- Email: [email protected]
Consumer preferences for pig welfare – Can the market accommodate more than one level of welfare pork?
Accepted Manuscript Consumer preferences for pig welfare – Can the market accommodate more than one level of welfare pork?
Sigrid Denver, Peter Sandøe, Tove Christensen PII: DOI: Reference:
S0309-1740(17)30241-3 doi: 10.1016/j.meatsci.2017.02.018 MESC 7192
To appear in:
Meat Science
Received date: Revised date: Accepted date:
23 December 2015 20 September 2016 20 February 2017
Please cite this article as: Sigrid Denver, Peter Sandøe, Tove Christensen , Consumer preferences for pig welfare – Can the market accommodate more than one level of welfare pork?. The address for the corresponding author was captured as affiliation for all authors. Please check if appropriate. Mesc(2017), doi: 10.1016/j.meatsci.2017.02.018
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Consumer preferences for pig welfare – can the market accommodate more than one level of welfare pork?
*
PT
Sigrid Denver*, Peter Sandøe* ** & Tove Christensen* Department of Food and Resource Economics, University of Copenhagen, Rolighedsvej 25,
Department of Large Animal Sciences, University of Copenhagen, Grønnegårdsvej 8, 1870
SC
**
RI
1958 Frederiksberg C, Denmark
NU
Frederiksberg C, Denmark
AC
CE
PT E
D
MA
E-mail corresponding author (Sigrid Denver): [email protected]
1
ACCEPTED MANUSCRIPT Abstract The purpose of the present paper is to investigate the market potential of pork labelled to indicate medium and high levels of animal welfare. The paper asks, in particular, whether there is a risk that Danish consumers will abandon high level welfare pork if less expensive products with a medium level of animal welfare became available. The study was based on an
PT
online questionnaire with a choice experiment involving 396 Danish respondents. The results
RI
indicated that the Danish market could accommodate more than one pork product with a
SC
welfare label but the price differential separating medium and high level animal welfare pork will have to be quite narrow. In addition, full willingness-to-pay of consumers who want to
NU
buy high level welfare pork cannot be relied upon to incentivise new consumers to buy medium welfare pork. Further, raising brand awareness in the shopping situation and
MA
improving consumer‟s understanding of brand attributes for high level welfare brands were
D
found to be vital.
PT E
Keywords: loose-housed sows; consumer demand; choice experiment; willingness-to-pay market driven animal welfare.
CE
1. Introduction
In EU member states and in other developed countries, the well-being of livestock is a
AC
serious issue for many people (Danish Agriculture & Food Council, 2016; Heerwagen, Mørkbak, Denver, Sandøe, & Christensen, 2015; Mayfield, Bennett, Tranter, & Wooldridge, 2007; Prickett, Norwood, & Lusk, 2010). As a consequence, there has been strong interest in identifying the market potential of specialty products focusing on animal welfare that exceed minimal requirements imposed by law (Christensen, Lawrence, Lund, Stott, & Sandøe, 2012; Heerwagen et al., 2015; The European Commission, 2006). Markets for such specialty products with welfare traits exceeding those in standard products would provide consumers with an opportunity to align their meat consumption with their concerns. 2
ACCEPTED MANUSCRIPT Several studies have pointed to the existence of three distinct consumer segments where preferences for food products with improved animal welfare are concerned: one segment is very interested in animal welfare issues, a second shows no particular interest in animal welfare, while the third segment, lying somewhere between these two, encompasses consumers who care about animal welfare to an extent but are also focused on other attributes,
PT
including gustatory quality, impacts on their own health, and price (Heerwagen et al., 2015;
RI
Meuwissen & Van Der Lans, 2005; Mørkbak, Christensen, & Gyrd-Hansen, 2010;
SC
Vanhonacker & Verbeke, 2014).
Kehlbacher, Bennett, and Balcombe (2012) investigated British consumer attitudes to a
NU
tiered welfare score indicating relative animal welfare levels on livestock products. They found that consumers were willing to increase their monthly meat expenditure by 26% and
MA
32% for meat produced according to the welfare scores “enhanced” and “excellent” respectively. The authors concluded that consumer willingness-to-pay (WTP) for improving
D
animal welfare scores was lower for higher levels of welfare which could indicate the
PT E
existence of a satiation level beyond which additional units of scored welfare had zero marginal utility for the consumers.
CE
In relation to pig welfare, it has commonly been found in stated preference studies that consumers have a positive attitude to increased welfare for pigs, and in particular outdoor
AC
production systems and systems allowing the pigs more space (Dransfield et al., 2005; Lassen, Sandøe, & Forkman, 2006; Mayfield et al., 2007; Mørkbak et al., 2010; Norwood & Lusk, 2011; Tawse, 2010; Taylor & Signal, 2009; Tonsor, Olynk, & Wolf, 2009; Tonsor & Wolf, 2011; Uzea, Hobbs, & Zhang, 2011). In addition, Uzea et al. (2011) found a positive WTP for group housing of sows, and Tonsor and Wolf (2011) found that consumers were willing to pay 20% higher prices for pork from production systems where sows were not
3
ACCEPTED MANUSCRIPT confined during gestation (pregnancy). Norwood and Lusk (2011) found that free-range husbandry was highly valued if accompanied by shelter and pasture. In Danish markets, three levels of animal welfare friendly pork can be identified. In the present study, medium level of animal welfare described production systems that guaranteed improved indoor conditions for sows or finisher pigs, while high level of animal welfare
PT
referred to production systems which, as a minimum, required outdoor sows and outdoor
RI
access for finisher pigs. Most pork in Denmark is sold as standard meat, which must meet the
SC
welfare requirements set in national legislation. Danish supermarkets have their own private labels for pork with medium levels of welfare. High levels are offered by so-called free-range
NU
and organic pork.
Not all stakeholders would agree that the marketing of medium level animal welfare
MA
products can be used to increase animal welfare. The main animal protection agency in Denmark, „Dyrenes Beskyttelse‟, sharing this scepticism, certifies animal-based food products
D
only where they guarantee high levels of animal welfare. The agency has expressed a concern
PT E
that animal products offering medium levels of animal welfare might poach market shares from the free-range or organic markets if consumers are unable to distinguish between
CE
different degrees of improved animal welfare (Brandt, 2014). The purpose of the present study was therefore to investigate the market potential of
AC
medium and high levels of welfare meat, with a particular focus on the risk that Danish consumers will abandon high level welfare pork if less expensive medium level products became available. The study focused on three dimensions of pig welfare: reducing the use of sow confinement, allowing sows outdoor access, and increasing space for finisher pigs. While there may be differences over the magnitude of the welfare benefits various stakeholders ascribe to less confinement, outdoor access and more space, there is little disagreement that
4
ACCEPTED MANUSCRIPT these factors improve welfare (EFSA, 2007; Humane Society International, 2104; Turner, Ewen, Rooke, & Edwards, 2000). 2. Materials and methods 2.1. Choice Experiment and perception of pig welfare
PT
A literature review and two focus group interviews were used to design a questionnaire, which was subsequently tested in a smaller group before the main survey was conducted. The
RI
questionnaire included questions regarding buying behaviour in relation to pork consumption,
SC
brand-knowledge, and perceptions of the importance of various welfare aspects in pig production. In addition, the questionnaire included a choice experiment where respondents
NU
were asked to trade-off between different animal welfare attributes of pork. By asking
MA
respondents to choose between different combinations of attributes it was possible to obtain an estimate of consumers‟ WTP for each individual product attribute (Hoyos, 2010).
D
Given the limited knowledge of pig welfare among those interviewed (Christensen, Tveit,
PT E
& Sandøe, 2014a), the hypothetical choice tasks in the experiment were deliberately kept simple. Accordingly, only two animal welfare attributes, together with a price vector, were included. Thereby, it was possible to identify the trade-off between the two animal welfare
CE
attributes as well as between animal welfare and price.
AC
Housing conditions for sows, including outdoor access, and additional space for finisher pigs were selected as welfare attributes in the choice experiment, because living in a natural environment, and in particular outdoor access and space, were mentioned repeatedly in the focus groups as well as in earlier literature as important aspects of animal welfare (Christensen et al., 2014a). Three levels of welfare condition for sows and finisher pigs were included in the analysis. An overview of all of the attributes used in the choice experiment is provided in Table 1. In the subsequent analysis, the lowest improvements of the two welfare attributes were categorized as the medium level welfare improvements on standard 5
ACCEPTED MANUSCRIPT production, and the largest improvements as the high level welfare improvements. All pork attribute choices were related to minced pork because that is the most popular cut of pork (Danish Agriculture & Food Council, 2014). The price vector was selected so that the lowest hypothetical price matched the lowest price seen for minced pork, while the highest price was
PT
set above the prices observed on the market.
RI
Please, insert Table 1 here
All respondents were provided with a fact sheet briefly explaining the production cycle in
SC
pork production with specific reference to the welfare attributes included in the experiment.
NU
Respondents were asked to act as though they were in a real shopping situation and prompted with a „cheap talk‟ script reminding them that they had the same amount of money available
MA
for food in the experiment as they had in real life, and that in experiments people tend to act differently from the way they do in real life. Inclusion of a cheap talk script has been shown
D
to reduce hypothetical bias (Cummings & Taylor, 1999).
