
61] can potentially increase knowledge about consumer behaviour following price changes, such as introduction of taxes and/or subsidies. Alternatively, discrete choice experiments (DCE) and/or willingness-to-pay (WTP) assessments could be used for studying healthier and more sustainable foods choices.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.