Commercializing Beef Sub-Sector in Arusha and Dar-es-Salaam Cities, Tanzania: Consumers’ Preferences and Willingness to Pay for Quality Beef

Analytical Framework

Conceptually, the consumer preference analytical technique assumed the utilitarian consumer theory whereby the consumer was assumed to be rational such that s(he) consumes quality beef with intrinsic and extrinsic attributes that yield him/her the maximum utility. Previous literature suggested that Tanzanian beef consumers’ preference attributes include tenderness, freshness, fat content, price, and hygiene of retailing outlets and safety of the product [26, 27]. Other meat quality attributes were revealed to include meat colour, sources of beef (warm butchers vs. supermarkets and modern butchers), and presence of municipal stamps catering for beef being inspected and hence its suitability/safety for human consumption [28]. Hence intrinsic features of quality beef included in the analysis are freshness, fat content, tenderness and colour of beef, while extrinsic features of quality are hygienic condition of retailing outlets, price and presence of municipal stamps on the beef sold (Equation 1).

Ụi= f (Pi, INTi, EXTi, M, α)…………….…………… (1)

whereby: Ụi is the utility of consumer, Pi, is the vector of prices, INTi, is the vector of intrinsic characteristics of beef, EXTi, is the vector of extrinsic characteristics of beef, M is the vector of expenditure levels of consumers and α is the vector of household characteristics influencing consumers’ preferences. The intrinsic characteristics included freshness, fat contents, color and tenderness of beef sold in retail outlets, while extrinsic characteristics included in the model are levels of hygienic condition of the retail outlets, price and presence of municipal stamps on the retailed beef.

Likert – scale measures were used to quantify importance of quality attributes preferred by beef consumers: 1 - strongly important to 5 – strongly unimportant) on the quality attributes stated. Then frequency distribution was done on beef consumers’ preference attributes stated. Percentages for respective quality attribute were computed and presented in charts.

Principal Component Analysis (PCA) and Wards Method of hierarchical analysis were used to cluster consumers in terms of attitudes and/or preferences towards major beef quality attributes.

The willingness to pay has been estimated through two major approaches: revealed preference (RP) and stated preference (SP) techniques. The RP techniques involve observation of prices that people pay for goods and services in various markets; and expenditures of their economic resources such as money, time and labour to obtain the goods and/or services or to avoid their losses [29]. The WTP for quality beef was elicited using stated preference technique that entails the use of respondents’ statements about their willingness to pay for a particular good or service [30]. The strength of using SP approaches is due to their applicability to wider disciplines in which consumer behavioral responses are either not available or was difficult to assess substantially attributes associated with Revealed Preference techniques [31, 32].

Stated preference techniques can further be classified into direct and indirect models of estimating WTP. Direct approach is executed when respondents are asked directly to indicate their maximum WTP for a good or service offered. Alternatively, indirect approach is done when respondents are presented a good or service with different profiles accompanied by different price levels to indicate whether or not he/she would pay for a product. Direct approaches have been noted to have serious shortcomings in eliciting consumers’ willingness to pay [33, 34]. Therefore, indirect approaches have been adopted in this study.

Two major approaches of eliciting WTP indirectly are Conjoint Analysis (CA) and Contingent Valuation Method (CVM). The Conjoint Analysis involves the stated choice experiment whereby respondents are asked to make choices on goods with different attributes and levels accompanied with various price levels [35]. The Conjoint Analysis is known to be incentive neutral in revealing the true WTP of respondents since a respondent has a tradeoff among entire price/performance of bundles [36]. However, the approach is acknowledged to have some shortcomings in eliciting respondents’ WTP: (i) In CA, WTP values are elicited using combination of different product attributes that might lead to inaccurate estimate if the choice of estimation technique is wrongly specified. (ii) The CA estimates part-worth for discrete price levels while WTP is a continuous variable that necessitates deduction of the respondents’ utilities other than those specified in the research design [37]. Moreover, the CA approach treats price as an attribute in estimating part- worth utilities. The inclusion of price as an attribute violates neoclassical economic theory of consumer behavior whereby price is treated as exogenous variable that can be included in the model through budget constraint [38].