PT E
Each choice set consisted of two hypothetical types of meat, Pork A and Pork B, with an additional option to choose Neither of these. The exact combinations of attribute levels in
CE
each choice set were determined using D-efficiency as a decision criterion. D-efficiency seeks to minimize the generalized variance of the parameter estimates of a pre-specified model
AC
(Vermeulen, Goos, Scarpa, & Vandebroek, 2011; Zwerina, 1997). A pilot study involving 100 respondents was used to identify a prior distribution of parameter values and for testing the time it took to complete the questionnaire. The combinations of attribute levels in the choice sets were designed with the software package Ngene 1.0. Attribute level swapping was used to avoid dominant alternatives (Zwerina, 1997). A design with D-error of 0.74 was considered acceptable and used in the main survey. The design involved nine choice sets which presented a feasible task to the respondents without inducing fatigue during the choice tasks (Bech et
6
ACCEPTED MANUSCRIPT al., 2011). It also helped to keep the overall response time for the questionnaire to approximately 15 minutes. Respondents were asked to complete the same nine choice sets, but the ordering of the sets was varied. One choice set is shown in Figure 1. Please, insert Figure 1 here
PT
2.2. Statistical method
RI
Data from the choice experiment were analysed in a mixed logit model with an error component (McFadden, 1974). In each choice set t, respondent n was assumed to choose the
SC
type of pork j that provided him/her with the highest utility. Assuming a linear description of
MA
NU
the systematic part of the utility, the utility Unjt was specified as follows:
An alternative specific constant α captured systematic preferences for not choosing any of captured the levels of the two animal welfare
D
the hypothetical alternatives. The term
PT E
attributes which were modelled as discrete variables to allow for different preferences for animal welfare improvements from standard to medium levels of welfare as opposed to
CE
preferences for improvements from medium to high levels of welfare. The random parameter was assumed to be normally distributed, which allowed for both positive and negative
AC
preferences for the animal welfare attributes and allowed for heterogeneity among respondents‟ marginal utilities for the animal welfare attributes. The term utility associated with price, where
was assumed to be a fixed parameter and
captured the was the
price vector. The price parameter was modelled as a continuous variable which implicated constant marginal disutility of costs across respondents. This seemed to be a realistic assumption as food costs do not significantly affect overall household expenditures. Furthermore, this assumption facilitated a straightforward estimation of respondents‟ WTP for
7
ACCEPTED MANUSCRIPT welfare attributes as -
. The term
was an error component with
specifying the individual specific zero-mean normally distributed error component that allows different correlations between unobservable components of the utility of the two hypothetical alternatives Pork A and Pork B, on the one hand, and between one of the hypothetical alternatives and the alternative Neither of these, on the other. The error term
was assumed
PT
to consist of independent and identically distributed random variables representing the
RI
unsystematic and unobserved element of individual n‟s choice (Hanley, Adamowicz, &
SC
Wright, 2005; Holmes & Adamowicz, 2003). Econometric estimation was performed using Nlogit 5 (Greene, 2010) and using 1,000 Halton draws to simulate the random parameter
NU
distributions.
In the statistical analysis, the respondents were divided into two groups. One group
MA
consisted of those who claimed to often or always buy high level welfare pork products in real shopping situations and the other consisted of respondents who usually bought standard or parameters for the two groups, the study could capture
D
medium pork. By allowing different
PT E
differences in WTP measures between the groups. 2.3. Data
CE
Consumer data on pork consumption and perceptions of pig welfare were obtained using
AC
a web-based questionnaire. The questionnaire was distributed to the online Danish consumer panel of the market research institute Userneeds in June 2013. The respondent sample was constructed to reflect the Danish population in respect of gender, age, and education. Two reinvitations were sent to respondents belonging to socio-demographic groups that tended to be underrepresented. The questionnaire was completed by 396 respondents, corresponding to a response rate of 15%. The sample consisted almost exclusively of people who were familiar with shopping decisions, in that 50% of the sample did all food shopping and 42% did part of the shopping. 8
ACCEPTED MANUSCRIPT The socio-demographic distributions of the resulting sample and the Danish population are shown in Table 2. Please, insert Table 2 here 3. Results
PT
3.1. Purchasing behaviour and knowledge of pork welfare attributes
RI
Two out of three respondents stated that they ate pork at least twice a week and only 4% of the respondents claimed never to eat pork. In addition, 14-35% of the respondents did not
SC
know how often they bought the various brands available on the Danish market. Further, one
NU
out of five respondents stated that they did not think at all about animal welfare while one out of five claimed to think a lot about animal welfare when buying minced pork. The
MA
respondents‟ knowledge of the pork welfare attributes of the following brands was tested: the brand „100% Danish‟, categorized as standard pork; the brands „Antonius‟, „Den Go‟e Gris‟,
D
and „Bornholmergrisen‟, categorized as medium level welfare pork; and the brands „Friland‟
PT E
(the Danish word for free-range) and „Organic‟ pork, categorized as high level welfare pork. Some of the brands were sold only in a limited number of retail chains, while others could be
CE
purchased more widely. Table 3 summarizes respondents‟ knowledge of the brands.
AC
Please, insert Table 3 here The results indicated a considerable lack of knowledge of the welfare attributes promised by the brands categorized as high welfare pork. For example, only around one third of the respondents associated the high level welfare pork with loose-housed sows, and only around a half associated the two high animal welfare brands with outdoor access. 3.2. The importance of loose-housed sows and more space for finisher pigs
9
ACCEPTED MANUSCRIPT The survey shed light on respondents‟ attitudes to a number of welfare-improving initiatives in pig production (not shown in table). More than half of the respondents considered it to be very positive that sows were always loose-housed, a quarter considered it to have small positive impact, and one respondent in ten did not care about this aspect of husbandry. Ranking the attributes according to the share of respondents who found that the
PT
attributes had great positive effect provided the following result: loose housing of sows (56%
RI
of respondents) was valued as important by less respondents than was more space for finisher
SC
pigs (73%), access to outdoor areas (70%), supply of straw (68%), short transport time to the slaughterhouse (67%), and the opportunity to socialize with other pigs (57%). On the other
NU
hand, the loose housing of sows was considered important by more respondents than organic production (36% of respondents) and the avoidance of tail docking (21%) and castration
MA
(8%).
Respondents who claimed often or always to buy high welfare pork in real shopping
D
situations, comprised 20% of the sample (N1=79). The remaining 80% never or seldom
PT E
bought high welfare pork (N2=317). Results from the choice experiment are shown in Table 4, which also shows marginal utilities reformulated as WTP.
CE
Please, insert Table 4 here
AC
The model fit measured by the McFaddden pseudo ρ2 was 0.29. As noted by McFadden (1978), ρ2 values between 0.2 and 0.4 represent acceptable fit. The confidence intervals of the estimates were rather large, indicating that there was substantial variation in the respondents‟ valuations of the attributes. Hence, allowing randomness in these parameters improved model fit. The standard deviations of the error components were insignificant, indicating that the choice between the two hypothetical packages of pork was not considered to differ in character from the choice between the hypothetical packages and Neither of these.
10
ACCEPTED MANUSCRIPT In the choice experiment, respondents who usually bought standard or medium pork were found to have a WTP for medium levels of the welfare attributes. However, they did not place additional value on sows being allowed outdoor access as compared with being allowed to be loose indoors. Neither did they value additional space for finishers from 30% to 100%. The average respondent who usually bought standard or medium pork claimed to be willing to pay
PT
a price premium of around € 1.6 per kg for minced pork from pigs born by sows which are
RI
loose-housed or roam outdoors; and around € 2.8 extra per kg for minced pork from pigs with
SC
more space. It is also important, though, to acknowledge the significant variations in the WTP represented by the large confidence intervals.
NU
By contrast, those respondents who usually bought high level welfare pork products showed a significantly stronger preference for the high levels of welfare than the medium
MA
levels in the experiment and had a generally stronger preference for pig welfare than the other
D
respondents.
PT E
4. Discussion
4.1. The Danish market for pork with different welfare levels The hypothetical welfare improvements used in the choice experiment broadly matched
CE
the range of pork products available on the Danish market categorized according to welfare
AC
levels (standard, medium, high). The estimated WTPs could therefore be used to assess marketing implications. Details of the three categories of pork on the Danish market are shown in Table 5, together with their estimated market shares and market prices. Please, insert Table 5 here 4.2. Estimated WTP for loose-housed sows and more space for finisher pigs When people were asked directly in the questionnaire, they valued more space for finisher pigs more highly than the housing of sows without confinement. A guarantee that sows are 11
ACCEPTED MANUSCRIPT loose-housed at all times seems to be a welfare attribute that was not assigned the highest priority by respondents. Nevertheless it was considered as a positive attribute which was ranked in the middle of the field compared with other welfare-improving conditions for pigs. In Table 6, price premiums based on estimated WTPs for hypothetical medium and high level welfare pork are compared with price premiums based on observed market prices. The
PT
group of respondents who usually bought standard or medium pork were on average willing
RI
to pay up to 80% more for pork with medium levels of the two tested welfare attributes.
SC
Members of the group of respondents who usually bought organic or free-range pork were on average willing to pay up to 170% more than the price of the standard product for pork
NU
offering medium levels of the two welfare attributes that were included in the experiment. Furthermore, they were on average willing to pay 15% more for high levels of the two
MA
welfare attributes, as compared with medium levels of these.
D
Please, insert Table 6 here
PT E
4.3. Putting the results in perspective
Let us for a moment assume that the estimated WTPs can be taken as true reflections of
CE
market behaviour. As the WTPs were estimated as normally distributed random variables, a mean WTP of 80% for the group of respondents usually buying standard or medium level
AC
welfare pork suggests that around half of these buyers would be willing to pay this price. In addition, half of the respondents usually buying high level welfare pork would be willing to pay a price premium of 15% for high relative to medium level welfare. The results therefore appear to suggest that, with price premiums for medium welfare pork of 75-80%, it should possible to attain a market share of around 50%, which is a considerably higher market share than the current of 12%. In addition, the market share for high level welfare pork should, at present price premiums, be able to attract around 10% of respondents.
12
ACCEPTED MANUSCRIPT A novel implication of the study is that the WTP for medium and high level welfare pork among the group of respondents who usually bought high level welfare pork could not be fully exploited. In the hypothetical setting they were willing to pay almost three times more for premium pork than standard pork. However, in any price adjustments designed to attract new consumers to medium welfare products the price premiums would be restricted by price
PT
differences of 80% between medium welfare and standard pork and 15% between medium
RI
and high level welfare pork. This insight was obtained by the categorization of respondents
SC
into two groups with different WTPs.
Further considerations need to be taken into account, however, before jumping to
NU
marketing conclusions. First, when interpreting the results, it should be kept in mind that both the medium and high welfare pork on the Danish market had welfare attributes which were
MA
not included in the choice experiment. On the other hand, the medium level welfare pork in the choice experiment assumed that sows were loose-housed at all times while the brands
D
offering medium level welfare pork on the market guaranteed 80% loose-housed sows.
PT E
Further, respondents in the choice experiment were likely to be better informed than consumers in real shopping situations. In addition, the elderly, those with higher levels of
CE
education, and women, were all found to be overrepresented in the sample relative to the Danish population as a whole. These consumers tend to be more concerned about animal
AC
welfare than the average consumer (Wolf, Tonsor, McKendree, Thomson, & Swanson, 2016). The income distribution of the sample is reasonably representative. However, 14% did not state their income which induces some uncertainty. Further, as the WTP estimates were obtained in a hypothetical setting, they are likely to differ from what would be observed at the meat counter (Fifer, Rose, & Greaves, 2014; Loomis, 2011). These factors point in opposite directions in terms of WTP bias, possibly leading to overestimation of the hypothetical WTP as compared with that in a real shopping situation.
