Impact of Dairy Production on Smallholder Households Food Security in the Central Gondar Zone,Ethiopia

Abstract

This study investigated how dairy production affected the food security of households in the Central Gondar Zone, Ethiopia. To collect primary data, 313 households were selected for interviews based on a multistage sampling approach. A review of published and unpublished documents was also conducted for secondary data collection. According to inferential statistics, about 73.08% (household dietary diversity scores (HDDs)) and 72.53% (HFCS) of dairy producer households were food secure, while 32.06% (HDDs) and 33.59% (HFCS) of dairy non-producer households were food secure. Furthermore, the Endogenous Switching Probit Regression model showed that dairy production increased food security among smallholders. Dairy production increased dairy producer households’ food security by 48.4% (HDDs) and 45.9% (HFCS). Also, it would have improved food security by 10.4% (HDDs) and 7.1% (HFCS) for dairy non-producer households. In conclusion, dairy production contributes significantly to enhancing food security for smallholders. Hence, stakeholders must pay attention to the dairy sector to improve its production and reduce household food insecurity.

Keywords

Household dietary diversity score Endogenous Switching Probit Regression food consumption score

Introduction

Ethiopia is one of the world’s most famine-prone and food-insecure countries. Chronic and temporary food insecurity affects a large population, and according to the WFP and CSA (2019) report, 26 million people were food-insecure in Ethiopia. In Ethiopia, according to a recent study by Dagninet and Adugnaw (2020), 33 million people are chronically malnourished and food-insecure; 25% need urgent assistance due to desert locusts, the spread of COVID 19, conflict and other major causes of food insecurity. Of these, the highest percentage of the food-insecure population is in the Amhara regional state of Ethiopia, with 52.3% of its population being food-insecure (Negash 2019). Conflict displacements, floods and droughts in the East Dembia and West Dembia Districts in this study area exacerbated food insecurity (FEWS NET & WFP 2019).¹

Improvement of Ethiopia’s agriculture sector that integrates both livestock and crops is key to minimising food insecurity. Hence, economic growth is highly constrained by its success in agriculture (Alilo 2019). The agriculture sector contributes about 35% of Ethiopia’s Gross Domestic Product (GDP); employs 68.2% of its labour force and generates 90% of its export earnings (FAO 2019). The livestock sector, meanwhile, contributes 40% of agriculture GDP (Asresie and Zemedu 2015).²

As one of the benefits gained from livestock production, dairy production in Ethiopia generates income and provides food and organic fertiliser, mainly sourced from cattle (Hunde 2018). The country produces 5.2 billion litres of milk annually (Shapiro et al. 2017). Over 3.8 billion litre milk is produced from cattle (FAO 2019).

In the study area, the East and West Dembia districts have huge potential (East Dembia—14680 tons, West Dembia—15150 tons) for dairy development due to large livestock populations and comfortable environments (Getabalew et al. 2019). Hence, the Ethiopian development agenda and International Dairy Development (IDD) have made significant efforts to enhance dairy production to combat the food insecurity of subsistent smallholder farmers (Hailu and Abate 2016). The term smallholder farmer describes farmers with less than two hectares of land on which they grow subsistence crops mainly by family labour. They also own a few herds of livestock and use outdated agricultural technologies (George 2015; Salami et al. 2010). As part of government initiatives to optimise dairy production, different technologies are introduced, including improved breeds, improved feeds and milk collector cooperatives (Yonas et al. 2018).

In Ethiopia, several studies analysed livestock production, feed resources availability, Fogera cattle reproductive and productive performance, and dairy cattle breeding practices (Fekade et al. 2020; Girma et al. 2016; Gizaw et al. 2017; Kebede et al. 2015; Yonas et al. 2018). In their study, they found that cattle produce 4 litres of milk on an average over a lactation period of 10–12 months. Cattles are constrained by scarce feed, low adoption of technologies to improve feed quality, disease outbreaks, droughts, lack of healthy water during winter and flooding during summer. In Zambia, Jodlowski et al. (2016) found that livestock ownership enhances dietary diversity and increases the consumption expenditures of households. As a result, previous studies have failed to provide context-specific evidence of how cattle dairy production improves the food security of smallholder households. This study was, therefore, aimed to investigate how cattle dairy production impacts smallholder households’ food security, which provides context-specific evidence for policymakers, practitioners and academics to better inform policy formulation and innovation.

Methods

This research was conducted in the north-western part of Ethiopia. Smallholder households were selected as the unit of analysis based on a multi-stage sampling step. First, two districts of northern Amhara National Regional State were chosen from the central Gondar Zone (East Dembia and West Dembia). Second, two kebeles were chosen per district according to their cattle dairying potential and the government’s involvement in the sector in the past. Third, dairy producers and non-producers were identified in each kebele through the agricultural offices. Finally, in total, 313 households were selected for the sample (182 producers and 131 non-producers) randomly proportionate to the sample size. The sample size was calculated using Cochran’s (1975) formula using the population’s limited size, heterogeneity and finiteness (Kothari 2004):

n = n_{0} [1 + (n_{0} - 1 / N],

where n₀ is Cochran’s sample size recommendation for infinite populations, N is the population’s size where the sample drowns and n is the new, adjusted sample size.

Primary data were collected directly from sample respondents and development agents (animal experts) in the study area. To obtain second-hand evidence, secondary information was collected from relevant government offices. To collect primary data from sampled households, the household interview schedule was used. The interview schedule questions were pre-tested and managed by trained enumerators.

Using descriptive analysis, the socioeconomic characteristics and food-security status of sample households were described. As part of the inferential analysis, the t-tests and chi-squares tests measured the significance of variances between the mean values of continuous and categorical variables in the two groups of cattle dairy producers and dairy non-producers, respectively. The Endogenous Switching Probit Regression (ESPR) model analysed the effects of dairy production on a household’s food security.

Endogenous Switching Probit Regression Model

Due to the empirical and theoretical complexity of impact studies, different approaches have been used, depending on the availability of data and study objective. The well-known approaches have included Simple Descriptive Statistics, Propensity Score Matching (PSM), Heckman’s selectivity model and Exogenous Switching Regression models (Kimty 2016; Tesfamicheal et al. 2017).

In the context of dairy production, smallholder households decide to produce dairy if they believe that the benefits in terms of food security exceed those of non-producers. Hence, the study aimed to analyse the effect of dairy production on household food security. It is difficult to do so without considering the unobserved heterogeneity and possible selection biases. This is because there are observed and unobserved differences between dairy producers and dairy non-producers (Tesfamicheal et al. 2017).

