1 Introduction

The development of agriculture plays a vital role in the overall economic growth of Ethiopia [1, 2]. Agriculture is the source of employment, income, food, raw materials, foreign currency, and livelihoods for most rural smallholders [3]. It is predominantly produced by smallholders and over 90% of total agricultural products are obtained from these smallholders [4, 5]. Ethiopian agriculture is subsistence due to insufficient capital, small landholdings, poor rural infrastructure, little market linkage, incompetent farm management, inadequate extension service, and environmental degradation [1, 6,7,8].

Transforming agriculture from subsistence to market-oriented production plays a vital role in the economic growth of developing countries [5, 9,10,11]. This transformation helps to ensure food security [12,13,14] and has important implications for farmers’ risk exposure and management [2, 10]. The study of Mgomezulu et al. [15] indicates that poverty and food insecurity are higher for subsistent farm households. Agricultural commercialization has multifaceted effects on improving farm productivity, income, food security, and livelihoods [8, 13, 14, 16,17,18]. Household market participation increases dietary diversity and economic growth and reduces vulnerability to poverty [11, 13, 19, 20]. It has a significantly positive effect on the household’s welfare and subjective well-being [9, 11,12,13]. Even though the market and its access improve smallholder livelihoods in Africa [21], smallholder market participation is much less [8, 9]. Likewise, in Ethiopia, many households produce farm products for home consumption on a subsistence basis, and the farming system is not market-oriented [6]. This low market orientation can adversely affect the livelihood of rural farmers and contribute to rural poverty [22]. So, agricultural commercialization is considered an important pathway out of poverty and a useful approach for improving the livelihoods of rural smallholders [23, 24].

Within the agriculture sector, vegetable production and marketing can play imperative roles in improving the livelihood of smallholders in Ethiopia [25]. It has a significant role in poverty reduction [26, 27]. That is why the Ethiopian government has given prior attention to growing high-value horticultural crops and vegetables with the motto “Yelemat Trufat.Footnote 1” Gosa et al. [28] noted that vegetables grown in various parts of the country are economically important. For example, vegetables create job opportunities, generate income, contribute to foreign exchange earnings, have nutritional and medicinal values, and ensure food security [25, 26, 28, 29]. According to Rahiel et al. [30], vegetable production has several advantages over other crops, such as requiring a shorter production duration, giving higher yields, and generating a greater income per unit of inputs. The vegetable sub-sector in Ethiopian agriculture has been providing substantial benefits to those involved in the vegetable market channels. Participating in the vegetable market is a crucial economic activity that helps farmers earn better income and improves their welfare [27, 31]. Other studies highlight that vegetable commercialization significantly improves household income and per capita annual consumption expenditure [32,33,34]. Thus, commercializing vegetables has paramount advantages in improving smallholders' economy [33]. The commercialization of vegetable production and market participation are mutually reinforcing. As farmers transition to market-oriented production, they depend on market participation to sell surplus crops and generate income. This creates a positive feedback loop, where commercialization boosts market engagement, and successful market participation encourages further commercialization [35, 36].

As the Ethiopian Statistical Services (ESS) report indicates, vegetable production in Ethiopia is not market-oriented. This is because many smallholders produce vegetables mostly in their home gardens as a subsidiary production. According to the ESS [37], at the national level in Ethiopia, the percentage of vegetable crops consumed at home is higher than that of other food crops, such as cereals and pulses. The report indicates that only 20% of vegetables produced are sold in the Meher (main) production season. This implies that vegetable marketing at the national level is minimal. Many previous studies [25, 26, 28, 38] confirm that although vegetable production is profitable, the marketing system for vegetables in Ethiopia is inefficient and lacks competitiveness. The fluctuating prices of vegetables in the market are significantly impacted by the strategic maneuvers of traders and the ever-shifting demands of consumers. Even though some farm households are growing larger volumes of vegetables, they have little market access and linkages [7, 25, 33].