13
ACCEPTED MANUSCRIPT Secondly, an informational problem concerning the animal welfare attributes of pork was identified in the market. This supports earlier findings by Kehlbacher et al. (2012), Taylor & Signal (2009) and Tawse (2010). In particular, the survey revealed a general uncertainty among respondents as to which brand of pork they actually bought as well as unawareness of the animal welfare requirements behind the labels for high level welfare pork (Table 3). This
PT
lack of knowledge could be due in part to the fact that not all brands were marketed in all
RI
shops, but still there seems to be a considerable lack of awareness of animal welfare
SC
parameters in actual shopping situations. If welfare differences between the various labels in a real market setting are not transparent, consumers will always tend to choose the cheapest
NU
product. The lack of awareness among Danish consumers regarding animal welfare in the shopping situation might shed light on the findings by Verbeke, Pérez-Cueto, de Barcellos,
MA
Krystallis, and Grunert (2010) who identified only a weak relationship between attitudes towards animal welfare and purchasing behaviour.
D
Hence, if efforts were to be made to expand market shares for pork with welfare traits
PT E
above the standard levels, it would seem to be crucial for the marketing of medium and high levels of welfare products that consumers are made aware of the differences between products
CE
and encouraged to become attentive to information set out on packaging and in labels when they choose their pork products. In particular, it is noteworthy that only around half of the
AC
respondents perceived the high level animal welfare products to guarantee outdoor access since this welfare trait is promoted heavily in the two brands considered. As the present study did not address whether detailed information on welfare attributes versus overall categories such as „enhanced‟ or „excellent‟ are most likely to attract consumers‟ attention, this would be an important topic for future studies. These insights support the findings in Kehlbacher et al. (2012) and expand them as the present study narrowed down the satiation level beyond which a group of respondents‟ WTP
14
ACCEPTED MANUSCRIPT for additional welfare improvements is zero. It may be wondered whether the results simply reflect a familiar pattern of decreasing marginal utility – in this case, of animal welfare – in line with standard assumptions in consumer economics. However, even though the vast majority of respondents involved in this study were partly or fully responsible for the daily shopping, there was a marked lack of awareness both of purchasing habits associated with the
PT
various labels and of the welfare characteristics of the brands. Hence the results could equally
RI
well represent a WTP for animal welfare that could be strengthened if consumers were better
SC
informed.
Both the hypothetical and the observed market prices indicated tolerance of a price
NU
premium of only 15% between medium and high level welfare pork. Hence, to develop the market for free-range and organic pork it would seem to be important to focus on how to
MA
increase consumer WTP for high level welfare pork. Future studies on consumer interest in the addition of further attributes to these pork products (such as larger improvements in
D
welfare attributes, and more focus on organic provenance, and on improved characteristics of
PT E
relevance to human health, convenience, locality, and so on) could be very useful in guiding further speculation about how to increase market shares for high level welfare products.
CE
As a final and more general observation, it should be emphasized that the demand side of the pork market is not represented and shaped by consumers alone. Retailers are increasingly
AC
powerful in framing consumer choice. Hence, an important question is whether – given consumer WTP for animal welfare – retailers have an interest in supporting markets for medium levels of animal welfare. 5. Conclusions As expected, respondents who usually bought high level welfare pork stated a higher WTP for animal welfare attributes in the hypothetical settings than other respondents. More interestingly, though, these respondents additionally displayed a WTP for high levels of 15
ACCEPTED MANUSCRIPT animal welfare as opposed to medium levels of animal welfare, albeit that it was a relatively small WTP. In contrast, the group of respondents who usually bought standard or medium level welfare pork showed no additional WTP for high level welfare products as compared with medium level welfare pork– even in the hypothetical settings of the choice experiment. These findings suggest that the market can accommodate three levels of welfare pork, and that
PT
the market share of medium level welfare pork could be increased considerably.
RI
However, the risk that Danish consumers will abandon free-range and organic pork if less
SC
expensive products with an animal welfare label become available is certainly real: this change in purchasing behaviour could well occur if the price differential separating medium
NU
and high level animal welfare pork is too great. Efforts to enlarge the market share of medium welfare pork also face another challenge. This is that, in attempting to incentivise new
MA
consumer groups to buy medium level welfare pork, it will not be possible to exploit the full WTP among consumers who generally buy high level welfare pork. The finding that the
D
respondents had a limited understanding of the animal welfare traits associated with high
PT E
level welfare products presently being sold and were often unable to recall which brands they bought suggests that measures designed to raise the visibility of welfare attributes will be as
AC
differentials.
CE
important in attracting consumers to medium or high level welfare products as price
Acknowledgements
Based on the project „Market driven animal welfare – the case of loose-housed sows 20122013‟ funded by the Danish Pig Levy Fund (http://www.svineafgiftsfonden.dk/). Additionally, the authors want to thank four anonymous referees for very constructive comments that have considerably improved the paper.
16
ACCEPTED MANUSCRIPT References Bech, M., Kjaer, T., & Lauridsen, J. (2011). Does the number of choice sets matter? Results from a web survey applying a discrete choice experiment. Health Econonomics, 20(3), 273-86. Brandt, I. (2014). So-called “welfare products” harm the sales of organic meat. Organic &
PT
Business 554 October 2014. [In Danish]. http://okologi.dk/media/635431/554-okt-l.pdf
RI
(accessed 09/09 2016).
SC
Christensen, T., Lawrence, A., Lund, M., Stott, A., & Sandøe, P. (2012). What can economists do to help improve animal welfare? Animal Welfare, 21(S1), 1–10.
NU
Christensen, T., Tveit, G., & Sandøe, P. (2014a). Loose-housed sows – a cross-disciplinary investigation on market driven animal welfare. Project report 19. CEBRA - Centre for
MA
Biosafety and Risk Assessment, Copenhagen. [In Danish]. Christensen, T., Denver, S., Hansen, H.O., Lassen, J., & Sandøe, P. (2014b). Animal welfare
D
labels – comparing experiences from six EU countries. Commissioned Work 2014/10,
English summary].
PT E
Institute of Food and Resource Economics, University of Copenhagen. [In Danish with
CE
Cummings, R.G. & Taylor, L.O. (1999). Unbiased value estimates for environmental goods: a cheap talk design for the contingent valuation method. American Economic Review,
AC
89(3), 649–666.
Danish Agriculture & Food Council (2014). Consumption of pork, beef and chicken meat – available, bought or eaten? Danish Agriculture & Food Council, August 2014. [In Danish]. Danish Agriculture & Food Council (2016). Consumers choose Danish when they want good meat. Market analysis from Danish Agriculture & Food Council 06/24/2016 [In Danish].
17
ACCEPTED MANUSCRIPT Dransfield, E., Ngapo, T. M., Nielsen, N. A., Bredahl, L., Sjödén, P. O., Magnusson, M., Campo, M. M., & Nute, G. R. (2005). Consumer choice and suggested price for pork as influenced by its appearance, taste and information concerning country of origin and organic production. Meat Science, 69(1), 61-70. EFSA (2007). Animal health and welfare aspects of different housing and husbandry systems
PT
for adult breeding boars, pregnant, farrowing sows and unweaned piglets. Scientific
RI
Opinion of the Panel on Animal Health and Welfare. The EFSA Journal, 572, 1-13.
SC
Fifer, S., Rose, J., & Greaves, S. (2014). Hypothetical bias in Stated Choice Experiments: Is it a problem? And if so, how do we deal with it? Transportation Research Part A, 61,
NU
164–177.
Greene, W. (2010). NLogit Version 4.0. Econometric software Inc.
MA
http://pages.stern.nyu.edu/~wgreene/DiscreteChoice/Software/NLOGIT-Short-StudentManual.pdf (accessed 09/09/2016).
PT E
D
Hanley, N., Adamowicz, W., & Wright, R. E. (2005). Price vector effects in choice experiments: an empirical test. Resource and Energy Economics, 27, 227–234. Heerwagen, L. R., Mørkbak, M. R., Denver, S., Sandøe, P., & Christensen, T. (2015). The
CE
role of quality labels in market-driven animal welfare. Journal of Agricultural &
AC
Environmental Ethics, 28, 67-84. Holmes, T. P., & Adamowicz, W. (2003). Feature based methods. In: Champ, P.A., Boyle, K.J., Brown, T.C. (Eds.), A primer on nonmarket valuation. Kluwer Academic Publishers. Hoyos, D. (2010). The state of the art of environmental valuation with discrete choice experiments. Ecological Economics, 69, 1595–1603. Humane Society International (2014). An HSI Report: The Welfare of Animals in the Pig Industry. http://www.hsi.org/assets/pdfs/hsi-fa-whitepapers/welfare_of_animals_in_the_pig.pdf (accessed 09/09/2016). 18
ACCEPTED MANUSCRIPT Kehlbacher, A., Bennett, R., & Balcombe, K. (2012). Measuring the consumer benefits of improving farm animal welfare to inform welfare labelling. Food Policy, 37(6), 627-633. Lassen, J., Sandøe, P., & Forkman, B. (2006). Happy pigs are dirty! – conflicting perspectives on animal welfare. Livestock Science, 103(3), 221–230. List, J. A., & Gallet, G. A. (2001). What Experimental Protocol Influence Disparities
PT
Between Actual and Hypothetical Stated Values? Evidence from a Meta-Analysis.
RI
Environmental and Resource Economics, 20, 241–254.
SC
Loomis, J. B. (2011). What's to Know About Hypothetical Bias in Stated Preference Valuation Studies? Journal of Economic Surveys, 25(2), 363-370.
NU
Mayfield, L. E., Bennett, R. M., Tranter, R. B., & Wooldridge, M. J. (2007). Consumption of welfare-friendly food products in Great Britain, Italy and Sweden, and how it may be
MA
influenced by consumer attitudes to, and behaviour towards, animal welfare attributes. International Journal of Food and Agriculture, 15(23), 59-73.
D
McFadden, D. (1974). Conditional Logit Analysis of Qualitative Choice Behavior. In P.
PT E
Zarembka, (Ed.), Frontiers in Econometrics (pp. 105-142). New York: Academic, New Press.
CE
McFadden, D. (1978). Quantitative methods for analyzing travel behaviour in individuals: Some recent developments. In D. Hensher, & P. Stopher (Eds.), Behavioural Travel
AC
Modelling (pp. 279-218). London: Croom Helm. Meuwissen, M. P. M., & Van Der Lans, I. A. (2005). Trade-offs between consumer concerns: An application for pork supply chains. Food Economics - Acta Agriculturae Scandinavica, Section, C2(1), 27-34. Mørkbak, M. R., Christensen, T., & Gyrd-Hansen, D. (2010). Choke price bias in choice experiments. Environmental & Resource Economics, 45(4), 537-551.
19
ACCEPTED MANUSCRIPT Norwood, F.B., & Lusk, J.L. (2011). A calibrated auction-conjoint valuation method: Valuing pork and eggs produced under differing animal welfare conditions. Journal of Environmental Economics and Management, 62, 80–94. Prickett, R. W., Norwood, F. B., & Lusk, J. L. (2010). Consumer preferences for farm animal welfare: results from a telephone survey of US households. Animal Welfare, 19, 335-347.
