Unlocking the monetary value: investigating the importance of quality information in Australian red meat chains

Potential benefits and costs of sharing product quality information

Zapata and Carpio (2014) demonstrated that a producer’s WTP for a new production input is equal to the perceived difference between anticipated profit levels with and without the change: in economic terms, a net marginal benefit. Accordingly, Olynk et al. (2010a) suggested that farmers can provide information about production process attributes, to maximise profits. Farmers can pursue a price premium through provision of extra information to signal the quality of sires, animals and the meat produced, which is enabled by a merit- or grid-based pricing mechanism (Johnson and Ward 2006). IQQ improvement is an extension of this principle.

Information in the FB reports in the beef industry contain a number of eating quality-related factors, including hump height, pH, fat colour, meat colour and MSA marbling score (Polkinghorne et al. 2008), and can be received and reviewed by MSA registered cattle suppliers through an online feedback system (known as myMSA) in Australia. Although the benefits of FB information and associated IQQ improvement do not directly increase revenue, producers can use this information to improve their cattle’s compliance and eating quality performance (Bowler 2014; Meat and Livestock Australia 2021), ultimately benefitting processors by improving processing efficiency and quality consistency (Goers and Craig 2008). Participants incur no extra costs beyond MSA membership fees, but utilising this information for decision-making remains a challenge due to variations in farmers’ knowledge, and capacity to utilise and convert this information to inputs for livestock management and production. This process necessitates that farmers diligently record pertinent on-farm data, and establish a meaningful connection between the FB information received and their recorded information. As MSA sheepmeat standards are still at an early stage of development, this is included in the current study as hypothetical information.

The current paper defines PM as information that encompasses the particular methods and strategies utilised in the rearing and management of cattle and sheep for meat production. This information encompasses various facets, such as geographic origin, health and veterinary care, and feeding regimens, all of which contribute to the potential enhancement of value through product differentiation. Consumers’ WTP for this information incentivises producers to provide it as a competitive tool in product marketing and market access (Verbeke and Ward 2006; Napolitano et al. 2010). However, providing PM information can come with costs, such as audit fees, skilled labour and compliance costs. For value delivery, PM information must be consistently perceived, and processed along the SC, and used by end consumers.

Basis of producers’ WTA and WTP

Based on Lusk and Hudson (2004), a producer can choose the optimal level of information inputs, i; facing vectors of input prices, w; and output prices, p. The indirect restricted profit function is described as: π(p, w, i). Assuming that the producer initially faces an information inflow (information receiving) level, i⁰, and then information I is introduced, the new inflow information level is denoted by i⁰ + I. Given the restricted profit function, a profit maximiser will pay for the new information I, if:

The perceived profit derived from receiving new information I is greater than or equal to the amount foregone. Otherwise, that producer will not be interested in receiving the information. The WTP for the new information I or for the change is then:

The payment received for providing a piece of information can be treated as the compensation for the cost generated by associated production changes (such as compliance); information technology adoption; costs. such as certification; and labour inputs. Although it should be noted that producers may be compelled to provide information (Olynk et al. 2010b), only the case where information is shared voluntarily is addressed.

Denoting q⁰ as the initial outflow of information (i.e. information provision), and q⁰ + I as the new outflow, the producer will find it acceptable to provide the new information I, if:

which means that the compensation accepted in exchange for provision of information I is greater than or equal to the costs of provision. The WTA to provide the new information I or for the change is then:

Factors affecting farmer WTA and WTP

Storer (2006) proposed that food producers could enhance the value of their products by improving quality, yield and grades through better utilisation and sharing of information. Sharing information vertically in the red meat SC is an innovative agricultural practice (Griffith et al. 2010). However, the attitudes of farmers towards such innovations may be influenced by their individual characteristics, and research on the subject has produced inconsistent results. The impact of these factors on farmers’ perceived value of product quality information is not well understood.