Vegetable production and market supply are insufficient to meet its demand in Ethiopia, and vegetable growers have not fully benefited [26, 27, 31, 38]. Different demographic, socioeconomic, institutional, and environmental factors can affect vegetable market participation decisions and their effects can vary from one region to another depending on the study area’s specific conditions [3, 11, 25, 27, 32, 33, 39, 40]. Thus, it is crucial to understand the factors influencing smallholder participation in the vegetable market, especially in areas like Lemo District, where limited insight exists into low household participation rates. Thus, using a Heckman two-stage model, this study intends to investigate major drivers influencing market participation and its intensity among vegetable producers in Lemo District, south-central Ethiopia. Other similar studies also employed this model [20, 21, 27, 39, 41, 42]. The study findings may aid farm households, development agents, policymakers, and other concerned stakeholders by providing pertinent information that helps to develop comprehensive agricultural policies and make informed decisions so that vegetable-growing households can gain better benefits through increased market participation. The findings also serve as a reference for future studies in research institutions and Universities.

2 Methodology

2.1 Description of the study area

This investigation was undertaken in Lemo District, Ethiopia. It is among the 13 districts and 7 town administrations constituting the Hadiya Zone. Its geographical location is shown in Fig. 1. The district has 33 rural kebelesFootnote 2 and 33,101 household heads. The main livelihood for households is mixed farming [23]. Agricultural production, including vegetables, is the main source of livelihood for inhabitants. The agroecological setting of Lemo District is conducive to farm activities. Many rural households usually grow vegetables on their farm for consumption and sale. The vegetables mainly grown include potatoes, tomatoes, onions, cabbage, and carrots in their home gardens and other farm plots. However, the level of vegetable market participation among farmers in the district is low compared to its potential. Therefore, analyzing the fundamental drivers of vegetable market participation decisions among growers is imperative.

Fig. 1
figure 1

The study area map

Full size image

2.2 Sampling procedures

Both probability and non-probability sampling methods were used to select the respondent vegetable producers. This study used a multi-stage sampling technique. In the first stage, Lemo District was purposefully chosen because it has a high potential for vegetable production and marketing compared to other districts of Hadiya Zone. According to the Lemo District Office of Agriculture and Natural Resources, vegetables are grown in all rural kebeles of the district. In the second stage, out of 33 rural kebeles of the district, 4 kebeles: Gora Tume, Masbira, Shecha Roma, and Shurmo Witbira were selected using a simple random sampling technique, the lottery method. Then, a list of vegetable growers in each kebele was obtained from their agricultural extension offices to determine the sample frame. In the third stage, 345 vegetable growers were chosen using a simple random sampling technique based on their proportionate size to the entire number of vegetable growers in each sample kebele (Table 1).

Table 1 The procedure of selecting vegetable-growing respondents
Full size table

The sample size was determined by using Taro Yamane’s formula as specified below:

$$n = {\raise0.7ex\hbox{$N$} \!\mathord{\left/ {\vphantom {N {1 + N\left( {e^{2} } \right)}}}\right.\kern-0pt} \!\lower0.7ex\hbox{${1 + N\left( {e^{2} } \right)}$}} = {\raise0.7ex\hbox{${2499}$} \!\mathord{\left/ {\vphantom {{2499} {1 + 2499\left( {0.05^{2} } \right)}}}\right.\kern-0pt} \!\lower0.7ex\hbox{${1 + 2499\left( {0.05^{2} } \right)}$}} = {\raise0.7ex\hbox{${2499}$} \!\mathord{\left/ {\vphantom {{2499} {7.2475}}}\right.\kern-0pt} \!\lower0.7ex\hbox{${7.2475}$}} = 345$$
(1)

where, n = Sample size.

N = study population.

e = Precision level (5%).

