How Australian consumers value intrinsic and extrinsic attributes of beef products

Accepted Manuscript How Australian Consumers Value Intrinsic and Extrinsic Attributes of Beef Products Ardeshiri Ali, Rose John PII: DOI: Reference:

S0950-3293(17)30257-4 https://doi.org/10.1016/j.foodqual.2017.10.018 FQAP 3414

To appear in:

Food Quality and Preference

Received Date: Revised Date: Accepted Date:

3 March 2017 24 October 2017 24 October 2017

Please cite this article as: Ali, A., John, R., How Australian Consumers Value Intrinsic and Extrinsic Attributes of Beef Products, Food Quality and Preference (2017), doi: https://doi.org/10.1016/j.foodqual.2017.10.018

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.

How Australian Consumers Value Intrinsic and Extrinsic Attributes of Beef Products Ardeshiri, Ali1*; Rose, John2

1- Institute for Choice, University of South Australia, 140 Arthur St, North Sydney, NSW 2060 2- University of Technology, 15 Broadway, Ultimo, NSW 2007

* Corresponding author Dr. Ali Ardeshiri Post-doctoral Research Fellow Institute for Choice - UniSA Business School Level 13, 140 Arthur Street, North Sydney 2060 Tel +61 8 8302 1650 Email [email protected]

Page |2

Abstract The purpose of this paper is to determine which information cues on beef labels actually attract consumer interest. We applied a discrete choice experiment to investigate 1002 Australian consumer preferences and willingness-to-pay (WTP) for different beef products. Consumers were presented with a novel experiment in which they indicated “how many” they would purchase for mince, diced, roast, and four cuts of steaks (rump, porterhouse, scotch fillet and eye fillet). The results from an ordered logit model showed that cues related to healthy option purchases play a stronger influential role on Australian consumers decision making compared to other beef attributes. Australian consumers have a stronger preference for less marbled beef. Moreover, white fat colour is more desirable than yellow colour. Furthermore, in relation to labelling information, origin of the beef is a key indicator in consumer’s evaluation process. We observed a highly inflated WTP for origin of the beef. For example a WTP of $5.76 for Scotch fillet steak from “Tasmania” compared a WTP of $14.22 for the same cut from “China”. This finding may be due to Australian consumers using origin as a cue for food safety or quality. We concluded that preferences for beef products are not similar across consumers from different nations and country-specific research is required to illustrate consumer’s preference. Finally, this study provides managerial and policy implication and recommendations to better understand the relative value to the Australian consumer of beef product appearance and labelling information.

Keywords Discrete Choice Experiments, Product Appearance, Labelling Information, Information Cues, Beef Preference, Ordered logit

Page |3

1 INTRODUCTION Australia is one of the world’s largest beef exporter (Meat & Livestock Australia, 2016), nonetheless domestically Australians beef consumption has decreased (OECD/Food and Agriculture Organization of the United Nations, 2015). While beef has traditionally taken the pride of place at the centre of the Australian dinner table, Australians are now eating two-thirds the amount of beef compared to the 1980s, but nearly 2.5 times as much chicken and twice as much pork (Figure 1). Different ‘consumer focus’ campaigns have been launched by the Australian meat industry, such as “You’re Better on Beef” in 2014 and “Australian beef is the greatest” campaign in 2017 to continue generating value of beef for consumers. While an early analysis of the companies show a rise in weekly beef serving amongst the targeted audience, the challenge is to maintain the positive results given in the face of slow but steady long-term decline in consumption of beef (Meat & Livestock Australia, 2016). To maintain a positive growth in the long-term consumption levels, it is necessary to gain a deeper understanding of consumers’ values and trade-offs for beef products. This paper contributes to our understanding of how consumers value judgement for beef products are formed by studying consumers’ decision making over intrinsic and extrinsic information cues on several beef products. Understanding the causal factors underpinning these decisions can benefit the modern food industry from facing a high failure rate in introducing and entering new beef products to the super market shelves and increase beef consumption (Asioli et al., 2017; Dijksterhuis, 2016). Although research has extensively addressed the effects of retail atmospherics such as scents, displays and sounds on consumer behaviour (Turley & Milliman, 2000) the effect of packaging design on consumer behaviour has only recently started to receive substantial attention (van Ooijen, Fransen, Verlegh, & Smit, 2017). To the author’s knowledge, little scientific evidence exist regarding consumer preferences for intrinsic and extrinsic cues on packages of different beef cuts in Australia. From a marketing perspective, the variety of information cues that can be used to target the final consumer raises the question: ‘which information cues do consumers prefer over the others?’ These questions are relevant for producers, processors and retailers in the beef industry for new product development which has a too-high fail rate (Dijksterhuis, 2016). From a supply side perspective, it is important to know which information cues are most effective on product packaging and in marketing communication measures. Space in product packages is, however, often limited in particular on the front side. 45 40 35

Kg Per person

30 25 20 15 10 5 0 1978

1983

1988

1993

Beef

Pork

1998

Chicken

2003

2008

2013

Lamb

Figure 1: Australian meat consumption per person comparison (1979-2014). Source: OECD-FAO 2015

Page |4

From an empirical point of view this study contributes to the current literature by providing significant empirical findings that product developers can benefit from when improving existing products and developing new products. In addition, it identifies the main drivers of consumer choice when purchasing beef products in Australia. Furthermore, the findings will inform different functional departments within the food industry to effectively meet consumer needs (De Pelsmaeker, Dewettinck, & Gellynck, 2013; Fiszman, 2012; Jacobsen et al., 2014). From a study design perspective, this study is innovative in the elicitation of the Discrete Choice Experiment (DCE), by replacing the typical “pick a product” mechanism used in traditional choice experiment surveys by further exploring the quantity aspect i.e., “how many” (including zero) products will be purchased. Furthermore, from a methodological point of view, this study applies an econometric model that specifically accounts for the underlying ordered structure of the purchasing behaviour to gain an understanding of the preferences beyond current studies utilising DCEs. The objective of this study is to determine the relative value Australian consumers place on intrinsic and extrinsic attributes of different beef products using a novel DCE. Attributes considered for this study included product appearance, such as packaging type, meat colour, fat colour and fat content (measured as fat marbling, fat rim and their combination) as well as a set of information that appears on the label of the product, which includes origin, price, brand, weight, traceability, type of feed, organic status, expiry date and the presence of a Quick Response1 (QR) code. The results indicates that amongst all the studied cues, fat colour and fat content as intrinsic cues, and origin as extrinsic cue play a key role in Australian beef purchasing judgment.

2 LITERATURE ON BEEF LABELLING AND CONSUMERS VALUE Beef labelling gained momentum following the discovery of the bovine spongiform encephalopathy (BSE) disease in the U.S. in 1996, that led to widespread discussion in the popular media about the possibility of bioterrorism related to food safety and trade policy. Following the discovery of BSE, it was argued that mandatory country of origin labelling (COOL) would increase consumer demand for beef by allowing both domestic and international consumers to discriminate between BSE and BSE-free regions (Ikenson, 2004; Jin, Skripnitchenko, & Koo, 2004; Umberger, 2005). Since then a new system for identification and registration of beef and beef products has been introduced. In Australia a labelling declaration is mandatory and The Australia New Zealand Food Standards Code prescribes a minimum standard for the necessary information for meat products. A licensee’s food safety program must contain procedures and data collected to allow validation and verification of any statements or claims made on a product label. Some of the labelling information required by the code are: name of the product, legibility – in clear English, date marking, details of the manufacturer or importer, country of origin, ingredient listing and percentage declaration, directions for use and storage conditions and nutritional 1

Quick Response code is a type of two dimensional barcode that contains information about the item to which it is attached.

Page |5 and health claims (Authority, 2002). For producers, processors and retailers to know which information cues - in addition to the mandatory ones - should be presented on the package to better target the final consumers - bearing in mind the limited space available on the front side of the package – persuade them to know precisely how consumers value judgement for beef products are formed. The literature has revealed that when consumers form a value judgement as to their quality perceptions, it becomes necessary to break the concept of quality down into two major groups of factors (Asioli et al., 2017; Steenkamp, 1997). The first group are intrinsic attributes that permit objective measurement of quality. These qualities imbue the product with its functionality and relate to its physical aspect. According to Olson and Jacoby (1972), intrinsic attributes are specific to each product, disappear when it is consumed and cannot be altered without changing the nature of the product itself. Relevant intrinsic cues that unequivocally define a given category of beef includes sensory (i.e. colour, visible fat, cut of the meat) and nutritional attributes (Acebrón & Dopico, 2000). Extrinsic attributes, as the second group, are aspects that are related to the product but are not physically a part of it and can be changed without altering the physical product characteristic. Examples of extrinsic attributes that can significantly influence consumers in their choices are brand, price, package-layout and health claims (Jaeger, 2006; Lähteenmäki, 2013). Visual impressions based on perceived intrinsic and extrinsic cues, such as label information and appearance of a product, are important inputs that may generate beef quality expectations. Numerous studies have been conducted looking at the importance of intrinsic and extrinsic cue for beef products using different models to understand consumer expectations (Acebrón & Dopico, 2000; Caputo, Scarpa, & Nayga, 2017; Chung, Boyer, & Han, 2009; De Pelsmaeker et al., 2013; Endrizzi et al., 2015; Grunert, 2015; Hoppert, Mai, Zahn, Hoffmann, & Rohm, 2012; Loebnitz, Schuitema, & Grunert, 2015; Reicks, 2006; Sánchez, Beriain, & Carr, 2012; Sasaki & Mitsumoto, 2004; Van Wezemael, Caputo, Nayga, Chryssochoidis, & Verbeke, 2014; Verbeke & Ward, 2006; Xue, Mainville, You, & Nayga Jr, 2009). Research that combines both intrinsic (sensory) and extrinsic factors makes it possible to obtain more complete and realistic information about consumer behaviour in real life buying and eating situations (Köster, 2009). Discrete choice experiments have become a commonly used technique to elicited preferences for attribute related to labelling information as well as appearance of beef products. There are several reasons for the increased interest in discrete choice experiments. It reduces some of the potential biases of contingent valuation methods, more information is elicited from each respondent and the possibility of testing for internal consistency (Adamowicz, Boxall, Williams, & Louviere, 1998; Hanley et al., 1998). DCE can create decision scenarios very similar to the real-world decision making situation where the decision maker behaviour can be examined (Mark & Swait, 2004). DCE does a better job than contingent valuation in measuring the marginal value of changes in the characteristics of the goods (Hanley et al., 1998). This is often a more useful focus from a management/policy perspective than focussing on either the gain or loss of the good, or on a discrete change in its attributes. For this research we have specifically focused at the existing literature that have used Choice Experiment (CE) to elicited consumer preference for intrinsic and extrinsic attributes for beef products (refer to Table A.1 in the Appendix for a full list and summary of findings