The Contingent Valuation Method (CVM) is a survey/questionnaire based approach which asks respondents their WTP for market and non-market goods and services [39]. The CVM approach elicits stated preferences from a sample of consumers using either open ended questions that directly asks for WTP, or referendum (closed ended) questions that presents a bid to the respondent and asks for a Yes or No vote on the bid subjects to WTP [40]. In this approach, the stated preference is commonly used as a mean value of WTP [41].

The CVM has proved to be a reliable approach in eliciting peoples’ WTP through dichotomous choice model [42]. It has been applied in agricultural and food economics to elicit WTP for foods regarding safety measures [43, 44, 45, 46, 47]. For that case, this study applied CVM to examine the WTP of quality beef in Arusha and Dar-es- Salaam cities among household beef consumers.

Therefore, estimation of WTP for quality beef was done using contingent valuation method (CVM). This CVM was done through double bounded model of dichotomous choice questions. It was assumed that a respondent i had unobserved true WTP for quality beef equal to Yi*. The true WTP Yi* for quality beef had some relationship with the household head characteristics Xi as specified in equation 2.

Yi* = Xiβ + εi …………………. (2) Whereby εi is assumed to be normally distributed with mean zero and variance δ2. The functional form of true WTP was specified as Tobit model since the dependent variable cannot assume negative values because the respondent would either decide to buy quality beef (1) or decide not to buy quality beef (0).

Each respondent was initially asked whether he/she was willing to purchase conventional beef and/or quality beef at the same price, that is, the price differential (PDI) equals to zero. If the respondent indicated a preference on quality beef, s(he) was further asked whether s(he) is willing to pay a premium bid, that is, the price differential (PDH) is greater than zero [48]. Bids incremental were 33%; 66% and 100% due to the fact that, there are increased costs in fattening of cattle and ageing of beef to make it tender leading to an increased price of quality beef marketed. The true WTP Yi* for quality beef is not known but lies between PDI and PDH due to responses given to the series of contingent valuation questions; and is specified in equation.

Pr (PDI ≤ Yi* ≤ PDH) …………....... (3) When replacing the true WTP value for quality beef Yi* with equation (3), then the function can be written as in equation (4).

Pr (PDI ≤ Xiβ + εi ≤ PDH) ………………. (4)

The outcome of bidding procedures is presented in equation 5 due to three expected outcomes of the contingent valuation questions. That is, (i) A ‘No’ to the first bid (i.e. no preference for quality beef vs. conventional beef) at 0 premium. (ii) A ‘Yes’ followed by ‘No’ (i.e. there is a preference at 0 price difference but not at higher bid) and (iii) A ‘Yes’ to both bids (i.e. preference at 0 premiums and at higher bids) Yi* ≤ PDI (outcome 1) Di = PDI≤Yi*<PDH (outcome 2)………………. (5) PDH ≤ Yi* (outcome 3) The choice probabilities for true WTP for equation 4 are expressed in equations 6 to 8. P (Yi*≤ PDI) = P(εi ≤ PDI - βi Xi ) = F(PDI - βi Xi ) ……... (6) P(PDI ≤ Yi*< PDH) = P(PDI - βi Xi ≤ εi< PDI - βi Xi) = F(PDH - βi Xi) – F(PDI - βiXi …….... (7) P(PDH ≤ Yi*) = P(εi> PDH - βi Xi) = 1 – F(PDH - βi Xi) ….... (8) The stated probabilities specified in equations 6-8, yield the log likelihoods presented in equation 9. L=∑ lnF[(PDI - βiXi)]+∑ lnF[(PDH - βiXi) – F(PDI - βiXi)]

1 D

2 D

+ ∑ ln[1– F(PDH -βiXi)]........ (9)

3 D

Where: Di – is a group of respondents in the ith bidding process outcome, F – is a cumulative distribution function with mean zero and variance δ2 and P – is the choice of probabilities corresponding to equation 5 Estimation of model parameters employed censored regression approach for the estimation of closed ended questions of contingent valuation surveys as developed by Juma GP, et al. [49]. Specifically, Tobit model was used to examine factors influencing WTP for quality beef consumers and was specified as equation 10. Yi* = β0 +β1HHI+β2HHV+β3HHS+β4HHA+β5HHO+β6HHE+β7HHM+ εi …. (10) Whereby; β0 is the equation constant term, βi is the parameter coefficients of the model to be estimated, HHI – is the expenditure of household head, HHV – the size of household, HHS - sex of household head, HHA - age of household head, while HHO, HHE and HHM are occupation, educational level, and marital status of household heads respectively; and εi is the error term that caters for the effects of variables not included in the model. Table 1 presents model explanatory variables and the expected relationship with WTP for quality beef.