PT
Statistics Denmark (2013). Statistical Yearbook 2013. Statistics Denmark. [In Danish]
RI
http://www.dst.dk/pukora/epub/upload/17956/saa2013.pdf (accessed 09/09 2016).
SC
Tawse, J. (2010). Consumer attitudes towards farm animals and their welfare: a pig production case study. Bioscience Horizons, 3(2), 156-165.
NU
Taylor, N., & Signal, T. D. (2009). Willingness to pay: Australian consumers and “on the farm” welfare. Journal of applied animal welfare science: JAAWS, 12(4), 345-59.
MA
The European Commission (2006). Report on a Community Action Plan on the Protection and Welfare of Animals 2006–2010 (2006/2046(INI), 19/09/2006). Committee on
D
Agriculture and Rural Development 19/09/2006.
PT E
Tonsor, G. T., Olynk, N., & Wolf, C. (2009). Consumer Preferences for Animal Welfare tributes: The Case of Gestation Crates. Animal Welfare, 3, 713-730.
CE
Tonsor, G. T., & Wolf, C. (2011). On mandatory labelling of animal welfare attributes. Food Policy, 36, 430–437.
AC
Turner, S. P., Ewen, M., Rooke, J. A., & Edwards, S. A. (2000). The effect of space allowance on performance, aggression and immune competence of growing pigs housed on straw deep-litter at different group size. Livestock Production Science, 66(1), 47–55. Uzea, A. D., Hobbs, J. E., & Zhang, J. (2011), Activists and Animal Welfare: Quality Verifications in the Canadian Pork Sector. Journal of Agricultural Economics, 62, 281– 304.
20
ACCEPTED MANUSCRIPT Vanhonacker, F., & Verbeke, W. (2014). Public and Consumer Policies for Higher Welfare Food Products: Challenges and Opportunities. Journal of Agricultural and Environmental Ethics, 27(1), 153-171. Vermeulen, B., Goos, P., Scarpa, R. & Vandebroek, M. (2011). Bayesian Conjoint Choice Designs for Measuring Willingness to Pay. Environmental Resource Economics, 48,129–
PT
149.
RI
Verbeke, W., Pérez-Cueto, F.J.A., de Barcellos, M.D., Krystallis, A., & Grunert, K.G. (2010).
SC
European citizen and consumer attitudes and preferences regarding beef and pork. Meat Science, 84, 284–292.
NU
Wolf, C.A., Tonsor, G.T., McKendree, M.G.S., Thomson, D.U., Swanson, J.C. (2016). Public and farmer perceptions of dairy cattle welfare in the United States. Journal of Dairy
MA
Science, 99(7), 5892-5903.
Zwerina, K. (1997). Discrete Choice Experiments in Marketing: Use of Priors in Efficient
D
Choice Designs and Their Application to Individual Preference Measurement. Berlin
AC
CE
PT E
Heidelberg: Springer Verlag (Chapter 4).
21
ACCEPTED MANUSCRIPT
Pork A
Pork B
Current legislation for sows
Sows are always loose with outdoor access
Finisher pigs have 100% more space Finisher pigs have 30 % more space € 7/kg of minced pork
Which of the products do you prefer?
○ Neither of these
SC
○ Pork B
NU
○ Pork A
RI
PT
€ 5.4/kg of minced pork
Fig 1 An example of a choice set used in estimation of willingness-to-pay for pork welfare
AC
CE
PT E
D
MA
attributes.
22
ACCEPTED MANUSCRIPT Table 1 Overview of attributes and levels included in the choice experiment Animal Label in choice Attributes
welfare
Description experiment
level Standard
Current legislation for sows
PT
Housing
Sows are loose during gestation Indoor loose-housed sows at all
Medium
Sows are always loose
times
sows
Sows are always loose with
Loose-housed sows at all times
SC
High
RI
conditions for
outdoor access
with opportunity to go outdoors
for finisher
MA
for finisher pigs
Indoor space
NU
Current legislation of space Standard
Minimum 0.65 m2 per pig
Finisher pigs have 30% Medium
Minimum 0.85 m2 per pig
more space
pigs
D
Finisher pigs have 100% Minimum 1.3 m2 per pig
PT E
High
more space Four prices of minced pork
Market observations were used
€/kg: 5.4; 7.0; 10.2; 16.1
to identify a realistic price vector
CE
N/A
AC
Price
23
ACCEPTED MANUSCRIPT Table 2 Socio-demographic distribution of respondents regarding gender, age, education and income (in %, N=396).
Variable
Respondent
National
distribution
distributiona
Female
55
50
Male
45
50
18-34
15
30
Levels
RI
35-49
24
SC
Age (years)
PT
Gender
50-70
31
61
39
21
27-37
44
38-48
34
19-29
11
21
26,810-80,430
47
51
> 80,430
28
27
No answer
14
N/A
NU
Primary school (10 years schooling)
High school level (additional 3 years) Education
MA
University level (additional 2 or more
< 26,810
PT E
Household income, before
CE
tax (€/year)
Information on national distribution is based on Statistics Denmark (2013).
AC
a
D
years)
24
ACCEPTED MANUSCRIPT Table 3 Respondents‟ knowledge of pork brands (in % of sample, N=396). Animal welfare
Standard
Medium Den Go’e
100% Pork brand
High Born-
Antonius Danish
Friland
Organic
40
32
6
57
44
7
25
40
Gris
holmer
4
Sows always loose2
5
3
3
7
4
4
15
PT
housed
Pigs
have
SC
access
RI
Pigs have outdoor
more
9
NU
space indoor
Grey shading indicates that requirements were not guaranteed by the label and that answers
MA
thereby were incorrect. The exact wording of the question to respondents was „Which attributes do you particularly associate with the following brands of pork. Several answers per
AC
CE
PT E
D
brand can be stated?
25
ACCEPTED MANUSCRIPT Table 4 Model results and estimated price premiums based on respondents‟ willingness-topay (WTP) for welfare attributes in pork (N=396). Estimated Variable
parameters Estimated WTP price
Mean (95% C.I.)
premiums €/kg
[std. dev. (95% C.I.)]a
Mean (95% C.I.)b
PT
Respondents usually buying standard or medium level welfare pork (normally distributed
0.71 (0.42, 0.99)
1.6 (1.0, 2.2)
SC
Sows loose-housed indoors
RI
random parameters)
[0.88 (0.60, 1.16)]
NU
0.71 (0.36, 1.05) Sows loose-housed and outdoors
1.6 (0.8, 2.4)
MA
[1.60 (1.24, 1.96)] 1.19 (0.89, 1.48)
Pigs 30% more space indoors
2.8 (2.0, 3.4)
PT E
Pigs 100% more space indoors
D
[0.81 (0.40, 1.22)] 1.25 (0.86, 1.64) 2.8 (2.0, 3.8)
[1.44 (1.17, 1.72)]
Sows loose-housed indoors
CE
1.11 (0.51, 1.71)
4.2 (2.2, 6.2)
Sows loose-housed and outdoors
1.50 (0.70, 2.29)
5.2 (2.4, 7.8)
AC
Respondents usually buying high level welfare pork (non-random parameters)
Pigs 30% more space indoors
0.84 (0.23, 1.46)
4.8 (2.6, 6.8)
Pigs 100% more space indoors
1.45 (0.75, 2.15)
6.2 (3.8, 8.8)
Non-random parameters, all respondents Price
-0.12 (-0.13,-0.11)
Alternative specific constant
-5.28 (-5.84, -4.73)
Std. dev. of error component, σ
2.66 (-9.55, 14.86)
a
All parameter estimates were significant at 0.01% level except σ which was not significant. 26
ACCEPTED MANUSCRIPT b
The confidence intervals of WTP estimates were calculated by dividing marginal utility
AC
CE
PT E
D
MA
NU
SC
RI
PT
confidence levels by the price parameter.
27
ACCEPTED MANUSCRIPT Table 5 Estimated market shares and market prices for minced pork in Denmark categorized according to animal welfare levels (data from 2015). Type
Market a
share
Market price (€/kg)
Description
b
Requirements in accordance with the Danish
PT
legislation:
Sows are loose-housed 60% of the time 84%
5.4 - 8.0 c
RI
Standard
SC
Finisher pigs have 0.65 m2 Tale docking if necessary
NU
Age at weaning at least 21 days Sows are loose-housed 80% of the time
Medium
12%
MA
Finisher pigs have 30% more space
9.4
No tale docking
D
Age at weaning at least 28 days
PT E
High 2%
Sows are 100% loose with outdoor access
10.7
Finisher pigs have 100% more space in the
Organic
AC
CE
Free-range
Plus a few other requirementsd
2%
stables with access to an outdoor veranda No tale docking
13.2 Age at weaning at least 30 days Plus other requirementse
a
Market shares from Christensen et al. (2014b).
b
Market prices were Googled in November 2015; various homepages for online shopping and
supermarket advertisements were consulted. Dyrenes Beskyttelse and Danish Crown (the
28
ACCEPTED MANUSCRIPT largest abattoir in Denmark) provided valuable information via e-mail correspondence (September 2015) which helped to confirm the estimated market prices. c
The standard price of minced pork was € 8/kg, but standard minced pork was routinely sold
at a discount price of € 5.4/kg. d
These involve more straw, no dispensation for a max of 8 hours transportation time to
Organic pig production has additional requirements, as compared with free-range
RI
e
PT
slaughter, no use of electrified stick at slaughterhouse.
SC
production, in terms of later weaning, more space for sows and finisher pigs, and organic
AC
CE
PT E
D
MA
NU
feed.
29
ACCEPTED MANUSCRIPT Table 6 Market observations and model-results for willingness-to-pay (WTP) for welfare pork (N=396). Stated WTP for welfare porkb
Market price
Respondents usually
standard or medium level
buying high level
welfare pork
welfare pork
0%
Medium level
Base (WTP not estimated)
RI
Standard
Respondents usually buying
PT
premiumsa
Product
17% - 75%
80% higher higher
SC
relative to standard High level relative
0% higher
NU
14% higher to medium
15% higher
Estimates of market price premiums in per cent were based on values from Table 5. Price
MA
a
170% higher
premium for medium relative to standard were specified as an interval based on discount price
Price premiums for stated WTP were based on results in Table 4. Example: estimated WTP
PT E
b
D
and standard price.
price premium for two medium levels of animal welfare relative to the lowest price level used
AC
CE
in the choice experiment: (€ 1.6 + € 2.6)/ € 5.4 = 80%.
