Experimental Design
The stated choice experiment is part of an online survey.
In a first step, respondents representative for the German population in terms of age, gender and region of living (on a federal state level) are screened for two criteria to have access to the survey: 1. being at least partially responsible for grocery shopping in own household; 2. buyers of pork meat. If respondents fulfill the screening criteria, they are randomly allocated to one of the two experimental groups (fresh meat "Schnitzel" treatment or processed meat "Würstchen" treatment) and to one of the two scenarios to see first ("No tax" scenario or "Tax" scenario). Thus, there are four combinations to which they can be assigned: 1. "Schnitzel" treatment + "No tax" scenario first; 2. "Schnitzel" treatment + "Tax" scenario first; 3. "Würstchen" treatment + "No tax" scenario first; 4. "Würstchen" treatment + "Tax" scenario first.
The stated choice experiment starts with an introduction of the (hypothetical) choice setting, namely that respondents should imagine to go grocery shopping for meat products (either 350g of fresh meat "Schnitzel" or 350g of processed meat "Würstchen") in a supermarket. Before they see the first choice set, respondents are provided with information on the three attributes. If they are in the "Tax" scenario, they are informed that all prices include an animal welfare consumption tax of 0.40 EUR per kg, i. e. 0.14 EUR per 350g.
Respondents receive twelve choice sets in total, grouped into two blocks, which are shown in random order: one block with six choice sets in a scenario without a tax and another block with six choice sets in which the price contains the animal welfare consumption tax.
After the stated choice experiment, debriefing questions on perception of consequentiality, attitudes towards animal farming and own moral obligations, knowledgeability about and perception of the state animal welfare label as well as socio-demographics follow.
The design is supposed to test the following hypotheses:
To measure consumers’ WTP for labels and for their underlying attributes, choice experiments are common methods. Labels have shown to increase WTP for certain (public good) attributes of food products, for example for certain husbandry conditions, carbon footprint, organic, health and nutritional facts, quality grades or a combination of those (e.g., Macdiarmid et al. 2021; Carlsson et al. 2020; Lusk et al. 2018; Gracia and de-Magistris 2016; Van Loo et al. 2014; Koistinen et al. 2013). Research also finds that the state is the preferred and credible institution to provide labels in the food context as opposed to retailers, industry organizations, NGOs and consumer groups (Gracia et al. 2011; Olynk et al. 2010; Vanhonacker et al. 2010). We thus hypothesize:
H1a: Respondents have a higher WTP for any level of the state animal welfare label compared to no label.
With regard to the three label levels, previous research on multi-level labels finds decreasing marginal WTP with increasing animal welfare levels, indicating demand for medium animal welfare conditions as opposed to only premium levels (Weinrich and Spiller 2016; De Jonge et al. 2015; Heerwagen et al. 2015). We hence expect similar findings, amending the first hypothesis to:
H1b: The increase in respondents' WTP is marginally decreasing from the lowest to the highest animal welfare level.
If the objective is to resolve the information asymmetry, labels should be used on all products impacting the public good of animal welfare, independent from their degree of processing. A rational agent should have the same WTP for animal welfare independent from the labeled product if it originates from the same animal. However, psychological research provides evidence that the processing level of the meat product makes a difference: The larger the dissociation of the meat product from the animal it stems from, the lower the empathy for the animal (Kunst and Hohle 2016; Hoogland et al. 2005). Following this logic, the dissociation from the animal should be larger in the case of sausages, a highly processed meat product, compared to fresh meat. We therefore divide participants into two treatment groups, the fresh meat "Schnitzel" treatment and the processed meat "Würstchen" treatment and test the following hypothesis:
H2: Respondents' WTP for any label level is higher in the "Schnitzel" treatment than in the "Würstchen" treatment.
While labeling could (partly) resolve information asymmetry, it does not solve the problem of free-riding (Lusk et al. 2007). Three previous studies explicitly question if such an externality in the case of animal welfare actually exist. Carlsson et al. (2007) and Tonsor et al. (2009) do not find a free-rider deficit in the case of battery cages in egg production and gestation crates for pigs respectively, while Uehleke and Hüttel (2019) detect free-riding behavior in their contingent valuation study. The political plans to introduce a consumption tax for animal welfare also raises the question if this introduction impacts or reveals free-riding behavior. As the tax is supposed to be imposed on every kilogram of meat sold, independent from its type and the animal welfare conditions under which it has been produced, it has to be paid by every consumer who decides to buy a meat product. Two reactions are possible, which we test confronting each participant with two scenarios, one without a tax and one with the animal welfare tax included in the price.
On the one hand, every individual now knows that others have to pay a tax, too. As the tax is earmarked to be paid out only to farmers who provide higher animal welfare standards, consumers automatically contribute to the public good of animal welfare when buying a meat product. This could lead to consumers being motivated to additionally contribute to animal welfare by buying labeled products and free-riding might be reduced. Their WTP for the animal welfare label would increase. The result would be similar to the effect found in Uehleke and Hüttel (2019) for the voting on product bans or Mulder and Zomer (2017) for providing information to respondents on market shares of animal welfare products.
On the other hand, people could perceive the payment of the consumption tax as some kind of indulgence. By paying the tax, they already contribute to the public good of animal welfare. Thus, there is no need to conduce further by buying products with labeled animal welfare. In this case, the tax would encourage free-riding behavior. Consumers’ WTP for the animal welfare label would decrease compared to the “No tax” scenario. Both directions are conceivable. We thus hypothesize:
H3: Respondents' WTP for any label level is different between the "Tax" treatment and the "No tax" treatment.