Table 1: Definition of variables in WTP model and expected relationship Household size was assumed to have a negative influence on willingness to pay for quality beef. This is because large households were expected to spend their resources on basic needs since the households require huge basic needs and other protein foods that are less expensive than quality beef. Findings of empirical studies suggest that there is a negative relationship between household size and WTP for quality foods [50, 52, 53].

With regard to sex of household head, previous studies suggest a positive relationship between WTP for quality foods and female headed households [44, 51, 54]. In this study, a negative relationship was hypothesized between female headed household and WTP for quality beef. This is due to the fact that women have poor access to the means of production and high paying jobs [55] that could allow them to earn higher incomes and thus increase the likelihood of paying premium prices for quality beef.

The body of literature portrays conflicting relationship between WTP and quality foods. Some authors revealed a positive relationship between age and WTP for foods [48], while others revealed negative associations between age and WTP for quality foods [50, 56, 57]. In this study, age of the household head was expected to have a positive relationship with WTP because as one gets older, s(he) develops skills to perform his/her job better, gets promotion in his/her carrier and increases his/her purchasing power. The increase in purchasing power is associated with higher likelihood of paying premium prices for quality beef.

Education level of household head was expected to have a positive influence on WTP for quality beef. This is because an increase in education level leads to more exposure of consumers that create awareness on quality beef attributes. Moreover, highly educated people are likely to get well-paying jobs which can in turn lead to high likelihood of purchasing quality beef. Previous studies [43, 53, 56] revealed a positive association between education levels and WTP for premium priced foods.

Lyford C, et al. [56] found a negative relationship between marital status and WTP for quality foods. In this study, a positive relationship was hypothesized between marital status and WTP for quality beef because married couples can pull together resources accrued for common consumption of the household hence increased purchasing power. The increase in purchasing power increases the likelihood of the household’s willingness to pay a premium price for quality beef.

Sampling

A multi-stage cluster sampling technique was used. In the first stage, municipals and districts were chosen from the sampled regions. In Dar es Salaam; Kinondoni municipal was selected as a case study because it has the most diverse society with different income levels including high income earners who are likely to consume quality beef. On the other hand, in Arusha region; Arusha

municipality was selected based on the fact that its residents have higher incomes compared to other districts in the region. This is because, the municipal serves as the Arusha regional economic base whereby most organizations that employ highly paid expatriates exist; and it is the centre for the regional business activities that allow income multiplier effect among its residents.

In the second stage, wards under study were selected. In Kinondoni; Tandale and Manzese wards were chosen to cater for low income households while Sinza and Msasani wards were chosen to cater for households with middle and high income respectively. In Arusha, Unga Ltd, Kaloleni, and Lemara wards were included in the sample to represent wards with low, middle and high income households respectively.

In the third stage, proportionate stratified sampling technique was used to select household respondents. Proportionate stratified sampling was used to obtain a sample comprising of clusters of respondents with different characteristics such as sex, education, occupation and income levels.

Data Collection

Primary data were collected using structured questionnaires for five months between February and June, 2012. Appointments were made two days before the date of interview through Wards Executives. Interviews with household heads/representatives took place at their homes.

Consumer Preferences for Quality Beef

A sample of beef consumers were asked to rank different quality attributes taken into consideration when buying beef. Figure 1 displays consumers’ ranking of quality attributes according to their order of importance. ‘Freshness’ attribute scored the highest rank of being ‘strongly important’ as it was considered by consumers when purchasing beef. This finding supports the finding by Teklebrhan T, et al. [59] who found that freshness was among quality attributes preferred by consumers of beef in Germany. The ‘less fat content’ quality attribute was ranked second to ‘freshness’ among sampled household beef consumers in the study area (Figure 1). This finding supports the finding by CIRAD [60] who found that beef consumers in Ethiopia preferred lean and red meat color than fat and white meat.