Page |6 of all research articles using choice experiments to study consumer preference for intrinsic and extrinsic attributes for beef products). Based on this review the information about production systems and beef marbling can modify expectations about beef, influencing consumer-purchasing decisions. Beef quality traits such as colour, freshness and marbling of beef can influence consumer purchasing decisions. For instance, a bright red colour, thought to indicate freshness or wholesomeness, is preferred and beef which darkens in the supermarket is less sought after (Acebrón & Dopico, 2000; Carpenter, Cornforth, & Whittier, 2001; Hood & Riordan, 1973; Realini et al., 2014; Zanoli et al., 2013). Fat content has an impact on consumer visual attention and choice of beef products, with consumers paying more attention and choosing more often beef with lower fat content (Acebrón & Dopico, 2000; Banović, Chrysochou, Grunert, Rosa, & Gamito, 2016; Realini et al., 2014; Van Wezemael et al., 2014). On other hand, it has been shown that beef marbling is an important positive expectation generator in several markets because there are consumers who relate marbling with eating quality (Egan, Ferguson, & Thompson, 2001). Conversely, in some European markets, consumers tend to reject beef with high levels of marbling (Morales, Aguiar, Subiabre, & Realini, 2013; Scozzafava, Corsi, Casini, Contini, & Loose, 2016). Furthermore, food packaging has been repeatedly found to be a strong driver for consumers’ food choice and packaging characteristics lead to significant market price differences (Carpenter et al., 2001; Loose & Szolnoki, 2012). Recent studies have been conducted focused on the effect of animal welfare information on beef (Caputo et al., 2017; Lewis, Grebitus, Colson, & Hu, 2017; Ortega, Hong, Wang, & Wu, 2016; Risius & Hamm, 2017) while other studies have related origin and production system (organic vs conventional) to beef expectation (Colella & Ortega, 2017; Lagerkvist, Berthelsen, Sundström, & Johansson, 2014; Lagerkvist & Hess, 2014; Ortega et al., 2016; Peterson & Burbidge, 2012; Risius & Hamm, 2017; Zanoli et al., 2013). Thus, label information and appearance of a product not only triggers consumers’ subconscious symbolic associations and valuations (Becker, van Rompay, Schifferstein, & Galetzka, 2011) but also affects consumers’ ability to inspect food characteristics. Although DCE have become popular for modelling consumer behaviour, however, welfare value estimates obtained with DCE are sensitive to study design (Hanley, Mourato, & Wright, 2001). For example the choice of attributes, the levels chosen to represent them, and the way in which choices are relayed to respondents (for example, through the use of photograph pairs) may all impact on the values of estimates of consumers' surplus and marginal utilities (Hanley et al., 2001). Another concern with DCE is the choice complexity. Swait and Adamowicz (1996) found an inverted V-shaped relationship between choice complexity and variance of underlying utility amounts; whilst Mazotta and Opaluch (1995) found that increased complexity leads to increased random errors. Bradley and Daly (1994) have found that respondents become fatigued the more choices they are presented with, whilst Hanley et al. (2002) found that value estimates for outdoor recreation changed significantly when respondents were given eight rather than four choice pairs. Ben-Akiva and Morikawa (1990) and Ardeshiri (2014) found evidence of inconsistent responses that increase as the number of rankings increase. This implies that, whilst the researcher might want to include many attributes, and also interactions between these attributes, unless very large samples are collected, respondents will be faced with daunting choice tasks. This may lead them into relying on shortcuts to provide answers, rather than solving the underlying utility-maximisation problem. Finally, Lancaster and Swait (2014) argue that it is essential that the analyst chooses a representative process validity when analysing a DCE. Lancaster and Swait explain further that by process validity they mean that the decision process

Page |7 described by a mathematical and/or statistical model should be plausible/valid at the desired level of representation because it bears a semblance to the actual decision process(es). For example, if decision makers are actually using threshold-based satisficing as their decision rule, while the mathematical representation of the process employs instead utility maximisation, then we would understand that the process validity of the model is lower than if it were to represent the actual decision rule.

3 METHOD One set of approaches for enhancing understanding of consumer behaviour involves analyses of consumer choices. These choice studies primarily rely on modelling consumer behaviour using either a random utility theory framework (McFadden, 1974) or Lancaster’s (1966) consumer utility maximization model. Discrete Choice Experiments (DCE) (Louviere & Hensher, 1982; Louviere, Hensher, & Swait, 2000), based on the random utility theory, now have a mature microeconomic foundation that allows for measurement of the relative importance of various attributes in consumer behaviour through participants’ repeated selection of goods with different combinations of attributes, thus assessing the participants’ preferences for the attributes by analysis (Hanemann & Kanninen, 1998). While relationships between individual consumer attitudes, preferences and actual purchasing behaviours are complex (McEachern, Seaman, Padel, & Foster, 2005), DCEs and their use of consumer panels opens up the possibility of exploring multiple attributes influencing purchasing decisions across populations of consumers.

3.1 EXPERIMENT DESIGN AND MATERIALS Consumers’ preferences were achieved using a novel choice modelling framework. Individuals could select how many of each (including none) given beef product would they most likely purchase. For this study we investigate these attributes on mince, diced, roast and four cuts of steak (rump, porterhouse, scotch fillet and eye fillet)2 using a discrete choice experiment. Each individual was given four sets to complete and in each set four random alternatives (In a given choice set alternatives were allowed to be repeated) that contained the relevant attributes and levels were present. For the experimental design we used the D-efficiency criteria (Scarpa & Rose, 2008). D-efficiency design strategies produce significantly improved results, in a statistical sense of relative efficiency, than the more traditional orthogonal design (Rose, Bliemer, Hensher, & Collins, 2008). Furthermore, it enhances the relevance and comprehension ability to the attribute levels being assessed. From the literature summary above, it is evident that consumers’ purchasing decisions are influenced by interactions amongst a complex of dynamic considerations related to meat safety, nutritional and dietary values, production ethics, as well as individual aesthetic and taste related expectations. Therefore, for this study we used a diverse list of attributes such as fat colour, meat colour, marbling (mince and diced beefs were excluded) types of packaging, origin, claim attributes, brand, weight, nutritional attributes, the presence of a QR code and price (please see Table 1 for the full list of attributes and levels). Ngene software was used to generate the design for this study. The final design had a D-error of 2

The selection of the beef cuts was developed in consultation with the industry partner involved in the ARC grant associated with this research.

Page |8 0.0253 and include 96 choice tasks in 24 blocks providing each participant with 4 repeated choice occasions. We replaced the typical “pick a product” mechanism used in traditional choice experiment surveys with an ordered logit structure and asked individuals to respond to “how many” (including zero) products they will purchase. It is believed that this is the first study whereby an ordered choice model has been applied in beef consumer studies and at the time of writing this manuscript the authors could not find any other DCE studies that have used an ordered choice structure (please refer to Table A.1 in the Appendix for the list of estimation models used in the previous DCE studies on beef preferences). Figure 2 presents a snapshot of the DCE task used for this study.

Figure 2: Sample choice experiment task.

Page |9 Table 1. Attributes and levels in the choice experiment Attribute Fat Colour*

Levels 1-White (0)

2-Light yellow (4)

3-Yellow (8)

Meat Colour*

1-Pink (1A)

2-Red (3)

3-Brown (6)

Marbling*

1-Not marbled (0) (9)

Type of Packaging

1-Tray Packed (TP)

2-Somewhat marbled (4)

3-Very much marbled

Tray packed meat is when the meat is packed into an open container or tray, and covered with a film. This is mainly used in smaller primal cuts or portioned meat.

2-Modified Atmosphere Packed (MAP) Modified Atmosphere Packed indicates that packs (primal cuts or retail ready tray) are wrapped and flushed with a mixture of gases to remove the oxygen. The packs are impermeable to gases and retain the modified gas atmosphere around the meat to preserve the meat quality and shelf life by restricting the amount of bacteria growth.

3-Vacuum Packed (VP) Vacuum Packed involves the removal of air and oxygen from the packaging. This creates a vacuum and assists in the preservation of meat and improvement in meat quality due to the lack of oxygen around the meat that promotes bacterial growth.

