Smallholder farmers’ intention to use insect-based feed in dairy cattle diet in Kenya

Abstract

Limited access to good quality, adequate and affordable livestock feed impose a major challenge to livestock production in developing countries. In order to improve access to good quality and adequate livestock feed, policymakers, practitioners, and researchers are promoting the utilization of alternative feed sources. While insects have been promoted as an alternative source of protein, their production and utilization is low across smallholder livestock systems in sub-Saharan Africa. This study assessed smallholder farmers’ intention to use insect-based feed to supplement dairy cattle diets in Murang’a County in Kenya. The study employed the Theory of Planned Behaviour (TPB) and collected data from a random sample of 378 dairy farming households. A heteroscedastic probit (hetprobit) regression model was used to assess determinants of smallholder dairy farmers’ intention to use insect-based feed. Findings show that while only a small proportion of dairy farmers (11%) were aware of the use of insects as an alternative source of livestock feed, a considerable proportion (76%) were willing to use insect-based feed when they become available. The results of the hetprobit model revealed that the three TPB constructs; attitude, subjective norm and perceived behavioural control positively and significantly determined the likelihood of farmers’ intentions to use insect-based feed. Of the three constructs, attitude had the highest influence on the farmers’ intention to utilise insect-based feed, followed by perceived behavioural control and subjective norms. While age of the farmer, flock size, access to extension services and wealth status were positively associated with farmers’ intention to use insect-based feed, gender (being a male-headed household) of the farmer and farming experience had a negative influence on the likelihood of farmers’ intention. The study discusses the implications of these findings in scaling up the production and utilization of sustainable alternative protein feed sources.

Annex

Annex 1 Description and measurement of variables used in the study

Variables	Description and measurement
Dependent variable
Intention	Farmers’ willingness to use insect-based feeds measured on a five-point likert scale (1) very unlikely, (2) unlikely, (3) neutral, (4) likely, (5) very likely
Independent variables
Age	Age of the household head in years
Gender	Sex of the household head (1 if male, 0 otherwise)
Education level	The highest level of education attained by the household head (1 if had post-primary education, 0 if had primary education and below)
Farming experience	Experience of household head in dairy farming in years
Household size	Total number of members in a household
Farm size	Total amount of land owned by the farm household in acres
Dairy cattle	Total number of dairy cattle kept by the household during the time of the survey
Milk yield	Milk yield of lactating dairy cows kept by the household in litres per cow per day
Distance to market	Distance from the farm to the nearest market in kilometres
Household income	Annual income from all income sources in the household in Kenya shillings (KES) in the last year preceding the survey
Access to extension service	Household access to extension service within the last year preceding the survey (1 if had access, 0 otherwise)
Access to credit	Household access to credit during the last year preceding the survey (1 if had access, 0 otherwise)
Access to insurance service	Household access to insurance service (1 if had access, 0 otherwise)
Group membership	Membership to a farmer group or association (1 if belongs to a farmer group or association, 0 otherwise)
Household wealth category	Measure of a household’s cumulative living standarda (Poorest, Middle, Wealthiest)
Awareness	Farmer awareness of insects as source of livestock feed (1 if yes, 0 otherwise)

1. ^a The study employed (Hjelm et al. 2017) approach for computation of a household wealth index, which was arrived at through Principal Component Analysis (PCA). Households were classified into three categories. The wealthiest category- those with a wealth index above the mean value plus standard deviation, middle category- those with a wealth index within the range of mean and standard deviation and poorest households- those with a wealth index below the mean and standard deviation

Annex 2 Coefficient estimates of determinants of farmers’ intention to use insect-based feed

Variable	Probit model			Hetprobit model
	Coef	S.E	p-value	Coef	S.E	p-value
Awareness of insect-based feed	-0.128	0.280	0.647	-0.140	0.207	0.499
TPB Constructs
Attitude	0.483***	0.107	0.000	0.532***	0.116	0.000
Subjective norms	0.420***	0.107	0.000	0.359***	0.090	0.000
Perceived behavioural control	0.497***	0.190	0.009	0.391***	0.148	0.008
Household characteristics
Gender of household head	-0.132	0.258	0.609	-0.512**	0.247	0.038
Age of household head	0.006	0.008	0.430	0.012*	0.007	0.099
Farming experience	-0.009	0.008	0.310	-0.014*	0.008	0.067
Education of household head	0.242	0.190	0.202	0.130	0.147	0.374
Household size	-0.030	0.053	0.565	-0.046	0.043	0.285
Farm size	-0.085**	0.043	0.050	-0.024	0.022	0.276
Number of dairy cattle	0.063	0.063	0.320	0.110*	0.060	0.065
Distance from the farm to the nearest market	-0.032	0.042	0.448	-0.046	0.038	0.230
Household income	-0.012	0.121	0.924	-0.070	0.095	0.462
Wealth index (Wealthiest)	0.396	0.295	0.180	0.507**	0.239	0.034
Wealth index (Middle)	0.130	0.208	0.533	0.581**	0.282	0.039
Institutional arrangements
Access to credit	0.117	0.219	0.594	-0.652***	0.175	0.000
Access to extension service	0.416*	0.218	0.057	0.414**	0.180	0.022
Access to insurance service	0.070	0.205	0.733	0.223	0.146	0.127
Log-likelihood	-129.14			-120.07
Pseudo-R2	0.34
LR	132.03***
Homoskedasticity (LM Test)	26.05*
Wald test (χ2 with 18 df)				67.61***
Het-test (χ2 with 2 df)				18.14***

1. Coef is coefficient, S.E. is standard error
2. ***, ** and * indicate statistically significant level at 1%, 5% and 10% respectively; Poorest wealth category is the reference level