Dairy product market participation and its impacts on household food security of smallholder farmers in Jimma zone Ethiopia: A PSM approach

Abstract

This paper investigates the impact of dairy product marketing on household food security in Jimma Zone Ethiopia. A quasi-experimental research design was applied through the use of a cross-sectional data type. In order to identify the study kebeles and 266 respondents, a multi-stage sampling procedure was conducted, and the data was gathered through interview schedules. Descriptive statistics, Binary logit, and Propensity Score Matching (PSM) models were used to analyze the data. From the randomly selected households about 54.9% were participated in milk marketing. The study discovered that family size, livestock size, off-farm engagement, access to dairy input, ownership of crossbreed cows, and cooperative membership significantly and positively influenced dairy product market participation. Conversely, age, education level, and market distance had a negative effect. The results of PSM indicated that dairy product market participation had a positive impact on the household food consumption score and dietary diversity, increasing them by 16.38% and 15.42%, respectively. All in all, the research found that milk market participation has a meaningful effect on household food security, and so measures like Innovations to increase dairy production and productivity, and cooperative membership should be promoted to strengthen the participation of smallholder dairy farmers in the area. The study can offer empirical evidence on the contribution of dairy product market participation to household food security.

PUBLIC INTEREST STATEMENT

Dairy farming is a way of life for many people in rural Ethiopia, especially for women who reside there. It provides food, money, and work opportunities. Dairy enables them diversify their food offerings and provide women and children of marginal farmers with a healthy diet. Ethiopia’s demand for dairy products is rising as a result of the country’s fast urbanization. Nonetheless, farmers who live distant from an urban milk marketing hub participate in the market less. Therefore, in order to inform policy makers, an examination of how rural smallholder dairy producers affected in milk marketing and the effects of milk marketing is required. This publication will be important knowledge and information improving the position of smallholders’ dairy farmers to actively engage in the milk market that will have vital role for households’ food security.

Keywords:

• Dairy product
• market participation
• impact
• household food security
• PSM
• Ethiopia

SUBJECTS:

• Environment & agriculture
• agriculture & environmental sciences
• agriculture and food
• environmental studies & management
• environment & gender

REVIEWING EDITOR:

• Pedro González-Redondo [email protected] Senior Editor
• University of Seville, SPAIN

1. Introduction

Livestock is a resource that allows smallholders to thrive resiliently in numerous developing countries, (Herrero et al., 2015). They make up a significant portion of global agricultural income sources (Nabarro & Wannous, 2014), and supplying 33% of worldwide proteins. From the sector dairy ­provides sustenance to as many as 1 billion smallholder individuals. The dairy sector can make a larger contribution to nourishing poor diets particularly in developing countries (Muehlhoff et al., 2013) and the demand of dairy product has been increasing marketing of dairy products in urban and rural towns.

Ethiopia has one of the most prominent livestock populations in Africa and anywhere between 35 and 49% of its agricultural GDP was generated by livestock (Endalew & Ayalew, 2016). Once factors such as the processing, marketing, organic fertilizer production and animal traction are taken into consideration, the contribution of livestock to the GDP rises to 25.3% (Endalew & Ayalew, 2016). This sector also supports the livelihood of 80% of all rural poor, granting either all or some of their sustenance to over 11.3 million rural households (CSA, 2018). Additionally, dairy production is a crucial element of smallholder income and poverty reduction, providing food security to the country (Derbe et al., 2022; Mihret et al., 2017). Dairy production also offers job opportunities which could trigger economic growth in Ethiopia (Andeweg et al., 2020).

In order to enhance the dairy sector output, the Ethiopian government has made efforts to address feed shortages, improve animal health and genetics, and increase the productivity of livestock. Nonetheless, such interventions may not bear fruit unless viable market services are made available for the product. Participation in the market is an effective means of alleviating poverty and guaranteeing food security (Berhanu et al., 2014; Paul et al.,2018; Reardon & Timmer, 2014), making market integration a paramount strategy to enhance rural smallholder livelihoods in developing countries (Olwande et al., 2015). By participating in the market, farmers have access to the profits of their products, thereby giving them the ability to invest in technologies (Asfaw et al., 2012; Barrett, 2010; Chege et al., 2015). They can produce more of what they excel at and use the surplus to obtain goods they are not familiar with (Barrett, 2008). Dairy product consumption can alleviate malnutrition deficiency and the marketing of such products allows farmers to purchase additional food from the market, broadening the dietary variety of households. All in all, a sustained dairy product market has substantial benefits for both producers in particular and the national GDP in general (Bekuma et al., 2018). For the purpose of our study dairy product market participation is refers to the involvement of smallholder dairy farmers in the sale of milk for income generation to meet food security of households.

