Consumers’ willingness to pay for pork produced with different levels of antibiotics

Abstract

1 Introduction

Since the 1940s, animal agriculture in Europe and the USA has been using antibiotic drugs in feed and drinking water at subtherapeutic levels to increase weight and improve feed efficiency (Brorsen et al. 2002; Lusk et al. 2006; McBride et al. 2008; Landers et al. 2012; Van Boeckel et al. 2015). Antibiotics in animal agriculture treat clinical diseases, prevent and control common diseases, and enhance animal growth¹ (NRC 1980; Landers et al. 2012). Globally, the quantity of antibiotics used in animal agriculture is already almost twice as high as that used by humans (Aaestrup 2012). Yet, global consumption of antibiotics in food-producing animals is projected to rise by 67 per cent, from 63,151 tons in 2010 to 105,596 tons in 2030 (Van Boeckel et al. 2014). Between 2010 and 2030, the use of antibiotics in animal agriculture in countries such as Brazil, Russia, India, China, and South Africa (BRICS countries) is expected to grow by 99 per cent. The growth is driven by the increasing demand for animal protein, and is seven times the projected population growth in the BRICS countries (Van Boeckel et al. 2015).

According to Monnier et al. (2018) and WHO (2012), the projected increase in the use of antibiotics in animal agriculture will have negative implications on livestock production, animal health, human health, and the overall economy. In light of this, studies have examined the cost and welfare implications of a ban on antibiotic use in animal agriculture in the USA on producers’ cost and net returns. They have found that a ban will negatively impact producer profitability (Brorsen et al. 2002; Miller et al. 2005; McBride et al. 2008; McBride and Key 2013; Dee et al. 2018). However, concerns regarding the development of antibiotic resistance led the US Federal Drug Administration (FDA) to restrict the use of all medically important antibiotics for growth promotion in food-producing animals. Under Guidance for Industry (GFI) #213 of the Veterinary Feed Directive (VFD), antibiotics that are important for human medicine can no longer be used for growth promotion or feed efficiency in cows, pigs, chickens, turkeys, and other food animals in the USA (AVMA 2018). In addition to the VFD restricting the use of medically important antibiotics for growth promotion purposes, it puts antibiotics used for disease prevention under the control of veterinarians. Therefore, the objectives of this study are two-fold. First, we investigate consumer willingness to pay (WTP) for pork labeled for antibiotic use. Second, we test whether WTP is affected by information on the use of antibiotics and the VFD.

Though pork production without antibiotic use is favorable for a number of reasons, research by Dee et al. (2018) highlighted the difficulties of raising livestock completely antibiotic-free under a significant disease challenge, such as porcine reproductive and respiratory syndrome virus. Instead, the authors argued for responsible use of antibiotics. Although antibiotic-free pork products are currently available in the market, no pork products are available that differentiate the use of antibiotics for disease treatment from disease prevention, or the conventional use of antibiotics from responsible use of antibiotics as outlined in Dee et al. (2018). This leads to the question whether consumers who lately have favored ‘free-from’ production (Grebitus et al. 2018) would prefer pork produced with minimal use of antibiotics over pork produced with standard antibiotic use, and how this would compare with pork produced without use of antibiotics. Differentiating meat products into more than two categories based on antibiotic-use levels offers producers and processors new market opportunities and caters to consumers who might be opposed to standard use of antibiotics but might be willing to purchase pork with minimum use of antibiotics. Given the implementation of the VFD and increasing consumer demand for reduced antibiotic use, it will be of value for livestock producers to know consumer WTP for meat from animals raised under the following production process categories: (1) raised without the use of antibiotics, (2) raised with antibiotic use for disease treatment only, and (3) raised with conventional use of antibiotics (pork from pigs that have received growth promotion levels of antibiotics).