2.3 Data collection

For this investigation, the required data were taken from primary and secondary sources. Primary data were acquired by utilizing pre-tested semi-structured interview schedules. Since most older farmers in the study area cannot read and write, nine enumerators were recruited to collect the data under continuous follow-up with the primary researcher. A one-day training was provided to enumerators regarding how they could handle the questionnaire, conduct interviews, and record farmers’ responses. The data were collected regarding the main factors hypothesized to determine market participation and the level of participation among vegetable growers in the study area. Two focus group discussions were conducted in each selected kebele, involving eight participants, including model farmers, community and kebele leaders, and local vegetable traders. In addition, key informant interviews were undertaken with relevant stakeholders at different levels of governmental organization and subject matter experts. This was made for the triangulation of findings. Furthermore, data validation was conducted using both manual and automated data validation techniques. Data were reviewed manually for accuracy, completeness, and consistency and validated by automated software tools such as SPSS, Excel, and Stata to check data for errors or inconsistencies based on pre-set criteria. Data validation was undertaken using various techniques such as outliers check, range check, type check, consistency check, completeness check, cross-validation, and duplicate check. Secondary data were obtained from various sources, including published documents, websites, books, journals, and unpublished documents and reports prepared by local governmental offices, such as sample kebeles and Lemo District.

2.4 Method of data analysis

Statistical software such as SPSS version 22 and Stata 14.2 were used as analytical packages. The data were analyzed using different statistical and econometric analytical techniques. Descriptive statistics including mean and standard deviation were employed to describe the respondents’ socio-economic characteristics. Inferential statistics such as X2- and t-tests were used to examine the significant correlation between variables. Graphs and tables were used to present the results. Furthermore, an econometric model was also employed to determine the main drivers of vegetable market participation and its extent. Sample selection biases can lead to inconsistent estimates in econometric analysis, as finding an absolute random sample is difficult. This can be commonly addressed by applying Heckman’s selection model. Thus, the study utilized Heckman's selection regression model to analyze drivers of market participation among vegetable producers. Many prior similar studies also used a Heckman two-stage model [5, 18, 20, 21, 27, 41,42,43,44].

2.5 Specification of the econometric model

The study runs two models with Heckman selection regression. In the first stage of the model, Probit regression was applied to assess the main drivers of vegetable market participation given that the response variable (selection equation) is dichotomous or binary. The various hypothesized independent variables included in the Probit model are sex, age, educational level, family size, vegetable farming experience, off/non-farm income level, land size, livestock ownership, perception towards the weather conditions, soil fertility situation, access to credit, adoption of improved technologies, lagged market information, extension contact, and market distance. The model examines the outcome of all explanatory variables on vegetable market participation among farmers. The effects of various explanatory variables on vegetable market participation are estimated in Probit regression. It can be specified as follows:

$$M_{i} = aZ_{i}{\prime} + u_{i}$$
(2)

Equation 2 designates the likelihood of a respondent’s vegetable market participation (\({M}_{i}^{*}\)) that is influenced by a vector of independent variables \({Z}_{i}{\prime}\), \(\alpha\) stands for an estimate of \({Z}_{i}{\prime}\) coefficients and \({u}_{i}\) is an error term. \({M}_{i}^{*}\) is not seen, but one can see whether the respondent has participated in the vegetable market or not, whereby \({M}_{i}=1\) when \({M}_{i}^{*}>0\) and \({M}_{i}=0\) when \({M}_{i}^{*}\le 0\). Here, the selection dependent variable is vegetable market participation, which took “1” for the respondents who participated in vegetable marketing, and “0” for those who did not participate.

The second model is the outcome equation model. That means the drivers of the intensity of vegetable marketing are estimated using ordinary least squares (OLS) in the second stage. In this stage, a correction factor or inverse Mill’s ratio computed from the selection equation is added to the OLS model. Here, the outcome variable is the intensity of vegetable market participation by farm households, which has various positive values for market participant households. In the second stage, the OLS regression rigorously analyzes the effect of explanatory variables on the outcome variable, which is the intensity of vegetable market participation. Each stage yields a residual for every observation, meticulously addressing non-random samples and decisively testing for bias by examining the relationship between the residuals for the two stages. Selection bias can be considered as a form of omitted-variable bias. In the second stage, drivers of the intensity of vegetable market participation are explored. In our case, drivers considered to influence the level of vegetable market participation include sex, age, educational level, family size, vegetable farming experience, off/non-farm income, total land size, livestock ownership, perception towards weather conditions, soil fertility situation, access to credit, access to extension contact, adoption of improved vegetable seeds, lagged market information, market distance, and Lambda. The outcome equation, the OLS regression model used in the study, can be specified as:

$$Y_{i} = b_{i} X_{i} + v_{i}$$
(3)

where \({X}_{i}\) designates the vector of explanatory variables determining the intensity of vegetable market participation, \({b}_{i}\) are the estimates of \({X}_{i}\) coefficients, \({v}_{i}\) is the disturbance term.

