1 Introduction

Commercialization of the smallholder agriculture is considered as means of the poverty reduction in less developed countries like Ethiopia [5]. It represents an attempted to alter the subsistence production to the highly market- oriented agricultural production [6]. However, smallholder commercialization to be successful and to bring structural transformation, it requires the capitals [23], access to market information [5], technical knowledge [21], and access to the modern agricultural technologies [4]. These constraints results the quality of the input demanded, yields inquired, sales (quality, quantity and price of sales) and reduces the incomes of the smallholder households, which, in turn, leads to reinforce and perpetuate subsistence- orientation [33].

Contract farming (CF) is one of the contested issues in the literature on the commercialization of smallholder agriculture. Some scholars argue in favor of the CF as it increases farmers’ income [13], improve access to technology and inputs [31], provides secured outlet and price for products [15], improve productivity [14], increase the quality and quantity demand for the market [24] and it also offers more profitable business activity to the farmers [3]. Other scholars refute the CF for several reasons. CF can increase the smallholder farmers’ indebtedness, loss of control over their lands; causes the gender inequality, increase labour dependence, lack of the fairness and low participation of smallholder farmers in contract enforcements (22 and 32). Moreover, one side favourism in contract agreement, unclear, incomplete, and misguiding contracts clues leads to the diminishing the possibility to improve the smallholder farmers’ income and their livelihood through such institutional arrangement [2, 30]. These results there were mixed results regarding role of the CF in the context of smallholder farmers. The evidence base remains inconclusive and debatable. Hence, study was initiated to contribute to this debate by producing empirical evidence at the household level in the context of smallholder farmers in Ethiopia.

The research area is known for the irrigated vegetable production mainly tomato, onion, cabbage, and pepper. However, there is widely held view that the smallholder producers were not able to get fair share of the benefit from the high-value vegetable value chain [13]. Perishability of the product, lack of the local processing facilities, long market chain dominated by the brokers, market information asymmetry, and season based price fluctuation are among the factors that negatively influencing the benefit accrued from the vegetable producers [10]. Hence, CF was considered as the appropriate institutional options to help the vegetable producers through promoting market linkage among smallholder vegetable producers and facilitate institutional arrangement [6].

Recent empirical studies in the Ethiopia found out that, the appropriate CF arrangement could enhance smallholder producers’ access to inputs, credit, technical advice, and remunerative market leading to increased income [6, 13]. It has also been found that, CF might be fail to deliver the expected benefits to the smallholder farmers due to both internal factors mainly related to contract and enforcement mechanism [13, 18, 31] and external factors including climate change, soil fertility degradation and pests and diseases [6]. These may exclude the poor farmers from parts of the CF and increases social inequality in the community [33]. There is a particular need for the roboust studies for the clear understanding of the how and to what extent the CF benefits smallholder farmers and impacts on their income. Therefore, the study aimed to identify the determinants of the smallholder participation in the vegetables’ CF and their impacts on households’ income based on the household’s data in Central rift-valley, Ethiopia.

2 Methods and materials

2.1 Data and description of study area

The study was carried out in the East Shewa Zone which is located in the central rift valley of Ethiopia (Fig. 1). Of the total ten districts in the zone, Dugda and Bora districts were selected purposively as the predominantly vegetable growers [9]. Lake Danbal and Meki River are located/ cross the selected districts, in addition to the availability of the sub-surface water at the depth of ranging from 15 to 43 m. Small scale irrigation has been practiced around the coast of the Lake Danbal and Meki River. Farmers produce vegetables like tomato, onion, peppers and cabbage 2- 3 times per year using irrigation scheme.

Fig. 1
figure 1

Source: Own computation

Map of the study area.

Full size image

The study used both primary and secondary data sources. Quantitative and qualitative data were gathered through sampled household survey using the semi-structured interview schedule, focused group discussion, and key informant interview. Multistage sampling procedure was used to select the districts, kebeles, and finally vegetable producer sample households. Bora and Dugda districts were selected as research area for the prominent irrigated vegetables production. Secondly, eight kebeles known for their vegetable production and participation in vegetable CF were identified in consultation with local experts. Third, vegetable producing households were stratified into two strata based on their status of participation in the CF over the last two years to draw representative of the sample households from both vegetable producer CF participants and non-participant. The total sample size was determined using a Cochran formula, which provides the maximum size to ensure the desired precision in the following [8]