While there is a growing body of literature that has examined consumers’ WTP for specific attributes of livestock production (McKendree et al. 2013; Lewis et al. 2017; Grunert et al. 2018), so far no study has examined US consumers’ WTP for meat produced from animals raised (1) completely antibiotic-free, (2) with antibiotic use for disease treatment purpose only, and (3) with conventional use of antibiotics for growth promotion. In this study, we address this gap in the literature by investigating the difference in US consumers’ WTP for pork chops produced with the aforementioned levels of antibiotic use. In addition, we estimate consumers’ WTP for growth hormone use and production method attributes (pasture-raised/confinement-raised). Furthermore, this study analyzes the effect of information related to non-therapeutic use of antibiotics and the VFD on WTP. To do so, we conducted a national online survey with 660 US consumers in 2019 where respondents were randomly split into control and information treatments. Results from discrete choice experiments show that WTP for antibiotic-free pork chops was significantly higher than WTP for conventional use of antibiotics in the control and information treatments. However, only in the information treatment was WTP for pork chops produced with minimal use of antibiotics significantly higher than WTP for conventional use of antibiotics. Our findings suggest that market outreach and communication efforts can be successful when pork products are labeled based on antibiotic-use level.

2 Background information

2.1 Use of antibiotics in livestock production and related regulations

The USA is the second largest user of antibiotics in food-producing animals (Van Boeckel et al. 2014). The non-therapeutic use of antibiotics in food-producing animals for growth promotion and disease prevention facilitates year-round large-scale production of livestock in close confinements in the USA. It contributes to lower production cost, productivity gains, and higher net returns for producers, and results in low meat prices for consumers (Brorsen et al. 2002; Lusk et al. 2006; Dee et al. 2018). Though the amount of medically important antimicrobials used in food-producing animals decreased from 62 per cent in 2015 to 51 per cent in 2017, the use of antibiotics is still high in US animal agriculture (USFDA 2016, 2018).

Using antibiotics in food production is linked to serious consequences for human and animal health. The use of antibiotic drugs increases selection pressure on bacteria to become resistant (Michael et al. 2014), and evidence links subtherapeutic use of antibiotics in food animals to antimicrobial resistance (Sipahi 2008; Marshall and Levy 2011). Given that many of the antibiotics used in animal agriculture are also used for treating infections in humans, there is concern that resistant bacteria may pass from food-producing animals to humans, e.g. through food handling and consumption (Landers et al. 2012; Telliant and Laxminarayan 2015). Due to the potential health consequences, antibiotic resistance has become a global concern. It can lead to financial burdens on households, strain national health care budgets, and entail economic consequences far beyond the health sector, such as restrictions on international travel and trade resulting from cross-border spread of resistant infections (WHO 2012; Monnier et al. 2018).

Investigating the use of antibiotics in pork production and related consumer WTP is of importance given the tremendous effect it has on human lives. Currently, at least 700,000 people die globally each year due to drug-resistant diseases (WHO 2019). In the USA, two million people are infected by antimicrobial-resistant pathogens every year, and an estimated 23,000 people die annually as a direct result of these infections (Golkar et al. 2014). Antimicrobial-resistant infections in the USA cost an estimated |${\$}$|20 billion annually in medical services, and |${\$}$|35 billion in lost productivity (Golkar et al. 2014; Lushniak 2014; Michael et al. 2014). The UN Ad hoc Interagency Coordinating Group on Antimicrobial Resistance warns that without taking drastic steps, drug-resistant diseases could cause 10 million deaths each year by 2050 (de Kraker et al. 2016; WHO 2019).

Concerned by the strong epidemiological evidence of an association between antibiotic use in food animals and antibiotic resistance in humans, many countries have restricted the use of antibiotics in livestock (Aaestrup 2012; Allen 2014; Sneeringer 2019). For example, Denmark, the largest exporter of pork, banned the use of antibiotics in finishing pigs and in nursery pigs in 1998 and 2000, respectively. Producers adjusted to the ban by making management changes, such as later weaning and a variety of dietary additions. Although there was an increase in the therapeutic use of antibiotics for a short period after the ban, the total use of antibiotics in animal production has decreased, with a reduction in antibiotic resistance in food-producing animals. Evidence also suggests that the ban did not harm Danish swine productivity long term (Aaestrup 2012; Levy 2014; Teillant and Laxminarayanan 2015). The EU, led by Nordic countries, banned the use of antibiotics for growth promotion in 2006. In 2018, the European Parliament adopted legislation that would limit the use of antibiotics to prevent disease (prophylactic use) to individual animals, and only in cases where a veterinarian identified a high risk of infection. As per new European legislation that will become law in 2022, treating a group of animals when only one shows signs of infection (metaphylactic use) should be a last resort (Dall 2018). The British Parliament prohibited the use of antibiotics at subtherapeutic level in animal feeds after a salmonella outbreak in the mid-1960s. Since then, proposals to ban antibiotic use in animal agriculture were also considered in the USA (Burbee et al. 1985; Lusk et al. 2006).