Unfortunately, rural smallholder dairy producers in Ethiopia have limited access to the market, due to various social, economic and demographic barriers like remoteness to market, lack of market information, cultural taboos of dairy product selling, low products due to poor adoption of dairy technology and constraints of institutional intervention (Bekuma et al., 2018, Kena et al., 2022; Kidanu, 2010; Kuma et al. 2013; Somano, 2008; Tadesse et al., 2016).

Analysis of rural milk market participation is vital for policy information that needs to design programs and projects to enhance rural productivity and benefits of smallholder dairy farmer. In contrast, the majority of milk market participants are from nearby urban and town farmers. Dairy products predominate in rural and remote areas. However prior studies like Dairy Producers’ Market Participation Decision and Volume of Milk Supply in Mekelle City, Ethiopia (Girmay et al., 2020), Factors affecting milk market participation and volume of supply in Ethiopia (Berhanu et al., 2014), Determinants of Market Participation Decision and Level of Participation of Dairy Farmers in Tigray, Ethiopia (Kebede et al., 2015), and Determinants of dairy farmers’ market participation in the major dairy producing towns of Jimma Zone of Southwest Ethiopia (Tadesse et al., 2016) have examined the factors that determine the dairy producers to partake in the market and the extent of their market participation mainly focusing on urban and town dairy holders.

A single study by Lenjiso et al. (2016) indicates that milk market participation contributed to improvement of nutritional status of children in rural Ethiopia. In anticipation of now scant attention has been given to rural milk producers’ market participation and the impacts of dairy product marketing on their household food security in Ethiopia. The aim of this study is to fill the information and knowledge gap on milk producers’ market participation and its effect on their household food security in rural southwest Ethiopia.

The hypotheses of this research are the following:

H1: A major determinant of whether or not dairy product producers in the study area join the market is the surrounding socio-economic (sex, age, education level, family size, land size, Total HH Income, Off-farm participation, Cross breed cow own), institutional (Extension contact, access to dairy input, credit services, cooperative access, training access), and geographical features (market distance).

H2: Participating in the dairy product market has a positive and considerable effect on households’ food security.

2. Materials and methods

2.1. Description of the study area

The research was conducted in the Jimma Zone of the Oromia Regional State of Ethiopia, which is situated roughly 346 kilometers away from Addis Ababa, the country’s capital. Situated 1,704 meters above sea level, this region is situated between 7°4 North latitude and 36°5 East longitudes. One of Ethiopia’s main growing regions for coffee is this region. As per the Jimma zone Agricultural and Rural Development Office, the production of coffee in Jimma accounts for over 20 percent of Ethiopia’s total share of coffee exports. In addition, 570,241 bee hives and more than 5.5 million livestock reside in the zone (CSA, 2014/2015). Three main climate zones can be found in the Jimma Zone: subtropical (78 percent), temperate (12 percent), and tropical or thermal (10 percent). Two towns and two districts from the zone were selected for the investigation.

2.2. Sampling procedure and sample size

Multi-stage sampling was applied in order to get the required respondents. Firstly, two districts Gomma and Gera were purposively selected based on their potential for dairy production. Then, three kebeles from each district and two kebeles from agaro towns were chosen randomly from each district. Households were then identified in the selected kebeles, stratified into two groups – those who did and did not participate in dairy marketing using the stratified sampling technique. Ultimately, sample respondents were selected from each stratum via simple random sampling, taking into account the probability proportional to sample size of the identified households in the nine kebeles. The Cochran’s sample size formula was used to determine sample size for this study (Cochran, 1963) $$\mathrm{\text{no}}=\frac{z^2\mathit{\text{pq}}}{{\left(\mathrm{e}\right)}^{\left(2\right)}}$$

Where· no is sample size; z is the value of the t-distribution corresponding to the chosen alpha level for 0.05 is 1.96; p the estimate of population proportion is the desired level of precision; p is the estimated proportion of an attribute that is present in the population, and q is 1-p $$\text{no}=\hspace{0.17em}\frac{(1.96)2\ast 0.5(1-0.5)}{{(0.06)}^{\left(2\right)}}=\underset{¯}{266}$$

Accordingly 266 respondents were stratified into dairy market participant and non-participant.

2.3. Data type, sources and methods of collection

The research utilized a cross-sectional survey and obtained both qualitative and quantitative data from primary and secondary sources. The primary data was obtained through an interview utilizing a schedule to collect information from households engaged in dairy production. Structured questionnaires were formulated and distributed to 266 respondents in order to obtain quantitative data.

2.4. Methods of data analysis

To analyze the collected data, a combination of different methods were used; descriptive statistics and econometric model for quantitative data.