In this regard, using a simple descriptive statistics approach is unsuitable for this study. Because it fails to address selectivity bias and endogeneity issues. Heckman’s selectivity model is also not able to fully solve the problem. Although the approach controls the selection bias, it does not control systematic differences between groups due to the assumption that the consumption functions would differ between dairy producers and non-producers by only a constant term (Kimty 2016). Furthermore, the model can only be applied when one regime (in this case, dairy non-producers) is missing.

In PSM, dairy producers and non-producers are matched by using a propensity score, and the average differences in outcome variables are measured between the treated and untreated households. Accordingly, self-selection bias is mitigated (Tesfamicheal et al. 2017). However, the robustness check of PSM does not consider unobserved factors heterogeneity, and thus, it still yields biased estimates due to unobserved factors like personal characteristics (Kimty 2016). Therefore, the robustness check using the ESPR model is more accurate since it considers both observable and unobservable heterogeneity. Furthermore, this robust model was tested for multicollinearity, heteroscedasticity and goodness of fit using specific tests.

Therefore, an ESPR model was applied in this study to address the aforementioned two problems: (a) self-selection bias and (b) unobserved heterogeneity. As a result, in this study, both the unobservable and observable characteristics of dairy producers and non-producers were taken, which affect the participation decision and outcome variables. In addition, in contrast to Heckman’s selectivity model, the ESPR model observes both regimes (dairy producers and non-producers) (Tesfamicheal et al. 2017). For this reason, this study employed an ESPR model to examine whether cattle dairy production increased smallholder households’ food security.

An ESPR consists of two stages. First, a probit model-identified factors that determine household involvement decisions in dairy productions. It is estimated as follows:

Z * i = α + γ Q_{i} + ε_{i},

(1)

where Z_i* reflects whether the household produces dairy; otherwise, 0; α is an intercept; Q_i represents the vector of exogenous variables influencing smallholder households’ decisions whether to produce dairy; γ is a vector of coefficients; and $ε_{i}$ represents the disturbance terms with a zero mean and constant variance. In the second stage of the ESPR, the regression model is used to consider the potential selection biases. A household only produces dairy if it gains a significant net benefit than if it did not. While the researcher was unaware of the perceived net benefits of dairy production, he observed characteristics of farmers and technology attributes during the survey. In this case, the net benefits of dairy production can be represented by a latent variable Ii, which is not observed; however, it can be expressed as a function of observed characteristics and attributes, represented by the variable m according to the model below:

I_{i} = m_{i} Y + ε_{i}, I_{i} = 1, if I_{i} > 0; I_{i} = 0 if I_{i} < 0;

The dummy variable Ii is equal to 1 for dairy producers’ households, and zero otherwise. The binary outcomes (household food security) conditional on producing dairy were represented as switching regimes as follows:

Regime 1 (for dairy producer households) : Y_{1 i} = β_{1} X_{i} + ε_{1 i}, if I i = 1 and

(2a)

Regime 2 (for dairy non-producer households) : Y_{2 i} = β_{2} X_{i} + ε_{2 i}, if I i = 0

(2b)

Assume that the error terms ε₁_i, ε₂ _i and 𝑢_𝑖 have trivariate normal distributions, with a zero mean vector and a covariance matrix:

Cov (u_{i}, ε_{1 i}, ε_{2 i}) = [[\begin{matrix} σ_{u}^{2} & . \\ σ_{ε 1 u}^{2} & σ_{ϵ 1}^{2} \\ σ_{ϵ 2 u}^{2} & . σ_{ϵ 2}^{2} \end{matrix}]] .

Here in the selection equation, $σ_{u}^{2}$ stands for the variance of the error term, whereas $σ_{ϵ 1}^{2}$ and $σ_{ϵ 2}^{2}$ stand for the variances of the error terms in the continuous equations. $σ_{ϵ 2 u}^{2}$ and $σ_{ε 1 u}^{2}$ are the covariances of 푢푖 and ε₂ _i and ε₁ _i , respectively. The covariance of the corresponding error terms cannot be calculated since Y₁ _i and Y₂ _i are not observed instantaneously. With this structure, the outcome equation’s error terms are interrelated with the selection equation’s error term, which results in non-zero expected values of ε1i and ε2i given the selection equation’s error term, ui. Therefore, the anticipated values of the truncated error terms (휀₁ | 퐼 = 1) and (휀 | 퐼 = 0) are represented as

\begin{array}{l} (ε_{1} ∣ I = 1) = (ε_{1} ∣ u > - Z α) = δ ε_{1 u} \frac{φ (\frac{z α}{σ})}{Φ (\frac{z α}{σ})} \equiv σ ϵ_{1 u} λ_{1} and \\ (ε_{2} ∣ I = 0) = (ε_{2} ∣ u \leq - Z α) δ ε_{2 u} \frac{- φ (\frac{z α}{σ})}{1 - Φ (\frac{α α}{σ})} \equiv σ ϵ_{2 u} λ_{2} . \end{array}

In the standard normal distribution, Φ denotes the cumulative distribution and φ represents the probability density function, respectively. The ratio of Φ and φ assessed at Zα is referred to as the inverse Mills ratio 휆2 and 휆1 (selectivity terms). If the predicted covariances σε1u and σε2u are statistically significantly different, then the decision to produce dairy and household food security are correlated. This indicates evidence of endogenous switching as well as sample selectivity biases (Irina & Jeffrey 2015).

Additionally, treatment effects are estimated. Using the results for estimated values of the explanatory variable for producers and non-producers in actual and counterfactual scenarios, the average treatment effects on treated and untreated (ATT and ATU) are calculated:

E (\frac{Y_{1 i}}{I i} = 1, X_{1 i}) = β_{1} X_{1 i} + σ ϵ_{1} u ρ_{1} \frac{φ (Z α)}{Φ (zα)}

(3)

E (\frac{Y_{2 i}}{I i} = 0, X_{2 i}) = β_{1} X_{2 i} - σ ε_{2 u} ρ_{1} \frac{φ (Z α)}{(1 - Φ (Z α))}

(4)

E (\frac{Y_{2 i}}{I i} = 1, X_{1 i}) = β_{2} X_{1 i} + σ ε_{2 u} ρ_{2} \frac{φ (Z α)}{Φ (Z α)}

(5)

E (\frac{Y_{1 i}}{I i} = 0, X_{2 i}) = β_{2} X_{2 i} - σ ε_{1 u} ρ_{2} \frac{φ (Z α)}{(1 - Φ (Z α))} .

(6)

Where 휌1 and 휌2 are correlation coefficients between the selection equation error term 푢푖 and the error terms of the outcome equations 휀1 and 휀2. ATT is estimated from the variance among the expected values of the dairy producers (Equation (3)) and if the dairy producers were not producing (Equation (5)). ATU is estimated from the variance among the expected values of the dairy non-producers (Equation (4)) and if dairy non-producers were producing (Equation (6)).