Based on Heckman’s (1979) equation, the expected intensity of vegetable market participation was explained:

$$E\left( {Y_{i} {|}Z_{i} = 1} \right) = E\left( {Y_{i} {\text{|Z}}_{i}^{*} > 0} \right) = E\left( {Y_{i} {|}u_{i} > - a_{i} M_{i} } \right)$$
$$= b_{i} X_{i} E\left( {v_{i} {|}u_{i} > - a_{i} M_{i} } \right) = b_{i} X_{i} + \beta \sigma_{v} \mu_{i} \left( {\alpha_{u} } \right)$$
(4)

The function \({\mu }_{i}({\alpha }_{u})\) is called the inverse Mill’s ratio or lambda. In Eq. (2), lambda was used as the independent variable. A highly significant lambda shows selection bias, and Heckman's two-stage selection model needs to be applied. In the study, the lambda was highly significant (Table 6).

2.6 Description of hypothesized variables

In the study, the dependent variable is the market participation decisions made by vegetable-growing households. Several predictors were supposed to determine market participation among vegetable-growing households in Lemo District as described in Table 2.

Table 2 Description of hypothesized variables
Full size table

3 Results and discussion

This section presents the findings of descriptive statistics and econometric models. The first and second subsections present the statistical analysis. The third subsection presents the results of the econometric analysis.

3.1 Statistical summary of sample households

3.1.1 Demographic characteristics

As the descriptive findings indicate, among all respondent households, about 79% were male-headed, while about 21% were female-headed (Table 3). It revealed that 94.9% of male-headed households participated in the vegetable market, while 66.7% of female-headed households participated. This implies that families led by women had lower vegetable market participation than their counterpart families led by men. There was a significant proportion variation between both groups in vegetable market participation at a 1% significance level. The finding agreed with the prior study results [21]. The average age of vegetable-growing respondents was 46.17 years, and the average age of market participants was less than their counterpart non-participants. It is observed that more elder household heads were not participating in vegetable marketing as likely as younger household heads. Limited transportation options or physical challenges related to age may prevent older household heads from actively engaging in marketing activities. The t-test result has shown a statistically significant average age variation between both groups at a 1% significance level (Table 4). The respondents’ education level has shown a negative association with vegetable market participation. A significant mean education level variation between the two groups has been observed at a 1% significance level. This happens because more educated households may engage in other alternative income-generating activities and prefer to consume more home-produced vegetables instead of selling them in the market. This result is similar to other findings [15, 21]. The respondents’ average family size was 5.65 adult equivalents. The mean family size for vegetable market participants was less than for their counterpart non-participants as revealed in Table 4. A significant family size variation has been shown between the two groups at a 5% significance level. This result is similar to the previous finding [44].

Table 3 Statistical summary of dummy variables
Full size table
Table 4 Continuous variables description
Full size table