Descriptive statistics. Descriptive statistics was used to provide a summarized view of the sample’s demographic and socioeconomic behavior. In particular, mean, standard deviation, frequency, charts, t-test, and chi-square test were utilized to draw pertinent conclusions.

2.4.1. Econometric model

A binary logit model and Propensity Score Matching (PSM) econometric model were utilized to assess the factors influencing the dairy market participation, as well as the effects of such participation on household’s food security. Following Gujarati (1995), the logit model utilized was in the form: the dependent variable milk market participation. For such binary variables, a binary logistic regression model and a probit model were appropriate. The logistic distribution, on the other hand, is useful for analyzing dichotomous outcomes because of its adaptability and simplicity, which enable it to yield meaningful interpretations, as described by Hosmer et al. (1989), so for this investigation, the binary logistic was chosen. Following Gujarati (1995), the functional form of the binary logit model is specified as follows: (2) $$p_i=E\left(Y=\frac{1}{X_i}\right)=\frac{1}{1+e^{-({\beta }_{0+}{\beta }_j{x}_{\mathit{\text{ji}}})}}$$(2)

This probability of the milk market participant is expressed as: (3) $$p_i=\frac{1}{1+e^{-\left(Z_i\right)}}$$(3)

Conversely, the probability that household will not participate is (4) $$1-\mathit{\text{pi}}=\frac{1}{1+e^{Z_i}}$$(4)

This can be written as (5) $$\frac{p_i}{1-p_i}=\frac{1+e^{z_i}}{1+e^{{-z}_i}}=e^{z_i}\mathrm{ }$$(5)

$\frac{\mathit{\text{pi}}}{1-\mathit{\text{Pi}}}$. This implies that the ratio of the likelihood of household becoming milk market participant to the likelihood of they being a non-participant is equal to the odds ratio in favor of will be a participant. Ultimately, we can obtain the log of odds ratio by transforming Eq. (5)(5) $$\frac{p_i}{1-p_i}=\frac{1+e^{z_i}}{1+e^{{-z}_i}}=e^{z_i}\mathrm{ }$$(5) into the natural log. (6) $$\mathit{\text{Li}}=\mathit{\text{Ln}}\mathrm{ }\left(\frac{\mathit{\text{pi}}}{1-p_i}\right)=z_i={\beta }_0+{\beta }_1{x}_1+{\beta }_2{x}_2+\dots {\beta }_n{x}_n,$$(6)

Incorporating Ui, the disruption variable, into the logit model produces the following equation: (7) $$Z_i={\beta }_0+{\beta }_j{X}_{\mathit{\text{ji}}}+U_{\mathrm{i}}$$(7)

Where n is sets of all available data points. The probability of dairy product market participation, denoted as pi, can be balanced by the probability of non-participation (1-pi). β0 is an intercept, while β1, β2, and so on, up to βn are the slopes of the equation utilized in the model. Li is the log of the odds ratio, a function which is dependent on both Xi and the corresponding parameters. Xi, on the other hand, stands for the vector of factors specific to farmers which can affect the market participation. This method is used to identify the key elements influencing dairy product market participation.

2.4.1.1. Propensity score matching (PSM)

PSM is used when a group of subjects is receiving treatment and is used to compare the results of the treatment group to those of the control group. This study used the PSM model to estimate the impact of dairy product market participation on household food security. Participants scored 1 and non-participant scored 0, because market participant and non-participants are binary in nature. This model helps smooth out early differences between market participant and non-participants by matching each unit in two groups based on similar observable characteristics (Rosenbaum & Rubin, 1983). In PSM, the first step is to predict the trend value for each observation that indicates treatment potential. The next step after predicting scores is to impose common areas of support. The identification of correctly matching estimators is then the next step following the implementation of common support regions. Various match estimators are discarded, including nearest neighbor match, caliper match, radius match, and kernel match (Caliendo & Kopeinig, 2008). Finally, we check the matching quality to whether the matching procedure can even out the distribution of the various variables. The impact of dairy market participation on household food security was measured by FCS and HDDS described as follows: (8) $$\tau i=Y_i\left(D_i=1\right)=Y_{\mathrm{i}}\left(D_i=0\right)$$(8)

Where, τi effect because of market participation, Yi is the impact of dairy market participation on household food security, which is the outcome variable and Di, is dairy market participant or non-participant household i. estimating the average treatment effects of the population important than the person was analyzing individual treatment effect. According to Heckman and Vytlacil (2007), average treatment effect on treated (ATT) described as: (9) $$\tau \_\mathit{\text{ATT}}=E\left(\tau /D=1\right)=E\left[Y\left(1\right)/D=1\left]-E\right[Y\left(0\right)/D=0\right]$$(9)