Food-security Measurement

Conceptually, food security is complicated and includes many aspects such as food access, availability, utilisation and stability. Hence, despite the development of a wide variety of measurement indicators, no single one has captured the multidimensional aspects of food security. Consequently, studies use a combination of food-security measures in accordance with study objectives and study areas. This study applied the food consumption and household dietary diversity scores (HDDs) to assess smallholder households’ food security.

Household Dietary Diversity Scores

FANTA II released HDDs, in 2006, as a qualitative tool for measuring household food access. In different countries, it has been validated as a method of measuring food availability and access aspects of food security (Huluka and Wondimagegnhu 2019). The method measures how many food groups are eaten by the household (12 commonly eaten food groups, in this study case [Table 1]) during the previous 24-hour recall period (Gonete et al. 2020). Food groups were given a score of 1 (if consumed) or 0 (if not consumed), which is equal to their total intake. The HDDs range from 0 to 12, counting the food groups eaten in the past 24 hours (Jodlowski et al. 2016). That is:

HDDs = Ce + R + Ve + Fr + Pl + Mt + Eg + Fs + Mk + Ob + Sh + Ct .

The scores of HDDs were used as an indicator to determine whether household food is secure or insecure. A household that consumed six or more food groups was considered food-secure, and otherwise, food-insecure (Arias 2018).

Household Food Consumption Score

The food consumption score (FCS) was primarily used by the WFP, which is a combination of a long period and sum-frequency of consumption for in-depth food-security analysis (Baliwati et al. 2015). For this study, 12 commonly eaten food groups (Table 1) were collected on how often each household consumed a particular food item over the past 7 days. As indicated in Table 1, the values of each food group were multiplied by their weight to compute household FCS. Finally, the weighted scores for each food group were added to calculate household FCS. Consequently, those households with at least 35 FCS were categorised as food-secure, while those with less FCS were categorised as food-insecure (WFP & CSA 2019).

Table 1.

Food Consumption Multiplication Weight.

No.	Food Group	Weight		No.	Food Group	Weight
No.	Food Group	HDDs	HFCs	No.	Food Group	HDDs	HFCs
Ce	Cereals	0/1	2	Eg	Eggs	0/1	4
R	Root/tubers	0/1	2	Fs	Fresh and dry fish	0/1	4
Ve	Vegetables	0/1	1	Mk	Milk and milk products	0/1	4
Fr	Fruits	0/1	1	Ob	Oils/fats/butter	0/1	0.5
Pl	Pulses/legumes/nuts	0/1	3	Sh	Sugar or honey	0/1	0.5
Mt	Meat	0/1	4	Ct	Condiments, coffee, tea	0/1	0

Source: WFP and CSA (2019).

Control Variable

The complexes of variables in the study area are expected to have an impact on the food security level of smallholder households. These factors are hypotheses as follows (Table 2).

Age: Household head’s age indicates the general experience they possess. Older farmers have more experience and resources to handle dairy production (Idahe et al. 2018). In contrast, younger farmers have more awareness compared to the older generation (Melesse 2018). Since older farmers are inactive to receive information. Consequently, it was expected that household head’s age would either be positively or negatively correlated with household food security. Gender: it refers to a household head’s maleness or femaleness. Socio-cultural norms and values restrict women’s freedom of mobility, access to resources and access to information and technological innovations (Shibeshi 2017). Conversely, it enables men to obtain better livestock and improved breeds (Lombebo and Wosoro 2019), which improves their food security. Hence, the study predicted that households headed by men would have a positive impact on food security.

Education: The study predicted that education would have a positive impact on households’ food security (Tefera and Tefera 2014). Since, education can help people to obtain and apply relevant information to their farming practices (Lidetu 2019).

Livestock ownership: The study measured livestock ownership as a TLU (Storck et al. 1991). Livestock production is a significant means of income for smallholders (Derbe 2020). Since a household’s food security was expected to be positively impacted by TLU.

Farm size: It refers to the amount of land owned by households, measured in hectares. A large farm size provides access to grazing land for dairy production and provides a secure livelihood for households (Idahe et al. 2018). The study, therefore, expected the size of the farm to affect household food security positively.

Family size: It is determined by the number of individuals living in the households. A large family size increases the number of family dependency ratios (Maharjan and Joshi 2011) and increases food consumption and expenditure, thereby negatively affecting household food security (Agidew and Singh, 2018). Therefore, it had been anticipated that family size would negatively correlate with household food security.

Family labour: Its effect is defined as the amount of productive and active family members in a household measured in adult equivalents (Storck et al. 1991). Farming activities, including dairy products, are labour-intensive in developing countries (Shibeshi 2017), meaning a household that has a larger adult equivalent can produce diversified food items. So, it was hypothesised that a higher adult equivalent would improve household food security.

Dependency ratio: A ratio was calculated using the number of people under 18 plus the number of people over 65 plus the number of people chronically ill and disabled adults 19–64 years of age to the total number of households (Miller et al. 2011). A higher dependency ratio means that inactive members of the household are fed by a few workers in the household, resulting in the lowest calorie intake. Consequently, food security decreases as the dependency ratio increases (Maharjan and Joshi, 2011). So, it was expected that the dependency ratio would be negatively correlated with food security.

Veterinary service: It was hypothesised that access to veterinary service would positively relate to household food security because it helps to protect the health of dairy cattle breeds.

Market: The variable measured the distance between the nearest market and the household home in kilometres. Access to the markets is essential for purchasing farm equipment and farm inputs, and for selling dairy and dairy products that can maximise returns on dairy production and food security for farm households closer to markets than those living far away from markets (Lijalem et al. 2015). Thus, it was anticipated that distance to the market would be positively associated with food security.

Extension: Extension agent contacts are helpful for farmers to learn, become technique proficient and develop their confidence to manage integrated dairy production sustainably (Dehinenet et al. 2014). It also inspires households to adopt improved technologies (Muluye 2016). Therefore, it was expected that extension contact would increase food security.

Non-farm income: It was believed that receiving additional income would help households to access dairy production inputs and enhance their confidence to participate in dairy production to enhance their food security. As such, it was believed that non-farm income would improve household food security.

Feed: Feed availability is a determinant of milk quality and yield in Ethiopia, regardless of the dairy production systems and environments (Getabalew et al. 2019). In other words, a lack of sufficient feed negatively impacts milk production (Lijalem et al. 2015). Hence, the study hypothesised that access to adequate feed would improve household food security.

