Consumer preferences for farmed tilapia in Tanzania: A choice experiment analysis

ABSTRACT

This study used a choice experiment to analyze the preferences of consumers for the attributes of tilapia (price, mode of production, product form, and size), a major aquaculture product in Tanzania. The results showed that consumers were willing to pay a price 665.020 Tanzanian shillings (TZS) lower for farmed tilapia than for wild tilapia; a price TZS 833.210 and TZS 1799.110 higher for medium-size and large-size tilapia respectively than for small-size tilapia; and a price TZS 1214.090 higher for fresh tilapia than for smoked tilapia. Consumers’ willingness to pay less for farmed tilapia than for wild tilapia is mainly attributed to issues related to taste and availability. It was also found that consumers were heterogeneous in their preferences for all the tilapia attributes (mode of production, size, and form) considered in this study.

KEYWORDS:

• Attributes
• choice experiment
• consumer preference
• Tanzania
• tilapia
• willingness to pay

Notes

¹ Inland (Lakes Victoria, Tanganyika, and Nyasa) and marine waters account for about 86% and 14% of domestic fish production respectively, and the rest (less than 0.5%) is accounted for by aquaculture.

² Apart from the fact that population growth exceeds growth in fish production, the cause of the decline in per capita fish consumption is unclear.

³ Heterogeneity has also been accounted for with latent class models (LCM) (Krishnakumar and Chan-Halbrendt 2000). The LCM assumes that the population consists of unobservable (latent), finite, and discrete classes across which preference patterns of consumers are heterogeneous. Within each class of the population, however, the preference patterns of consumers are assumed to be homogenous.

⁴ Morey and Rossman (2003) reported that, by allowing for correlation over choice alternatives, two pair-wise choices (one from each of the two individuals) provide more information than two choices from the same individual.

⁵ Mean WTP measures the trade-off between the various attributes and the price of tilapia.

⁶ The mean WTP is expressed with Equation (2) because the attributes are effects coded.

⁷ The respondents did not hesitate to declare their food expenditures. The main challenge was with household income; many of them were not willing to disclose that. For those who were not willing to disclose their household income levels, we asked for the occupation of all the household members and estimated their household incomes from the average incomes of people in the city involved in similar occupations as the household members.

⁸ The effects-coded levels of attributes take on 1 when applicable, −1 when the omitted level applies, and 0 when otherwise.

⁹ We assumed that heterogeneity of the non-price attributes are uncorrelated. Because this could be a strong assumption, we estimated the corresponding model in which the assumption was relaxed and performed a log-likelihood test to assess the assumption. The test results rejected the hypothesis that they are correlated, so we maintained the original results.

¹⁰ The consumers were asked to indicate their preference for farmed and wild tilapia on a Likert scale of 1 to 4: 1 = not preferred, 2 = least preferred, 3 = indifferent, and 4 = most preferred.

¹¹ The reasons for lack of preference for farmed tilapia were stated relative to wild tilapia.

¹² Consumers indicated their preference for the various tilapia sizes on a Likert scale of 1 to 3: 1 = most preferred, 2 = least preferred, and 3 = not preferred.