Origin

1- Unknown(not mentioned) 2-Victoria 3-Tasmania 4-Queensland 5-New south wales, 6-China Claim 1-Grass fed/Grass finished 2-Grain fed 3-Traceable back to the farm attributes 4-Antibiotic/Hormone free 5-Organic 6-Angus Brand 1-Cape Grim 2-Certified Angus 3-King Island 4-Coles Finest 5-Riverine 6-Hopkins River 7- Super market packed Best before 1-One day to expiry 2-Three days to expiry 3-Seven days to expiry 4-Fortheen days to expiry Weight Mince and diced: 500gr, 700gr Roast: 500 gr - 1 kg Rump, Porterhouse & Scotch fillet: 250gr, 450gr Eye fillet: 150gr, 250gr & 450gr Nutritional Energy: 500, 527, 553 (KJ/100gr) factors Fat: 3.6, 3.8, 4 (gr/100gr) Protein: 21.5, 22.7, 23.85 (gr/100gr) QR code 1 if the QR code was present. Price Mince & diced: $12.99, $16.99, $19.99 ($AUS/per Roast: $13.99, $18.99, $24.99 kg) Rump: $14.99, $20.99, $26.99 Porterhouse: $28.99, $37.99, $45.99 Scotch fillet: $32.99, $40.99, $49.99 Eye fillet: $38.99, $50.99, $61.99 *Note that we used the handbook of Australian meat standards to present the levels for these attributes. The number between the parentheses refers to the reference standard score. Please see the following link for more information. https://www.ausmeat.com.au/custom-content/cdrom/Handbook-7th-edition/English/DA71F4DE-F68A-11DA-AA4B000A95D14B6E.html

P a g e | 10

3.2 PARTICIPANTS In November 2015 an online survey focused on labelling preferences for beef products was completed by 1,002 Australian residents located at the eastern border of Australia (QLD, NSW, VIC and TAS). Summary data of selected demographic attributes of survey respondents are provided in Table 2. The survey was conducted amongst Australians aged 18 years and above, who had primary or shared responsibility for grocery shopping for their household as well as for their household’s meat purchases. No gender-based weighting was applied, as we wanted to speak to qualified meat purchasers (the sample was 602 female, 400 male and one person as other). Participant were from different types of households with the majority as “Couple family with no children” or “Couple family with children”. The Average age was 49 years and the mode was 60 years old. Average household income was $65,000 with the median of $57,200. More than 80% of the respondents were earning income. They were either working as full-time (36%), part-time (21%) or retired (25%). Only 34% respondents indicated that they have a degree from university and 33% have higher diploma or a certificate from TAFE. The remaining have a degree from school or completed few years of school. Majority of respondents (51.8%) purchase beef once a week whereas 12% purchase it twice a week or more and 23.7% purchase beef 2 or 3 times a month. Table 2. Demographic variables and summary statistics of choice experiment participants

Variable

Definition

Statistics

Male Female Total participants

40% 60% 1,002

Average age in years

49

Couple family with no children Couple family with children One parent family Single person household Group household

36.4% 33.2% 6.7% 15.6% 8.1%

Average in dollars

$65,000

Full-time Part-time Retired Un-employed

36% 21% 25% 18%

A degree from university Higher diploma or a certificate from TAFE A degree from school or completed few years of school

34% 33%

2+ per week Once a week 2 or 3 a month Once a month

12% 51.8% 23.7% 12.5%

Gender

Age Household type

Household income Employment

Education

33%

Beef consumption frequency

P a g e | 11

3.3 D ATA ANALYSIS As mentioned previously, for this study we are interested in seeing how many of each (including none) given beef product each respondent is willing to purchase, which is different to the “picked products” mechanism used in traditional choice experiment surveys. In the survey respondents had the option of choosing between zero, one, two, three and four plus of the presented products. Having said that, only a few respondents chose the three or four plus option and therefore we accumulated the two, three and four plus options to “two and plus”. Thus, for data analysis, we examined the data for three options; zero purchase, one purchase and two or more purchases. Attributes in DCE with qualitative levels have typically been handled in the food economics literature by a number of dummy coded variables. For this study we have used effects coding which is an alternative to dummy coding, wherein the effects are uncorrelated with the intercept (Louviere et al., 2000). When effects coding is applied the constant term can only reflect the utility associated with the fixed comparator and misinterpretation is not possible (Bech & Gyrd-Hansen, 2005). For any individual, we might reasonably hypothesize that there is a continuously varying strength of preference for beef purchasing that would underline the “how many” they submit. For convenience and consistency with what follows, we will label that strength of preference “utility,” V. Given that there are no natural units of measurement, we can describe utility as having the following range: −∞ <  ≤ ∞

where i indicates the individual and m indicates the beef product. Individuals are invited to choose “how many” of the product they want on an integer scale from 0 to 2. Logically, then, the translation from underlying utility to a rating could be viewed as a censoring of the underlying utility,  = 0 − ∞ <  ≤  (1)  = 1  <  ≤   = 2  <  ≤ ∞ The crucial feature of the description thus far is that the viewer has (and presumably knows) a continuous range of preferences that they could express if they were not forced to provide only an integer from zero to two. Therefore, the observed rating represents a censored version of the true underlying preferences. Note that the thresholds,  , are specific to the person and number (J-1) where J is the number of possible ratings (here, Three) J-1 values are needed to divide the range of utility into J cells. The thresholds are an important element of the model; they divide the range of utility into cells that are then identified with the observed ratings. One of the admittedly unrealistic assumptions in many applications is that these threshold values are the same for all individuals. Importantly, difference on a rating scale (e.g., 0 compared to 1, 1 compared to 2) are not equal on a utility scale; hence we have a strictly nonlinear transformation captured by the thresholds, which are estimable parameters in an ordered choice model.

The model as suggested thus far provides a crude description of the mechanism underlying an observed choice. But it is simple to see how it might be improved. Any individual brings their own set of preferences for the beef attributes to the utility function, such as meat

P a g e | 12 colour, fat colour, packaging, etc. which we denote ,  , … ,  . They also bring their own aggregate of unmeasured and unmeasurable (by the statistician) idiosyncrasies, denoted  . How these features enter the utility function is uncertain, but it is conventional to use a linear function, which produces a familiar random utility function,  =  1  1 +  2  2 + ⋯ +     +  

(2)

In addition to estimating the main effects, for the analysis of the choice experiment for this study we considered effects of other beefs present when respondents where making choices. Therefore, we let  , i ≠ j , denote the availability effects of another beef j on beef i (Anderson & Wiley, 1992). Respondents were given a combination of four tasks (also allowing for repetition) from the seven possible beef alternatives. Designs in which some alternatives are present/absent allow the analyst to estimate all alternative-specific effects for each alternative independent of one another, as well as to estimate the attribute effects within each alternative, independently of one another. The key difference lies in the fact that presence/absence designs allow the variances of each alternative to differ with presence/absence. It would be possible to parametrise the variance difference as a function of presence/absence and hence, directly embed the effects in both means and variances (Louviere et al., 2000). For this study we allowed the cross-effects to be random components, , and normally distributed with zero means and variance of one. The assumption of normality will allow us to attach probabilities to the choice option. On this basis, the utility function is described as below:  =   +  + ⋯ +   +

 + 

(3)

The latent regression model would describe an underlying continuous, albeit unobservable, preference for the legislation as  . The surveyed individual, even if they could, does not provide ! , but rather, a censoring of  into three different ranges, one of which is closest to their own true preferences. By the laws of probability, the probabilities associated with the observed outcomes are "#$%&  = '|) * = "#$%& ≤  −  +  * − "#$%&, −  +  *, ' = 0,1, … , -.

(4)

A couple of normalisations are needed to identify the model parameters for this study. First, in order to preserve the positive signs of all of the probabilities, we require  > , . Second, if the support is to be the entire real line, then , = −∞ and  = +∞. Having these normalisations in place, the likelihood function for estimation of the model parameters is based on the implied probabilities, "#$%&  = '|) * = [12 − 3′56 7 − 12, − 3′56 7] > 0 , ' = 0,1, … , -.

(5)

Figure 3 shows the probabilities for an ordered choice model with three outcomes, "#$%&  = 0|) * = 1&0 − 3′56 * − 1&−∞ − 3′56 * = 1&−3+ 56 * "#$%&  = 1|) * = 1&−3′56 * − 1& − 3′56 * "#$%&  = 2|) * = 1&−∞ − 3′56 * − 1& − 3′56 * = 1 − 1& − 3′56 *

(6)

P a g e | 13

Figure 3: Underlying Probabilities for an Ordered Choice model. Estimation of the parameters is a straightforward problem in maximum likelihood estimation (Greene & Hensher, 2010). The log Likelihood function is 9$:; =

?

<

= =>

 logC12 − 3+ 56 7 − 12, − 3+ 56 7D

(7)

4 RESULTS Simulations were based on 1000 Halton draws using Python Biogeme 2.4. (Bierlaire, 2016). Attributes were measured for beef specific alternatives, however some attributes were allowed to be estimated as generic across all or a range of beef products. Coefficients that are not significant at the 90% level of accuracy have been removed. All coefficients presented are statistically significant at conventional critical levels, and the relationship with the utility function is as expected.

4.1 PRODUCT APPEARANCE Table 3 presents the estimated parameters for attributes related to product appearance on beef products. According to the consumers’ responses, the fat content of beef was the most important attribute, followed by marbling, then the type of packaging, and finally the meat colour. Australian respondents prefer a beef product with white fat colour, red meat colour, not marbled and in a modified atmosphere packed. The fat colour for diced beef has more impact on a respondents purchasing behaviour than does for mince, scotch fillet and roast beef. Beef products that are red coloured are respectively valued more than having a pink colour. However, there is no difference in preferences for pink and brown coloured beef products. With regards to level of marbling, participants were significantly consistent with all beef cuts in placing a negative value for products that are “somewhat marbled”. This negative value increases if the product is “very much marbled” (Please refer to Table 3 for the estimated parameters values).