3.1.2 Resource endowment and economic variables

The land is an important scarce resource for farmers. The land resource is a great bottleneck to smallholder agricultural production. Most farmers in the study area were producing their farm outputs relying on rainfed farming systems on small plots of land. With smaller land size, the smallholder transaction costs such as transportation, information, and negotiation costs are high and can reduce profitability. As a result, smallholder farm production is not viable and remains subsistent. Descriptive results revealed in Table 4 that the average landholding in the study area is less than 1 hectare (ha), which is 0.68 ha. The average land size for market participants is 0.73498 ha with a standard deviation of 0.361 and for non-participants is 0.52 ha with a standard deviation of 0.175. As described in Table 4, the size of landholding is positively and significantly associated with market participation. In rural areas, smallholder households obtain income from miscellaneous sources. Alternative sources of income and livelihoods contribute greatly to supplementing traditional farming, enhancing resilience to climate change, and improving household income. Some of the key alternative income sources and livelihoods in the study area: diversified agriculture, off-farm employment, remittances, livestock rearing, small-scale trading, handcrafts, artisan works, and natural resource-based livelihoods such as charcoal making, timber harvesting, and sale of non-timber forest products. In the study, off/non-farm income was negatively and significantly associated with vegetable market participation among smallholder farm households. The level of off-farm income obtained per year shows a negative association with vegetable market participation. This could happen because people earning higher incomes from off/non-farm or non-farm activities choose to devote more time to these pursuits and less to growing vegetables. It has shown a statistically significant off/non-farm income mean difference between the two groups at a 1% significance level. This finding is similar to the earlier findings [44]. Similarly, livestock ownership measured in Tropical Livestock Units (TLU) was positively associated with participation in the vegetable market. This was significant at a 1% significance level. Regarding livestock, the respondents owned 3.5 TLU on average, and it was higher for market participant households than non-participants.

3.1.3 Ecological factors

In the study, two ecological factors were considered to influence the market participation of vegetable-growing households. Both the household’s perception regarding weather conditions and the soil fertility situation of farmland were significantly associated with vegetable market participation in the study area. When households perceive that the weather conditions are favorable and their land is fertile enough to grow vegetables, they will grow greater quantities of vegetables, consequently, their likelihood of involving in the market rises. In contrast, if households perceive that weather conditions are not favorable and their farmland is infertile and not suitable for growing vegetables, they will grow smaller quantities of vegetables to reduce the risk of output loss. As indicated in Table 3, both ecological factors have shown a statistically significant difference between vegetable market participants and non-participants and this result is similar to the earlier findings [15].

3.1.4 Institutional factors

Many institutional factors have shown a strong association with vegetable market participation in the study area as indicated in Table 3. Access to credit is very crucial for smallholder farm households to fulfill the need for cash [51]. In this study, the majority of respondent households (51.9%) affirmed that they did not have access to formal credit at the time cash was required to purchase inputs. There was a statistically significant proportion difference between vegetable market participants and non-participants. This result is similar to the earlier findings [21]. Adoption of improved technologies such as improved seeds, fertilizers, and agrochemicals is also significantly associated with vegetable market participation. Those households that adopted improved technologies have participated more in the vegetable market than non-adopters and there was a significant adoption of technology proportion difference between both groups. This result is similar to the earlier findings [15, 23]. Market information, particularly price information, could greatly influence the respondent’s choice of whether to be involved in the vegetable market or not participate. In this study, the result implies that lagged market information has a strong association with vegetable market participation. This result agrees with the earlier findings [5, 21]. The study of Sori and Adugna [52] shows that access to market information significantly influenced ground nut market outlet choices. Another institutional factor that is strongly associated with vegetable market participation in the study area was extension service contact. Those households having more access to extension contact showed greater market participation than those who did not contact extension service workers per month. It was positively and significantly associated with market participation. This result is similar to the earlier findings [44]. Market distance also showed a negative association with vegetable market participation. The result revealed a statistically significant distance difference between both groups at a 10% significance level (Table 4). This result agrees with the prior findings [44].

3.2 Status of vegetable market participation by smallholder households

Out of the total respondent households, about 89% of households participated in the vegetable market and the remaining 11% did not participate in the vegetable output selling market (Fig. 2). The intensity of vegetable market participation by vegetable-producing households in the study area was 34.32% (Table 5). That means out of total vegetable production, about 65.68% did not pass through output market transactions. About nearly two-thirds of the vegetables produced were consumed by producers themselves. This implies that vegetables are not satisfactorily marketed in Lemo District. That means the proportion of produced vegetables transacted through the market in the study area is quite low. The reports from FGDs and KIIs supplement this finding. The reports imply that most rural households produce various vegetables for home consumption and sell small amounts to cover miscellaneous expenditures. This might happen due to subsistence farming and a lack of awareness about the benefits of high-value cash crops like vegetables. This result agrees with the report of Ethiopian Statistical Services [37] which implied that vegetables and root crops sold in 2021 in the Hadiya Zone, where Lemo District is encompassed, were only 4.25% and 11.82%, respectively [37]. Thus, concerned bodies need to design sound strategies that enhance the intensity of smallholder vegetable market participation.