$$n=\frac{-\mathrm{ Z }2 *{\text{p}}*\mathrm{q }}{e2} \frac{(1.96) 2 (0.5)(0.5)}{\left(0.05\right)2}\approx 384$$

where n is desired sample size; Z is confidence level; e is the desired level of precision; p is estimated proportion (values of 0.5) and q represents the total population with the values of 0.5 as suggested by [16] to get the minimum sample size of the households at 95% confidences level and \(+\) 5% precision. Additionally, 40 respondents (10.42%) were kept as a reserve considering possible errors, omissions, and non-response rates and thus, 424 household heads (194 participants and 230 non-participants) were selected from the eight kebeles and interviewed through semi-structured interview schedule. In addition, 16 focused group discussion were conducted with individuals who have depth knowledge and experiences in relation to the topic at village. Key informant interviews were also conducted from kebele administrative, elders, experts and bureau of agricultural experts through check list.

2.2 Methods of data analysis: propensity score matching

The Propensity Score Matching (PSM) method was preferred for the impact assessment. PSM is commonly used for the impact assessment where baseline data is not available and where randomized design is not feasible [26, 27]. The model can reduce the bias in the observational studies by the identifying the non- participants and participants who were similar in all characteristics except for status of participation in vegetable CF [6, 13].

Recommended steps were followed in the applying PSM method to analyze the impact of participation in the CF on smallholder vegetable producers’ income. First the probability model of the participation in vegetables’ CF was estimated to calculate the propensity score of the each sample household heads simply by running Psmatch2. Second, the selection of the best matching algorithm was made from their alternatives (the nearest neighbor, calliper or radius matching and kernel and matching methods) [7], based on their performance criteria like number of insignificance variables after matching, low pseudo R2, large sample size and lower standard bias were selected. Third, check for the common support region. It helps to ensure the comparable observations of the treated and untreated distribution overlaps and then the predicted observation of the propensity scores is discarded, if it falls outside the range of the common support. Fourth, test of the matching quality was conducted through the standard bias, t-test, joint significant and pseudo R2 are suggested. After the score matching quality was checked, the impact of the participation was made by using matched sample. The parameter value is called ATT (The average treatment effect on the treated). Finally, check result for the sensitivity confound. Sensitivity analysis gives answers to whether the inferences about outcomes can be altered by unobservable or confounders [25].

The explanatory and matching variables were identified based on the review of relevant empirical literatures on CF and impact assessment [6, 13, 17, 19, 20, 22, 28, 30, 32]. The explanatory variables selected were age of the Household Heads, sex of the Household Heads, education level of the Household Heads, dependency ratio of the Household Heads, farm experience of the Household Heads, off-farm and non-farm income of the Household Heads, proportion of vegetable crops to the total farm land of the Household Heads, access credit, extension contact, membership in saving group ‘equib’ (is a kind of voluntary association of informal financial institution that deposit money and withdrawal for members through lottery methods for all in amounting equal contribution regularly), access to the market information, market distances, and disease and pest infestation.

3 Results and discussions

3.1 Socio- economic and demographic characteristics of respondents

Demographic, socio-economic and the institutional variables were hypothesized to determine the smallholder producer’s participation in the CF in the study area. Of the total of the sample households, 46% were participating in the vegetable CF, while 54.25% were not participating in the scheme. In terms of the sex of the households, 80.42% were male-headed households. From the total of the female-headed sample households (83), 39.76% were participating in the vegetable CF in the study area. There is no significant difference between the contract farmers and non- contract farmers in terms of their sex of the household heads (Table 1). This shows that, there is no sex restriction for participation in the vegetable CF in the study area.

Table 1 Descriptive statistics of dummy variables
Full size table

Actual or perceived price is another factor that may influence the decision to join CF. In the 2021/22 production year, of total sample households who perceived uncertain price (193), 88.7% of the household heads were participated in the vegetable CF in order to reduce risks. A chi-square value of 243.4 test indicated that, there is statistically significance difference at 1% probability level between contract participants and non-participants in terms of their perception on price uncertainty. This implies that, majority of smallholder farmers participate in the CF due to fear of the price uncertainty. Access to credit and other institutional services were also expected to improve the smallholder farmer’s production and their welfare. The average of the households’ access to the credit for participants and non- participants in the vegetable CF were 34.93% and 39.56%, respectively. The chi-square value of 8.3 tests for independence indicated that, there is a significance difference between participants and non-participants at 1% probability level to the credit institutions.