Previous studies have found mixed results regarding the influence of farmers’ age and education level on their WTP for innovative agricultural practices, with some studies (e.g. D’Souza et al. 1993; Kim et al. 2008) showing a negative association between age and WTP, and a positive association between education level and WTP. Women have been found to have lower levels of education and knowledge about agricultural production than men in some cases (Doss and Morris 2000), but they may have more positive attitudes towards collaboration and conservation technology adoption (Druschke and Secchi 2014; Hay and Pearce 2014). Farmers’ experience in agricultural production may not have a positive impact on their use or provision of novel agricultural information (Rehman et al. 2013; Carrer et al. 2017), as more experienced farmers may make decisions based on their experience (Magne et al. 2010) and trusted sources of information. Farm size is often related to farm financial status (e.g. debt level, income; Kallas et al. 2010; Ainembabazi et al. 2017), number of employees (Kallas et al. 2010) and property size (e.g. in ha; Finger and Lehmann 2012). It has been hypothesised to influence farmers’ willingness to invest in new technologies (Knowler and Bradshaw 2007), but results have been mixed (as illustrated by Adrian et al. 2005; Finger and Lehmann 2012). Third-party certification programs may influence producers’ WTA and WTP for product quality information, as these programs facilitate information sharing between SC members and are associated with increased concern for product quality (King et al. 2005; Prado and Woodside 2015).

A conceptual framework for eliciting producers’ WTA and WTP for information

A conceptual framework is developed to elicit livestock producers’ WTA and WTP for information (Fig. 1), especially cattle and sheep producers’ WTA for providing PM information and WTP for receiving FB information. The framework is based on previous research exploring the value and benefits of information sharing in the SC (i.e. Zhang et al. 2020). The framework considers whether farmers would provide or pay for information, and the valuation of information at two levels of IQQ.

Fig. 1.

Conceptual framework for eliciting producers’ willingness to accept or willingness to pay for information. Source: Developed by the author based on Zhang et al. (2020). WTA, willingness to provide information; WTP, willingness to pay for information; IQQ, information quality and quantity; $WTA/WTP, WTA/WTP in dollar amount. The factors affecting producers’ WTA/WTP for information at low and premium IQQ levels, shown as grey arrows, are not tested in this paper. Logical flow of eliciting producers’ WTA/WTP for information.

The framework for measuring the value of information in the upstream SC considers both information provision and receipt, and uses the CV approach to elicit producers’ WTA or WTP for information. Where IQQ is unknown, producers’ willingness to provide or pay for information, or share it (onwards and outwards), is identified. WTA or WTP in dollar amounts are elicited from those producers who were willing to provide or willing to pay for that information. This is considered at two levels of IQQ: (1) low, and (2) premium levels to obtain the WTA or WTP in dollar amount at each level of IQQ separately.

To prevent ‘protest responses,’ where respondents simply answer ‘no’ to the initial question about their WTA or WTP, a preliminary question is posed. Those who respond negatively to this initial inquiry are considered as not interested in the information, and no further questions regarding monetary values are asked. This approach serves the purpose of avoiding an excessively high rate of zero responses, a common issue in CV studies (Jorgensen et al. 1999; Szabó 2011), although it may reduce the sample size and constrain available methodological options.

Characteristics of producers and farms are included as factors moderating their WTA or WTP for information. This framework is not limited to cattle and sheep producers, and can be applied to other stages of the SC to compare the values of information and diagnose potential chain failures due to asymmetry in the valuation of information.

Data collection and survey design

The study collected data from two online surveys of Australian beef and sheep meat industry actors from November 2017 to February 2018, and from May to October 2018. Survey advertisements were distributed to cattle and sheep producers in Australia via various channels, including livestock producer groups, certification programs, cattle and sheep exhibitions, and university social media. The current study used a subset of the data about livestock producers’ WTA and/or WTP for information, and the self-reported characteristics of the survey respondents. Data extracted from a total of 104 valid respondents was used in this study: 50 from the first survey, and 54 from the second. There were several respondents involved in both beef and sheep industries (n = 26). These were treated as members of each group, as the questionnaire addressed activities in separate industry groups. The resulting number of cattle and sheep producers were 91 and 39, respectively.