P a g e | 14 Dissimilar to results from Carpenter et al. (2001) where US consumers preferred vacuum skin pack over modified atmosphere, Australians tend to place higher value on modified atmosphere packed over tray packed and vacuum packed product, respectively. Their premium range starts from $0.52 for mince products to $2.05 for scotch fillet steak. It is only with roast beef that Australian prefer tray packed over modified atmosphere packed product and they value a premium of $1.92 for this. Table 4 provides all the WTP estimates for product appearance attributes. Table 3: Ordered Logit Results for Beef Product Appearance Using Discrete Choice Experiment Attributes/levels Fat Colour B White Light yellow

Mince

Diced

Roast

0.276 -0.276 (-2.41)

0.295 -0.295 (-3.08)

0.122 -0.122 (-2.42)

Rump

Porterhouse

Marbling B Not marbled Somewhat marbled Very much marbled Type of Packaging B Tray packed Modified atmosphere packed Vacuum packed Sample size Number of draws

Eye

0.204

Yellow Meat Colour B Pink Red

Scotch

-0.204 (-2.61) -0.071 0.071 (2.77)

-0.071 0.071

-0.071 0.071

-0.071 0.071

-0.071 0.071

-0.071 0.071

-0.071 0.071

-0.082*

0.227 -0.082*

0.227 -0.082*

0.227 -0.082*

0.358

-0.082* (-2.36) -0.145 (-4.26)

0.227 -0.082*

-0.145

-0.145

-0.145

-0.145

-0.358 (-4.78)

-0.145

0.041 0.112 (3.37) -0.153 (-3.79) 1002 1000

0.041 0.112

0.194 -0.194 (-3.49)

0.041 0.112

0.041 0.112

0.041 0.112

0.041 0.112

-0.153

-0.153

-0.153

-0.153

-0.153

Notes: The T- ratio values are presented in the parentheses. *are statistically significant at 95% level The remaining are statistically significant at 99% level B indicates the reference level in effects codding

P a g e | 15 Table 4: Willingness to Pay Estimates for Product Appearance. Attributes/levels

Mince

Diced

Roast

Fat Colour B White Light yellow Yellow

$1.28 -$1.28

$4.10 -$4.10

$1.21 -$1.21

Meat Colour Pink B Red

B

Porterhouse

Scotch

Eye

$3.75 -$3.75

-$0.33 $0.33

-$0.98 $0.98

Marbling B Not marbled Somewhat marbled Very much marbled Type of Packaging B Tray packed Modified atmosphere packed Vacuum packed

Rump

-$0.70 $0.70

-$0.69 $0.69

-$0.74 $0.74

-$1.29 $1.29

-$1.21 $1.21

$2.24 -$0.81 -$1.43

$2.22 -$0.80 -$1.42

$2.39 -$0.86 -$1.52

$6.56 -$6.56

$3.88 -$1.40 -$2.48

$0.19

$0.57

$1.92

$0.40

$0.43

$0.75

$0.70

$0.52

$1.56

-$1.92

$1.10

$1.18

$2.05

$1.92

-$0.71

-$2.13

-$1.50

-$1.61

-$2.80

-$2.62

indicates the reference level in effects codding

4.2 LABELLING INFORMATION3 Table 5 presents the estimated parameters for attributes related to labelling information on beef products. Within Australian states included in this study, Australians value beef from New South Wales (β=0.204) and Queensland (β=0.203) relatively equally but have respectively greater value if the beef is from Tasmania (β=0.314) and then Victoria (β=0.292). In addition, Australian have a significant and negative perception if the origin of the beef product is not mentioned (β=-0.237) and this will intensify if the origin of the beef product is from China (β=-0. 776). In relation to brands, Australians place a relatively lower value on beef products branded as “Certified Angus” than other brands included in this study. Australians place a significant and positive value on beef products that are labelled grass fed and grain fed, with a slightly greater WTP premium towards the latter (β=0.098 for grain fed and β=0.083 for grass fed). Australian place a positive and significant value if the beef product claims that is an “Angus” beef. Furthermore, an organic beef production claim has the least (but positive) impact on consumers purchasing behaviour in comparison with other claim attributes included in this study (β=0.046). Australians prefer to purchase products that weight more, have longer expiry dates and indicate they contain less calories. Interestingly, products with QR codes are relatively less likely to be purchased. As expected, the price coefficient was negative for all beef products. Australians are more sensitive to a price increase for mince products than for eye fillet or scotch fillet steak. Table 6 provides all the WTP calculated values for product labelling attributes.

3

Except for price, all other labelling information attributes were considered generic across all beef products.

P a g e | 16 Table 5: Ordered Logit Results for Labelling Information Using Discrete Choice Experiment Attributes/levels Origin B Not mentioned Victoria Tasmania Queensland New south wales China Brand Certified angus Claim attributes Grass Fed Grain Fed Angus Antibiotics free Organic Generic attributes Best before Energy Weight QR code Price Sample size Number of draws

Mince

Diced

Roast

Rump

Porterhouse

Scotch

Eye

-0.237 0.292 (6.38) 0.314 (6.66) 0.203 (4.39) 0.204 (4.53) -0.776 (-15.04)

-0.237 0.292

-0.237 0.292

-0.237 0.292

-0.237 0.292

-0.237 0.292

-0.237 0.292

0.314

0.314

0.314

0.314

0.314

0.314

0.203

0.203

0.203

0.203

0.203

0.203

0.204

0.204

0.204

0.204

0.204

0.204

-0.776

-0.776

-0.776

-0.776

-0.776

-0.776

-0.174 (-4.08)

-0.174

-0.174

-0.174

-0.174

-0.174

-0.174

0.083 (3.21) 0.098 (3.67) 0.072 (2.84) 0.064 (3.08) 0.046* (2.12)

0.083

0.083

0.083

0.083

0.083

0.083

0.098

0.098

0.098

0.098

0.098

0.098

0.072

0.072

0.072

0.072

0.064

0.064

0.064

0.064

0.064

0.064

0.046

0.046

0.046

0.046

0.046

0.046

0.013 (3.02) -0.004 (-3.59) 0.002 (7.75) -0.374* (-2.1) -0.215 (-6.57) 1002 1000

0.013

0.013

0.013

0.013

0.013

0.013

-0.004

-0.004

-0.004

-0.004

-0.004

-0.004

0.002

0.002

0.002

0.002

0.002

0.002

-0.374

-0.374

-0.374

-0.374

-0.374

-0.374

-0.072 (-2.88)

-0.101 (9.47)

-0.102 (-6.01)

-0.095 (-7.78)

-0.055 (-3.75)

-0.058 (-6.08)

Notes: The T-ratio values are presented in the parentheses. *are statistically significant at 95% level The remaining are statistically significant at 99% level B indicates the reference level in effects coding

0.072

P a g e | 17 Table 6: Willingness to Pay Estimates for Labelling Information. Attributes/levels

Mince

Diced

Roast

Rump

Porterhouse

Scotch

Eye

-$1.10 $1.36 $1.46 $0.94 $0.95 -$3.61

-$3.30 $4.06 $4.38 $2.82 $2.85 -$10.80

-$2.34 $2.88 $3.11 $2.00 $2.02 -$7.67

-$2.32 $2.86 $3.08 $1.99 $2.00 -$7.60

-$2.50 $3.07 $3.31 $2.13 $2.15 -$8.17

-$4.34 $5.34 $5.76 $3.71 $3.75 -$14.22

-$4.06 $4.99 $5.38 $3.47 $3.50 -$13.28

Certified Angus

-$0.81

-$2.42

-$1.72

-$1.70

-$1.83

-$3.18

-$2.97

Claim attributes Grass Fed Grain Fed Angus Antibiotics free Organic

$0.39 $0.45 $0.33 $0.30 $0.21

$1.15 $1.36 $1.00 $0.89 $0.64

$0.82 $0.96 $0.71 $0.63 $0.45

$0.81 $0.96 $0.70 $0.62 $0.45

$0.87 $1.03 $0.75 $0.67 $0.48

$1.52 $1.79 $1.17 $0.84

$1.42 $1.67 $1.22 $1.09 $0.79

Generic attributes Best before Energy Weight QR code

$0.06 -$0.02 $0.01 -$1.74

$0.19 -$0.05 $0.02 -$1.74

$0.13 -$0.04 $0.02 -$1.74

$0.13 -$0.03 $0.02 -$1.74

$0.14 -$0.04 $0.02 -$1.74

$0.25 -$0.06 $0.03 -$1.74

$0.23 -$0.06 $0.03 -$1.74

Origin Not mentioned B Victoria Tasmania Queensland New south wales China Brand

B

indicates the reference level in effects codding

Table 7, provides the estimation of the threshold properties as well as the cross effects of the available alternatives in the choice task. The threshold parameters are all incrementally increasing ( > , ), in order to preserve the positive signs of all of the probabilities. For example in Table 7 the mince cut threshold (1) is equal to -3.622 and threshold (2) is -0.767 representing an increase value of 2.855. In terms of the effects of other available beef products on consumer’s choice, Table 7 also indicates that consumers intend to purchase diced beef, in situations where porterhouse steak and scotch fillet steak is present. They are likely to purchase roast beef when mince is presented however roast beef is likely to be chosen when scotch fillet and eye fillet is presented. Rump steak is considered when mince, diced and scotch fillet is presented but less expected when porterhouse steak is available. Porterhouse is also preferred in a task in which the roast beef is presented as well. Scotch fillet steak has priority over porterhouse steak but a disadvantage when eye fillet is available. Finally, eye fillet is more likely to be chosen over a roast beef though it nonetheless has a disadvantage when rump steak is available.

P a g e | 18 Table 7: Estimation of threshold properties and Cross effects in the ordered logit model Threshold’s &* Threshold 1 Threshold 2

Mince

Diced

Roast

Rump

Porterhouse

Scotch

Eye

-3.622 (-5.59) -0.767 (17.32)

-1.799 (-3.08) 0.277 (21.05)

-1.977 (-3.6) 0.497 (29.13)

-1.593 (-2.83) -0.154 (24.99)

-1.897 (-3.41) 0.06 (21.57)

-1.05 (-1.74) 1.218 (18.95)

-1.526 (-2.72) 0.386 (20.92)

1.264 (5.3)

1.950 (7.07) -1.258 (-8.08) 1.577 (9.55)

-1.116 (-8.29)

Cross effects &* Mince Diced Roast Rump

0.502 (2.49) 2.127 (9.59)

Port

0.821 (5.85)

-1.333 (-10.37) 0.907 (8.52)

Scotch Eye

1.181 (6.95) 0.981 (8.36)

-1.025 (-7.26)

-1.152 (-9.91)

Estimation Report Final log likelihood -13222.2 Number of 63 parameters Note: The T-ratio values are presented in the parentheses.