Access to credit: Credit facility services are a means to bridge households’ financial gap for purchasing new improved technologies, such as improved breeds of dairy cattle and other household necessities (Mekuria et al. 2017). Credit facilities also serve as a buffer against food insecurity (Nwokolo 2015). Thus, access to credit was hypothesised to improve food security.

Table 2.

Control Variables Explanations and Measurements.

Variable	Variable’s Explanation	Measurement
Age	Household head age	Continuous (years from birth)
Gender	Household head gender	0 = female, 1 = male
Education	Household head’s education	0 = illiterate, 1 = literate
Family size	Number of people living in the household	Continuous (number of people)
Dependency ratio	Dependency ratio in the household	Continuous (0–1)
Family labour	Family labour in the adult equivalent	Continuous (adult equivalent)
Tropical Livestock Unit (TLU)	Livestock owned by a household	Continuous (TLU)
Farm size	The size of land in hectares	Continuous (hectare)
VETS	Access to veterinary services	0 = no, 1 = yes
Market	Distance from the market in kilometres	Continuous (km)
Extension	Contact with extension agents	0 = no, 1 = yes
Income	Income from non-farm sources	0 = no, 1 = yes
Feed	If the household has access to enough feed	0 = no, 1 = yes
Credit	If the household has access to credit	0 = no, 1 = yes

Source: The authors.

Results and Discussions

Food-security Status of Sampled Households

It can be seen from Table 3 that in both food-security measurements most sampled households were food-secure. Though every household consumes a variety of cereals, fats and oils, food-secure households tend to eat different cereals, sugar, milk, pulses and oil or fats with a higher frequency than their food-insecure counterparts. As displayed in Table 3, the highest percentage (73.08%) of dairy producer households was food-secure. However, only 32.06% (HDDs) and 34.35% (FCS) of dairy non-producer households were food-secure, and the remaining higher percentage of dairy non-producer households were food-insecure. Additionally, the chi-square results revealed that there were significant variations among dairy producers and dairy non-producers in terms of food-security status at less than a 1% significance level (χ² = 54.34 HDDs and χ² = 49.68 FCS). Therefore, dairy producer households had better food security. This may be because households who produce dairy products consume more milk and milk products, and that affects their consumption of other foods as well.

Table 3.

Food Security Status of Respondents.

Measurement	Food Security Status	Dairy Production Participation			χ² Value
Measurement	Food Security Status	Dairy Producer (n = 182)	Dairy Non-producer (n = 131)	Total (n = 313)	χ² Value
HDDs	Food-secure	133 (73.08%)	42 (32.06%)	175 (55.91%)	54.34***
HDDs	Food-insecure	49 (26.92%)	89 (67.94%)	138 (44.09%)	54.34***
FCS	Food-secure	133 (73.08%)	45 (34.35%)	178 (56.87%)	49.68***
FCS	Food-insecure	49 (26.92%)	86(65.65%)	135 (43.13%)	49.68***

Source: The authors.

Notes: *** indicates statistical significance at less than 1% significance level.

Impacts of Dairy Production on Household Food Security

This study examined the effect of dairy production on food security in a household using an ESP regression (ESPR) model since food security was a dummy variable (Lokshin and Sajaia, 2011). Before running the regression, the problem of multicollinearity was checked through the variance inflation factor (VIF) and contingency coefficient. Consequently, multicollinearity between the family size and dependency ratio was detected. As a result, the variable dependency ratio was eliminated and a re-run of the regression confirmed the absence of a multicollinearity problem. Lastly, 13 control variables in the selection equation (treatment equation) were used in the ESPR model to estimate the determinants of household participation in dairy production.

In addition, a simple falsification test was applied. The instrument candidate variables that were significant in the selection model and insignificant in the outcome model were used as valid instruments in the ESPR model (Oparinde 2021). The outcome variable of this study is food security for households. Getting enough feed was selected as an instrumental candidate variable. Because, for dairy producers to benefit from dairy production to increase food security, they must obtain enough feed for their dairy cattle to become dairy producers. In the first stage, households’ participation decisions to become dairy producers are significantly influenced by getting enough feed, but in the second stage, it does not significantly affect households’ food security. Therefore, the variable that gets enough feed was chosen as an instrumental variable in the ESPR model as it matched the criterion (Min et al. 2017). In addition, the Wald test was highly significant, indicating that the ESPR model is well-fit.

Determinants of Household Food Security

As presented in Table 4, the food security of dairy producers and non-producers was affected by different variables. The dairy producer household’s food security was significantly and positively impacted by the distance from the market (Market), access to veterinary service (VETS), as well as farm size. The dairy non-producer household’s food security was significantly and positively correlated with household head educational level (Education) and with livestock ownership (TLU). In contrast, the relationship between other hypothesised variables and food security is insignificant.

Table 4.

Result of ESPR Model on the Impact of Dairy Production on Household Food Security.

Variables	HDDs				FCS
	Dairy Producer		Dairy Non-producer		Dairy Producer		Dairy Non-producer
	Coef.	Std Err	Coef.	Std Err	Coef.	Std Err	Coef.	Std Err
Age	.039	.013	.029	.007	.029	.014	.012	.006
Education	.168	.234	.520*	.234	.162	.225	.487*	.202
Gender	.162	.258	.048	.275	.162	.306	.082	.234
Family size	–.008	.010	–.012	.035	–.006	.012	–.019	.065
Family labour	.013	.060	.035	.066	.014	.067	.023	.008
TLU	.016	.042	.133***	.036	.024	.053	.132***	.045
Farm size	.316*	.199	.002	.162	.319*	.191	–.007	.153
VETS	.916***	.355	.223	.217	.976***	.359	.230	.103
Market	.008*	.003	.001	.003	.008*	.004	–.001	.003
Extension	.324	.269	–.222	.213	.256	.269	–.338	.209
Income	.045	.287	–.405	.255	.035	.271	–.271	.251
Feed	.240	.258	.071	.198	.223	.254	.149	.160
Credit	–.232	.308	.014	.247	–.181	.256	.031	.231
_cons	–.502	.812	–1.432	.481	–.546	.781	–1.611	.459
athrho1	–.382	.659			–.489	.676
athrho0	–12.685	567.53			–15.89	1819.201
Number of obs. = 312Wald χ² (14) = 47.21Prob. > χ² = 0.0000Log likelihood = –325.36846LR test of indep. eqns (rho1 = rho0 = 0): χ²(2) = 6.13 Prob > χ² = 0.0482					Wald χ² (14) = 46.62Prob. > χ² = 0.0000Log likelihood = –337.20594LR test of indep. eqns (rho1 = rho0 = 0): χ²(2) = 5.79

Source: The authors.