Fig. 2
figure 2

Source: Authors’ survey result, 2024

The vegetable market participation among smallholder households.

Full size image
Table 5 Descriptive summary of vegetable market participation
Full size table

3.3 Econometric analysis

In this study, a Heckman two-stage selection model was employed to recognize the major drivers of vegetable market participation among vegetable-growing households in the study area. Before applying the model, several tests were undertaken to address model assumptions and the issue of the risk of misspecification based on the study’s datasets. To address the risk of misspecification, the authors conducted several tests. The Ramsey RESET test was used to verify the appropriateness of the functional forms used for both the selection and outcome equations. Alternative model specifications or functional forms such as non-linear models or non-parametric methods were used to assess the robustness of the results. A sensitivity analysis was also conducted to check if the estimates change significantly under different functional forms. It was conducted by re-estimating the model by including or excluding certain variables to test whether the results are sensitive to the choice of explanatory variables or remain stable. The valid instrumental variables were used to address potential misspecification in the selection equation by ensuring that the instruments used are strongly correlated with participation but not directly related to the outcome variable. Model fit diagnostics such as the Wald test, likelihood ratio test, or goodness-of-fit measures were employed to validate the model's assumptions and that the specified functional forms are appropriate for the data. A Variance Inflation Factor (VIF) test was undertaken to confirm that multicollinearity is not a serious problem in the study. In this study, VIF was less than 5 for all explanatory variables, implying multicollinearity is not a concern.

Vegetable farmer’s market participation decisions and level of participation can be affected by various demographic, socioeconomic, ecological, and institutional factors. For this study, fifteen variables were supposed to influence vegetable market participation and its extent among farm households. As shown in Table 6, the Probit model in the first stage of the Heckman selection model has indicated that vegetable market participation is affected by eight factors. The OLS regression model in the second stage has shown that seven factors influence the intensity of vegetable market participation in the study area. In addition, the Inverse Mill’s ratio has shown strong significance in the study, indicating the relevance of employing Heckman's selection regression to capture selectivity biases.

Table 6 Econometric model results
Full size table

3.3.1 Vegetable market participation among smallholder households

The sex of the household head plays a vital role in the market participation decisions among vegetable growers. As seen from the marginal effect value of the Probit regression, being a male-headed household increases the likelihood of vegetable market participation by approximately 6%, and was statistically significant. It agrees with other earlier findings [18, 21, 41]. Age has shown a negative association with market participation decisions among vegetable growers. As the age of the household head rose by a year, the likelihood of market participation was decreased by approximately 0.55% at a 1% significance level. The result is similar to earlier findings [15, 17, 34], but contrasts with another finding [18]. Education level plays a significant role in determining vegetable market participation. It was implied that a unit rise in the schooling year increased the likelihood of market participation by 1.8% and was found statistically significant. The result is similar to the prior finding [46] but contrasts with other findings [21, 48].

The land is a vital resource to undertake farm activities like vegetable production. As the landholding size increases the quantity production is also expected to increase, ceteris paribus. This is because the larger landholding enhances the productivity and profitability of vegetable production through increasing economies of scale and reducing transaction costs. In this study, a hectare increase in landholding increased the likelihood of vegetable market participation by 22.17% at a 1% significance level. It agrees to prior results [11, 17, 22, 30, 32].