5 DISCUSSION With the growing interest in beef, which is considered a luxury purchase when compared with other types of meat consumption (Wong, Selvananthan, & Selvananthan, 2013), consumers are increasingly interested in beef attributes (Verbeke, 2000) and labels are a key player in consumer decision making processes. Changes to the attributes available in the label create an additional dimension of consumer utility which may be traded for other quality indicators. The literature has revealed that in real markets, consumers are faced with consumption choices over bundles of attributes that can be modelled in a stated preference framework and then a WTP measure can be calculated for each attribute. In other words, it confirms the adaptation of using DCEs among researchers to determine the share of preference a given attribute has in a particular market. Therefore, stated choice experiments provide a richer description of the attribute trade-offs that consumers are willing to make than do more traditionally used contingent valuation methods (Lusk & Schroeder, 2004). This research provides industry and policy-makers with additional information to better understand the relative value of beef product appearance and labelling information to Australian consumers. In the context of our results, although the empirical findings supports the majority of the claims from the previous studies, some controversy has been indicated. In relation to product appearance, the finding of this study was consistence with the literature that white colour fat is more preferred than light yellow and yellow fat colour; and consumers are WTP

P a g e | 19 a price premium of $1.21 for roast beef and up to $4.10 for diced beef. Respondents also attach higher value for red coloured meat for all beef products. These results are aligned with the work of Zanoli et al. (2013) and Carpenter et al. (2001). The highest premium for having red coloured meat is $1.29 for scotch fillet beef while the lowest premium is mince beef with $0.33. Marbling or intramuscular fat content has been mentioned as the primary determinant of the quality grading system. Highly marbled beef, specifically steaks, typically have better taste but more fat. Australian consumers’ taste for beef marbling is very similar to that of the United Kingdom and United States consumers (mentioned in Lusk et al. study 2003) where they prefer the least amount or no marbling (In this study marbling attribute was not considered for mince and diced products). The highest premium for having no intramuscular fat (no marbling) was at $6.56 for scotch fillet cut and the lowest was $2.22 for rump steak. In relation to labelling information, there is an inflated WTP for Australian beef origins in comparison to beef originating from China. One reason for these inflated values is that consumers may have placed a high value on origin because they used it as a cue for food safety or quality. Thus, the WTP values are highly likely to be artificial and industry and policy makers should not be misdirected to focus on costly labelling policy rather than on improving quality or safety of the beef. Australian consumers do not possess as strong a value as U.S. consumers (Abidoye, Bulut, Lawrence, Mennecke, & Townsend, 2011) have for a grass-fed claim, and the finding is in contrast with Mennecke et al. (2007), where they found no valuation for grass-fed cattle. Angus is a breed of cattle that has traditionally been associated by consumers with quality, flavour, juiciness, and tenderness because of its natural marbling (Froehlich, Carlberg, & Ward, 2009). These characteristics of Angus beef were also mentioned in a sensory assessment study by Chambaz et al. (2003). In Froehlich et al. study (2009), Canadians were WTP $1.31 for Angus beef. In this study, Australian’s also place higher value when it is claimed that the beef is Angus (β=0.072). The premium range starts with $0.33 for mince beef, around $0.70 for roast and rump steak, $0.75 for porterhouse steak, $1.00 for diced beef and $1.22 for eye fillet steak. In relation to an organic claim, similar results to our finding were concluded in Lagerkvist et al. (2014) in which they concluded that an organic claim is unlikely to stand as a relevant beef labelling attribute. Having said that, in another study by Zanoli et al. (2013), Italian consumers attached higher value to organic beef. This may perhaps indicate that the demand for organic beef products in Europe is different to Australia, due to issues related to genetically modified beef in Europe, US and Canada.

5.1 MANAGERIAL AND POLICY IMPLICATIONS The policy implications of the current study is potentially wide-ranging influencing consumers, businesses and the government. The results from this study yield in Australian consumer welfare enhancing information by better understanding consumer’s utility and the information cues on beef products that actually attract consumer interest and will be processed for subsequent use in their decision making. Typically, a unique bundle of beef intrinsic and extrinsic attributes maximizes the

P a g e | 20 consumer's satisfaction with their purchased commodity. Thus, targeted information provision is proposed as a potential solution to market failure of products. This will make information meaningful, useful and effective and will enable consumers to navigate between products more efficiently and consequently increase their satisfaction from their shopping trip. Given the large number of food choices that consumers make each day and the diversity of products, it seems unlikely that individuals allocate substantial cognitive effort and time to each decision. Furthermore, food consumers face uncertainty and demand high quality and safe food products. However this doesn’t mean that consumers are asking for the provision of evermore and too detailed information as it entails a risk of information overload, resulting in consumer indifference or loss of confidence (Verbeke, 2005). These results are also useful for business firms’ with their policy implementation and product differentiation strategy. Consumers may use heuristics to screen out whether or not to investigate a product category in detail (Swait, 2001). The results from this study not only informs business firms’ on which intrinsic and extrinsic attributes of beef products will trigger changes in decision strategy, but also highlighting the importance of the layout and positioning of the beef products relative to each other in the supermarket chiller shelves. Furthermore, results will extend industry firms understanding of a “better product design” from the consumer’s perspective. This information will increase the efficiency gain in an oligopolistic/monopolistic competition market where for example firm A and B produces the exact same beef cut but with different packaging design. Finally results from this study will complement the collective efforts of governments in influencing the decision-making environment of food producers, food consumers and food marketing agents in order to further social objectives. These objectives nearly always include improved nutrition for inadequately nourished citizens and more rapid growth in domestic food production. In this study it has been revealed that consumers WTP decreases by increased marbling density in the beef product. For example, for a “not marbled” roast consumers are willing to pay $2.24 per kilo, however, this value decreases to -$0.81 if the roast is “somewhat” marbled. If the roast beef is “very much” marbled then the WTP drops to -$1.43 per kilo. The fact that Australian consumers behaviour toward the amount of marbling is homogenous for all the studied beef cuts highlights that Australian consumers are avoiding relatively unhealthy beef products although the may provide a better taste. Knowing this information will reinforce government with their plans and policies for promoting health claims on the products. Moreover, obtained results supports the mandatory labelling (such as country of origin, name of the product, date marking, details of the manufacturer, nutritional and health claims) policy that the Australian government is currently undertaking. This can help government to strengthen the promotion of scientific knowledge regarding the beef production system and to improve the general public’s knowledge about the system, in order to generate effective market demand. Finally, the outcome of this research, and its comparison with similar studies around the world, informs the Australian government that homogeneity in preferences for beef products does not hold across consumers from different nations and country-specific research is required to illustrate consumers’ preference.

P a g e | 21 5.1.1 HYPOTHETICAL SCENARIO In the following paragraphs we will show an example of how the results from this study can help business firms with their cost-benefit estimation for a newly developed product and its relative market impact within the same beef cuts and across all other cuts. Bellow a hypothetical scenario is presented for a better explanation of the latter statement. Imagine firm X is investigating the development of a new product and is seeking to know which new product would be attractive to consumers and thus has the highest probability of being purchased in the Australian market. For this reason the firm investigates nine different beef cuts where seven of them comes from a farm located in Tasmania and the remaining two products are sourced from a New South Wales farmer. The products details are listed in Table 8. Other attributes such as brand and the nutrition information, estimated from our model, has been considered to be the same among all the nine products and we have avoided adding them to Table 8. Table 8: Product detailed information for the hypothetical scenario Mince (B) Yellow

Diced

Roast

Fat Colour

Mince (A) White

White

Rump steak White

Porterhouse steak White

Scotch steak (A) White

White

Marbling

N/A

N/A

Meat Colour Origin Type of Packaging

Red TAS VP

Type of fed Antibiotic free Organic Package weight (gr) Price ($AUS/per kg) Presence of QR code

Eye fillet steak White

Red TAS VP

Scotch steak (B) Light yellow Very much marbled Brown NSW TP

N/A

Not marbled

Not marbled

Not marbled

Not marbled

Brown NSW MAP

Red TAS VP

Red TAS VP

Red TAS VP

Red TAS VP

Grass Yes Yes 500 16.5

Grain No No 700 15

Grass Yes Yes 500 16.5

Grass Yes Yes 500 19.5

Grass Yes Yes 250 21

Grass Yes Yes 250 37.5

Grass Yes Yes 250 41.5

Grain No No 450 39

Grass Yes Yes 150 50.5

Yes

No

Yes

Yes

Yes

Yes

Yes

No

Yes

Not marbled Red TAS VP

Figure 4 illustrates the probability of purchasing zero, one or two or more for each of the nine products in this hypothetical scenario. The business firm can observe from this figure that on average the new roast product is more attractive as it has the highest accumulative probability of being purchased one, two or more and the eye fillet steak product has the lowest. Furthermore mince (A) has slightly a higher probability of being purchased in comparison to mince (B). The same comparison can be done for scotch fillet steak (A) and (B). Furthermore, by using the product weights, market share of these new products can be estimated based in total kilogram sold for each cut of beef. Figure 5 demonstrates the market shares for the hypothetical new beef products and shows that the new roast beef product, with 28.3% of the market share, has the highest among all other products.

P a g e | 22

8.7%

5.1%

6.4% 19.4%

15.2%

8.3%

4.5%

2.2% 12.7%

2.8% 9.4%

85.1%

87.8%

19.0%

24.6%

27.1%

28.7%

23.7% 21.7%

47.8%

62.6%

76.5%

70.3%

66.5%

63.1%

68.0%

32.8%

Mince (A)

Mince (B)

Diced

Zero purchase

Roast

Rump steak Porthouse Scotch fillet Scotch fillet steak steak (A) steak (B)

One purchase

Eye fillet steak

Two & plus purchases

Figure 4: Probability of purchasing none, one and two & plus of each beef products.

28.3%

14.7%

13.9%

14.4%

9.1% 6.7% 4.7%

Mince (A)

Mince (B)

Diced

Roast

4.3%

Rump steak Porthouse Scotch fillet Scotch fillet

3.9%

Eye fillet

Figure 5: Market shares using the product weights Figure 6 presents purchase probability trends for the roast product when price is ranging from $15 to $65 per kilogram. It can be observed that as the price increases the probability of zero purchase also increases. This chart will also allow business firms to predict revenue gain for different pricing mechanisms such as identifying the optimal price that will generate the highest revenue. Figure 7 has been produced to show the revenue gained for an increment in the price value for the new roast product. Results are generated using the probabilities for purchasing zero, one and two plus purchases (average quantity purchases

P a g e | 23 mentioned by the responses has been used for the two plus category) in Figure 6 for each specific price and a hypothetical market size of 1000 consumers. The graph highlights that the maximum revenue occurs if the product is priced at $27 per kilogram.