Notes: * and *** indicate statistical significance at less than 10% and 1% significance levels, respectively.

The household head’s education level (Education) of the dairy non-producer household was statistically significant and positively linked with food security at less than a 10% significance level. Compared to illiterate households, the probability of literate non-producer households being food-secure increased by 48.7%. Literate households are more likely to use improved practices, diversifying their sources of income, thereby increasing yields and food security (Gebre 2012). In line with this study, Khanyiswa et al. (2017), Maitra and Rao (2015) and Nwokolo (2015) found that household head’s education is positively associated with food security. Also, Maharjan and Joshi, 2011) note that food insecurity is most prevalent among illiterate households.

The results suggest, as hypothesised, that livestock ownership (TLU) significantly and positively impacts the food security of dairy non-producer households at less than a 1% significance level. It is illustrated that the increment of a TLU raises the likelihood of dairy non-producer households being food-secure by 13%. Ownership of livestock is a more viable means of availability and access to food in households, as it provides both animal-based food items and income to purchase diversified food items in the market (Derbe et al. 2018; Mekuria et al., 2017; Tefera and Tefera, 2014). Thus, livestock ownership is a double-pronged response to food-security enhancement.

As expected, farm size has an encouraging influence on dairy producers’ food security at less than a 10% significance level. Accordingly, as the farm size of dairy producer households is increased by one hectare, the likelihood of being food-secure would increase by 31.6%, assuming other variables remain the same. This may be due to the size of the farm allowing households to produce more and more diversified food groups (crops and/or livestock products) for direct consumption, as well as exchange food items required from the market. In agreement with this study, Welderufael (2014) found that farm size reduces food insecurity.

As hypothesised, the results indicated a significant and positive association between access to veterinary services (VETS) and dairy producer household food security at less than a 1% significance level. It means that dairy producer households who have access to veterinary services are more food-secure than dairy producer households without such access. Holding other variables constant, the coefficient also showed that dairy producer households who have access to veterinary services have a 91.6% higher likelihood of being food-secure than those without. Since households with access to veterinary services can increase dairy cow productivity, thereby increasing household food security (Mebrate et al. 2019).

The findings revealed a positive correlation between distance from the market (Market) and the food security of dairy producer households. It suggests food insecurity increases as dairy producer households become closer to the market. It may be attributed to the fact that households living near markets specialise in some food items that they sell to generate income. This hinders them from producing and eating diversified food items to ensure food security. They also participate in off-farm activities in the marketplaces to generate income instead of growing diversified crops and keeping livestock. In addition, they may not purchase diversified food items from the market due to a lack of knowledge about the importance of eating diversified food. In contrast, households located far from the market may cultivate a variety of crops and keep livestock to minimise their failure risk, thus improving their food security. In contrast to this finding, Welderufael (2014) found that an increase in the time to get to the market negatively impacts food security.

Based on the results of the study, household head’s age (age) has a positive and insignificant relationship with food security. Hence, the variation of households in terms of asset ownership is less because of asset sharing between older families and their children or younger households (Kosec et al. 2018). Additionally, the gender of the household head (gender) has a positive and insignificant correlation with food security. This is because Ethiopia’s government focuses on ensuring gender equality, so that women attend education and training, as well as participate in government administrations. Under Ethiopian law, women were also guaranteed equal access to assets (Kassa 2015). Therefore, the gender gap in terms of freedom of movement, access to resources and access to information and technology is less significant and has little impact on food security.

According to the results, there is an insignificant and inverse relationship between food security and family size. A large family size leads to higher family dependency ratios and increases food consumption, which negatively impacts household food security (Agidew and Singh 2018; Maharjan and Joshi 2011). To maintain a low dependency ratio, households in the study area use contraception to avoid increasing their family size beyond their household carrying capacity (Oumer et al. 2020). On the other hand, family labour has a direct relationship with food security, despite its insignificance. Having a larger adult equivalent in the household enables the household to produce more food, improving food security. Families in the study area, however, have smaller farms and cannot use all available family labour, which would have affected food security significantly. Therefore, during the summer season, household members migrate to cash crop-producing lowland areas of Quora, Metema, Humera, Tegedie, Lay Arimachiho and Tach Arimachiho to sell extra labour wages and gain an income that reduces the household’s food shortage (Derbe et al. 2021).

Contact with an extension agent (extension) to get advice about dairy production has a positive and insignificant influence on dairy producer households. As a result of the advice, dairy producer households can improve the quality of their dairy products and increase their incomes and food security (Dehinenet et al. 2014; Muluye 2016). On the other hand, due to the lack of dairy products, the advice dairy non-producer households receive about dairy production is negatively and insignificantly correlated with food security. Extension contact is rarely used in the study area to obtain dairy-related information (Assefa et al. 2016), so it has no significant impact on food security.

There is a positive and insignificant relationship between non-farm income (income) and food security for dairy producer households. The main source of non-farm income in the study area is the sale of labour, especially during the summer months when people migrate from rural-to-rural areas (Derbe et al. 2021). So non-farm income enables dairy producers to purchase feed for their dairy (Al-Amin and Hossain 2019), which is the primary challenge for dairy production in the study area. In contrast, non-farm income affects dairy non-producer households negatively and insignificantly. The main livelihood source of dairy non-producer households, crop management, is negatively constrained by non-farm income-generating activities. In the study area, however, non-farm income is small and does not significantly affect a household’s food security.

In the study area, livestock and dairy production are challenged by the lack of feed availability. The result confirmed that feed availability is insignificantly and positively correlated with food security. Crop residue is the primary source of feed (Gezahagn et al. 2017), which is less due to smaller cultivated farms. The availability of feed among smallholder farmers in the study area, however, is insufficient and similar, and as a result, its impact on food security is insignificant.

Estimation of Dairy Production Impact on Household Food Security

The study examined binary outcomes (food-secure = 1, and food-insecure = 0) and treatments (dairy producer = 1, and dairy non-producer = 0). The ESPR model was applied to the estimation of the impact of dairy production on the food security of households through a counterfactual analysis using a post-estimation method from observed and unobserved heterogeneities between dairy producer and non-producer households (Lokshin and Sajaia, 2011).

Cells (a) and (d) (see Table 5) present the estimated probabilities of food security based on the actual (observed) outcomes for dairy producer households and dairy non-producer households, respectively. The actual outcome (a) shows that the expected mean probability of food security for dairy producers was 0.771 HDDs and 0.675 FCS of mean probability level for food security. However, the observed outcome condition (d) shows 0.555 HDDs and 0.565 FCS of mean probability level for food security for dairy non-producer households. Counterfactual case (b) (Table 5) shows that if dairy producer households had not produced dairy, they would have had on average 0.287 HDDs and 0.307 FCS of the mean probability level for food security. Counterfactual case (c) (Table 5) shows that if dairy non-producer households had produced dairy, they would have had on average 0.658 HDDs and 0.636 FCS mean probability level of food security.