30
ACCEPTED MANUSCRIPT Highlights Many consumers will not pay for improving animal welfare beyond the medium level
Small difference between willingness-to-pay for medium and high level welfare pork
Full willingness-to-pay for high level welfare pork cannot be exploited
Increasing brand awareness and consumers‟ knowledge of brands is vital
AC
CE
PT E
D
MA
NU
SC
RI
PT
31
Sigrid Denver, Peter Sandøe, Tove Christensen PII: DOI: Reference:
S0309-1740(17)30241-3 doi: 10.1016/j.meatsci.2017.02.018 MESC 7192
To appear in:
Meat Science
Received date: Revised date: Accepted date:
23 December 2015 20 September 2016 20 February 2017
Please cite this article as: Sigrid Denver, Peter Sandøe, Tove Christensen , Consumer preferences for pig welfare – Can the market accommodate more than one level of welfare pork?. The address for the corresponding author was captured as affiliation for all authors. Please check if appropriate. Mesc(2017), doi: 10.1016/j.meatsci.2017.02.018
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Consumer preferences for pig welfare – can the market accommodate more than one level of welfare pork?
*
PT
Sigrid Denver*, Peter Sandøe* ** & Tove Christensen* Department of Food and Resource Economics, University of Copenhagen, Rolighedsvej 25,
Department of Large Animal Sciences, University of Copenhagen, Grønnegårdsvej 8, 1870
SC
**
RI
1958 Frederiksberg C, Denmark
NU
Frederiksberg C, Denmark
AC
CE
PT E
D
MA
E-mail corresponding author (Sigrid Denver): [email protected]
1
ACCEPTED MANUSCRIPT Abstract The purpose of the present paper is to investigate the market potential of pork labelled to indicate medium and high levels of animal welfare. The paper asks, in particular, whether there is a risk that Danish consumers will abandon high level welfare pork if less expensive products with a medium level of animal welfare became available. The study was based on an
PT
online questionnaire with a choice experiment involving 396 Danish respondents. The results
RI
indicated that the Danish market could accommodate more than one pork product with a
SC
welfare label but the price differential separating medium and high level animal welfare pork will have to be quite narrow. In addition, full willingness-to-pay of consumers who want to
NU
buy high level welfare pork cannot be relied upon to incentivise new consumers to buy medium welfare pork. Further, raising brand awareness in the shopping situation and
MA
improving consumer‟s understanding of brand attributes for high level welfare brands were
D
found to be vital.
PT E
Keywords: loose-housed sows; consumer demand; choice experiment; willingness-to-pay market driven animal welfare.
CE
1. Introduction
In EU member states and in other developed countries, the well-being of livestock is a
AC
serious issue for many people (Danish Agriculture & Food Council, 2016; Heerwagen, Mørkbak, Denver, Sandøe, & Christensen, 2015; Mayfield, Bennett, Tranter, & Wooldridge, 2007; Prickett, Norwood, & Lusk, 2010). As a consequence, there has been strong interest in identifying the market potential of specialty products focusing on animal welfare that exceed minimal requirements imposed by law (Christensen, Lawrence, Lund, Stott, & Sandøe, 2012; Heerwagen et al., 2015; The European Commission, 2006). Markets for such specialty products with welfare traits exceeding those in standard products would provide consumers with an opportunity to align their meat consumption with their concerns. 2
ACCEPTED MANUSCRIPT Several studies have pointed to the existence of three distinct consumer segments where preferences for food products with improved animal welfare are concerned: one segment is very interested in animal welfare issues, a second shows no particular interest in animal welfare, while the third segment, lying somewhere between these two, encompasses consumers who care about animal welfare to an extent but are also focused on other attributes,
PT
including gustatory quality, impacts on their own health, and price (Heerwagen et al., 2015;
RI
Meuwissen & Van Der Lans, 2005; Mørkbak, Christensen, & Gyrd-Hansen, 2010;
SC
Vanhonacker & Verbeke, 2014).
Kehlbacher, Bennett, and Balcombe (2012) investigated British consumer attitudes to a
NU
tiered welfare score indicating relative animal welfare levels on livestock products. They found that consumers were willing to increase their monthly meat expenditure by 26% and
MA
32% for meat produced according to the welfare scores “enhanced” and “excellent” respectively. The authors concluded that consumer willingness-to-pay (WTP) for improving
D
animal welfare scores was lower for higher levels of welfare which could indicate the
PT E
existence of a satiation level beyond which additional units of scored welfare had zero marginal utility for the consumers.
CE
In relation to pig welfare, it has commonly been found in stated preference studies that consumers have a positive attitude to increased welfare for pigs, and in particular outdoor
AC
production systems and systems allowing the pigs more space (Dransfield et al., 2005; Lassen, Sandøe, & Forkman, 2006; Mayfield et al., 2007; Mørkbak et al., 2010; Norwood & Lusk, 2011; Tawse, 2010; Taylor & Signal, 2009; Tonsor, Olynk, & Wolf, 2009; Tonsor & Wolf, 2011; Uzea, Hobbs, & Zhang, 2011). In addition, Uzea et al. (2011) found a positive WTP for group housing of sows, and Tonsor and Wolf (2011) found that consumers were willing to pay 20% higher prices for pork from production systems where sows were not
3
ACCEPTED MANUSCRIPT confined during gestation (pregnancy). Norwood and Lusk (2011) found that free-range husbandry was highly valued if accompanied by shelter and pasture. In Danish markets, three levels of animal welfare friendly pork can be identified. In the present study, medium level of animal welfare described production systems that guaranteed improved indoor conditions for sows or finisher pigs, while high level of animal welfare
PT
referred to production systems which, as a minimum, required outdoor sows and outdoor
RI
access for finisher pigs. Most pork in Denmark is sold as standard meat, which must meet the
SC
welfare requirements set in national legislation. Danish supermarkets have their own private labels for pork with medium levels of welfare. High levels are offered by so-called free-range
NU
and organic pork.
Not all stakeholders would agree that the marketing of medium level animal welfare
MA
products can be used to increase animal welfare. The main animal protection agency in Denmark, „Dyrenes Beskyttelse‟, sharing this scepticism, certifies animal-based food products
D
only where they guarantee high levels of animal welfare. The agency has expressed a concern
PT E
that animal products offering medium levels of animal welfare might poach market shares from the free-range or organic markets if consumers are unable to distinguish between
CE
different degrees of improved animal welfare (Brandt, 2014). The purpose of the present study was therefore to investigate the market potential of
AC
medium and high levels of welfare meat, with a particular focus on the risk that Danish consumers will abandon high level welfare pork if less expensive medium level products became available. The study focused on three dimensions of pig welfare: reducing the use of sow confinement, allowing sows outdoor access, and increasing space for finisher pigs. While there may be differences over the magnitude of the welfare benefits various stakeholders ascribe to less confinement, outdoor access and more space, there is little disagreement that
4
ACCEPTED MANUSCRIPT these factors improve welfare (EFSA, 2007; Humane Society International, 2104; Turner, Ewen, Rooke, & Edwards, 2000). 2. Materials and methods 2.1. Choice Experiment and perception of pig welfare
PT
A literature review and two focus group interviews were used to design a questionnaire, which was subsequently tested in a smaller group before the main survey was conducted. The
RI
questionnaire included questions regarding buying behaviour in relation to pork consumption,
SC
brand-knowledge, and perceptions of the importance of various welfare aspects in pig production. In addition, the questionnaire included a choice experiment where respondents
NU
were asked to trade-off between different animal welfare attributes of pork. By asking
MA
respondents to choose between different combinations of attributes it was possible to obtain an estimate of consumers‟ WTP for each individual product attribute (Hoyos, 2010).
D
Given the limited knowledge of pig welfare among those interviewed (Christensen, Tveit,
PT E
& Sandøe, 2014a), the hypothetical choice tasks in the experiment were deliberately kept simple. Accordingly, only two animal welfare attributes, together with a price vector, were included. Thereby, it was possible to identify the trade-off between the two animal welfare
CE
attributes as well as between animal welfare and price.
AC
Housing conditions for sows, including outdoor access, and additional space for finisher pigs were selected as welfare attributes in the choice experiment, because living in a natural environment, and in particular outdoor access and space, were mentioned repeatedly in the focus groups as well as in earlier literature as important aspects of animal welfare (Christensen et al., 2014a). Three levels of welfare condition for sows and finisher pigs were included in the analysis. An overview of all of the attributes used in the choice experiment is provided in Table 1. In the subsequent analysis, the lowest improvements of the two welfare attributes were categorized as the medium level welfare improvements on standard 5
ACCEPTED MANUSCRIPT production, and the largest improvements as the high level welfare improvements. All pork attribute choices were related to minced pork because that is the most popular cut of pork (Danish Agriculture & Food Council, 2014). The price vector was selected so that the lowest hypothetical price matched the lowest price seen for minced pork, while the highest price was
PT
set above the prices observed on the market.
RI
Please, insert Table 1 here
All respondents were provided with a fact sheet briefly explaining the production cycle in
SC
pork production with specific reference to the welfare attributes included in the experiment.
NU
Respondents were asked to act as though they were in a real shopping situation and prompted with a „cheap talk‟ script reminding them that they had the same amount of money available
MA
for food in the experiment as they had in real life, and that in experiments people tend to act differently from the way they do in real life. Inclusion of a cheap talk script has been shown
D
to reduce hypothetical bias (Cummings & Taylor, 1999).
PT E
Each choice set consisted of two hypothetical types of meat, Pork A and Pork B, with an additional option to choose Neither of these. The exact combinations of attribute levels in
CE
each choice set were determined using D-efficiency as a decision criterion. D-efficiency seeks to minimize the generalized variance of the parameter estimates of a pre-specified model
AC
(Vermeulen, Goos, Scarpa, & Vandebroek, 2011; Zwerina, 1997). A pilot study involving 100 respondents was used to identify a prior distribution of parameter values and for testing the time it took to complete the questionnaire. The combinations of attribute levels in the choice sets were designed with the software package Ngene 1.0. Attribute level swapping was used to avoid dominant alternatives (Zwerina, 1997). A design with D-error of 0.74 was considered acceptable and used in the main survey. The design involved nine choice sets which presented a feasible task to the respondents without inducing fatigue during the choice tasks (Bech et
6
ACCEPTED MANUSCRIPT al., 2011). It also helped to keep the overall response time for the questionnaire to approximately 15 minutes. Respondents were asked to complete the same nine choice sets, but the ordering of the sets was varied. One choice set is shown in Figure 1. Please, insert Figure 1 here
PT
2.2. Statistical method
RI
Data from the choice experiment were analysed in a mixed logit model with an error component (McFadden, 1974). In each choice set t, respondent n was assumed to choose the
SC
type of pork j that provided him/her with the highest utility. Assuming a linear description of
MA
NU
the systematic part of the utility, the utility Unjt was specified as follows:
An alternative specific constant α captured systematic preferences for not choosing any of captured the levels of the two animal welfare
D
the hypothetical alternatives. The term
PT E
attributes which were modelled as discrete variables to allow for different preferences for animal welfare improvements from standard to medium levels of welfare as opposed to
CE
preferences for improvements from medium to high levels of welfare. The random parameter was assumed to be normally distributed, which allowed for both positive and negative
AC
preferences for the animal welfare attributes and allowed for heterogeneity among respondents‟ marginal utilities for the animal welfare attributes. The term utility associated with price, where
was assumed to be a fixed parameter and
captured the was the
price vector. The price parameter was modelled as a continuous variable which implicated constant marginal disutility of costs across respondents. This seemed to be a realistic assumption as food costs do not significantly affect overall household expenditures. Furthermore, this assumption facilitated a straightforward estimation of respondents‟ WTP for
7
ACCEPTED MANUSCRIPT welfare attributes as -
. The term
was an error component with
specifying the individual specific zero-mean normally distributed error component that allows different correlations between unobservable components of the utility of the two hypothetical alternatives Pork A and Pork B, on the one hand, and between one of the hypothetical alternatives and the alternative Neither of these, on the other. The error term
was assumed
PT
to consist of independent and identically distributed random variables representing the
RI
unsystematic and unobserved element of individual n‟s choice (Hanley, Adamowicz, &
SC
Wright, 2005; Holmes & Adamowicz, 2003). Econometric estimation was performed using Nlogit 5 (Greene, 2010) and using 1,000 Halton draws to simulate the random parameter
NU
distributions.