Figure 6: Purchase probability trend for the roast product

Figure 7: Product revenue for the roast product (market size of 1000) and the optimal price

P a g e | 24

6 CONCLUSION The results obtained from the Australian household survey provide interesting information on the relative importance of intrinsic and extrinsic beef attributes to Australian consumers when selecting beef products. This study highlighted how choice experiments can provide insights for industry and policymakers with additional information to better understand relative value with regards to consumers’ safety, ethical, nutritional and aesthetic concerns. It benefits the policy makers and stock holders to introduce more consumer-desirable products as well as estimating the economic benefits of a given policy measure. It presents information to decision makers about how consumers might be balancing trade-offs inherent in the decision-making process. Extant literature from non-DCE domains reveals that consumers’ value for purchasing a beef product is driven not only by the beef attributes but by sets of other variables such as the location, socio-economic background, cultural beliefs, level of knowledge, purchasing behaviour, consumption attitudes, environmental sustainability preferences, religious beliefs and so on. Thus a limitation to this study and as a further research stream it is suggested to investigate individual differences (i.e. segmentation) and identify consumers segments for better targeting consumers as well as marketing strategies for beef producers and marketers. Finally, an alternative suggestion for future directions of research is to investigate consumers processing resources and cognitive efforts for each decision. Given the sheer number of decisions involved across the many facets of people’s lives, it seems unlikely that individuals allocate substantial cognitive effort and time to each decision. Indeed, decisions regarding small budget items like food or consumer packaged goods would seem more likely to be relegated to some form of habitual choice behaviour.

Funding This work has been funded through the “Pathways to market: transforming food industry futures through improved sensing, provenance and choice” Australian Research Council IH120100021 grant.

P a g e | 25 References Abidoye, B. O., Bulut, H., Lawrence, J. D., Mennecke, B., & Townsend, A. M. (2011). US consumers' valuation of quality attributes in beef products. Journal of Agricultural and Applied Economics, 43(1), 1. Acebrón, L. B., & Dopico, D. C. (2000). The importance of intrinsic and extrinsic cues to expected and experienced quality: an empirical application for beef. Food Quality and Preference, 11(3), 229-238. Adamowicz, W., Boxall, P., Williams, M., & Louviere, J. (1998). Stated preference approaches for measuring passive use values: choice experiments and contingent valuation. American Journal of Agricultural Economics, 80(1), 64-75. Anderson, D. A., & Wiley, J. B. (1992). Efficient choice set designs for estimating availability cross-effects models. Marketing Letters, 3(4), 357-370. Ardeshiri, A. (2014). Evaluating urban services using economic valuation techniques: towards better urban environmental quality and promotion of sustainable development. Asioli, D., Varela, P., Hersleth, M., Almli, V. L., Olsen, N. V., & Næs, T. (2017). A discussion of recent methodologies for combining sensory and extrinsic product properties in consumer studies. Food Quality and Preference, 56, 266-273. Authority, A. N. Z. F. (2002). Australia New Zealand Food Standards Code. Victoria: Anstat Pty Ltd. Banović, M., Chrysochou, P., Grunert, K. G., Rosa, P. J., & Gamito, P. (2016). The effect of fat content on visual attention and choice of red meat and differences across gender. Food Quality and Preference, 52, 42-51. Bech, M., & Gyrd-Hansen, D. (2005). Effects coding in discrete choice experiments. Health Economics, 14(10), 1079-1083. Becker, L., van Rompay, T. J., Schifferstein, H. N., & Galetzka, M. (2011). Tough package, strong taste: The influence of packaging design on taste impressions and product evaluations. Food Quality and Preference, 22(1), 17-23. Ben-Akiva, M., & Morikawa, T. (1990). Estimation of switching models from revealed preferences and stated intentions. Transportation Research Part A: General, 24(6), 485-495. Bierlaire, M. (2016). PythonBiogeme: a short introduction. Report TRANSP-OR 160706 ,Series on Biogeme. In: Transport and Mobility Laboratory, School of Architecture, Civil and Environmental Engineering, Ecole Polytechnique Fédérale de Lausanne, Switzerland. Bradley, M., & Daly, A. (1994). Use of the logit scaling approach to test for rank-order and fatigue effects in stated preference data. Transportation, 21(2), 167-184. Caputo, V., Scarpa, R., & Nayga, R. M. (2017). Cue versus independent food attributes: the effect of adding attributes in choice experiments. European Review of Agricultural Economics, 44(2), 211-230. Carpenter, C. E., Cornforth, D. P., & Whittier, D. (2001). Consumer preferences for beef color and packaging did not affect eating satisfaction. Meat Science, 57(4), 359-363. Chambaz, A., Scheeder, M., Kreuzer, M., & Dufey, P.-A. (2003). Meat quality of Angus, Simmental, Charolais and Limousin steers compared at the same intramuscular fat content. Meat Science, 63(4), 491-500. Chung, C., Boyer, T., & Han, S. (2009). Valuing quality attributes and country of origin in the Korean beef market. Journal of agricultural economics, 60(3), 682-698.

P a g e | 26 Colella, F., & Ortega, D. L. (2017). Where's the beef? Retail channel choice and beef preferences in Argentina. Meat Science, 133, 86-94. De Pelsmaeker, S., Dewettinck, K., & Gellynck, X. (2013). The possibility of using tasting as a presentation method for sensory stimuli in conjoint analysis. Trends in food science & technology, 29(2), 108-115. Dijksterhuis, G. (2016). New product failure: Five potential sources discussed. Trends in food science & technology, 50, 243-248. Egan, A., Ferguson, D., & Thompson, J. (2001). Consumer sensory requirements for beef and their implications for the Australian beef industry. Australian Journal of Experimental Agriculture, 41(7), 855-859. Endrizzi, I., Torri, L., Corollaro, M. L., Demattè, M. L., Aprea, E., Charles, M., et al. (2015). A conjoint study on apple acceptability: Sensory characteristics and nutritional information. Food Quality and Preference, 40, 39-48. Fiszman, B. P. (2012). More than meets the mouth: Assessing the impact of the extrinsic factors on the multisensory perception of food products. Froehlich, E. J., Carlberg, J. G., & Ward, C. E. (2009). Willingness-to-Pay for Fresh Brand Name Beef. Canadian Journal of Agricultural Economics/Revue canadienne d'agroeconomie, 57(1), 119-137. Greene, W. H., & Hensher, D. A. (2010). Modeling ordered choices: A primer: Cambridge University Press. Grunert, K. G. (2015). The common ground between sensory and consumer science. Current Opinion in Food Science, 3, 19-22. Hanemann, M. W., & Kanninen, B. (1998). The statistical analysis of discrete-response cv data, California agricultural experiment station giannini foundation of agricultural economics. Working Paper No 798. California Agricultural Experiment Station, Berkeley. Hanley, N., MacMillan, D., Wright, R. E., Bullock, C., Simpson, I., Parsisson, D., et al. (1998). Contingent valuation versus choice experiments: estimating the benefits of environmentally sensitive areas in Scotland. Journal of agricultural economics, 49(1), 1-15. Hanley, N., Mourato, S., & Wright, R. E. (2001). Choice modelling approaches: a superior alternative for environmental valuatioin? Journal of Economic Surveys, 15(3), 435462. Hanley, N., Wright, R. E., & Koop, G. (2002). Modelling recreation demand using choice experiments: climbing in Scotland. Environmental and Resource Economics, 22(3), 449-466. Hood, D., & Riordan, E. (1973). Discolouration in pre-packaged beef: Measurement by reflectance spectrophotometry and shopper discrimination. International Journal of Food Science & Technology, 8(3), 333-343. Hoppert, K., Mai, R., Zahn, S., Hoffmann, S., & Rohm, H. (2012). Integrating sensory evaluation in adaptive conjoint analysis to elaborate the conflicting influence of intrinsic and extrinsic attributes on food choice. Appetite, 59(3), 949-955. Ikenson, D. J. (2004). Uncool Rules: Second Thoughts on Mandatory Country of Origin Labeling. Jacobsen, L. F., Grunert, K. G., Søndergaard, H. A., Steenbekkers, B., Dekker, M., & Lähteenmäki, L. (2014). Improving internal communication between marketing and technology functions for successful new food product development. Trends in food science & technology, 37(2), 106-114.