Table 5.

The Average Probability of the Food Security Score (FCS) and Treatment and Heterogeneity Effects.

Outcome Variable	Category	Prediction		Treatment Effect
Outcome Variable	Category	Producer	Non-producer	Treatment Effect
HDDs	Producer	(a) 0. 771	(b) 0.287	0.484 (4.145***) = ATT
	Non-producer	(c) 0.658	(d) 0.555	0.104 (6.370***) = ATU
	Heterogeneity effect	0.113	–0.267	0.38
FCS	Producer	(a) 0.765	(b) 0.307	0.459 (4.578***) = ATT
	Non-producer	(c) 0.636	(d) 0.565	0.071 (6.266***) = ATU
	Heterogeneity effect	0.129	–0.258	0.388

Source: The authors.

Notes: *** indicates statistical significance at less than a 1% significance level.

For dairy producers, the difference between cells (a) and (b) revealed the ATT, while for non-producers, the difference between cells (c) and (d) revealed the ATU. Based on the ATT results in Table 5, dairy production increased food-security status in dairy producer households by 48.4% HDDs and 45.9% FCS on average; this difference is significant statistically at less than 1% significance level. The ATU results also show that dairy production would have increased food-security status in dairy non-producer households by 10.4% HDDs and 7.1% FCS on average if they had produced dairy; the difference is significant statistically at less than a 1% significance level. Additionally, there was a positive higher transitional heterogeneity effect of 38.8% for dairy producer households compared with dairy non-producer households. The findings of the study suggest that smallholder households’ participation in dairy production has contributed significantly to food security compared with dairy non-producer households. Consequently, food security in households has improved in the study area.

Conclusion

Study findings revealed that dairy production increased food diversification, which improved smallholder households’ food security. According to the ATU results (HDDs and HFCS), households that produce dairy have better food security than the counterfactual case if they do not produce dairy. The ATU (HDDs and HFCS) results also indicated that households who did not produce dairy would have a higher probability of food security if they did produce dairy. The conclusion is that participation in dairy production increases the likelihood of household food security. Furthermore, studies that were conducted in Malawi, Jordan, Zambia and Tanzania (Al-Atiyat 2014; Banda et al. 2021; Jodlowski et al. 2016; Lwelamira et al. 2010) confirmed that dairy farming increased household income, resilience to food shortages and improved food-security status. Thus, stakeholders should work together to exploit the area’s dairy production capacity to improve food security for smallholder households.

Footnotes

Acknowledgements

It is the authors’ pleasure to express their gratitude to the farmers of the study area for their information and assistance.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The authors received no financial support for the research, authorship and/or publication of this article.

Notes

ORCID iD

Tegegne Derbe

References

Agidew

A. Assefa

, and Singh

K. N.

2018. ‘Determinants of Food Insecurity in the Rural Farm Households in South Wollo Zone of Ethiopia: The Case of the Teleyayen Sub-watershed’. Agricultural and Food Economics 6 (10): 1–23.

Al-Amin

A. K. M. Abdullah

, and Hossain

M. J.

2019. ‘Impact of Non-farm Income on Welfare in Rural Bangladesh: Multilevel Mixed-effects Regression Approach’. World Development Perspectives (13 February), 95–102. https://doi.org/10.1016/j.wdp.2019.02.014

Al-Atiyat

Raed M.

2014. ‘Role of Small-scale Dairy sector in Food Security and Poverty Alleviation’. Journal of Food, Agriculture and Environment 12 (2): 427–33.

Alilo

Afras A.

2019. ‘Review on Breeding Objectives and Practices of Dairy Cattle Production in Ethiopia’. Journal of Cell and Animal Biology 13 (1): 1–7. https://doi.org/10.5897/JCAB2018.0457

Arias

Frances Pola S.

2018. ‘Validity of Household Dietary Diversity Score as a Measure of Food Insecurity among Households in Lucena City, Quezon’. Philippine Journal of Health Research and Development 22 (4): 1–8.

Asresie

, and Zemedu

Lemma

. 2015. ‘Contribution of Livestock Sector in Ethiopian Economy: A Review’. Advances in Life Science and Technology 29: 79–90.

Assefa

A. H.

, Tegegne

Azage

, and Gebremedhin

2016. ‘Agricultural Extension Service for Dairy Production in Bure District, West Gojjam Zone, Amhara Region of Ethiopia’. International Journal of Original Research 2 (2): 12.

Baliwati

Y. F.

, Briawan

, and Melani

2015. ‘Validation Household Dietary Diversity Score (HDDS) to Identify Food Insecure Household in Industrial Area’. Pakistan Journal of Nutrition , 14 (4): 234–38.

Banda

L. J.

, Chiumia

, Gondwe

T. N.

, and Gondwe

S. R.

2021. ‘Smallholder Dairy Farming Contributes to Household Resilience, Food, and Nutrition Security Besides Income in Rural Households’. Animal Frontiers 11 (2): 41–46. https://doi.org/10.1093/af/vfab009

10.

Bekuma

, Galmessa

, and Fita

2018. ‘Review on Adoption, Impacts and Determinant Factors of Dairy Technology in Ethiopia’. Journal of Dairy & Veterinary Sciences 74 (2): 2224–6088.

11.

Dagninet

, and Adugnaw

2020. ‘Status of Food Insecurity in Dryland Areas of Ethiopia: A Review’. Cogent Food & Agriculture 6 (1). https://doi.org/10.1080/23311932.2020.1853868

12.

Dehinenet

, Mekonnen

, Kidoido

, Ashenafi

, and Bleich

E. G.

2014. ‘Factors Influencing Adoption of Dairy Technology on Small Holder Dairy Farmers in Selected Zones of Amhara and Oromia National Regional States, Ethiopia’. Discourse Journal of Agriculture and Food Sciences 2 (5): 126–35.

13.

Derbe

Tegegne

. 2020. ‘Determinants of Smallholder Farmers’ Livelihood Strategies in Wogera District of Ethiopia’. Ethiopian Renaissance Journal of Social Sciences and Humanities 7 (1): 1–11.

14.

Derbe

, Melak

, and Derso

2021. ‘Social Vulnerability Indices of Seasonal Rural-to-rural Migrant: A Comparative Study of Migrant-sending and Non-migrant Households in Northwest Ethiopia’. Ethiopian Renaissance Journal of Social Sciences and Humanities 7 (2): 1–11.

15.