Perception towards weather conditions also influenced the vegetable-producing households’ market participation decisions. When farmers perceive the weather conditions to be good for vegetable production, they produce more vegetables, increasing the likelihood of market participation among farm households. On the contrary, when farmers perceive the weather conditions as not good for vegetable production, they likely reduce the quantities of vegetable production, which directly diminishes the probability of vegetable market participation. In this study, the marginal effect result verified that a good perception of weather conditions increased the probabilities of vegetable market participation among smallholders by 10.34% at a 1% significance level. It happens because most farmers in the study area produce their farm products relying on rainfed and greatly depend on weather conditions. Therefore, access to modern irrigation schemes for smallholders is crucial to increase their farm productivity and profitability. It enhances vegetable production and productivity and thus improves the livelihoods of rural people. Concerned bodies should focus on activities that allow farmers to produce agricultural products multiple times without relying on rainfall.

Credit access can boost the production and productivity of smallholders in developing countries like Ethiopia. When farmers face a cash shortage, credit access could facilitate the procurement of essential inputs in vegetable production and marketing. The probit regression model result confirmed that credit utilization increased the likelihood of market participation among smallholders by 12.4% at a 5% significance level. Prior studies also reached a similar conclusion [12, 19, 25, 30, 32]. The adoption of improved technologies enhances farm productivity and increases farm income. Adopting improved technologies like high-yielding and disease-resistant seeds, fertilizers, and agrochemicals helps to increase vegetable yields and increase the likelihood of market participation. The probit model has revealed that adopting improved technologies increased the possibility of vegetable market participation by 4.8% at a 10% significance level. It shares the outcome of prior studies [23, 50]. The market distance negatively and significantly affected vegetable market participation among smallholder households. The model result revealed that a unit increase in market distance decreased the likelihood of vegetable market participation by 2.33% at a 10% significance level. A similar conclusion was observed in other prior studies [5, 40, 44, 46, 53].

3.3.2 Level of vegetable market participation

Likewise, for vegetable market participation, the education level of farm household heads plays an essential role in determining the extent of vegetable market participation. The study revealed that a rise in the household head’s schooling by a year had increased the likelihood of vegetable market participation level by 1% and was significant at a 1% significance level. In other expressions, education level has positively contributed to the intensity of vegetable marketing. More educated people might better understand the benefits of marketing farm products. This result agrees with the previous findings of [13, 23, 25]. However, Nambafu et al. [40], Olutumise [48], and Wakaba et al. [11] found a different result. Family size was shown to reduce the intensity of vegetable market participation. That means family size has negatively affected the level of vegetable market participation among farm households. The coefficient of OLS regression in the second stage of the Heckman selection model revealed that an increase of family size by 1 AE would lessen the likelihood of the level of vegetable market participation by 1.96%. It was significant at a 1% significance level. It implied that households with a greater family size would prefer to consume vegetables at home rather than supplying them to the market center. This directly reduces the amount of vegetable output supply brought for sale. The result agrees with the earlier finding [18]. However, it contrasts with the findings of [46, 47], which described larger family sizes increasing agricultural commercialization through greater labor contribution in the production and marketing processes.

The off/non-farm income level can highly determine the intensity of vegetable market participation among farm households. It negatively influenced the intensity of vegetable market participation. That means households earning higher incomes might prefer to produce vegetables mainly for home consumption and engage in off-farm or non-farm activities. The OLS regression model confirmed that getting more off/non-farm income by 1% would likely reduce the vegetable market participation level by 10.53%. The result agrees with earlier findings [50]. However, a contrasting result is reported by other studies [47, 48].

The household decisions regarding what to produce and how much to supply can be greatly influenced by their perceptions or expectations of the weather conditions. In this study, household’s optimistic perceptions positively influenced the level of vegetable market participation. A good perception regarding weather conditions increased the vegetable market participation level by 5.14%. It happens because, if households perceive the weather conditions in a certain production season are favorable, they may want to produce greater quantities of vegetables and participate more in the market. In contrast, if farmers perceive that the weather conditions for the production season are unfavorable for vegetable production, they are more likely to lessen the amount of vegetable production or they may not produce vegetables in that particular season. These can directly affect the market participation level.