P a g e | 27 Jaeger, S. R. (2006). Non-sensory factors in sensory science research. Food Quality and Preference, 17(1), 132-144. Jin, H. J., Skripnitchenko, A., & Koo, W. W. (2004). The effects of the BSE outbreak in the United States on the beef and cattle industry: Center for Agricultural Policy and Trade Studies, Department of Agribusiness and Applied Economics, North Dakota State University. Köster, E. P. (2009). Diversity in the determinants of food choice: A psychological perspective. Food Quality and Preference, 20(2), 70-82. Lagerkvist, C. J., Berthelsen, T., Sundström, K., & Johansson, H. (2014). Country of origin or EU/non-EU labelling of beef? Comparing structural reliability and validity of discrete choice experiments for measurement of consumer preferences for origin and extrinsic quality cues. Food Quality and Preference, 34, 50-61. Lagerkvist, C. J., & Hess, S. (2014). Consumer acceptance of an EU/non-EU label of origin of beef: An analysis of attribute attendance in a discrete choice experiment. In, presentation at the EAAE 2014 Congress ‘Agri-Food and Rural Innovations for Healthier Societies’, August. Lähteenmäki, L. (2013). Claiming health in food products. Food Quality and Preference, 27(2), 196-201. Lancaster, K. (1966). A new approach to consumer theory. Journal of Political Economy, 74, 132-157. Lancsar, E., & Swait, J. (2014). Reconceptualising the external validity of discrete choice experiments. Pharmacoeconomics, 32(10), 951-965. Lewis, K. E., Grebitus, C., Colson, G., & Hu, W. (2017). German and British Consumer Willingness to Pay for Beef Labeled with Food Safety Attributes. Journal of agricultural economics, 68(2), 451-470. Loebnitz, N., Schuitema, G., & Grunert, K. G. (2015). Who buys oddly shaped food and why? Impacts of food shape abnormality and organic labeling on purchase intentions. Psychology & Marketing, 32(4), 408-421. Loose, S. M., & Szolnoki, G. (2012). Market price differentials for food packaging characteristics. Food Quality and Preference, 25(2), 171-182. Louviere, J., & Hensher, D. A. (1982). On the Design and Analysis of Simulated Choice or Allocation Experiments in Travel Choice Modeling. Transportation Research Record, 890, 11-17. Louviere, J., Hensher, D. A., & Swait, J. D. (2000). Stated choice methods: analysis and applications: Cambridge University Press. Lusk, J. L., & Schroeder, T. C. (2004). Are choice experiments incentive compatible? A test with quality differentiated beef steaks. American Journal of Agricultural Economics, 86(2), 467-482. Mark, T. L., & Swait, J. (2004). Using stated preference and revealed preference modeling to evaluate prescribing decisions. Health Economics, 13(6), 563-573. Mazzotta, M. J., & Opaluch, J. J. (1995). Decision making when choices are complex: a test of Heiner's hypothesis. Land Economics, 500-515. McEachern, M., Seaman, C., Padel, S., & Foster, C. (2005). Exploring the gap between attitudes and behaviour: Understanding why consumers buy or do not buy organic food. British Food Journal, 107(8), 606-625. Meat & Livestock Australia, M. (2016). Australian Beef the Greatest. In. https://www.mla.com.au/marketing-beef-and-lamb/domestic-marketing/beefcampaigns/.

P a g e | 28 Mennecke, B. E., Townsend, A. M., Hayes, D. J., & Lonergan, S. M. (2007). A study of the factors that influence consumer attitudes toward beef products using the conjoint market analysis tool. Journal of Animal Science, 85(10), 2639-2659. Morales, R., Aguiar, A., Subiabre, I., & Realini, C. (2013). Beef acceptability and consumer expectations associated with production systems and marbling. Food Quality and Preference, 29(2), 166-173. OECD/Food and Agriculture Organization of the United Nations. (2015). OECD-FAO Agricultural Outlook 2015. In. OECD Publishing, Paris. Olson, J. C., & Jacoby, J. (1972). Cue utilization in the quality perception process. In, SVproceedings of the third annual conference of the association for consumer research. Ortega, D. L., Hong, S. J., Wang, H. H., & Wu, L. (2016). Emerging markets for imported beef in China: Results from a consumer choice experiment in Beijing. Meat Science, 121, 317-323. Peterson, H. H., & Burbidge, L. D. (2012). Japanese Consumers' Valuation of US Beef and Pork Products after the Beef Trade Ban. Journal of Agricultural and Resource Economics, 58-76. Realini, C., Kallas, Z., Pérez-Juan, M., Gómez, I., Olleta, J., Beriain, M., et al. (2014). Relative importance of cues underlying Spanish consumers’ beef choice and segmentation, and consumer liking of beef enriched with n-3 and CLA fatty acids. Food Quality and Preference, 33, 74-85. Reicks, A. (2006). Consumer motivations and the impact of brand on purchasing preferences of fresh beef. Texas Tech University. Risius, A., & Hamm, U. (2017). The effect of information on beef husbandry systems on consumers’ preferences and willingness to pay. Meat Science, 124, 9-14. Rose, J. M., Bliemer, M. C. J., Hensher, D. A., & Collins, A. T. (2008). Designing efficient stated choice experiments in the presence of reference alternatives. Transportation Research Part B: Methodological, 42(4), 395-406. Sánchez, M., Beriain, M., & Carr, T. (2012). Socio-economic factors affecting consumer behaviour for United States and Spanish beef under different information scenarios. Food Quality and Preference, 24(1), 30-39. Sasaki, K., & Mitsumoto, M. (2004). Questionnaire-based study on consumer requirements for beef quality in Japan. Animal Science Journal, 75(4), 369-376. Scarpa, R., & Rose, J. M. (2008). Design efficiency for non-market valuation with choice modelling: how to measure it, what to report and why*. Australian Journal of Agricultural and Resource Economics, 52(3), 253-282. Scozzafava, G., Corsi, A. M., Casini, L., Contini, C., & Loose, S. M. (2016). Using the animal to the last bit: Consumer preferences for different beef cuts. Appetite, 96, 70-79. Steenkamp, J.-B. E. (1997). Dynamics in consumer behavior with respect to agricultural and food products. In, Agricultural marketing and consumer behavior in a changing world: Springer. Swait, J. (2001). A non-compensatory choice model incorporating attribute cutoffs. Transportation Research Part B: Methodological, 35(10), 903-928. Swait, J., & Adamowicz, W. (1996). The effect of choice environment and task demands on consumer behavior: discriminating between contribution and confusion. Turley, L. W., & Milliman, R. E. (2000). Atmospheric Effects on Shopping Behavior. Journal of Business Research, 49(2), 193-211. Umberger, W. J. (2005). Will consumers pay a premium for country-of-origin labeled meat?

P a g e | 29 van Ooijen, I., Fransen, M. L., Verlegh, P. W., & Smit, E. G. (2017). Packaging design as an implicit communicator: effects on product quality inferences in the presence of explicit quality cues. Food Quality and Preference. Van Wezemael, L., Caputo, V., Nayga, R. M., Chryssochoidis, G., & Verbeke, W. (2014). European consumer preferences for beef with nutrition and health claims: A multicountry investigation using discrete choice experiments. Food policy, 44, 167-176. Verbeke, W. (2000). Influences on the consumer decision-making process towards fresh meat-Insights from Belgium and implications. British Food Journal, 102(7), 522-538. Verbeke, W. (2005). Agriculture and the food industry in the information age. European Review of Agricultural Economics, 32(3), 347-368. Verbeke, W., & Ward, R. W. (2006). Consumer interest in information cues denoting quality, traceability and origin: An application of ordered probit models to beef labels. Food Quality and Preference, 17(6), 453-467. Wong, L., Selvananthan, E., & Selvananthan, S. (2013). Changing Pattern of Meat Consumption in Australia. In: Griffith Business School, May. Xue, H., Mainville, D., You, W., & Nayga Jr, R. M. (2009). Nutrition knowledge, sensory characteristics and consumers’ willingness to pay for pasture-fed beef. In, Milwaukee, WI: Agricultural & Applied Economics Association’s 2009 AAEA & ACCI Joint Annual Meeting. Zanoli, R., Scarpa, R., Napolitano, F., Piasentier, E., Naspetti, S., & Bruschi, V. (2013). Organic label as an identifier of environmentally related quality: A consumer choice experiment on beef in Italy. Renewable Agriculture and Food Systems, 28(01), 70-79.

P a g e | 30

Appendix Table A.1: Research articles used choice experiments to study consumer preference for intrinsic and extrinsic attributes for beef products. Year

Country

Author

Estimation Model

Main Findings Consum er preference for beef colour were rated respectively red, purple and brown and for packaging, overwrap with polyvinyl chloride was the most preferred followed by vacuum skin pack and then modified atmosphere packaging. It was also concluded that consumer preferences for beef colour and packaging influenced likelihood to purchase, but did not bias eating satisfaction. On average, consumers were willing to pay a 30.6% premium for corn-fed beef. Sixty-two percent of the participants were willing to pay an average premium of $1.61 more per pound for the corn-fed beef, 23% of the consumers were willing to pay a premium of $1.36 more per pound for the grass-fed beef, only 15% of the consumers were indifferent.

2001

United States

Carpenter, Cornforth, and Whittier

Fishers least significant difference procedure

2002

United States

Umberger, Feuz, Calkins, & Killinger Mann

Multinomial logit model

2003

France, Germany, United Kingdom, and the United States

Lusk, Roosen, and Fox

Random parameters logit model

French consumers place a higher value on beef from cattle that have not been administered added growth hormones than U.S. consumers. European consumers place a much higher value on beef from cattle that have not been fed genetically modified corn than U.S. consumers.

2003

Norway

Alfnes and Rickertsen

Ordered least square

Norwegian consumers preferred hormone free beef to hormone tested beef.

2004

Norway

Alfnes

Mixed logit model

2005

France, Germany, and United Kingdom

Tonsor, Schroeder, Fox, and Biere

Random parameters logit models

2007

United States, Canada, Japan, and Mexico

Tonsor, Schroeder, Pennings, & Mintert

Mixed logit models

2007

United States

Loureiro and Umberger

Conditional multinomial logit model

2009

Canada, Japan, Mexico, and the United States

Tonsor, Schroeder, Pennings, & Mintert

Mixed logit model

2009

United States

Umberger, Thilmany McFadden, & Smith

Probit model

Norwegian consumers on average, after domestic beef, prefer beef from neighbouring Sweden to beef from more distant countries, and beef from developed countries to beef from less developed countries. US hormone-free beef was perceived as being as good as Irish beef, whereas US hormone-treated beef was perceived as being inferior to Irish and Botswanan beef. European consumers are significantly heterogeneous in their preferences for beef steak attributes. French and German consumers have a higher willingness to pay to avoid genetically modified feed use than British consumers, while German and British consumers would pay more for growth hormone-free beef. French and German consum ers are willing to pay for farm-specific source verification. Japanese and Mexican consumers have WTP preferences that are nonlinear in the level of food safety risk reduction. Conversely, U.S .and Canadian consumers appear to possess linear preferences. U.S. consumers value certification of USDA food safety inspection more than any of the other choice set attributes, including country-of-origin labelling, traceability and tenderness. Consum ers in Canada, Japan, Mexico, and the United States have willingness to pay preferences that are nonlinear in the level of food safety risk reduction. In particular, consumers in Japan and Mexico have preferences that are convex and consumers in Canada and the United States have preferences concave in the level of food safety enhancement. The results indicate that the probability a consumer will pay more or less of a premium depends on purchase behaviour and shopping location, stated importance of production attributes, awareness and interest in private and civic agricultural issues, in addition to some typical demographic variables such as income.