Derbe

Tegegne

, Yehuala

, and Agitew

2018. ‘Factors Influencing Smallholder Farmers Adoption of Eucalyptus Woodlot in Wogera District, North Gondar Zone, Amhara Regional State of Ethiopia’. International Journal of Scientific Research and Management 6 (7): 567–74. https://doi.org/10.18535/ijsrm/v6i7.em07

16.

Famine Early Warning Systems Network (FEWS NET) and World Food Programme (WFP). 2019. ‘Ethiopia Food Security Outlook: The Level of Conflict-induced Displacement Needs Begin to Surpass Drought-induced Needs’. Situation Report .

17.

Fekade

, Bayssa

, and Nurfeta

2020. ‘Available Feed Resources and Nutritive Value of Major Browse Species in East Dembia District, Central Gondar, Ethiopia’. Journal of Biology, Agriculture and Healthcare 10 (13): 17–23. https://doi.org/10.7176/JBAH/10-13-03

18.

Food and Agriculture Organization of the United Nations (FAO). 2019. The Future of Livestock in Ethiopia Opportunities and Challenges in the Face of Uncertainty . Rome.

19.

George

2015. The Economic Lives of Smallholder Farmers. An Analysis Based on Household Data from Nine Countries . Rome: Food and Agriculture Organization of the United Nations.

20.

Getabalew

, Alemneh

, and Akeberegn

2019. ‘Dairy Production in Ethiopia: Existing Scenario and Constraints’. Biomedical Journal of Scientific & Technical Research 16 (April): 12304–309. https://doi.org/10.26717/BJSTR.2019.16.002903

21.

Gezahagn

, Getnet

, and Feyisa

2017. ‘Review on Major Feed Resources in Ethiopia: Conditions, Challenges and Opportunities Farming System Dynamism in Ethiopia View project African Chicken Genetic Gains View Project’. Academic Research Journal of Agricultural Science and Research 5 (3): 176–85. https://doi.org/10.14662/ARJASR2017.013

22.

Girma

, Alemayehu

, Abegaze

, and Kebede

2016. ‘Phenotypic Characterization, Population Structure, Breeding Management and Recommend Breeding Strategy for Fogera Cattle (Bos indicus) in Northwestern Amhara, Ethiopia’. In Animal Genetic Resources , edited by Baumung

, 13–29. Food and Agriculture Organization of the United Nations (FAO).

23.

Gizaw

, Ebro

, Tesfaye

, Mekuriaw

, Mekasha

, Hoekstra

, … Tegegne

(2017). Feed resources in the highlands of Ethiopia: A value chain assessment and intervention options . LIVES Working Paper 27. Nairobi, Kenya: International Livestock Research Institute (ILRI).

24.

Gonete

K. A.

, Tariku

, Wami

S. D.

, and Akalu

T. Y.

2020. ‘Dietary Diversity Practice and Associated Factors among Adolescent Girls in Dembia District, Northwest Ethiopia, 2017’. Public Health Reviews 41 (23): 1–13.

25.

Hailu

, and Abate

2016. Community Breeding Practice and the Challenges in Dairy Cattle Management in North Gondar, Ethiopia. International Journal of Environmental and Agriculture Research 2 (4): 76–83.

26.

Huluka

A. T.

, and Wondimagegnhu

B. A.

2019. ‘Determinants of Household Dietary Diversity in the Yayo Biosphere Reserve of Ethiopia: An Empirical Analysis Using Sustainable Livelihood Framework’. Cogent Food & Agriculture 5 (1): 1690829. https://doi.org/10.1080/23311932.2019.1690829

27.

Hunde

2018. ‘Dairy Cattle Breeding Program in Ethiopia: Lesson Learned from Case Studies in the Tropical Countries’. Academic Research Journal of Agricultural Science and Research 6 (2): 97–104. https://doi.org/10.14662/ARJASR2017.086

28.

Idahe

G. D.

, Geleta

F. T.

, and Amsalu

S. G.

2018. ‘Determinants of Dairy Product Market Participation of the Rural Households: The Case of Adaberga District in West Shewa Zone of Oromia National Regional State, Ethiopia’. Journal of Business & Financial Affairs 7 (4). https://doi.org/10.4172/2167-0234.1000362

29.

Irina

, and Jeffrey

2015. ‘A Control Function Approach to Estimating Switching Regression Models with Endogenous Explanatory Variables and Endogenous Switching’. Journal of Econometrics 190 (2): 252–66. https://doi.org/10.1016/j.jeconom.2015.06.014

30.

Jodlowski

, Winter-nelson

, and Baylis

2016. ‘Milk in the Data: Food Security Impacts from a Livestock Field Experiment in Zambia’. World Development 77: 99–114. https://doi.org/10.1016/j.worlddev.2015.08.009

31.

Kassa

2015. ‘Challenges and Opportunities of Women Political Participation in Ethiopia’. Journal of Global Economics 3 (4). https://doi.org/10.4172/2375-4389.1000162

32.

Kebede

, Alemayehu

, and Girma

2015. ‘Reproductive and Productive Performance of Fogera Cattle in Lake Tana Watershed, North Western Amhara, Ethiopia’. Journal of Reproduction and Infertility 6 (2): 56–62. https://doi.org/10.5829/idosi.jri.2015.6.2.95203

33.

Khanyiswa

, Dywili

, and Nwokolo

E. E

. 2017. ‘The Role of Education, Income in Determining Standard of Living and Food Security amongst the Residents of Mhlontlo Local Municipality Eastern Cape, South Africa’. Journal of Human Ecology 60 (1): 18–28. https://doi.org/10.1080/09709274.2017.1397929

34.

Kimty

2016. The Effects of Market Participation on Farm Households’ Food Security in Cambodia: An endogenous switching approach (MPRA Paper 69669). Germany: University Library of Munich.

35.

Kosec

, Ghebru

, Holtemeyer

, Mueller

, and Schmidt

2018. ‘The Effect of Land Access on Youth Employment and Migration Decisions: Evidence from Rural Ethiopia’. American Journal of Agricultural Economics 100 (3): 931–54. https://doi.org/10.1093/ajae/aax087

36.

Kothari

C. R.

2004. Research Methodology Methods & Techniques (2nd edition). Jaipur: New Age International Publishers.

37.

Lidetu

2019. ‘Review on Dairy Production System, Constraints and Opportunities in Ethiopia’. Journal of Scientific and Innovative Research 8 (4): 108–11.

38.

Lijalem

, Asefa

, and Sharo

2015. ‘Challenges and Opportunities of Dairy Cattle Production in Selected Districts of Sidama Zone, Southern Ethiopia’. Food Science and Quality Management 44: 5–12. https://doi.org/www.iiste.org

39.