A two-stage CV survey questionnaire was designed following the conceptual framework presented in Fig. 1. In the study, respondents were asked if they were willing to provide or pay for nominated pieces of information. Those who answered ‘yes’ were then asked their WTA or WTP for the information at low and premium IQQ levels. The IQQ indicators for FB information were consistent between MSA-licenced processors in Australia, making them constructive for the study. The information quantity was measured based on the basis of measurement method; that is, herd/mob or individual animal based (Guy et al. 2018). For both the FB and PM information, individual animal-based measurements were employed to represent a high quantity of information, whereas herd/mob-based measurements signified a lower quantity of information. Various information quality measures were developed for different types of information. The quality of FB was identified using mechanical measurement or objective carcass measurement technologies, such as dual-energy X-ray absorptiometry (Calnan et al. 2021), and the certification status indicated the quality of PM.

Examples of survey questions are presented in Appendix I. The primary distinction between the two surveys lies in the format of the questions used to elicit WTA and WTP responses (Q2 in Appendix I). In the first survey, we employed an open-ended question format to prevent starting point bias, as suggested by Balistreri et al. (2001). However, this format led to a high non-response rate (Frew et al. 2003), and a relatively large variance in WTA and WTP. In response to these challenges, we conducted a second survey using a five-point payment card to improve the design of payment intervals. The numeric WTA and WTP data from the first survey were converted into the interval format used in the second survey to create a unified dataset. Previous studies have shown that open-ended and payment scale approaches are equally valid in eliciting WTP (Frew et al. 2003; Grutters et al. 2009).

Description of survey respondents

The characteristics of survey respondents and their businesses in each sample group are presented in Table 1. These factors are important in determining farmers’ WTA and WTP for information, as highlighted in the study’s conceptual framework (Fig. 1). The majority of respondents were men, and their average age ranged from 53 to 55 years, with the youngest being aged 22 years and the oldest being aged 81 years. This closely aligns with the gender composition (77% men) and the average age (57 years) of Australian farmers, as reported by ABS (2019). The survey participants had an average tenure of about 18 years in senior management roles within agribusinesses or farms, spanning from 0 to 61 years. In contrast, the farm demographic report from the ABS (2019) indicated an average overall involvement in farming of 37 years.

The majority of respondents received formal education in agriculture, and approximately half of the respondents had a university degree. The surveys were conducted online and advertised through email, which implies that the respondents were expected to have access to internet-enabled devices, such as computers or mobile phones. It is likely that respondents with a university degree or formal education were more likely to participate in the survey. The average number of full-time employees employed on respondents’ farms was approximately 2.3. In addition, the majority of the sample consisted of farms with a gross revenue ranging from AU$100 000 to AU$499 000, suggesting that most surveyed farms were small or medium sized. Approximately 60% of cattle and sheep producers in the sample indicated that they were a member of at least one third-party certification program, but the distribution channels of the survey may lead to a high proportion of positive responses.

Most survey respondents are likely part of a specialised group of producers with significant experience in farm management and formal education. They may possess the resources, including a workforce, to dedicate to information collection and processing. Additionally, they often engage in certification programs and directly sell to meat processors. Although this subgroup has the potential to play a leading role in advancing the collection and utilisation of high-quality information within the industry, it is important to acknowledge that this may result in relatively high WTP or WTA values. Furthermore, it is worth noting that the sample size in the sheep industry is relatively small, potentially influencing the WTP and WTA results. However, given the exploratory nature of the current study, it has been decided to retain these data for further analysis.

Method

A two-step analysis was conducted to estimate the monetary value of information on product quality (FB and PM) at two IQQ levels and determine the factors that influence commercial livestock producers’ WTA or WTP for this information.

The mean and median WTA and WTP for each type of information were estimated using a non-parametric approach following Haab and McConnell (2002, p. 71–72) and Carson et al. (1994). The lower bound for mean WTA and WTP was obtained by multiplying the probability of each value interval by the lower limit of that interval, assuming that when a respondent selects a payment scale interval, their WTA or WTP is not less than the lower-bound of that scale. The t-statistic was used to determine the statistical significance of differences between the values of information assigned to two levels of IQQ.