The frequency of extension contacts expected to give farm household’s opportunities to get advisory service for their vegetable production. The average frequencies of the extension contracts for the total sampled household heads were 33.96% once for every fortnight. A chi-square value of 10.4 for independence indicates that, there is a statistically significant difference in the percentage of contract participants and non-participant farmers in terms of their frequency of the extension contacts at 5% probability level. Of the total sample households, about 53.3% were participated in the training related to the vegetable production. A chi-test result indicated that, there is a statistically significant difference at the 1% probability level between contract participants and non- participants in terms of their training attendance.

Access to the market information enables the smallholder farmers to search for and associate information available for different market channels to manage the cost benefit analysis and related factors in the vegetable CF in the study area. Of the total sampled household heads, 84.2% had access to market information and of these 75.2%% participated in vegetable CF. A chi-squared test result shows, there is statistically significance difference and 5% probability level in the percentage of the participant and non-participants vegetable CF in terms of access to the market information.

In the 2021/22 production year, the average experiences of the household heads in the vegetable farming were 10. 8 years (Table 2). The mean of the smallholder’s farm experience for the participants and non-participants in the vegetable farming were 11.35 and 10.36 years, respectively. The t-test result show that, there is a statistically significance mean difference between the two groups in terms of their farm experience at the 10% significance level.

Table 2 Descriptive statistics of continues variables
Full size table

The total average value of the asset owned by the sampled household heads from the off and non- farm income sources (income earned from composite of remittances earned, aid, safety net programs, sale of fuel wood and charcoal, sale of grass, land rental income, mining, and compensation received) in the study area were 55,128.95 ETB for the whole sample; this was estimated at 52,258.6 and 57,550 ETB for participants and non- participants, respectively. The t-test for equality of means for the household heads income from off-farm and non-farm income sources among two groups was statistically significant at the 5% probability level.

The average area of total farm land allocated to vegetable crops in 2021/22 of the entire sampled households was 1.24 hectares with 1.25 ha and 1.23 ha for participants and non-participants respectively. The t-test result shows that, there is a statistically significance mean differences between the two groups in terms of the proportion of total farm land to vegetable farming at the 1% probability level.

3.2 Propensity score matching results

The probit regression model was used to estimate the propensity score for the participant and non- participant households in the CF. The pre-intervention variables were taken as the explanatory variables and assumed to affect the participation in the vegetables CF. Before proceeding to the impact estimation, the variance inflation factor (VIF) was applied to test for the presence of the strong multicollinearity problem among explanatory variables. There was no serious problem of the multicollinearity and hence, no explanatory variable was dropped from the estimated model. Similarly, Breusch- pagan test for the heteroscedensity was used to check the existence of the heteroscedasticity of the variance and there was no heteroscedasticity problem in the model.

The estimated probit regression model (Table 3) appears to perform well for the intended matching exercise. The pseudo- R2 value is 0.0962, low R2 value shows that, the participants households do not have much distinct characteristics overall and implies there is a good match between contract participants and non- participants. The interest of the matching producer is to get the households from vegetable producer in non-participants CF with similar probability of participants in a CF given explanatory variables. The Probit model was used to calculate the propensity scores by running the psmatch2 command.

Table 3 The Probit regression estimates of vegetable CF participation on HH income (Psmatch2)
Full size table

If the numbers of explanatory variables affecting the participation decision are limited, it created a good opportunity for matching and it makes the matching producer less difficult since matching algorism is implemented to estimate significant differences of explanatory variables between participants and non-participant groups. The maximum likelihood estimates of the probit regression model result shows that, five variables out of the thirteen variables were significant and affect the participation of smallholder farmers in vegetables CF.

Before proceeded to the next steps, it is better to see the common support region that was imposed (the second steps of PSM). The common support helps to check and identify the region of common support between the treatment and comparison groups (see Fig. 2). In the literature review, different ways were suggested to analysis the common support region. The most used one is visual analysis of the density distribution of the propensity scores for both groups. Figure 2 illustrates as follows.

Fig. 2
figure 2

Source: Own computation result based on the survey (2022)

Distribution of propensity scores for untreated and treated group (HH income).