Having credit access can affect a household’s level of market participation. In the study, having access to credit increased the market participation level among vegetable growers by 3.74% at a 5% significance level. This finding is consistent with the prior findings [48, 50]. However, the result contrasts with the findings of [2, 39]. Another important variable in determining the intensity of vegetable market participation is technology adoption, such as improved seeds, chemical fertilizers, irrigation, and agrochemicals. Improved technologies can increase the production and profitability of vegetable farmers through the effective use of existing resources. The OLS regression showed that adopting improved technologies enhanced the vegetable market participation level by 3.68% at a 5% significance level. This increase was statistically significant and is consistent with the prior findings [23]. Unlike industrial products, producing agricultural commodities including vegetables is time-lagged by at least two to three months or a production season. The decisions concerning vegetable production and marketing can be typically influenced by lagged market prices. If farm households could have the prior market price information, they could make the right production decision and in return, determine the quantities of vegetables supplied to the market. In the study, market price information positively influenced the level of vegetable market participation. Having lagged market information increased the likelihood of the extent of market participation by 4.22% at a 5% significance level. The finding agrees with prior studies [34, 47,48,49]. Also, the lambda was highly significant at a 1% significance level. It indicates the relevance of employing Heckma’s selection model to control the effect of sample selection bias and get consistent estimates.

4 Conclusion and recommendations

The study has been carried out to analyze the main drivers of vegetable market participation and the level of market participation in Lemo District, Ethiopia. A multistage sampling procedure was used to choose respondent households. Necessary data were obtained from both primary and secondary sources. With the help of the SPSS version 22 and STATA 14.2 software packages, data were analyzed employing descriptive and inferential statistics, and Heckman’s two-stage econometric model. The results showed that although smallholder vegetable market participation was high in the number of participating households (89%), the extent of participation, measured by the volume of vegetables sold, was quite low (34.32%) in Lemo District. This is because smallholder farmers bring only a smaller portion of their produce to market, consuming the larger proportion at home. Consequently, overall vegetable market participation is low relative to total vegetable production in the study area. This indicates that a small share of the produce is marketed, highlighting the need for improvement interventions. The Heckman two-stage regression revealed various factors influencing vegetable market participation and its intensity in the study area. As the Probit model indicated, the factors that influence the vegetable market participation among producers are sex, age, education level, land size, perception regarding weather conditions, access to credit, adoption of improved technologies, and market distance. Likewise, the OLS regression revealed that the education level, family size, off/non-farm income, perception towards weather conditions, credit access, technology adoption, access to market information, and lambda mainly determine the intensity of vegetable market participation among smallholder farm households. By considering demographic characteristics, socioeconomic conditions, agroecological factors, institutional structure, policy context, and methodology, the findings of this study can be applied and generalized to other similar regions or areas beyond Lemo district.

To improve the intensity of vegetable marketing among smallholders, this study suggests that the concerned bodies, including governmental and non-governmental organizations working for the overall development of smallholder households, should design sound strategies that accentuate those demographic, socioeconomic, ecological, and institutional factors that are significantly affecting the intensity vegetable marketing among farm households. Specifically, implementing gender-sensitive policies that address the barriers women face in market participation can help create a more inclusive economic environment. Marketing strategies that give equal access to adult education, credit, and improved technologies for all genders and more intensive input utilization should be fostered to promote market-oriented profitable vegetable production among smallholders. Improving rural infrastructure, such as roads, storage facilities, and transportation services, enables farmers to bring larger produce to the market. This can reduce post-harvest losses and increase marketable surplus. Furthermore, improved access to market information and market linkages should be created alongside encouraging diversification into high-value crops or off-season vegetables. These could give farmers a competitive edge and increase their earnings when market prices are more favorable.

4.1 Limitations of the study and areas for future research

This study mainly emphasizes analyzing the drivers of market participation decisions and the level of participation among vegetable-producing smallholders in Lemo District. The study examined vegetable market participation from the output market side and did not cover the inputs market. Therefore, future research may consider analyzing smallholder market participation decisions and level of participation in a broader context in Ethiopia and other similar developing countries by considering other agricultural commodities alongside the input side of market participation. Future research can also examine market participation by incorporating additional potential predictor variables not considered in this study.