E[Y(1)D = 1] a market participant outcome variable which is observed and E[Y(0)/D = 1] a non-market participant outcome variables which is not observed. The average treatment effect on treated can be specified: (10) $$\begin{array}{l}E[Y\left(1\right)/D=1]=E[Y\left(0\right)/D=1]-E[Y\left(0\right)/D=0]\\ =\tau \_\mathit{\text{ATT}}+E[Y\left(0\right)/D=1]-\hspace{0.17em}E[Y\left(0\right)/D=0]\end{array}$$(10)

If E[Y(0)/D = 1]-E[Y(0)/D = 0] zero the average treatment effect on treated can be determined Finally the propensity score matching algorithm to estimate ATT can be described as: (11) $$\begin{array}{l}E[Y\left(1\right)/D=1]=E[Y\left(0\right)/D=1]-E[Y\left(0\right)/D=0]\\ =\tau \_\mathit{\text{ATT}}+E[Y\left(0\right)/D=1]-E[Y\left(0\right)/D=0]\end{array}$$(11)

The equation indicates the PSM estimators are the difference between the mean of outcomes over common support region.

2.4.2. Description of variables and working hypothesis

2.4.2.1. Dependent variables

Milk market participation: this is the likelihood that a household will participate in the milk market. For households that participate in milk markets, the dummy variable takes on the value 1, whereas for households that do not, it takes on the value 0.

2.4.2.2. Outcome variable

Household food security: To measure food security status of the households, Household Food Consumption Score (HFCS) and Household Dietary Diversity Scores were used as proxies for food security of the households. HFCS is a weighted score based on dietary diversity, food frequency and the nutritional importance of food groups consumed. The HFCS of a household is calculated by multiplying the frequency of foods consumed within 7 days before the survey with the weighting of each food group which determined by WFP according to the nutrition density of the food group (Bickel et al., 2000). Households can then be further classified as having "poor," "borderline," or "acceptable" (0-21: Poor; 21.5-35: Borderline; >35: Acceptable) food consumption by applying the WFP's recommended cut-offs to the food consumption score. HDDS is similar to HFCS with slight differences in the components of the various food clusters. The HDDS takes into account food items consumed within the last 24 hour. We considered HDDS less than or equal to three as low dietary diversity group and between four to six as medium category while HDDS greater than or equal to seven as high diversity score category (low HDDS of ≤3, medium HDDS of 4–6 and high HDDS of 7–12) in line with the study of Huluka and Wondimagegnhu (2019). According to the existing recommendation by different scholars households in high HDDS category are considered as food secured in line with the study of Jebessa et al. (2019).

Explanatory Variables: There is a complex web of factors that affect dairy product market participation, some of which include the socioeconomic characteristics of households, institutional characteristics, and geographic factors. Based on the studies of (Bardhan et al., 2012; Berhanu et al., 2014; Chelkeba et al., 2016; Girmay et al., 2020; Kidanu, 2010; Ordofa et al., 2021, Kena et al., 2022; Tadesse et al., 2016) different explanatory variables were identified. Consistent with the literature reviewed above, the study’s major hypothesis was that the socio-economic, institutional, and geographical characteristics of the study area would determine whether or not dairy product producers enter the market.

3. Result and discussion

In this sub-section, the data pertaining to the demographic and socio-economic characteristics of households has been presented. Results indicate that 54.9% of households provided dairy products to the market, whereas the remaining 45.1% had no involvement in dairy-related sales however they use it for house consumption.

3.1. Household characteristics

Table 1 revealed that 82% of the sample households were headed by males. Out of total households 33.4% were involved in Off-farm activities and only 10.5% household utilized credit services. About 52.6% households were access to dairy inputs and 43.2% owning crossbreed cows, as well as about 44.7% being members of agricultural cooperatives.

Table 1. Socioeconomic characteristics of respondents (dummy variables).

Variables	Categories	Participant		Non-participant		% total	Chi-square
		Count	%	N	%
Sex of household head	Male	122	45.9	96	36.1	82	0.565
	Female	24	9	24	9	18
Off-farm participation	No	81	30.5	96	36.1	66.6	17.787***
	Yes	65	24.4	24	9	33.4
Training participation	No	59	22.2	66	24.8	47	5.628**
	Yes	87	32.7	54	20.3	53
Credit use	No	129	48.5	109	41	89.5	0.429
	Yes	17	6.4	11	4.1	10.5
Access to dairy inputs	No	46	17.3	87	32.7	50	44.273***
	Yes	100	37.6	33	12.4	50
Cooperative membership	No	67	25.2	80	30.1	55.3	11.5***
	Yes	79	29.7	40	15.0	44.7
Cross breed cow own	No	58	21.8	93	35.0	56.8	38.294***
	Yes	88	33.1	27	10.2	43.2

***, **, *shows 1%, 5% and10% level of significance respectively.

Chi-square tests revealed that, household characteristics such as training participation, access to dairy inputs, cooperative membership, and crossbreed cow ownership indicated statistically significant differences between market participants and non-participants. While the chi-square result indicated that there is no variation between a market participant and non-participants in terms of sex of household head and credit use. This could be female headed household and users of credit services are very small in the study area.