Cleanliness of retailing outlet facilities attribute ranked third after ‘freshness’ and ‘less fat content’ attributes. The sampled household beef consumers relied on cleanliness of retailing outlet facilities as a safety measure of the beef they bought. This is because interviewed consumers had no alternative mechanism of measuring beef safety. Few beef consumers looked at municipal stamps which could act as a safety measure for the consumers. This finding supports the finding by Kamugisha PP [27] Nandonde et al [26] who found hygiene among retailing outlet facilities as the most important valued attribute among beef consumers in the southern highlands of Tanzania.

Click to enlarge

Table 2 shows that respondents in Dar-es-Salaam preferred to consume beef with less fat content as opposed to Arusha beef consumers who preferred beef with relatively higher fat content. Cold weather condition might be the reason behind Arusha consumers’ preference of beef with relatively higher fat content than their counter parts in Dar-es-Salaam. CIRAD [60] reported similar findings in Ethiopia whereby lean meat (red in color) was most preferred among meat consumers. Freshness was identified by most (61.8%) of Arusha beef consumers as a very important quality attribute as opposed to only 36.1% of Dar-es-Salaam beef consumers (Table 2). The proportion of beef consumers who preferred fresh beef in the study area is smaller than that reported by Delgado CL [61] who found that 92% of beef consumers in India preferred fresh beef. However, this proportion is higher than the finding by Teklebrhan T [59] who found that 58% of beef consumers in Germany preferred fresh beef.

NB: Components 1, 2, 3, and 4 are freshness, less fat content, cleanliness of retailing outlets and tenderness quality attributes. Table 3: Principal component analysis of beef quality attributes.

while heads of household in this group were mainly self- employed with primary and tertiary education levels (Table 4).

The third group comprised of consumers who considered freshness as the most important quality attributes. It is the largest group that accounted for almost 33% of the sampled consumers. This group can be termed as ‘fresh beef consumers’. About 42% of this group had family size of between 1and 4 household members; and the group was dominated by self-employed household heads (58%) with primary school education levels (Table 4). Their opinion on freshness was similar to that of the second group which associated freshness with safety and non-deterioration of meat. They were also unaware of post slaughtering handling practices of meat such as chilling it in cold rooms to prevent deterioration.

The second group comprised of consumers who preferred beef with both freshness and less fat content quality attributes. However, they were more sensitive to freshness than to fat contents. This group accounted for 22% of the sampled consumers. Household heads in this group were relatively highly educated and had high incomes. This group can be termed as ‘safety scares consumers’ because most of them were of the opinion that fresh beef was preferred to avoid eating deteriorated beef. Moreover, they were of the opinion that eating meat with large amount of fats causes health problems. However, consumers were unaware that meat quality could be maintained through post-slaughtering processes such as aging/chilling of meat.

NB: Numbers in parenthesis are percentages. Table 4: Beef consumers’ segmentation.

The fourth group comprised of beef consumers who were neither concerned with ‘freshness’ nor ‘less fat content’ of beef. It accounted for 22% of the sampled households. They were influenced by other factors especially the price of beef in making decision to buy beef. This group can be termed as ‘price conscious beef consumers’. Members of this group were low income earning households with low frequency of beef consumption. Heads of households in this group were young and had small household sizes.

Willingness to Pay for Quality Beef

Table 5 shows the proportion of interviewed consumers who were willing to pay for quality beef at different prices. At the price of 14,205Tsh/kg, only 9.35% of the sampled household respondents bought and consumed quality beef. Questions on higher bids of quality beef were set to determine WTP for the beef at higher price premiums. With an increase in price by a premium of 33% of the prevailing price, the proportion of respondents who were willing to pay for quality beef dropped to 5.75%, at 66% premium price only 5% of the sampled respondents were ready to pay for quality beef and at 100% price premium only 1.08% were willing to buy and consume quality beef (Table 2). The decline in the proportion of respondents who were willing to pay for quality beef explains why there is low consumption of quality beef among interviewed respondents. The price of quality beef of 14,205 Tsh/kg was more than 3 folds the price of conventional beef of 4,270 Tsh/kg. The findings that consumers of quality beef decrease as the premium price increase is supported by findings from other studies in China, South Carolina and Korea [45, 47, 48] respectively.