A recent meta-analysis of studies examining the association between the restriction in the use of antibiotics in food-producing animals and antibiotic resistance development in food-producing animals and humans showed that reducing veterinary antimicrobial use leads to a decrease in antimicrobial resistance development in animals and humans (Tang et al. 2017). Earlier studies that examined the cost and welfare implications of a ban on antibiotic use in animal agriculture in the USA found that a total ban on subtherapeutic use of antibiotics in animal agriculture would increase production costs and reduce producers’ net returns. Instead of a total ban, these studies suggested the importance of voluntary labeling as a mechanism to reduce the use of medically important antimicrobials in animal agriculture (Brorsen et al. 2002; Lusk et al. 2006). However, more recent USDA estimates regarding market-level effects of an antibiotics ban on the use of antibiotics for growth promotion in swine production showed more limited effects on production and wholesale price, and very small changes in consumer prices (Sneeringer, 2019).

Although the FDA restricted the use of some classes of antibiotics in food-producing animals in the 1990s, and the VFD Final Rule was placed into effect in fall 2015, it was only in early 2017 that the implementation of the Final Rule was completed. While the VFD eliminated subtherapeutic use of antibiotics for growth promotion, the use of antibiotics for the prophylactic purpose (disease prevention) is still possible. Many of the antibiotics approved for use in feed and drinking water are authorized for both growth promotion and disease prevention.

2.2 Consumer willingness to pay for livestock attributes

Several studies have estimated consumers’ WTP for livestock production attributes. For instance, Lusk et al. (2006) conducted a field experiment on consumer demand for a ban on antibiotic drug use in pork production and found that US consumers would be willing to pay a premium for antibiotic-friendly pork and would be willing to contribute money in support of a ban on subtherapeutic use of antibiotics in pork production. Olynk et al. (2010) found that US consumers would pay more for pork chops with USDA-verified antibiotic use. McKendree et al. (2013) found similar results for USDA-verified antibiotic use for smoked ham and ham lunchmeat. Tonsor et al. (2009) found a higher WTP for pork voluntarily produced without the use of gestation crates compared with pork produced with a ban on the use of gestation crates, though WTP for both was positive. Grunert et al. (2018) investigated consumer interest in modern pig production in Germany and Poland. They found that free mobility of the sow is the most important production attribute in Germany, whereas no microbial contamination is the most important production attribute in Poland.

Lusk et al. (2003) conducted a multicountry assessment of consumers’ WTP for beef steaks produced from cattle with no administration of growth hormones. They found a positive WTP for no administration of growth hormones in France, Germany, the UK, and the USA, respectively. Lewis et al. (2017) compared British and German consumers’ WTP for beef attributes and found significantly positive WTP for hormone-free beef in the UK and Germany. Syrengelas et al. (2018) examined US consumers’ WTP for beef steaks labeled as natural and found that consumers were willing to pay premiums for ‘natural’ beef and for beef produced with no use of antibiotics in the absence of a USDA definition of ‘natural’.

Findings from these studies show that consumers are interested in a number of production characteristics, and are willing to pay premiums when the meat they purchase is characterized by their preferred attributes. In particular, this review of the literature highlights consumers’ WTP for credence attributes related to production (see Supplementary Material 1 for an overview). We contribute to the literature by investigating consumers’ WTP for different usage levels of antibiotics in pork production. In addition, we take into account other production attributes, such as use of synthetic growth promoters and confinement- and pasture-raised production methods. We also test the effectiveness of antibiotic-use information on WTP. Our findings provide insights for US livestock producers, processors, and policymakers to develop production practices, marketing strategies, and labeling policies that mitigate antibiotic resistance development.