Table 2 provides the summary statistics of the continuous variables, with mean age, education level, and dairy experience almost identical for both market participants and non-participants. The market participants, however, showed larger family size, land size, livestock size, and income. Distance to the market for the participants was significantly lower than for the non-participants. The results from Table 2 t-test revealed that there were significant differences between participants and non-participants with regard to age, family size, market distance, land size, income, livestock size, and extension contact, while no differences were found with education level and dairy experiences.

Table 2. Socioeconomic characteristics of respondents (continuous variables).

Variables	Participant		Non-participant		t-test value
	Mean	Std. Dev	Mean	Std. Dev
Age	41.43	11.81	42.8	10.03	1.75*
Education level	3.95	3.15	4.12	3.15	1.492
Family size	4.75	1.70	4.26	1.85	−2.266**
Market distance	4.37	2.02	5.75	2.41	5.081***
Land size	2.32	1.60	1.86	1.40	−2.442**
Total HH Income	48200	54240	30903	38147	−2.942***
Livestock size	5.07	3.06	4.02	2.85	−8.334***
Dairy experience	10.23	6.05	10	6.90	−0.304
Extension contact	1.57	1.17	1.18	1.04	−2.869***

HH: Household.

***, **, *shows 1%, 5% and 10% level of significance respectively.

3.2. Dairy product marketing and food security status of households

In the research area, the majority of milk and milk products were used for domestic consumption and the surplus was supplied to the market. On average, households produced 3.5 liters of milk a day, with 1.45 liters being sold and the rest, 2.05 liters, retained for home use (58.6 percent) (Table 3). However, most respondents did not sell their produce, which led to the mean of milk supplied to the market being low. The producers mostly sold their dairy products to their neighbors and to urban areas. Milk production in the study area was comparatively low, as per discussions from Focus Group Discussions (FGDs), due to the scarcity of grazing land causing households to reduce their livestock, including dairy cows. The households that produced smaller amounts of milk for their own consumption did not enter the market, while those with more production sold a major portion of it.

Table 3. Milk marketing.

	N	Minimum	Maximum	Mean	Std. Dev
Milk produced (liters per day per household)	266	0.00	19.00	3.50	2.13
Milk marketed (liters per day)	266	0.00	15.00	1.35	1.75

Figure 1 illustrate the food security status of households according to the two measures employed (food consumption score and dietary diversity), respectively. Households were classified into three levels of food security which were poor, medium and acceptable for food consumption score and low, medium and high for dietary diversity. As shown on the graph more than half of the respondents were in acceptable range of FCS and high HDDS group this reflects those households were food secured. Moreover, the majority of households who participate in milk marketing were found food secured.

Figure 1. Milk market participation and food security.

3.3. Factors affecting dairy market participation of smallholder farmers

As the demand for dairy products continues to increase, it is crucial to understand the factors that influence smallholder farmers’ participation in the dairy market. A recent study has examined the impact of various factors on dairy market participation among smallholder farmers in Ethiopia. This article will discuss the findings of the study and shed light on the factors affecting dairy market participation. Table 4 reveals the outcomes of the binary logit analysis. Out of 16 speculated variables, nine are demonstrably associated with the dairy market participation of households. Of these noteworthy variables, six have been positively impacted, while the remaining three have had a negative effect on dairy market participation.

Table 4. Factors affecting dairy market participation.

Market Participation	Coef	Std. Err	z	P > z	Odd Ratios
Sex of respondent	0.1349	0.4511	0.30	0.765	1.14
Age of respondent	−0.0355	0.0162	−2.19	0.029**	0.96
Education level	−0.1411	0.0596	−2.37	0.018**	0.87
Family size	0.2131	0.1034	2.06	0.039**	1.24
Market distance	−0.1809	0.0805	−2.25	0.025**	0.83
Land size	0.0089	0.1176	0.08	0.939	1.01
Total household income	4.43e-06	4.20e-06	1.06	0.291	1.00
Livestock size own	0.2945	0.0629	4.68	0.000***	1.34
Off-farm engagement	1.3526	0.3955	3.42	0.001***	3.87
Extension contact	0.0655	0.1625	0.40	0.687	1.07
Training participation	−0.5226	0.3775	−1.38	0.166	0.59
Credit use	0.0525	0.5521	0.10	0.924	1.05
Dairy Experience	0.0254	0.0273	0.93	0.353	1.03
Access to dairy input	1.0302	0.5022	2.05	0.040**	2.80
Cross breed cow own	0.9529	0.5268	1.81	0.070*	2.59
Cooperative membership	0.7467	0.3724	2.01	0.045**	2.11
_cons	−1.3888	1.0469d	−1.33	0.185	0.23
Number of obs = 266 LR chi^2(16) = 138.78 Prob > chi^2 = 0.0000				Log likelihood = −113.708 Pseudo R^2 = 0.379