P a g e | 31

2009

United States

Gao & Schroeder

Mixed logit model

It was concluded that for different types of consumer’s WTP for beef steak attributes varies significantly and their responses to new attribute information are different. Over all, there was no significant difference between the responses to new information between consumer groups. However, in the case where cue attributes existed, consumers with small family size, less children, lower income, are single and younger, respond significantly intensive to the new information than other consumers.

Estimation Model

Main Findings

Table A.1: (continued) Year

Country

Author

2010

United States

Xue, Mainville, You, & Nayga

Probit model

Finding shows that palatability attributes play a central role in determining consumers’ preferences and WTP. Furthermore, consumers’ nutrition knowledge, beef consumption behaviour, health condition, living alone status and household size have significant impacts on their WTP for grass-fed beef.

2011

United States

Abidoye, Bulut, Lawrence, Mennecke, and Townsend

Conditional and random parameters logit model

U.S. consumers have strong valuation for traceability, grass-fed, and U.S. origin attributes in a standard rib-eye steak.

2012

Japan

Aizaki, Sawada, Sato, and Kikkawa

Random parameters logit

Japanese consumers are greatly influenced by the BSE-test status and they considered it a crucial food safety requirement for beef products.

2012

United States

Lim, Hu, Maynard, & Goddard

Mixed logit model and Conditional logit model

Results showed that consum ers are willing to pay a premium for traceable and BSE-tested beef. Furthermore, concerns about BSE, influence of food manufacturer/ retailers over food safety, risk perception and risk attitude were factors that influence consumers’ WTP for traceable and BSE-tested beef.

2012

Japan

Peterson & Burbidge

Random parameter logit model

Japanese consumers prefer meat that is produced with GMfree feed and also meets the full organic standards.

2013

United States and Germany

Grebitus, Jensen, & Roosen

Mixed logit models

Consumers prefer cherry red ground beef with a 14 day shelf life. Americans are willing to pay higher prices for longer shelf life than Germans. Germans show a significantly higher WTP for cherry red ground beef than US Americans.

2013

United States

Kar and Hu

Error component logit model

U.S. consumers are willing to pay significantly less for imported steaks. Other beef attributes such as traceable, BSE and tenderness are respectively important to U.S. consumers.

2013

United States

Tonsor, Schroeder, and Lusk

Interval-censored model

U.S. consumers are willing to pay premiums for products carrying origin labels such as “Product of North America” or “Product of United States”.

2013

Chile

Morales, Aguiar, Subiabre, & Realini

General Linear Model

2013

Italy

Zanoli et al.

Multinomial logit model and mixed logit model

2014

China

Wang, Wu, Zhu, Wang, and Xu

Mixed logit model and a latent class model

Three groups of consumers, ‘lean beef lovers’ (25.5%), ‘high expectation consum ers’ (53.4%) and ‘grass-fed beef lovers’ (21.1%), were identified based on their expected acceptability. Information about the marbling level and production systems generated positive expectations and increased acceptability of beef with low marbling levels and beef from grazing animals. Italian consumers attach respectively higher value to organic meat, Italian origin, free range, local breed and red colour beef product. In addition, WTP for GM-fed beef, which is not yet sold in Italy, is well below current conventional beef prices. Quality certification (“government certification” level) was the most important characteristic, followed by appearance (“fresh-looking” level), and traceability information (“covering farming, slaughter, and processing, and circulation and marketing” level).

P a g e | 32

P a g e | 33 Table A.1: (continued) Year

Country

Author

Estimation Model

Main Findings

2014

Spain

Realini et al.

Latent class model

In general consumers preferred beef that is, locally produced, enriched with omega-3, has slight visible fat and it has a bright red colour.

Random parameters logit model

Higher willingness to pay exists for verified animal health stamp in both Kumasi and Sunyani compared to assured nutritional label, food and drugs board food safety certification license. Willingness to pay estimates in Kumasi were higher for assured nutritional label, food and drugs board food safety certification license compared to Sunyani. Consumer preferences for food safety inspection and certification, and nutritional label are explained by age, incom e and education in Sunyani Municipality whereas preferences for verified animal health status, food safety inspection and certification, and nutritional label are influenced by age, income, education and gender in Kumasi Metropolis.

Hierarchical Bayesian models

Labels with specific country-of-origin information instead of a wider EU/non-EU designation were the most determining attribute in this study. Quality attributes related to the current trends of organic food, ethical food production and functional food were found to be of lower importance and therefore not likely to stand as relevant beef labelling attributes.

DomínguezTorreiro

Multinomial logit model

Preference estimates for beef attributes are found to be sensitive to the design generation process. With regards to the studied attributes Cantabrian consumers value origin, organoleptic quality and nutritional quality, respectively, to be the most important factors when purchasing a beef products. Italian beef consumers strongly prefers certified and local Italian origin (PGI or local brands) and dislikes foreign breeds, even if their organoleptic characteristics are usually very high. The prices ranged studied showed that it play a marginal role in interfering consumer choices.

2014

2014

2014

Ghana

Owusu-Sekyere, Owusu, & Jordaan

Sweden

Lagerkvist, Berthelsen, Sundström, & Johansson

Spain

2014

Italy

Scozzafava, Casini, & Contini

Conditional multinomial logit model

2014

Sweden

Lagerkvist & Hess

Generalised linear random effects panel model

Sweden consumers respectively place higher value for attributes such as animal welfare, organic production, traceability and type of anim al feed.

Van Wezemael et al.

Multinomial logit and error component models

They concluded that consumer valuation of nutritional and health claims varies across countries. In Belgium, the Netherlands and France, nutrition and health claims on saturated fat yielded higher utilities than claims on protein and/or iron, while the opposite was found among consumers in the UK. The results imply that marketing opportunities related to nutrition and health claims on beef are promising, but that different nutritional marketing strategies are necessary within different countries.

Spain

Kallas, Realini, & Gil

Error Component Model, the Heteroscedastic Extreme Value model and the Random Parameters Logit model

Spanish consumers who received inform ation attach higher preference for enriched meat with polyunsaturated fatty acids. The utility associated with the higher content of fat increase for informed consumers, showing a substitute effect. Informed consumers are willing to accept meat with a higher amount of visible fat if it is enriched with beneficial fatty acids.

Portugal

Viegas, Nunes, Madureira, Fontes, & Santos

Random parameter logit model

Portuguese consumers place the highest WTP value on food safety, followed by animal welfare and, finally, by environmental protection. Findings also suggest that the combination of the three attributes has effects on the estimated WTP, which are interpreted as substitution relationships.

Ortega, Hong, Wang, & Wu

Random parameter logit model

Beijing beef consum ers are willing to pay for quality attributes including enhanced food safety claims, animal welfare practice information, and organic food certification. Beijing consumers value food safety information the most, and are willing to pay more for Australian beef products than for US or

2014

2014

2014

2016

Belgium, France, the Netherlands, United Kingdom

China

P a g e | 34 domestic (Chinese) beef.

Table A.1: (continued) Year

Country

Author

Estimation Model

Latent class model

2016

Italy

Scozzafava, Corsi, Casini, Contini, & Loose

2016

Canada

Kar H Lim & Hu

Mixed logit model

2017

Argentina

Colella & Ortega

Latent class model and Random parameters logit

2017

Germany and United Kingdom

Lewis, Grebitus, Colson, & Hu

Random parameter logit model

2017

Germany

Risius & Hamm

Random parameter logit model

2017

United States

Caputo et al.

Error Component Model

Main Findings The results shows that meat cut is the most important factor when choosing bovine meat followed by quality certification (origin), production technique, the type of breed and price. In terms of consumption occasions, we observe significantly lower price sensitivity for marbled steaks and cutlets for special occasions compared to normal occasions. Segmentation analysis shows that while the choices of two segments (comprising about 40% of the sample) are mostly driven by extrinsic product attributes, the remaining segments are mostly driven by meat cuts. These varying preferences are also reflected in the purchase portfolios of the different segments, while less variability is detected from a socio-demographic perspective. Canadian consumers are mostly indifferent between products labelled generically as “local” and as “local: from within 160 km,” implying that the 160-km radius fits perception of local of the representative consumers. Additionally, consumers are willing to pay significantly more for home-province products over products generically labelled “local.” This study also found significant positive WTP for enhanced bovine spongiform encephalopathy tested beef as well as for grass- over grain-fed beef. Argentinean beef consumers were described in two groups of consumers with heterogeneous preferences for retailer attributes, Service and Convenience Oriented customers. Convenience Oriented customers are prone to be willing to pay for organic, origin, and family farm certification, and Service Oriented customers are not. British consumers had the lowest WTP for beef from Argentina and German consumers had the lowest WTP for beef from Great Britain. The hormone-free label was the relatively most preferred label by consumers in both countries, and by those who considered food safety issues to affect their meat consumption patterns. German consumers exhibited a high preference for enhanced husbandry conditions and organic production. Without further information about the husbandry conditions, ‘organic’ and ‘pasture-based’ production labelling was most likely to influence buying decisions. When informed about the conditions of ‘extensive suckler cow husbandry’, consumers were most likely to be motivated by the label ‘extensive suckler cow husbandry’, followed by ‘organic production’; accordingly, willingness to pay for a beef steak was highest for ‘extensive suckler cow husbandry’. Results suggest that the way a subject processes food attributes depends not only on the design dimensions but also on food attributes’ functional roles. When complexity of designs increases, models that account for different sources of heterogeneity have better fit to the data. In terms of beef attribute Americans have ranked US Certified, Tender and Lean respectively as the most important attributes when purchasing beef.

P a g e | 35 Highlights • • • • •

We analysed Australian consumer values for intrinsic and extrinsic beef attributes. The study design accounted for the ordinal nature of consumer purchases. An ordered logit model was applied to explore beef purchase behaviour. We provided managerial and policy implications for developing new beef products. Information cues regarding health influence Australian consumers’ decisions.