3 Methodological background

3.1 Design of the study

We conducted an online survey to collect data through an opt-in Internet panel maintained by the company Qualtrics. The survey was administered using the Qualtrics^XM survey platform. A total of 660 US consumers 18 years or older participated during June 2019. Respondents were prescreened by age, gender, ethnicity, and income using quotas to align with the Current Population Survey. Additional screening questions ensured that respondents were primary grocery shoppers, purchased pork in the past, and were meat eaters.

The study received Institutional Review Board approval from South Dakota State University. To examine the influence of information regarding the antibiotic-use level on WTP, we used a between-sample design where respondents were randomly assigned into an information or control treatment. Respondents in the control treatment of the survey received only experimental instructions (see Supplementary Material 2). This included, among others, the following definitions for antibiotic-use levels—antibiotic-free pork: pork from pigs raised with no use of antibiotics; minimal antibiotic-use pork: pork from pigs raised with antibiotics used only for treatment purposes; conventional antibiotic-use pork: pork from pigs that have received growth promotion levels of antibiotics; no information: no information on antibiotic use.

In the information treatment, respondents received experimental instructions, as well as additional information on the non-therapeutic use of antibiotics in pork production, current knowledge of its effect on antibiotic resistance development, and on the VFD (see Supplementary Material 3). In order to ensure that respondents would read the instructions carefully, we included timers. In the control group, participants had to stay at least 20 seconds on the page before being able to move on and in the information treatment they had to stay at least 45 seconds on the page before being able to move on. The timer in the treatment was longer given there was more text to read. However, we cannot say with absolute certainty whether or not participants read all of the provided information.

Specifically, respondents in the information treatment were shown the following:

Non-therapeutic antibiotic use in pork production

• As per the World Health Organization (WHO), non-therapeutic antibiotic use in pigs refers to the routine administration of antibiotics even if the pig does not pose any health defect. This administration of antibiotics contributes to the high growth rate and increases the feed conversion efficiency of feed thereby leading to lower production cost and lower pork prices.

• The repeated use of non-therapeutic antibiotics in pigs can result in antibiotic residues in pork meat; however, different research and evidence suggest that these residues have little to no human health threat consequences.

• Also, the administration of non-therapeutic antibiotics can result in antibiotic resistance, including multidrug resistance among a variety of bacterial pathogens that infect both humans and animals, which would increase the medical expenses and create more difficulty in treating the animal and human diseases.

Veterinary Feed Directive

• As per the Food and Drug Administration of the USA, effective from January 2017, all medically important antibiotics to be used in feed or water for food animals in the US require a Veterinary Feed Directive (VFD) or a prescription. The FDA uses the VFD as a policy framework for the judicious use of medically important antimicrobial drugs in food-producing animals.

A total of N = 328 respondents were in the control treatment, and N = 332 respondents were in the information treatment. Table 1 displays the sample characteristics. As shown in Table 1, key sociodemographic characteristics are comparable between the two treatments. Similar to the 2010 US Population Census Data, 52 per cent of our respondents were female and 68 per cent were white with a median income of |${\$}$|50,000–|${\$}$|74,999. Unlike the general population (33 per cent), the majority of the study's survey respondents had a college education (68 per cent) and were older, with a median age of 45–54 years (compared with the 37.2 years of the general population) (US Census Bureau 2010, 2015a,b).

3.2 Discrete choice experiment

We used a hypothetical online choice experiment to collect data on WTP toward different levels of antibiotic use in pork production. A discrete choice experiment (DCE) is a stated preference method used extensively in consumer economics to elicit WTP for various food attributes. The DCE simulates real-life purchasing situations, allows multiple attributes to be evaluated, allows hypothetical attributes (attributes not available in the market) to be included, offers respondents the option to choose among the alternatives or to opt out from the purchasing process, and allows researchers to estimate trade-offs among different alternatives (Lusk et al. 2006; Olynk et al. 2010; Syrengelas et al. 2018). In this study, we used a DCE to elicit US consumers’ WTP for several pork attributes.