In the statistical analysis, the respondents were divided into two groups. One group
MA
consisted of those who claimed to often or always buy high level welfare pork products in real shopping situations and the other consisted of respondents who usually bought standard or parameters for the two groups, the study could capture
D
medium pork. By allowing different
PT E
differences in WTP measures between the groups. 2.3. Data
CE
Consumer data on pork consumption and perceptions of pig welfare were obtained using
AC
a web-based questionnaire. The questionnaire was distributed to the online Danish consumer panel of the market research institute Userneeds in June 2013. The respondent sample was constructed to reflect the Danish population in respect of gender, age, and education. Two reinvitations were sent to respondents belonging to socio-demographic groups that tended to be underrepresented. The questionnaire was completed by 396 respondents, corresponding to a response rate of 15%. The sample consisted almost exclusively of people who were familiar with shopping decisions, in that 50% of the sample did all food shopping and 42% did part of the shopping. 8
ACCEPTED MANUSCRIPT The socio-demographic distributions of the resulting sample and the Danish population are shown in Table 2. Please, insert Table 2 here 3. Results
PT
3.1. Purchasing behaviour and knowledge of pork welfare attributes
RI
Two out of three respondents stated that they ate pork at least twice a week and only 4% of the respondents claimed never to eat pork. In addition, 14-35% of the respondents did not
SC
know how often they bought the various brands available on the Danish market. Further, one
NU
out of five respondents stated that they did not think at all about animal welfare while one out of five claimed to think a lot about animal welfare when buying minced pork. The
MA
respondents‟ knowledge of the pork welfare attributes of the following brands was tested: the brand „100% Danish‟, categorized as standard pork; the brands „Antonius‟, „Den Go‟e Gris‟,
D
and „Bornholmergrisen‟, categorized as medium level welfare pork; and the brands „Friland‟
PT E
(the Danish word for free-range) and „Organic‟ pork, categorized as high level welfare pork. Some of the brands were sold only in a limited number of retail chains, while others could be
CE
purchased more widely. Table 3 summarizes respondents‟ knowledge of the brands.
AC
Please, insert Table 3 here The results indicated a considerable lack of knowledge of the welfare attributes promised by the brands categorized as high welfare pork. For example, only around one third of the respondents associated the high level welfare pork with loose-housed sows, and only around a half associated the two high animal welfare brands with outdoor access. 3.2. The importance of loose-housed sows and more space for finisher pigs
9
ACCEPTED MANUSCRIPT The survey shed light on respondents‟ attitudes to a number of welfare-improving initiatives in pig production (not shown in table). More than half of the respondents considered it to be very positive that sows were always loose-housed, a quarter considered it to have small positive impact, and one respondent in ten did not care about this aspect of husbandry. Ranking the attributes according to the share of respondents who found that the
PT
attributes had great positive effect provided the following result: loose housing of sows (56%
RI
of respondents) was valued as important by less respondents than was more space for finisher
SC
pigs (73%), access to outdoor areas (70%), supply of straw (68%), short transport time to the slaughterhouse (67%), and the opportunity to socialize with other pigs (57%). On the other
NU
hand, the loose housing of sows was considered important by more respondents than organic production (36% of respondents) and the avoidance of tail docking (21%) and castration
MA
(8%).
Respondents who claimed often or always to buy high welfare pork in real shopping
D
situations, comprised 20% of the sample (N1=79). The remaining 80% never or seldom
PT E
bought high welfare pork (N2=317). Results from the choice experiment are shown in Table 4, which also shows marginal utilities reformulated as WTP.
CE
Please, insert Table 4 here
AC
The model fit measured by the McFaddden pseudo ρ2 was 0.29. As noted by McFadden (1978), ρ2 values between 0.2 and 0.4 represent acceptable fit. The confidence intervals of the estimates were rather large, indicating that there was substantial variation in the respondents‟ valuations of the attributes. Hence, allowing randomness in these parameters improved model fit. The standard deviations of the error components were insignificant, indicating that the choice between the two hypothetical packages of pork was not considered to differ in character from the choice between the hypothetical packages and Neither of these.
10
ACCEPTED MANUSCRIPT In the choice experiment, respondents who usually bought standard or medium pork were found to have a WTP for medium levels of the welfare attributes. However, they did not place additional value on sows being allowed outdoor access as compared with being allowed to be loose indoors. Neither did they value additional space for finishers from 30% to 100%. The average respondent who usually bought standard or medium pork claimed to be willing to pay
PT
a price premium of around € 1.6 per kg for minced pork from pigs born by sows which are
RI
loose-housed or roam outdoors; and around € 2.8 extra per kg for minced pork from pigs with
SC
more space. It is also important, though, to acknowledge the significant variations in the WTP represented by the large confidence intervals.
NU
By contrast, those respondents who usually bought high level welfare pork products showed a significantly stronger preference for the high levels of welfare than the medium
MA
levels in the experiment and had a generally stronger preference for pig welfare than the other
D
respondents.
PT E
4. Discussion
4.1. The Danish market for pork with different welfare levels The hypothetical welfare improvements used in the choice experiment broadly matched
CE
the range of pork products available on the Danish market categorized according to welfare
AC
levels (standard, medium, high). The estimated WTPs could therefore be used to assess marketing implications. Details of the three categories of pork on the Danish market are shown in Table 5, together with their estimated market shares and market prices. Please, insert Table 5 here 4.2. Estimated WTP for loose-housed sows and more space for finisher pigs When people were asked directly in the questionnaire, they valued more space for finisher pigs more highly than the housing of sows without confinement. A guarantee that sows are 11
ACCEPTED MANUSCRIPT loose-housed at all times seems to be a welfare attribute that was not assigned the highest priority by respondents. Nevertheless it was considered as a positive attribute which was ranked in the middle of the field compared with other welfare-improving conditions for pigs. In Table 6, price premiums based on estimated WTPs for hypothetical medium and high level welfare pork are compared with price premiums based on observed market prices. The
PT
group of respondents who usually bought standard or medium pork were on average willing
RI
to pay up to 80% more for pork with medium levels of the two tested welfare attributes.
SC
Members of the group of respondents who usually bought organic or free-range pork were on average willing to pay up to 170% more than the price of the standard product for pork
NU
offering medium levels of the two welfare attributes that were included in the experiment. Furthermore, they were on average willing to pay 15% more for high levels of the two
MA
welfare attributes, as compared with medium levels of these.
D
Please, insert Table 6 here
PT E
4.3. Putting the results in perspective
Let us for a moment assume that the estimated WTPs can be taken as true reflections of
CE
market behaviour. As the WTPs were estimated as normally distributed random variables, a mean WTP of 80% for the group of respondents usually buying standard or medium level
AC
welfare pork suggests that around half of these buyers would be willing to pay this price. In addition, half of the respondents usually buying high level welfare pork would be willing to pay a price premium of 15% for high relative to medium level welfare. The results therefore appear to suggest that, with price premiums for medium welfare pork of 75-80%, it should possible to attain a market share of around 50%, which is a considerably higher market share than the current of 12%. In addition, the market share for high level welfare pork should, at present price premiums, be able to attract around 10% of respondents.
12
ACCEPTED MANUSCRIPT A novel implication of the study is that the WTP for medium and high level welfare pork among the group of respondents who usually bought high level welfare pork could not be fully exploited. In the hypothetical setting they were willing to pay almost three times more for premium pork than standard pork. However, in any price adjustments designed to attract new consumers to medium welfare products the price premiums would be restricted by price
PT
differences of 80% between medium welfare and standard pork and 15% between medium
RI
and high level welfare pork. This insight was obtained by the categorization of respondents
SC
into two groups with different WTPs.
Further considerations need to be taken into account, however, before jumping to
NU
marketing conclusions. First, when interpreting the results, it should be kept in mind that both the medium and high welfare pork on the Danish market had welfare attributes which were
MA
not included in the choice experiment. On the other hand, the medium level welfare pork in the choice experiment assumed that sows were loose-housed at all times while the brands
D
offering medium level welfare pork on the market guaranteed 80% loose-housed sows.
PT E
Further, respondents in the choice experiment were likely to be better informed than consumers in real shopping situations. In addition, the elderly, those with higher levels of
CE
education, and women, were all found to be overrepresented in the sample relative to the Danish population as a whole. These consumers tend to be more concerned about animal
AC
welfare than the average consumer (Wolf, Tonsor, McKendree, Thomson, & Swanson, 2016). The income distribution of the sample is reasonably representative. However, 14% did not state their income which induces some uncertainty. Further, as the WTP estimates were obtained in a hypothetical setting, they are likely to differ from what would be observed at the meat counter (Fifer, Rose, & Greaves, 2014; Loomis, 2011). These factors point in opposite directions in terms of WTP bias, possibly leading to overestimation of the hypothetical WTP as compared with that in a real shopping situation.
13
ACCEPTED MANUSCRIPT Secondly, an informational problem concerning the animal welfare attributes of pork was identified in the market. This supports earlier findings by Kehlbacher et al. (2012), Taylor & Signal (2009) and Tawse (2010). In particular, the survey revealed a general uncertainty among respondents as to which brand of pork they actually bought as well as unawareness of the animal welfare requirements behind the labels for high level welfare pork (Table 3). This
PT
lack of knowledge could be due in part to the fact that not all brands were marketed in all
RI
shops, but still there seems to be a considerable lack of awareness of animal welfare
SC
parameters in actual shopping situations. If welfare differences between the various labels in a real market setting are not transparent, consumers will always tend to choose the cheapest
NU
product. The lack of awareness among Danish consumers regarding animal welfare in the shopping situation might shed light on the findings by Verbeke, Pérez-Cueto, de Barcellos,
MA
Krystallis, and Grunert (2010) who identified only a weak relationship between attitudes towards animal welfare and purchasing behaviour.