***, **, *Significant at 1%, 5% and 10% level respectively.

3.3.1. Age of household head

The study found that the age of the household head has a negative effect on the market participation decision of the sampled dairy households. Older dairy household heads are less likely to participate in the dairy market. The marginal effect also confirms that an increase in household age by one year decreases the probability of participating in the dairy market by 0.5%. These results are in agreement with the findings of Berhanu et al. (2014) and Kebede et al. (2015), who had also examined the adverse association between age and market participation of dairy farmers.

3.3.2. Level of education

The study also found that the level of education of the household head has a negative effect on dairy market participation. More educated household heads are less likely to participate in the dairy market. This finding is consistent with the notion that education improves the nutritional knowledge of the household to consume dairy products rather than selling them in the market. The marginal effect confirms that one formal year of education leads the dairy household to decrease market participation decision by 1.8%. However, this result is contrary to the findings of Ordofa et al. (2021), who had found a significant positive connection between the farmers’ education level and their involvement in the dairy market.

3.3.3. Family size

Contrary to expectations, the study found that family size has a positive effect on the probability of dairy household market participation decision. As dairying is labor-intensive, a larger family size provides higher labor to undertake dairy production and management activities, which increases daily milk production, leading to encourage dairy household market participation. The marginal effect of the variable also emphasizes that for every one adult equivalent increase in a family, the probability of dairy market participation of the household increases by 3.3%. This result aligns with the findings of Kebede et al. (2015) and Tadesse et al. (2016), while conflicting with the study of Kidanu (2010), Willy and Gemechu (2016) and Ngeno (2018).

3.3.4. Market distance

The study found that the distance to the market negatively and significantly influences the participation in the milk market. As travel time and transportation costs increase, households are more likely to consume their milk. The marginal effect also confirms that a one-kilometer increase market distance from the dairy farm owner reduces the probability of participation in the milk market by 2.4%. Similarly, Bardhan et al. (2012) and Tadesse et al. (2016) both investigated that distance to market had a negative effect on the likelihood of producers’ market participation and intensity of market participation.

3.3.5. Livestock size

One of the significant factors influencing dairy market participation is the livestock size. The study found that as the size of livestock increases by one TLU (Tropical Livestock Unit), the milk market participation of household’s increases by 4%. This result is consistent with Berhanu et al. (2014), studies that suggest households with a large number of livestock are likely to collect more dairy products, which increases their share of milk sales per day per household and Willy and Gemechu (2016), suggested that number of local dairy cows positively influence dairy market participation.

3.3.6. Off-farm engagement

The study also found that off-farm engagement is positively related to milk market participation. Off-farm engagement provides households with additional sources of income, which advances their ability and interest in market participation for their dairy products. Moreover, off-farm engagement provides farmers with access to information and knowledge that helps them increase their agricultural production and marketing of products. The study found that an increase in off-farm engagement of households increases their market participation by 18.7%.

3.3.7. Access to dairy inputs

Access to dairy inputs such as Artificial Insemination (AI), improved forage, and veterinary services has a positive effect on milk market participation. These inputs help farmers to produce more volume of milk, which increases their chances of selling their products. The study found that an increase in access to dairy inputs for dairy farmers results in an increase the market participation of the households by 4.2%. Therefore, policymakers and stakeholders should focus on improving smallholder farmers’ access to these inputs to increase their participation in the dairy market.

3.3.8. Cross breed cow

Owning cross-breed cows is positively related to dairy market participation. Farmers purchase cross-breed cows to produce a large volume of milk and supply it to the market, while local cows provide a small amount of milk that may not be enough to sell. The odd ratio result indicates that when the household accesses owning cross-breed cows, the probability of their dairy market participation increases by 12.7%. These results are in agreement with the findings of Bayan and Deka (2018), reveal that crossbreed cattle adopters has stronger and highly significant effect on increased marketed surplus due to adopter’s milk production activity strongly guided by market participation and adoption of crossbreed cattle is significantly rewarding in terms of higher farm income and gain in nutrition from increased self-produced milk consumption (Bayan & Dutta, 2017).