Lokshin

, and Sajaia

2011. ‘Impact of Interventions on Discrete Outcomes: Maximum Likelihood Estimation of the Binary Choice Models with Binary Endogenous Regressors’. The Stata Journal 11 (3): 368–85.

40.

Lombebo

W. A.

, and Wosoro

E. S.

2019. ‘Challenges and Opportunities of Urban Dairy Cattle Keeping and its Role in Poverty Reduction of Livelihoods in Hosanna Town, Southern Ethiopia’. Veterinary Science and Research , 1 (1): 11–17.

41.

Lwelamira

, Binamungu

H. K.

, and Njau

F. B.

2010. ‘Contribution of Small Scale Dairy Farming under Zero-grazing in Improving Household Welfare in Kayanga Ward, Karagwe District, Tanzania’. Livestock Research for Rural Development 22 (31).

42.

Maharjan

K. L.

, and Joshi

N. P.

2011. Determinants of Household Food Security in Nepal: A Binary Logistic Regression Analysis. Journal of Mountain Science 8 (3): 403–13. https://doi.org/10.1007/s11629-011-2001-2

43.

Maitra

, and Rao

D. S. P.

2015. ‘Poverty-Food Security Nexus: Evidence from a Survey of Urban Slum Dwellers in Kolkata’. World Development 72: 308–25. https://doi.org/10.1016/j.worlddev.2015.03.006

44.

Mebrate

, Alemneh

, and Akeberegn

2019. ‘Dairy Production in Ethiopia: Existing Scenario and Constraints’. Biomedical Journal of Scientific & Technical Research 16 (5): 12304–09. https://doi.org/10.26717/BJSTR.2019.16.002903

45.

Mekuria

, Negatu

, and Mekonnen

2017. ‘Adoption of Improved Dairy Cows and Implications for Household Food Security: Evidence in Central Highland of Ethiopia’. Global Journal of Science Frontier Research (GJSFR): (D) Agriculture & Veterinary 17 (3).

46.

Melesse

2018. ‘A Review on Factors Affecting Adoption of Agricultural New Technologies in Ethiopia’. Journal of Agricultural Science and Food Research 9 (3): 1–4.

47.

Miller

C. M.

, Tsoka

, and Reichert

2011. ‘The impact of the Social Cash Transfer Scheme on Food Security in Malawi’. Food Policy 36 (2): 230–8. https://doi.org/10.1016/j.foodpol.2010.11.020

48.

Min

, Waibel

, and Huang

2017. ‘Smallholder Participation in the Land Rental Market in a Mountainous Region of Southern China: Impact of Population Aging, Land Tenure Security and Ethnicity’. Land Use Policy 68 (November): 625–37. https://doi.org/10.1016/j.landusepol.2017.08.033

49.

Muluye

K. M.

2016. ‘Milk Production and Reproductive Performance of Local and Crossbreed Dairy Cows in Selected Districts of West Gojam Zone, Amhara Region, Ethiopia’. MSc thesis, Bahir Dar University.

50.

National Disaster Risk Management Commission (NDRMC). 2020. Flood Response Plan Ethiopia 2020 Kiremt Season Floods .

51.

Negash

2019. ‘Pictures of food insecurity in Amhara region’. Research Journal of Social Science & Management 9 (3): 39–46. www.theinternationaljournal.org

52.

Nwokolo

E. E.

2015. ‘The Influence of Educational Level on Sources of Income and Household Food Security in Alice, Eastern Cape, South Africa’. Journal of Human Ecology 52 (3GG), 208–17. https://doi.org/10.1080/09709274.2015.11906944

53.

Oparinde

L. O.

2021. Fish Farmers’ Welfare and Climate Change Adaptation Strategies in Southwest, Nigeria: Application of Multinomial Endogenous Switching Regression Model. Aquaculture Economics and Management 25 (4): 450–71. https://doi.org/10.1080/13657305.2021.1893863

54.

Oumer

, Manaye

, and Mengistu

2020. ‘Modern Contraceptive Method Utilization and Associated Factors Among Women of Reproductive Age in Gondar City, Northwest Ethiopia’. Open Access Journal of Contraception 11: 53–67. https://doi.org/10.2147/oajc.s252970

55.

Salami

, Kamara

A. B.

, Abdul

, and John

2010. Smallholder Agriculture in East Africa: Trends, Constraints and Opportunities (Working Papers Series No. 105). African Development Bank, Tunis, Tunisia. http://www.afdb.org/

56.

Shapiro

B. I.

, Gebru

, Desta

, Negassa

, Nigussie

, Aboset

, and Mechale

2017. Ethiopia Livestock Sector Analysis . ILRI Project Report. Nairobi, Kenya: International Livestock Research Institute (ILRI).

57.

Shibeshi

T. F.

2017. Factors Affecting Adoption of Dairy Technologies and Their Impact on Farm Household Income and Asset Holdings: The Case of Tehuledere District, South Wollo Zone, Amhara Region, Ethiopia . Ethiopia: Hawassa University.

58.

Storck

, Emana

, Adenew

, Borowiecki

and W/Hawariat

Shimelis

. 1991. ‘Farming Systems and Farm Management Practices of Smallholder in the Hararghe Highlands: A Baseline Survey’. In Farming Systems and Resource Economics in the Tropics , Vol. 11, pp. 1–195. Keil: Wiss.-Verlag Vauk.

59.

Tefera

, and Tefera

2014. ‘Determinants of Households Food Security and Coping Strategies for Food Shortfall in Mareko District, Guraghe Zone Southern Ethiopia’. Journal of Food Security 2 (3): 92–99. https://doi.org/10.12691/jfs-2-3-4

60.

Tesfamicheal

, Tahirou

, Arega

, Haile

Mekbib G.

, Shiferaw

, Adetunji

, and Victor

2017. ‘Impacts of Extension Access and Cooperative Membership on Technology Adoption and Household Welfare’. Journal of Rural Studies 54: 223–33. https://doi.org/10.1016/j.jrurstud.2017.06.022

61.

Welderufael

2014. Determinants of Households Vulnerability to Food Insecurity in Ethiopia : Econometric analysis of Rural and Urban Households . 5 (24): 70–80.

62.

World Food Programme (WFP), & Central Statistical Agency (CSA). 2019. Comprehensive Food Security and Vulnerability Analysis (CFSVA) . www.wfp.org

63.

Yonas

, Bidru

, Abebaz

, G/tsadik

, Kidane

, Wondifraw

… Wondale

2018. Participatory Agricultural Production Systems Analysis of AGP-II Districts in North Gondar Zone: Implications for Research and Development . Bahir Dar: Amhara Agricultural Research Institute.