D
Hence, if efforts were to be made to expand market shares for pork with welfare traits
PT E
above the standard levels, it would seem to be crucial for the marketing of medium and high levels of welfare products that consumers are made aware of the differences between products
CE
and encouraged to become attentive to information set out on packaging and in labels when they choose their pork products. In particular, it is noteworthy that only around half of the
AC
respondents perceived the high level animal welfare products to guarantee outdoor access since this welfare trait is promoted heavily in the two brands considered. As the present study did not address whether detailed information on welfare attributes versus overall categories such as „enhanced‟ or „excellent‟ are most likely to attract consumers‟ attention, this would be an important topic for future studies. These insights support the findings in Kehlbacher et al. (2012) and expand them as the present study narrowed down the satiation level beyond which a group of respondents‟ WTP
14
ACCEPTED MANUSCRIPT for additional welfare improvements is zero. It may be wondered whether the results simply reflect a familiar pattern of decreasing marginal utility – in this case, of animal welfare – in line with standard assumptions in consumer economics. However, even though the vast majority of respondents involved in this study were partly or fully responsible for the daily shopping, there was a marked lack of awareness both of purchasing habits associated with the
PT
various labels and of the welfare characteristics of the brands. Hence the results could equally
RI
well represent a WTP for animal welfare that could be strengthened if consumers were better
SC
informed.
Both the hypothetical and the observed market prices indicated tolerance of a price
NU
premium of only 15% between medium and high level welfare pork. Hence, to develop the market for free-range and organic pork it would seem to be important to focus on how to
MA
increase consumer WTP for high level welfare pork. Future studies on consumer interest in the addition of further attributes to these pork products (such as larger improvements in
D
welfare attributes, and more focus on organic provenance, and on improved characteristics of
PT E
relevance to human health, convenience, locality, and so on) could be very useful in guiding further speculation about how to increase market shares for high level welfare products.
CE
As a final and more general observation, it should be emphasized that the demand side of the pork market is not represented and shaped by consumers alone. Retailers are increasingly
AC
powerful in framing consumer choice. Hence, an important question is whether – given consumer WTP for animal welfare – retailers have an interest in supporting markets for medium levels of animal welfare. 5. Conclusions As expected, respondents who usually bought high level welfare pork stated a higher WTP for animal welfare attributes in the hypothetical settings than other respondents. More interestingly, though, these respondents additionally displayed a WTP for high levels of 15
ACCEPTED MANUSCRIPT animal welfare as opposed to medium levels of animal welfare, albeit that it was a relatively small WTP. In contrast, the group of respondents who usually bought standard or medium level welfare pork showed no additional WTP for high level welfare products as compared with medium level welfare pork– even in the hypothetical settings of the choice experiment. These findings suggest that the market can accommodate three levels of welfare pork, and that
PT
the market share of medium level welfare pork could be increased considerably.
RI
However, the risk that Danish consumers will abandon free-range and organic pork if less
SC
expensive products with an animal welfare label become available is certainly real: this change in purchasing behaviour could well occur if the price differential separating medium
NU
and high level animal welfare pork is too great. Efforts to enlarge the market share of medium welfare pork also face another challenge. This is that, in attempting to incentivise new
MA
consumer groups to buy medium level welfare pork, it will not be possible to exploit the full WTP among consumers who generally buy high level welfare pork. The finding that the
D
respondents had a limited understanding of the animal welfare traits associated with high
PT E
level welfare products presently being sold and were often unable to recall which brands they bought suggests that measures designed to raise the visibility of welfare attributes will be as
AC
differentials.
CE
important in attracting consumers to medium or high level welfare products as price
Acknowledgements
Based on the project „Market driven animal welfare – the case of loose-housed sows 20122013‟ funded by the Danish Pig Levy Fund (http://www.svineafgiftsfonden.dk/). Additionally, the authors want to thank four anonymous referees for very constructive comments that have considerably improved the paper.
16
ACCEPTED MANUSCRIPT References Bech, M., Kjaer, T., & Lauridsen, J. (2011). Does the number of choice sets matter? Results from a web survey applying a discrete choice experiment. Health Econonomics, 20(3), 273-86. Brandt, I. (2014). So-called “welfare products” harm the sales of organic meat. Organic &
PT
Business 554 October 2014. [In Danish]. http://okologi.dk/media/635431/554-okt-l.pdf
RI
(accessed 09/09 2016).
SC
Christensen, T., Lawrence, A., Lund, M., Stott, A., & Sandøe, P. (2012). What can economists do to help improve animal welfare? Animal Welfare, 21(S1), 1–10.
NU
Christensen, T., Tveit, G., & Sandøe, P. (2014a). Loose-housed sows – a cross-disciplinary investigation on market driven animal welfare. Project report 19. CEBRA - Centre for
MA
Biosafety and Risk Assessment, Copenhagen. [In Danish]. Christensen, T., Denver, S., Hansen, H.O., Lassen, J., & Sandøe, P. (2014b). Animal welfare
D
labels – comparing experiences from six EU countries. Commissioned Work 2014/10,
English summary].
PT E
Institute of Food and Resource Economics, University of Copenhagen. [In Danish with
CE
Cummings, R.G. & Taylor, L.O. (1999). Unbiased value estimates for environmental goods: a cheap talk design for the contingent valuation method. American Economic Review,
AC
89(3), 649–666.
Danish Agriculture & Food Council (2014). Consumption of pork, beef and chicken meat – available, bought or eaten? Danish Agriculture & Food Council, August 2014. [In Danish]. Danish Agriculture & Food Council (2016). Consumers choose Danish when they want good meat. Market analysis from Danish Agriculture & Food Council 06/24/2016 [In Danish].
17
ACCEPTED MANUSCRIPT Dransfield, E., Ngapo, T. M., Nielsen, N. A., Bredahl, L., Sjödén, P. O., Magnusson, M., Campo, M. M., & Nute, G. R. (2005). Consumer choice and suggested price for pork as influenced by its appearance, taste and information concerning country of origin and organic production. Meat Science, 69(1), 61-70. EFSA (2007). Animal health and welfare aspects of different housing and husbandry systems
PT
for adult breeding boars, pregnant, farrowing sows and unweaned piglets. Scientific
RI
Opinion of the Panel on Animal Health and Welfare. The EFSA Journal, 572, 1-13.
SC
Fifer, S., Rose, J., & Greaves, S. (2014). Hypothetical bias in Stated Choice Experiments: Is it a problem? And if so, how do we deal with it? Transportation Research Part A, 61,
NU
164–177.
Greene, W. (2010). NLogit Version 4.0. Econometric software Inc.
MA
http://pages.stern.nyu.edu/~wgreene/DiscreteChoice/Software/NLOGIT-Short-StudentManual.pdf (accessed 09/09/2016).
PT E
D
Hanley, N., Adamowicz, W., & Wright, R. E. (2005). Price vector effects in choice experiments: an empirical test. Resource and Energy Economics, 27, 227–234. Heerwagen, L. R., Mørkbak, M. R., Denver, S., Sandøe, P., & Christensen, T. (2015). The
CE
role of quality labels in market-driven animal welfare. Journal of Agricultural &
AC
Environmental Ethics, 28, 67-84. Holmes, T. P., & Adamowicz, W. (2003). Feature based methods. In: Champ, P.A., Boyle, K.J., Brown, T.C. (Eds.), A primer on nonmarket valuation. Kluwer Academic Publishers. Hoyos, D. (2010). The state of the art of environmental valuation with discrete choice experiments. Ecological Economics, 69, 1595–1603. Humane Society International (2014). An HSI Report: The Welfare of Animals in the Pig Industry. http://www.hsi.org/assets/pdfs/hsi-fa-whitepapers/welfare_of_animals_in_the_pig.pdf (accessed 09/09/2016). 18
ACCEPTED MANUSCRIPT Kehlbacher, A., Bennett, R., & Balcombe, K. (2012). Measuring the consumer benefits of improving farm animal welfare to inform welfare labelling. Food Policy, 37(6), 627-633. Lassen, J., Sandøe, P., & Forkman, B. (2006). Happy pigs are dirty! – conflicting perspectives on animal welfare. Livestock Science, 103(3), 221–230. List, J. A., & Gallet, G. A. (2001). What Experimental Protocol Influence Disparities
PT
Between Actual and Hypothetical Stated Values? Evidence from a Meta-Analysis.
RI
Environmental and Resource Economics, 20, 241–254.
SC
Loomis, J. B. (2011). What's to Know About Hypothetical Bias in Stated Preference Valuation Studies? Journal of Economic Surveys, 25(2), 363-370.
NU
Mayfield, L. E., Bennett, R. M., Tranter, R. B., & Wooldridge, M. J. (2007). Consumption of welfare-friendly food products in Great Britain, Italy and Sweden, and how it may be
MA
influenced by consumer attitudes to, and behaviour towards, animal welfare attributes. International Journal of Food and Agriculture, 15(23), 59-73.
D
McFadden, D. (1974). Conditional Logit Analysis of Qualitative Choice Behavior. In P.
PT E
Zarembka, (Ed.), Frontiers in Econometrics (pp. 105-142). New York: Academic, New Press.
CE
McFadden, D. (1978). Quantitative methods for analyzing travel behaviour in individuals: Some recent developments. In D. Hensher, & P. Stopher (Eds.), Behavioural Travel
AC
Modelling (pp. 279-218). London: Croom Helm. Meuwissen, M. P. M., & Van Der Lans, I. A. (2005). Trade-offs between consumer concerns: An application for pork supply chains. Food Economics - Acta Agriculturae Scandinavica, Section, C2(1), 27-34. Mørkbak, M. R., Christensen, T., & Gyrd-Hansen, D. (2010). Choke price bias in choice experiments. Environmental & Resource Economics, 45(4), 537-551.
19
ACCEPTED MANUSCRIPT Norwood, F.B., & Lusk, J.L. (2011). A calibrated auction-conjoint valuation method: Valuing pork and eggs produced under differing animal welfare conditions. Journal of Environmental Economics and Management, 62, 80–94. Prickett, R. W., Norwood, F. B., & Lusk, J. L. (2010). Consumer preferences for farm animal welfare: results from a telephone survey of US households. Animal Welfare, 19, 335-347.