3.3.9. Cooperative membership

Membership in farmers’ cooperatives also has a positive impact on dairy market participation. Agricultural cooperatives provide farmers with information and expose them to technologies that can increase their agricultural productivity and welfare. The study found that an increase in cooperative membership increases dairy market participation by 11.1%. Therefore, increased access to cooperative membership should be a major part of efforts aimed at promoting dairy market participation. The result is in line with the study of Kena et al. (2022).

3.4. Estimating propensity scores and common support condition

In this study, propensity scores were used to compare the food security of dairy market participant and non-participant households. Propensity scores are a statistical tool used to balance the characteristics of treatment and comparison groups in observational studies. The common support assumption is an important aspect of this method, which ensures that the propensity scores of both groups overlap in a certain range. The analysis found that there were households from both groups in the common support region, which is the range of propensity scores where the values of both groups overlap. The range of the common support region was between 0.069346 and 0.9705057, based on the rules of minima-maxima (Caliendo & Kopeinig, 2008). Any observations outside of this range were discarded from further analysis.

The study also provided the magnitude and distribution of the estimated propensity scores. The propensity scores ranged between 0 and 1, with a mean score of 0.5465796 for the total sample. The mean propensity score for market participants was higher at 0.747376 compared to non-participants at 0.3022772. The distribution of the estimated propensity scores is presented in Figure 2.

Figure 2. Density of propensity score distribution.

3.4.1. Choice of matching algorithm and matching

In this study, the choice of a matching algorithm was determined based on several selection criteria, including joint significance and pseudo-R2, standardized bias balancing test, matched sample size, and t-test (Caliendo & Kopeinig, 2008; Rosenbaum & Rubin, 1983). The authors noted that a low pseudo-R2 value and a large matched sample size are desirable qualities in a matching algorithm, according to Dehejia and Wahba (2002). The authors ultimately choose kernel matching as their algorithm, as it resulted in a low pseudo-R2 value of 0.062 and high distribution in covariates after matching (see Table 5). Additionally, the likelihood ratio tests were insignificant, indicating that the algorithm effectively balanced the covariates between the treatment and comparison groups.

Table 5. Chi-square test.

Matching method	Sample	Ps R^2	LR chi^2	p > chi^2	Matched sample size
kernel	Unmatched	0.388	142.02	0.000	226
	Matched	0.062	21.87	0.190
Caliper	Unmatched	0.388	142.02	0.000	207
	Matched	0.062	13.02	0.735
Neighbor	Unmatched	0.388	142.02	0.000	226
	Matched	0.094	23.40	0.010

3.5. Impact of dairy market participation on household food security

The study employed the Propensity Score Matching (PSM) method to estimate the impact of dairy market participation on household food security. The results of the matching estimators (kernel algorithm, nearest neighbor, and caliper matching algorithm) indicated that participation in the dairy market has a positive and statistically significant effect on household food consumption score (HFCS) and dietary diversity (HDD) (Table 6).

Table 6. Impact of dairy product market participation on household food security.

Outcome variables	Matching method	Treated	Controls	ATT Difference	S.E	T-stat	Percent change
FCS	kernel	36.54	31.47	5.06	3.63	0.084	13.84
	Neighbor	36.54	31.74	4.80	3.97	0.071	13.13
	Caliper	35.59	31.76	5.83	4.62	0.083	16.38
HDD	kernel	7.32	6.47	0.85	0.44	0.09	11.60
	Neighbor	7.65	6.47	1.18	0.64	0.022	15.42
	Caliper	7.65	6.47	1.18	0.64	0.022	15.42

ATT: Average Treatment effect on the Treated.

FCS: Food Consumption Score.

HDD: Household Dietary Diversity.

The mean HFCS ATT result was closest to 36.5 using kernel algorithm. The difference in HFCS between participants and non-participants was found to be 5.06 (13.84%) higher for market participants than non-participants using the kernel matching algorithms. This result was statistically significant, indicating that dairy market participation leads to a significantly higher HFCS for households.

Similarly, the mean difference in HDD between participants and non-participants was also expressed by the average treatment effect on the treated (ATT). The mean ATT difference in HDD was significantly higher for market participants, with an average difference of 0.85 using kernel algorithm. This translates to a maximum of 15.42% increment in HDD for market participants compared to their counterparts. The result was statistically significant at a 10% significance level, indicating that market-oriented dairy product households are better able to purchase different types of foods and have a more diverse diet. This result aligns with the findings of Ntakyo and van den Berg (2019) found the more market-oriented households are better able to purchase different types of foods and thus have a slightly more diverse diet.