Based on discussions with pork industry stakeholders and results from a pilot survey, we selected USDA-inspected boneless center-cut pork chops as the study product and selected the following attributes and levels for the DCE. We chose four price levels: |${\$}$|1.99/lb, |${\$}$|3.99/lb, |${\$}$|5.99/lb, and |${\$}$|8.99/lb. The price levels were selected based on grocery store prices for pork chops and USDA's online National Retail Report–Pork for boneless center-cut pork chops (USDA-AMS 2019). For instance, the average price for pork chops during the week of 26 April–2 May 2019 was |${\$}$|3.20/lb (USDA-AMS 2019). A recent study estimated that switching to antibiotic-free swine production will reduce US swine producers’ net income by a third (Dee et al. 2018). Hence, we included |${\$}$|8.99/lb as the highest price to account for the potential to transfer the increased cost of production from switching to antibiotic-free production to consumers.

As described earlier, respondents were provided definitions for antibiotic-use levels prior to completing the DCE. Complete elimination of antibiotics or reduced use of antibiotics might warrant changes in production practices and environmental control. In order to examine WTP for the antibiotic-use attribute relative to other production attributes, we included production method with the following levels: confinement, pasture raised, and no information, and use of synthetic growth promoters with the following levels: yes, no, and no information. Table 2 shows the attributes and attribute levels. Supplementary Material 2 contains the specific definitions of each attribute included in the study and the instructions given to the survey respondents.

We used Ngene software to generate a D-efficient design for the DCE using Bayesian priors. Parameter estimates from prior studies that were related to consumer WTP for meat attributes were used to select the priors in the first stage to design a DCE for a pilot survey. In stage two, priors were revised based on results from a pilot survey to design the final DCE. The use of Bayesian priors decreases the standard error of the estimates and increases the validity of design and choice situations (Sandor and Wedel 2001). Evidence from simulation studies shows that increasing the number of choice situations and the number of alternatives within a choice situation increases statistical efficiency (Vermeulen et al. 2010; Vanniyasingam et al. 2016). However, increasing choice situations may lead to response fatigue among participants. To address the issue of response fatigue, we used Ngene to generate a D-efficient design with 24 choice situations to be included in the DCE. The D-error was 0.32. The final choice sets of 24 were grouped randomly into two blocks of 12 to keep the choice selection task reasonable for individual participants (Tonsor et al. 2009; Olynk et al. 2010). Each participant was randomly presented with one of the two blocks of choice sets. Each choice set had three alternatives, Option A, Option B, and Neither. Option A and Option B presented boneless center-cut pork chops with varying attribute levels, and Neither represented the opt-out alternative, i.e. not to choose either product (Table 3).

To address the issue of lack of incentive compatibility in hypothetical stated preference methods where respondents may overstate the price that they are willing to pay, we included a cheap talk script for participants to read before answering the DCE questions. A cheap talk script makes respondents aware of the possibility of overstating WTP and requests that they answer true to their actual choices in the grocery store (Syrengelas et al. 2018).

3.3 WTP space model

The choice situation in a DCE presents competing goods in the form of alternatives where individuals have to make a decision based on their utility maximization. While making decisions, individuals make trade-offs between utilities obtained from different product attributes and attribute levels. The individual's utility is considered a random variable because the researcher has incomplete information about said utility. Hence, most empirical models using DCEs employ a random utility theoretical framework in the analysis.

Because each participant responded to 12 choice situations, we have panel data available for the analysis, where the cross-sectional element is the individual consumer i and the time-series component is the choice situation t (Louviere and Hensher 1983; Revelt and Train 1998; Tonsor et al. 2009). The between-subject design of our study produces a panel of stacked data that allows us to estimate variation in consumer utility and WTP for various pork attributes with and without information.