PT
Statistics Denmark (2013). Statistical Yearbook 2013. Statistics Denmark. [In Danish]
RI
http://www.dst.dk/pukora/epub/upload/17956/saa2013.pdf (accessed 09/09 2016).
SC
Tawse, J. (2010). Consumer attitudes towards farm animals and their welfare: a pig production case study. Bioscience Horizons, 3(2), 156-165.
NU
Taylor, N., & Signal, T. D. (2009). Willingness to pay: Australian consumers and “on the farm” welfare. Journal of applied animal welfare science: JAAWS, 12(4), 345-59.
MA
The European Commission (2006). Report on a Community Action Plan on the Protection and Welfare of Animals 2006–2010 (2006/2046(INI), 19/09/2006). Committee on
D
Agriculture and Rural Development 19/09/2006.
PT E
Tonsor, G. T., Olynk, N., & Wolf, C. (2009). Consumer Preferences for Animal Welfare tributes: The Case of Gestation Crates. Animal Welfare, 3, 713-730.
CE
Tonsor, G. T., & Wolf, C. (2011). On mandatory labelling of animal welfare attributes. Food Policy, 36, 430–437.
AC
Turner, S. P., Ewen, M., Rooke, J. A., & Edwards, S. A. (2000). The effect of space allowance on performance, aggression and immune competence of growing pigs housed on straw deep-litter at different group size. Livestock Production Science, 66(1), 47–55. Uzea, A. D., Hobbs, J. E., & Zhang, J. (2011), Activists and Animal Welfare: Quality Verifications in the Canadian Pork Sector. Journal of Agricultural Economics, 62, 281– 304.
20
ACCEPTED MANUSCRIPT Vanhonacker, F., & Verbeke, W. (2014). Public and Consumer Policies for Higher Welfare Food Products: Challenges and Opportunities. Journal of Agricultural and Environmental Ethics, 27(1), 153-171. Vermeulen, B., Goos, P., Scarpa, R. & Vandebroek, M. (2011). Bayesian Conjoint Choice Designs for Measuring Willingness to Pay. Environmental Resource Economics, 48,129–
PT
149.
RI
Verbeke, W., Pérez-Cueto, F.J.A., de Barcellos, M.D., Krystallis, A., & Grunert, K.G. (2010).
SC
European citizen and consumer attitudes and preferences regarding beef and pork. Meat Science, 84, 284–292.
NU
Wolf, C.A., Tonsor, G.T., McKendree, M.G.S., Thomson, D.U., Swanson, J.C. (2016). Public and farmer perceptions of dairy cattle welfare in the United States. Journal of Dairy
MA
Science, 99(7), 5892-5903.
Zwerina, K. (1997). Discrete Choice Experiments in Marketing: Use of Priors in Efficient
D
Choice Designs and Their Application to Individual Preference Measurement. Berlin
AC
CE
PT E
Heidelberg: Springer Verlag (Chapter 4).
21
ACCEPTED MANUSCRIPT
Pork A
Pork B
Current legislation for sows
Sows are always loose with outdoor access
Finisher pigs have 100% more space Finisher pigs have 30 % more space € 7/kg of minced pork
Which of the products do you prefer?
○ Neither of these
SC
○ Pork B
NU
○ Pork A
RI
PT
€ 5.4/kg of minced pork
Fig 1 An example of a choice set used in estimation of willingness-to-pay for pork welfare
AC
CE
PT E
D
MA
attributes.
22
ACCEPTED MANUSCRIPT Table 1 Overview of attributes and levels included in the choice experiment Animal Label in choice Attributes
welfare
Description experiment
level Standard
Current legislation for sows
PT
Housing
Sows are loose during gestation Indoor loose-housed sows at all
Medium
Sows are always loose
times
sows
Sows are always loose with
Loose-housed sows at all times
SC
High
RI
conditions for
outdoor access
with opportunity to go outdoors
for finisher
MA
for finisher pigs
Indoor space
NU
Current legislation of space Standard
Minimum 0.65 m2 per pig
Finisher pigs have 30% Medium
Minimum 0.85 m2 per pig
more space
pigs
D
Finisher pigs have 100% Minimum 1.3 m2 per pig
PT E
High
more space Four prices of minced pork
Market observations were used
€/kg: 5.4; 7.0; 10.2; 16.1
to identify a realistic price vector
CE
N/A
AC
Price
23
ACCEPTED MANUSCRIPT Table 2 Socio-demographic distribution of respondents regarding gender, age, education and income (in %, N=396).
Variable
Respondent
National
distribution
distributiona
Female
55
50
Male
45
50
18-34
15
30
Levels
RI
35-49
24
SC
Age (years)
PT
Gender
50-70
31
61
39
21
27-37
44
38-48
34
19-29
11
21
26,810-80,430
47
51
> 80,430
28
27
No answer
14
N/A
NU
Primary school (10 years schooling)
High school level (additional 3 years) Education
MA
University level (additional 2 or more
< 26,810
PT E
Household income, before
CE
tax (€/year)
Information on national distribution is based on Statistics Denmark (2013).
AC
a
D
years)
24
ACCEPTED MANUSCRIPT Table 3 Respondents‟ knowledge of pork brands (in % of sample, N=396). Animal welfare
Standard
Medium Den Go’e
100% Pork brand
High Born-
Antonius Danish
Friland
Organic
40
32
6
57
44
7
25
40
Gris
holmer
4
Sows always loose2
5
3
3
7
4
4
15
PT
housed
Pigs
have
SC
access
RI
Pigs have outdoor
more
9
NU
space indoor
Grey shading indicates that requirements were not guaranteed by the label and that answers
MA
thereby were incorrect. The exact wording of the question to respondents was „Which attributes do you particularly associate with the following brands of pork. Several answers per
AC
CE
PT E
D
brand can be stated?
25
ACCEPTED MANUSCRIPT Table 4 Model results and estimated price premiums based on respondents‟ willingness-topay (WTP) for welfare attributes in pork (N=396). Estimated Variable
parameters Estimated WTP price
Mean (95% C.I.)
premiums €/kg
[std. dev. (95% C.I.)]a
Mean (95% C.I.)b
PT
Respondents usually buying standard or medium level welfare pork (normally distributed
0.71 (0.42, 0.99)
1.6 (1.0, 2.2)
SC
Sows loose-housed indoors
RI
random parameters)
[0.88 (0.60, 1.16)]
NU
0.71 (0.36, 1.05) Sows loose-housed and outdoors
1.6 (0.8, 2.4)
MA
[1.60 (1.24, 1.96)] 1.19 (0.89, 1.48)
Pigs 30% more space indoors
2.8 (2.0, 3.4)
PT E
Pigs 100% more space indoors
D
[0.81 (0.40, 1.22)] 1.25 (0.86, 1.64) 2.8 (2.0, 3.8)
[1.44 (1.17, 1.72)]
Sows loose-housed indoors
CE
1.11 (0.51, 1.71)
4.2 (2.2, 6.2)
Sows loose-housed and outdoors
1.50 (0.70, 2.29)
5.2 (2.4, 7.8)
AC
Respondents usually buying high level welfare pork (non-random parameters)
Pigs 30% more space indoors
0.84 (0.23, 1.46)
4.8 (2.6, 6.8)
Pigs 100% more space indoors
1.45 (0.75, 2.15)
6.2 (3.8, 8.8)
Non-random parameters, all respondents Price
-0.12 (-0.13,-0.11)
Alternative specific constant
-5.28 (-5.84, -4.73)
Std. dev. of error component, σ
2.66 (-9.55, 14.86)
a
All parameter estimates were significant at 0.01% level except σ which was not significant. 26
ACCEPTED MANUSCRIPT b
The confidence intervals of WTP estimates were calculated by dividing marginal utility
AC
CE
PT E
D
MA
NU
SC
RI
PT
confidence levels by the price parameter.
27
ACCEPTED MANUSCRIPT Table 5 Estimated market shares and market prices for minced pork in Denmark categorized according to animal welfare levels (data from 2015). Type
Market a
share
Market price (€/kg)
Description
b
Requirements in accordance with the Danish
PT
legislation:
Sows are loose-housed 60% of the time 84%
5.4 - 8.0 c
RI
Standard
SC
Finisher pigs have 0.65 m2 Tale docking if necessary
NU
Age at weaning at least 21 days Sows are loose-housed 80% of the time
Medium
12%
MA
Finisher pigs have 30% more space
9.4
No tale docking
D
Age at weaning at least 28 days
PT E
High 2%
Sows are 100% loose with outdoor access
10.7
Finisher pigs have 100% more space in the
Organic
AC
CE
Free-range
Plus a few other requirementsd
2%
stables with access to an outdoor veranda No tale docking
13.2 Age at weaning at least 30 days Plus other requirementse
a
Market shares from Christensen et al. (2014b).
b
Market prices were Googled in November 2015; various homepages for online shopping and
supermarket advertisements were consulted. Dyrenes Beskyttelse and Danish Crown (the
28
ACCEPTED MANUSCRIPT largest abattoir in Denmark) provided valuable information via e-mail correspondence (September 2015) which helped to confirm the estimated market prices. c
The standard price of minced pork was € 8/kg, but standard minced pork was routinely sold
at a discount price of € 5.4/kg. d
These involve more straw, no dispensation for a max of 8 hours transportation time to
Organic pig production has additional requirements, as compared with free-range
RI
e
PT
slaughter, no use of electrified stick at slaughterhouse.
SC
production, in terms of later weaning, more space for sows and finisher pigs, and organic
AC
CE
PT E
D
MA
NU
feed.
29
ACCEPTED MANUSCRIPT Table 6 Market observations and model-results for willingness-to-pay (WTP) for welfare pork (N=396). Stated WTP for welfare porkb
Market price
Respondents usually
standard or medium level
buying high level
welfare pork
welfare pork
0%
Medium level
Base (WTP not estimated)
RI
Standard
Respondents usually buying
PT
premiumsa
Product
17% - 75%
80% higher higher
SC
relative to standard High level relative
0% higher
NU
14% higher to medium
15% higher
Estimates of market price premiums in per cent were based on values from Table 5. Price
MA
a
170% higher
premium for medium relative to standard were specified as an interval based on discount price
Price premiums for stated WTP were based on results in Table 4. Example: estimated WTP
PT E
b
D
and standard price.
price premium for two medium levels of animal welfare relative to the lowest price level used
AC
CE
in the choice experiment: (€ 1.6 + € 2.6)/ € 5.4 = 80%.
30
ACCEPTED MANUSCRIPT Highlights Many consumers will not pay for improving animal welfare beyond the medium level
Small difference between willingness-to-pay for medium and high level welfare pork
Full willingness-to-pay for high level welfare pork cannot be exploited
Increasing brand awareness and consumers‟ knowledge of brands is vital
AC
CE
PT E
D
MA
NU
SC
RI
PT
31