3.5. Sensitivity analysis

A sensitivity analysis is a process utilized to test the influence of an unmeasured variable on the selection process (Caliendo et al., 2008). The results of the Table 7 below indicate that the link between the dairy product market’s participation and household food security is starting to change. For instance, the p-values of outcome variable estimates are significant when compared to the gamma value. This means that the impact estimates are not influenced by unobserved selection bias.

Table 7. Sensitivity analysis.

Gamma	sig+	sig-	t-hat+	t-hat-	CI+	CI-
1	0	0	43.5	43.5	41	45.75
1.25	0	0	41.5	45.25	39.25	47.75
1.5	0	0	40	46.75	37.75	49.25
1.75	0	0	38.75	48.25	36.5	50.5
2	0	0	37.75	49.25	35.25	51.75
2.25	0	0	36.75	50.25	34.25	52.9
2.5	0	0	36	51	33.5	53.75
2.75	0	0	35.25	51.75	32.75	54.5
3	1.1e-16	0	34.5	52.5	32	55.25
3.25	2.2e-15	0	34	53.25	31.5	56
3.5	2.1e-14	0	33.5	53.75	31	56.5
3.75	1.4e-13	0	33	54.25	30.5	57.25
4	7.8e-13	0	32.5	54.75	30	57.75
4.25	3.5e-12	0	32	55.25	29.5	58.25
4.5	1.3e-11	0	31.75	55.75	28.85	58.75
4.75	4.4e-11	0	31.5	56	28.5	59.25
5	1.3e-10	0	31	56.5	28.25	59.75

*gamma - log odds of differential assignment due to unobserved factors.

sig+ - upper bound significance level.

sig- - lower bound significance level.

t-hat+ - upper bound Hodges-Lehmann point estimate.

t-hat- - lower bound Hodges-Lehmann point estimate.

CI+ - upper bound confidence interval (a= .95).

CI- - lower bound confidence interval (a= .95).

4. Conclusion and recommendation

The dairy industry in Ethiopia is predominately subsistence-oriented, meaning there is both low milk production and consumption. To evaluate determinants of dairy product market participation and its effects on household dietary diversity and food consumption score, 266 randomly selected households from the Jimma zone in Southwestern Ethiopia were studied. It was found that dairy product market participation plays a significant role in achieving food and nutritional security. Factors such as age, education level, and land holding size, market distance, and membership of a cooperative were discovered to be important in the market participation. This study also demonstrated that dairy product market participants had higher household dietary diversity and food consumption scores than those who did not partake in the market.

Limited access to dairy inputs, dairy technology, and infrastructure must be resolved for the government and its allies to continue supporting dairy production and dairy product market participation. Innovations to increase dairy production and productivity should be supported and encouraged in the study area. It is also imperative that cooperative membership is promoted in order to increase dairy technology in the area, and to make the community aware of these advancements in a timely and regular fashion.

A major limitation of this study was the investigation of the outcome variables, household dietary diversity and food consumption score, which were measured through 24-hour recall and seven day periods respectively. For future research, longitudinal data should be used and more outcome indicators such as poverty reduction and asset accumulation should be examined to further understand the implications of HDDS and HFCS in the study area.

Authors’ contributions

GM design the research, preparing drafts analyzed and interpreted the data. GB, EY and TC made significant contributions in review, comment and editing from the draft paper to the report. All authors read and approved the final manuscript.

Ethical approval and consent to participate

Ethical clearance was obtained from Jimma University and from the districts of study area. About the study was fully explained to the study participants to obtain consent and any information kept confidential.

Acknowledgments

This study authors would like to thank respondents of the study for their willingness and patience to be part of this research. We would also extend our grateful thanks to Jimma University for sponsoring the data collection and other expenditure for this research.

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The data will be available upon request from the first author.

Additional information

Funding

Jimma University sponsored for research work.

Notes on contributors

Geremew Motuma Jebessa

Geremew Motuma Jebessa (Mr) is an ­academic staff of Jimma University Department of Rural Development and Agricultural Extension. He is a PhD candidate at Jimma University, Ethiopia. He holds a bachelor’s degree in Rural Development and Agricultural Extension and MSc degree in Rural Development from Haramaya University and Jimma University respectively. His research focus includes food security, technology adoption, marketing and climate change.

Guduro Beriso

Guduro Bariso (Mr) is a lecture at Jimma University Department of Rural Development and Agricultural Extension. His research mainly focuses on food security, livelihood and gender.

Eskindir Yacoob

Eskindir Yacoob (Mr) is a lecture at Jimma University Department of Rural Development and Agricultural Extension. His research mainly focuses on sustainable environment and technology adoption.

Tamiru Chalchisa

Tamiru Chalchisa (Mr) is a lecture at Jimma University Department of Rural Development and Agricultural Extension. His research focus includes technology adoption, marketing and value chain.