The standard approach in DCE literature is to estimate WTP in preference space where marginal utility coefficients are estimated first, followed by a calculation of WTP values. Instead of specifying the distribution of coefficients in the utility function and deriving the distribution of WTP, Train and Weeks (2005) propose to specify the distribution of WTP and derive the distribution of WTP coefficients directly (WTP space model). Since mutually compatible distributions for coefficients and WTP can be represented in either preference space or WTP space, the two approaches are equivalent. Given this, we first estimate a preference space model (see Supplementary Material 4) and then estimate a WTP space model to be able to compare findings. That said, our focus is on the WTP space model, which we describe in the following in detail.

Eq. 4 is called utility in WTP space where variation in WTP is independent of scale. The coefficients derived from Eq. 4 are readily interpretable as marginal WTP values (Gilmour et al. 2019).

All attribute levels are compared with the no-information option. The variable Price represents the price of pork chops in |${\$}$|/lb; Conventional is a dummy variable equal to 1 if the pork chop was labeled as ‘Conventional antibiotic-use’, and 0 otherwise; MinimalUse is a dummy variable equal to 1 if the pork chop was labeled as ‘Minimal-use antibiotics’, and 0 otherwise; AntibioticFree is a dummy variable equal to 1 if the pork chop was labeled as ‘Antibiotic-free’, and 0 otherwise; Pasture is a dummy variable equal to 1 if the pork chop was labeled as ‘Pasture-raised’, and 0 otherwise; Confinement is a dummy variable equal to 1 if the pork chop was labeled as ‘Confinement production’, and 0 otherwise; Synthetic is a dummy variable equal to 1 if the pork chop was labeled as ‘Synthetic growth promoter use’, and 0 otherwise; Nonsynthetic is a dummy variable equal to 1 if the pork chop was labeled as ‘No growth promoter use’, and 0 otherwise. Neither indicates that the participant chose to opt out, the variable being 1 if none of the pork chop alternatives were chosen, and 0 otherwise. All attributes except price are considered as normally distributed random variables.

In addition, we used Krinsky and Robb's method of parametric bootstrapping to calculate 95 per cent confidence intervals for the WTP estimations (Krinsky and Robb 1986). We also used confidence intervals to examine the statistical significance of the difference in WTP estimations between attribute levels within each production method. Non-overlapping of confidence intervals between attribute levels suggests that WTP estimates are statistically significantly different (Olynk et al. 2010).

We used the Poe test (Poe et al. 2005) to test for significant differences across the treatments for all attributes. The Poe test provides unbiased estimates of the significance of difference of two (control and treatment) distributions.

4 Empirical results

4.1 Willingness to pay for antibiotic-free pork

We estimated WTP space and preference space models. Results from the preference space and WTP space models were very similar. Hence, we report the results from the preference space models as Supplementary Material 4. Table 4 reports the estimates from the WTP space models for both the control and information treatments. We start by describing findings for WTP in the control treatment. Note that all findings are compared with the no-information option as indicated by the attribute levels in the choice-experimental design.²

Results for different labels of antibiotic use suggest that consumers are willing to pay, on average, |${\$}$|0.51 more for pork labeled for conventional use of antibiotics, |${\$}$|0.66 more for pork chops labeled for minimal use of antibiotics (therapeutic purpose), and |${\$}$|2.88 more for pork chops labeled as antibiotic-free. Examination of 95 per cent confidence intervals in Table 4 shows that the WTP estimate for antibiotic-free pork chops is statistically significantly different from WTP estimates of conventional use and minimal use.

With regard to other production attributes, findings indicate that consumers are willing to pay, on average, |${\$}$|2.33 more for pork labeled as pasture-raised production, and |${\$}$|1.16 more for pork chops carrying the label ‘no use of synthetic growth promoter’. However, consumers were having a negative WTP for confinement-based production practices and use of synthetic growth promoters. Consumers were willing to pay, on average, −|${\$}$|1.07 less for pork labeled as confinement production methods, and −|${\$}$|1.19 less for pork carrying the label ‘use of synthetic growth promoter’. Examining the confidence intervals for WTP estimates shows that the WTP estimate for pasture-raised production is significantly different from the confinement production method. Similarly, the WTP estimate for no use of synthetic growth promoter is significantly different from use of synthetic growth promoters.

