Poster prepared by Nadhem Mtimet (ILRI), Emily Ouma (ILRI) and Derek Baker (ILRI) for the Agrifood chain toolkit conference: Livestock and fish value chains in East Africa, Kampala, 9-11 September 2013.
What attributes do Uganda pig traders consider to maximize returns and utility?: A choice experiment application
What attributes do Uganda pig traders
consider to maximize returns and utility?: A
choice experiment application
Nadhem Mtimet, Emily Ouma and Derek Baker
International Livestock Research Institute
Emily A. Ouma
email@example.com ● Box 24384 Kampala ● +256 39-2-081154/5
Kampala Uganda ● ilri.org
This project is funded by IFAD/EU
This document is licensed for use under a Creative Commons Attribution –Non commercial-Share Alike 3.0
Unported License June 2012
Actors in a Typical Pig Value Chain
Farmers working on the seasonal calendar
Background and rationale
Pig production is increasingly becoming an important economic activity for
smallholder famers, especially women: about 32% of women currently
involved in pig production compared to only 15% 10 years ago (UBOS, 2009).
They lack voice and bargaining power as they are not organised and
individually sell their pigs to traders at farm gate.
Main market outlets include neighbor-hood butchers (60%) and live pig
Price is determined by traders based on visual estimates of carcass-weight
equivalent of the live pigs and other factors that are never clear to the
farmers. Has resulted in mistrust by the farmers who often lack capacity to
make such estimates
Purpose: Investigate the attributes preferred by the pig traders in order to
inform farmers to produce pigs that meet the needs of the market.
Butchery traders in Gulu district
Choice experiment methodology
Based on Lancasterian consumer framework and random utility
o Focus on attributes that a good posses
Step 1 : 6 attributes were identified by traders as important.
o Weight and purchase price of the pig
o Apparent health
o Pig type (female, castrated male, un-castrated male)
o Pig back-fat layer
Step 2: Different levels associated with each of the attributes
Step 3: Run a fractional factorial experimental design to come up
with the optimal attribute level combinations. Result was a total
of 12 choice sets each with 3 choice alternatives.
Step 4: Choice experiment surveys administered to 42 pig traders
in Mukono and Kamuli districts between May and July 2013.
Step 5: Modeling using conditional logit to obtain estimates of the
values of each attribute.
Pig trader participating in a choice experiment
interview in Ntenjeru in Mukono district
Attribute Coefficient Standard
Good apparent health 1.922 0.176 0.000
Thin fat layer 0.464 0.144 0.001
Age of the pig -0.072 0.012 0.000
Female -0.041 0.159 0.799
Un-castrated male -0.440 0.205 0.032
Price (in ‘000’) 0.016 0.008 0.041
Price2 -0.000 0.000 0.084
Log likelihood function = -262.377; Pseudo-R2 = 0.3137;
Conditional logit model results
Preferred pig attributes:
o Apparent good health - strongly preferred.
o Thin fat layer compared to thick layer.
o Young (6-12 months compared to 24 months)
o Castrated males most preferred followed by the females
Price linked to quality of pigs (+positive first order coefficient and –
negative quadratic form): utility associated with increasing prices up
to a point where further increment results in diminishing utility)
Interact choice variables with socio-demographic characteristics of