Full size image

To choose the best matching algorithm, the most commonly used methods are nearest neighbor, kernel and radius or caliper matching methods. To select the best matching algorithms, different criteria are recommended and used by different scholars. As [11] states, the equal means test referred as balancing test, pseudo- R2, mean standard bias and matched sample size were recommended as best criteria to selected best common support region in propensity matching scores. Table 4 shows, all matching algorithms were undertaken and offer the same results. The caliper radius matching with radius (0.1) has relatively low pseudo with best balancing test (all explanatory variables insignificant) and large sample size as compared to the other alternatives in both outcome variables.

Table 4 Performance of the different matching algorithms on Household income
Full size table

The third stage is to conduct the balancing test to know whether there is statistically significant difference in the mean value of the two groups of the sampled households. It is better to compare the influence of the background characteristics of both treated and comparison groups before matching for variable selection. Table 5 shows that, the t-test of covariate balance test resulted in the statistically insignificant difference between treated and comparison groups in selected variables.

Table 5 Covariate balance test for the impact of participation CF on HH income
Full size table

Once the difference between the outcomes of participants in CF and non- participants was computed, the next stage is to provide the evidences of the impact of participation in vegetable CF on the household’s incomes (Table 6). After controlling the pre-participation differences, we found out that, participation in the CF has decreased average income of the participant by 2,941.95 ETB in 2021/22 production year. This result shows that, participating in CF decreases the household’s gross annual income by 3.8%. This implies that, Local context matters for the success of CF as institutional arrangement to facilitate smallholder commercialization and enhance producers’ income. In addition, information from key informant interview and focus group discussion confirms that, smallholder farmers were willing to enter into CF in order to access production inputs, advice on market specification and production management to reduce risks and uncertainty. This finding is consistent with the finding of [20] who found out that, smallholder participation in CF lower the income of farmers in rural Benin. Similarly, [1] reported that, participation in soybean CF in Ghana decreased the average annual income of farmers and [29] reported similar finding in his study, comparing participation in CF and out- grower scheme in Wageningen in Netherland.

Table 6 ATT effect of vegetable CF on the household ‘s income
Full size table

Finally, sensitivity test was computed to check the robustness of the estimation covariates to show whether the hidden bias affects the estimated ATT or not. Therefore, a sensitivity test was used to investigate whether the causal effect estimated from the PSM is susceptible to influence unobservable covariates. Table 7 shows, the sensitivity analysis of hidden bias for the impact of the vegetable CF on household’s income. To check for unobservable biases, Rosenbaum bounding approaches were used. As reported in Table 7, the inference for the effect of the vegetable CF is not changing, though the participant and non-participant households have been allowed to differ their odds of being treated up to gamma = 5 in terms of unobservable covariates. This shows, all outcome variables estimated at various levels of critical values of gamma and p- values are significant.

Table 7 Sensitivity analysis of the hidden bias for the impact of CF on the household ‘s income
Full size table

4 Conclusion and policy implication

Contract farming (CF) is expected to promote agricultural commercialization in small-scale vegetable farming by reducing transaction costs in supplying agricultural production and solving market imperfections by linking smallholder farmers to the markets. The result from the probit model shows that, market information, perception about price uncertainty, access to credit, frequency extension contacts, training on vegetable farming, experience of the household head, off- farm and non- farm income and proportion of vegetables to total farm land were significantly affecting CF participation. The finding of the PSM shows that, vegetable CF reduces the gross annual income of smallholder farmers. This implies that, Local context matters for the success of CF as institutional arrangement to facilitate the smallholder commercialization and enhance producers’ income. Nonetheless, smallholder vegetable producers who engaged in CF benefited in terms of access to agricultural inputs, technical advice and share of risks and uncertainty. The smallholders who participated in the CF complained about the cost of inputs, unequal power relations, unpaid family labours and unfair market price. If the smallholder farmers continue like this, the benefit that added to their current resources will be washed away and the performance of the vegetable CF will become worse in the future. In general, local context matters for success of vegetables CF and it is not a panacea that benefits smallholder farmers in the studies area. Therefore, concerned bodies should base their choices on empirical evidence and local context to consider alternative and complementary intervention options to enhance smallholder commercialization and ensure sustainable livelihood. In addition, there is a need for holistic approaches and effective collaboration between policymakers, researchers, and donor’s agencies to enhance smallholder’s farmer’s participation in vegetables CF to the way of getting fair share of the benefits from the values chain. Moreover, further study should be conducted in relation to the contract farming structure and its performance in Ethiopia.