Next, we discuss findings for WTP from the information treatment. Again, all findings are compared with the no-information option. When provided with information on antibiotic use and the VFD, consumers are willing to pay, on average, |${\$}$|0.51 more for pork labeled as ‘conventional use of antibiotics’, |${\$}$|1.27 more for pork labeled as ‘minimal use of antibiotics’ (therapeutic purpose), and |${\$}$|3.46 for pork labeled as antibiotic-free. Examination of confidence intervals of WTP estimates shows that unlike in the control treatment, there is a significant difference in WTP between antibiotic-free, minimal-use, and conventional-use attribute levels.

With regard to other production attributes, participants in the information treatment are willing to pay, on average, |${\$}$|2.38 more for pork labeled as pasture-raised production, and |${\$}$|1.19 more for pork carrying the label ‘no use of synthetic growth promoters’. Similarly to the control treatment, participants in the information treatment are also discounting the price for confinement-based production practices (−|${\$}$|1.29) and use of synthetic growth promoters (−|${\$}$|1.06). As in the control treatment, examination of confidence intervals for WTP estimates in the information treatment shows that the WTP estimate for the pasture-raised production method is significantly different from the confinement production method. Similarly, the WTP estimate for no use of synthetic growth promoters is significantly different from the one for use of synthetic growth promoters.

Table 5 presents the results from the Poe test. The P-values indicate whether the WTP for the attributes listed is significantly larger for the control treatment compared with the information treatment. In our analysis, the Poe test results show no significant differences in WTP estimates between the control and information treatments.

4.2 Pooled model

Next, we pooled the data from the control and information treatment and included interaction effects between each attribute variable and a treatment dummy variable that took the value of 1 if the data came from the treatment group, and 0 otherwise. Results are presented in Table 6. Findings show that only the interaction between antibiotic-free and the treatment dummy variable is significant at 5 per cent level, indicating that the WTP between the two groups differs. All other estimates are similar in sign, significance, and magnitude to the original models.

4.3 Discussion of results

In the control treatment (Table 4), WTP for pork chops was 1.3 times higher for pork produced with minimal antibiotic use and 5.6 times higher for antibiotic-free pork, compared with the WTP for conventional antibiotic use. In the information treatment, WTP for pork chops increased, being 2.5 times and 6.8 times higher for minimal antibiotic-use and antibiotic-free pork, respectively, compared with the WTP for conventional antibiotic use. This indicates that providing information on the different usage levels of antibiotics results in an even higher premium. While WTP for the conventional usage level was similar in the control and treatment group, providing information increased WTP for lower usage levels, i.e. antibiotic-free pork. This difference is also displayed in the pooled model. Overall, these results are in line with those from Lusk et al. (2006), who conducted non-hypothetical experiments involving grocery shoppers in Oklahoma and found that US consumers place a substantial premium (76.7 per cent) on pork produced without antibiotics. It is evident from Table 4 that providing information on the non-therapeutic use of antibiotics in pork production increased consumer WTP for pork chops from pigs raised with minimal use of antibiotics and from pigs raised without any antibiotics, compared with the control treatment. No other attributes listed in Table 4 displayed such a large difference in WTP estimates between the control and information treatments.

The importance of providing information becomes also evident when examining the confidence intervals in Table 4. The WTP estimate for antibiotic-free pork chops is significantly different from conventional-use antibiotic pork chops in both the control and information treatments. However, the WTP estimate for minimal-use antibiotic pork chops is only significantly different from pork chops produced with ‘conventional-use antibiotics’ in the information treatment. This means that producers who switch from conventional use of antibiotics to minimal use of antibiotics would have to promote this change and provide additional information to shoppers to capitalize on the change in usage level. Overall, these results from the information treatment imply that providing information on the non-therapeutic use of antibiotics in pork production and its relation to antibiotic resistance development has a positive effect on consumer WTP for meat products with reduced use of antibiotics, something that was not found for information on natural labels (Syrengelas et al. 2018).

While results from the Poe test (Table 5) did not produce statistically significant WTP estimates between control and information treatments with regard to providing additional information on antibiotic use and its relation to antibiotic resistance development and VFD, the results from the pooled model (Table 6) show that the information provision has a significant and positive effect on WTP for antibiotic-free pork chops.