Table 5: Distribution WTP among interviewed respondents When consumers were asked behind their low rate of consuming quality beef, they stated that the beef is too expensive to afford. Apart from expensiveness of the quality beef, interviewed consumers stated other reasons behind low consumption of quality beef as indicated in Figure 2. Unfreshness and chemical composition attributes were found to be other major reasons scaring beef consumers from eating quality beef. About 51% of consumers rated unfreshness as a ‘strongly important’ and ‘important’ quality attribute’ scaring them from consuming quality beef. This is because most consumers had experienced deterioration of conventional beef which is normally sold warm in butchers whereby meat deterioration increases as time elapses. Loss of natural flavours in meat which increases with an increase of time the meat stays in the shelves before being consumed after slaughtering and lack of information on chilling procedures of quality beef leads to low WTP for quality beef.

Click to enlarge

High cost attribute was rated as ‘strongly important’ by 56% of the interviewed household beef consumers. The high cost of quality beef reflects the fact that high price and low income were reasons limiting the consumption of quality beef. When asked as to whether they could consume quality beef if their income rise, or if the price for quality beef decreased, 43.86% reported that, they would consume quality beef if their income rose. This indicates that, with the expansion of middle-income class in developing countries including Tanzania as projected by various authors [62, 63, 64] is likely to lead to an increase of the consumption of quality beef. Other two major reasons restricting respondents from consuming quality beef are unfreshness and the perception that quality beef contain chemicals for preservation. Wrong perception on quality beef can be associated with the lack of awareness among consumers regarding quality beef attributes. Awareness has been noted by various authors as a major influencing factor in making purchasing decisions [65, 66, 67, 68].

Chemical composition was also rated high among respondents as the reason behind non consumption of quality beef (Figure 2). A big proportion of consumers (53%) who were unaware that chilling preserves beef and increases its tenderness; perceived that the beef stays longer because they contain chemicals for preservations that could be hazardous to their health. If awareness on the benefits of chilling beef could be created, consumers who were skeptical to consume quality beef believing that the beef is preserved with chemicals could be potential consumers of quality beef. The finding on skepticism of consuming quality beef on the belief that the beef is preserved with chemicals is more or less similar to the findings of study by Phillip B, et al. [51] Alphonce R and Alfenes F [50] who found that Tanzanian consumers are Expenditure coefficient was significant within 95% confidence interval. Expenditure levels of households had a positive influence on WTP for quality beef (Table 6). This observation conforms to the Engel’s model whereby increased expenditure levels relate to increased consumption of goods. Moreover, it was revealed that, females had a likelihood of 0.23 times higher for WTP for quality beef than was the case with males counterparts. This observation supports the finding by Mapunda FM [25] who found that, women had high probability of buying tender chilled beef than men in southern highlands of Tanzania.

Variable Coefficient Std error Z p>IzI 95% conf. interval Lower Upper Sex 0.2368458 0.41908 -0.6 0.57 -1.058228 0.5845364 Age 0.0062292 0.016112 0.39 0.7 -0.025349 0.0378071 Family size -0.09767 0.095629 -1 0.31 -0.2851 0.089759 Expenditure** 1.05E-07 1.70E-08 6.18 0 7.15E-08 1.38E-07 Constant -2.631568 1.156952 -2.3 0.02 -4.899153 -0.363984 willing to pay for safety inspected and organically produced tomatoes. Inspection and organic produced tomatoes cater for safety attribute which consumers looked for, which is similar to skepticism of consuming quality beef, for consumers who perceive that chemicals are used in preserving the beef.

Selling of different types of meat particularly pork in the same retailing outlet (supermarkets and modern butchers) was found to limit the consumption of quality beef by 7% of consumers who were Muslims. These consumers were of the opinion that selling of different meat types including pork in the same retailing outlet (though in different containers) might make quality beef loses its halaal status if it comes in contact with pork, something which is unacceptable in Islamic belief.

Table 3 displays summary statistics of variables used in WTP analytical model. The model for estimated parameters was able to describe 91% variation for factors influencing WTP for quality beef and was found to be significant with 2= 46.91 and desirable Fadden pseudo R2 = 0.2037.