The initial survey question on WTA/WTP was a binary choice. To investigate the factors influencing this decision, such as the characteristics of the farmer and farm, binomial logistic regression following the method of Burton et al. (1999) was used. The probit model is another approach used to examine the relationship between binary dependent variables and explanatory variables, but the logistic model was chosen due to its simplicity and interpretability. This approach has been widely used in the literature to analyse the factors affecting farmers’ WTP decisions on technology, services or innovation (e.g. Asrat et al. 2004; Mariano et al. 2012).

The binomial logistic regression distribution is expressed (Brooks 2014, p. 562) as:

where P_j denotes the probability that the jth farmer is willing to provide or willing to pay for a piece of information (Y_j = 1). The binary response variable, Y_j, takes the value 1 if the jth farmer is willing, and 0 otherwise. It is assumed that the values taken by Y_j are determined by a latent variable, Y_j*, given by

where β_k is a vector of coefficients; ε_j is the error term; and x_k is a vector of observed covariates representing the characteristics of jth farmer and farm; and Y_j* is the unobserved binary variable that corresponds to the observed dichotomous variable for the jth farmer, Y_j. The relationship between Y_j and Y_j* is expressed as:

so that Eqn 5 is then transformed as:

To maximise the utilisation of the available data, the data from sheep meat and beef industries was merged, and ‘cattle producers’ (denoted by ‘BEEF’) added as a dummy variable in the logit model to maintain the distinction between the two industries. As the development of information measurement in the Australian beef industry (e.g. MSA) is more advanced than that for sheep meat, cattle producers were expected to have a better understanding of the benefits and costs of information sharing. The other explanatory variables and their measures are listed in Table 1. Two binominal logit models are presented, differentiated by type of information:

The parameter estimates were used to generate average partial effects (APEs; also known as average marginal effects (AMEs)) to estimate the expected partial effect for each identified variable by averaging across the marginal effects of jth farmer’s observed values on the explanatory variables. APE is expressed as (Greene 2018, p. 735):

The authors chose APEs over partial effect at the average, as the mean values of explanatory variables (covariates) computed by partial effect at the averag may not exist in the dataset or in reality (Wooldridge 2015, p. 600). The differences between partial effect at the averag and APEs are discussed in Wooldridge (2015, p. 600) and Greene (2018, p.735).

Bootstrapping in the binominal models and in calculating robust standard errors for generation of APEs was applied. This approach addresses sample variation in cases such as these, where the sample cannot cover the entire population or is limited to a certain point in time. The statistical analyses were conducted using the Stata 15 software package.

Descriptive statistics

The frequency of respondents that are willing to provide and willing to pay for each type of information, and the associated distributions of monetary WTA and WTP at the low and premium IQQ levels are presented in Tables 2 and 3. The 0 subscript for each WTA or WTP denotes the question on willingness, and the 1 and 2 subscripts denote the IQQ levels of ‘low’ and ‘premium’ respectively.

One notable result from Table 2 is that approximately one-third of the cattle producers and sheep producers expressed interest in receiving FB information. This finding is unexpected, given that the MSA system has been available for the Australian beef industry since 1998 and research has demonstrated the benefits of implementing MSA, including consumers’ WTP and improving producers’ competitive advantage (e.g. Griffith and Thompson 2012; Morales et al. 2013). However, these findings may be indicative of a strong demand among sheep producers for information related to eating quality and the development of associated information systems to aid in their decision-making processes, given that implementation procedures for MSA for sheep carcasses remain at an early stage.

For the provision of PM information (Table 3), a higher proportion of producers expressed interest, with 49.4% of cattle producers and 35.9% of sheep producers willing to provide it. This is not surprising, as farmers providing PM information are usually seeking a price premium. The difference in results between species is expected due to varying levels of benefits available to sheep and cattle producers. As previously noted, studies by Zhang et al. (2018, Scenario 11) and Mounter et al. (2019, Scenario 5) found that an increase in domestic consumers’ WTP would generate a larger share of total surplus value for cattle farmers compared with sheep farmers.