Taken together, the results suggest that product differentiation and labeling based on antibiotic-use levels offer livestock producers an opportunity to mitigate the potential costs associated with reduced or no use of antibiotics in production. The higher WTP estimates for pork chops produced with minimal use of antibiotics and/or antibiotic-free pork chops in both treatments suggest a market potential for pork products that can be differentiated from those with prophylactic use of antibiotics. This offers an additional way to market these products in an environment where alternatives to antibiotics, such as prebiotics, probiotics, phytochemicals, and organic acids, are available for disease prevention and growth promotion in food-producing animals (Pew 2017; Sneeringer 2019). Given the commercial availability of alternatives to antibiotics, findings from our study imply that with the VFD in place, US pork producers could contribute to efforts to reduce the development of antibiotic resistance voluntarily. To do so, they could use labeling to differentiate their products with minimal use of antibiotics for treatment only from those produced conventionally. Therewith, they could then leverage consumer demand for meat with reduced use of antibiotics. The use of labeling might also increase their access to export markets that have stringent regulations on antibiotics for growth promotion in food-producing animals, such as the EU and Taiwan (Telliant and Laxminarayan 2015).

To facilitate such product differentiation in the market place, regulations and standards that define antibiotic-free meat production and therapeutic use (disease treatment) of antibiotics in meat production are needed. Although there has been some progress in redefining drug labels since the FDA Guidance for Industry #213 in 2013, more work needs to be done (Pew 2014) to create marketing opportunities based on antibiotic-use levels. Currently, there is no single definition for ‘antibiotic-free’ food labels in the USA, and the definition of antibiotic use for disease prevention is even less clear (Van Boeckel et al. 2015). The FDA considers uses that are associated with the treatment, control, and prevention of specific diseases to be therapeutic uses, implying that the FDA does not differentiate prophylactic use (prevention) from treatment-purpose use. Removing confusion from the definitions of ‘antibiotic-free meat’ and ‘therapeutic-use of antibiotics in food-producing animals’ could incentivize livestock producers and processors to leverage consumer demand for meat products with reduced use of antibiotics.

5 Conclusions and policy implications

Due to the non-therapeutic use of antibiotics in food-producing animals for growth promotion, the USA is the second largest consumer of antibiotics in animal agriculture. Because of the evidence on the linkage between antibiotic use and antibiotic resistance development, there is increasing consumer demand for antibiotic-free food production. Previous studies have shown that US consumers are willing to pay a premium for antibiotic-free meat. Nevertheless, there were concerns that the elimination of antibiotics for non-therapeutic purposes would increase production cost for US food animal producers and, hence, affect their profitability. Furthermore, there were concerns that this could increase consumer prices for meat. Therefore, the FDA enacted the VFD to eliminate the use of medically important antibiotics for growth promotion in food-producing animals. Given this, US livestock producers have the opportunity to leverage consumers’ increasing demand for reduced use of antibiotics in meat by not only advertising the complete elimination of the use of antibiotics, but also promoting the reduced use of antibiotics in food-producing animals for disease treatment purposes.

In this study, we used a national online survey of US pork consumers to estimate their WTP for antibiotic-free pork chops, pork chops with minimal use of antibiotics, and pork chops with conventional use of antibiotics. Results from the study showed a higher WTP for antibiotic-free pork chops, which was statistically significantly different from those produced with minimal use and conventional use of antibiotics in both control and information treatments. The study also showed a higher WTP for pork chops with minimal use of antibiotics in the control treatment. Furthermore, results indicated that providing information on different levels of use of antibiotics increased WTP for minimal use of antibiotics and antibiotic-free production even more.

Combining policy efforts to curb the development of antimicrobial resistance with market incentives has the potential to address producers’ concerns related to cost of regulations and consumers’ food safety concerns. Our findings support stakeholders in developing production practices, marketing strategies, and labeling policies that reduce antibiotic use in animal agriculture.

Acknowledgments

Deepthi Kolady acknowledges the financial support from the South Dakota State University Agricultural Experiment Station for conducting the survey.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Footnotes

References