At the low IQQ level (denoted by subscript 1 in Tables 2 and 3), the majority of respondents valued the first two payment ranges for all information categories. At the premium IQQ level (denoted by subscript 2 in Tables 2 and 3), the majority of respondents assigned value to the studied information, except for a few sheep producer respondents who expressed that they would provide PM information without any payment. This may be because their purposes of sharing this information were to receive non-monetary benefits, such as customer retention or improving or maintaining a long-term relationship with buyers. Another potential explanation is that the improvement of IQQ was not expected to generate substantial costs for them for that type of information. Another finding is that the variance of respondents’ valuation was higher at the premium IQQ level compared with the low IQQ level.

Values for FB information in the beef industry were more evenly distributed at the higher level of IQQ than the lower (see Table 2). Although the majority of sheep producers who were willing to pay for FB information valued it at AU$0.1–$1/carcass at both IQQ levels, a substantial number of respondents still selected higher values for the high IQQ than the low IQQ level. For PM information, the first two payment intervals comprised one-third of respondents willing to provide PM information at the IQQ level, whereas another one-third were in the highest payment interval for both beef and sheep meat industries.

Estimation of mean and median WTA and WTP

Table 4 presents the mean and median values of WTA and WTP for farmers at two IQQ levels. For the beef industry, it is observed that the highest mean value was associated with PM information at the premium IQQ level, amounting to AU$20.49 per head. In contrast, the lowest mean value was found for FB information at the low IQQ level, totalling AU$0.83 per carcass. Similarly, in the sheep meat industry, PM information at the premium IQQ level exhibited the highest mean value of AU$10.13 per head, whereas the lowest mean value was recorded for FB information at the low IQQ level, totalling AU$0.14 per carcass.

An interesting result from comparing the value of information at two levels of IQQ is that the median and mean values of information were higher at the premium IQQ level than at the low IQQ level, for all types of information. This difference was statistically significant, as indicated by the t-statistics results, and highlights the potential for value addition due to higher levels of IQQ.

Logit model results

The study estimated two logit models to explore the determinants of farmers’ WTP for FB information and willingness to provide PM information. The results of estimated coefficients and APE analyses, followed by their bootstrapped standard errors (s.e.s), are presented in Tables 5 and 6. All models delivered a good statistical fit, as indicated by likelihood-ratio Chi-squared (P < 0.01) and pseudo R². The models provided different results, and with each discussed separately.

The factors influencing farmers’ interest in receiving FB information included gender (GEN), experience (EXP), formal education in agriculture (EDU_AG) and gross farm income (FZ_GR), all of which were statistically significant at the 5% level. The corresponding APE results showed that a marginal reduction in farmers’ experience in farming was associated with a 1.2% increase in the likelihood of being willing to pay for FB information, whereas formal education in agriculture was associated with a 29% higher probability of being willing to pay. The effect of farm size was positive and significant, indicating that an increase in farms’ gross income resulted in a 9.4% increase in the likelihood of being willing to pay for FB information.

The logit model results in Table 6 indicated that experience (EXP) had a positive and highly significant impact on farmers’ willingness to share PM information. This is in contrast to the model for farmers’ WTP for FB information. Farmers with more experience in farming are more likely to be familiar with data collection procedures and quality management practices on their farms, enabling them to process raw data into information and subsequently provide it. University qualifications (EDU_UNI) were positive and highly significant, whereas EDU_AG was not significant. This may be due to some respondents having a university degree in subjects unrelated to agriculture, and those with a university degree may have a better understanding of the value addition processes involved in providing PM information, making them more willing to share information to improve financial performance. Farm size (FZ_EMP and FZ_GR) was not a significant factor in the provision of PM information. However, small and medium-sized farms in the sample were more likely to depend on off-farm income than those with larger farms, which may limit their engagement in activities, such as data collection and provision of PM information. Respondents involved in cattle production were more willing to provide PM information than those in the sheep meat industry, possibly due to the more advanced information measurement development in the beef industry and stronger quality-related price signals in beef markets.