1 Introduction

Issues related to food and nutrition security have far-reaching consequences for individuals and their surroundings, especially in low and middle-income countries. As urban populations increasingly grapple with the food dilemma, urban agriculture (UA) emerges as a proposed solution. It is seen as a means to provide food and income sources for the urban poor; while also enhancing the urban environment and alleviating the strain on limited farmland [1]. Urban agriculture plays a crucial role in addressing urban food insecurity, which will become increasingly important as the general trend of poverty and population urbanization persist in emerging economies. The discussion surrounding how UA can contribute to the food security and nutrition of cities/towns as a whole, as well as of households involved in farming inside urban limits, is a prominent issue [2].

The urbanization rate in Ethiopia is high, averaging 4.3% annually. The current 12 million urban population will triple to 32 million by 2030 [3]. This rapid increase will have a broader environmental, food, and livelihood impact [4]. Therefore, tackling urban food insecurity has become a critical element of Ethiopia's development program. The implementation of UA can be taken as one measure for combating urban food insecurity [5]. Encouragingly, there have recent developments in UA practices in Ethiopia, with the adoption of UA becoming increasingly popular and promoted to sustain the lives of poor and unemployed urban dwellers [6, 7].

Though smallholder urban farmers are familiar with UA practices, the integration of production and marketing of urban agricultural products is limited [5, 8, 9]. Households engaged in UA practices often rely on informal markets due to a lack of marketing knowledge and weak or non-existent links with official markets. As a result, the majority of their products are sold at lower prices at their farm gate [10]. Despite a small amount of urban agricultural products being provided and sold in the market, the level of participation of smallholder urban farmers in the UA product market remains low due to various constraints [11,12,13]. Fundamentally, their decisions on the amount of output to sell are heavily influenced by various factors. To the best of our knowledge, no empirical study has been conducted to verify the factors responsible for the low market participation of urban farmers in Ethiopia in general and the study areas in particular. Moreover, it is unclear the extent of households' participation in the UA product market and the benefits from participation in these market opportunities.

Moreover, despite their undeniable importance, most studies that examine market participation and extent of participation have found mixed results indicating that whereas some factors are significant drivers, other factors are not. Additionally, studies on the determinants of UA output market participation have methodological flaws in that they only capture the revealed marketing decisions of households while ignoring the volume of supply [13,14,15,16]. Furthermore, their findings differ within and among nations because of the variability of conditions encountered by smallholder farmers, UA market participation decisions, and factors impacting the intensity of supply in eastern Ethiopia that were not addressed. There have been relatively few studies addressing the status and factors of market involvement, as well as the consequences of obstacles on the decision-making of urban farmers [17]. In this regard, the current study contributes to filling this information gap for market participation and intensity in the studied region.

The socioeconomic, demographic, and institutional factors that influence the participation and intensity of the UA output market in Ethiopia, particularly in the study area, have not been thoroughly studied before, and no empirical research has been done to address this limitation. As a result, this study considered various UA products, such as vegetable production, livestock rearing, and crop and fruit production, and employed a Double Hurdle (DH) Model to determine UA market participation and intensity. Furthermore, the inability of planners and policymakers to create evidence-based, effective policies and strategies is hampered by the lack of information on the critical factors influencing UA output market participation and intensity. Thus, this study provides crucial information to planners, policymakers, and other governing bodies so they can create strategies and policies that will increase the number of participants and intensity of the UA output market in an efficient and long-lasting manner.

2 Organization of the paper

There are five chapters in this paper. The first chapter discusses the background of the study. The second chapter presents the review of the literature. The third chapter discusses the study's methodology, which includes a description of the study areas, sample size and sampling technique, and methods of data analysis. The fourth chapter presents the study’s major findings and discussion. The first section of this chapter discusses the descriptive results, while, the second section presents the estimation results of the DH Model. The fifth chapter contains the study’s conclusion and recommendations. Finally, references and an appendix are included at the end of this paper.

3 Review of market participation decision

According to Taylor and Adelman [18], agricultural household models provide a steady source of micro-research in less developed countries. The availability of markets, or perfect and imperfect markets, determines the separation of production and consumption. In a separable market (a perfect market), income is the single factor mediating the relationship between output and consumption; in a non-separable market (a market failure scenario), there is a complicated interplay between household production and consumption. Food produced more than needed for family consumption is sold (purchased), and labor generated more than needed for household production is sold (bought) [19].

The product market sells extra food produced for the household, while the labor market sells labor provided by family members not needed on the home plot. If output falls short of consumption, the household may be a net buyer of food, requiring money from other sources of revenue. Farm earnings act as a hinge between the two issues, allowing separability and addressing the producer side before the consumer side [19]. However, agricultural households in lower-income nations often find this division less clear-cut. Both production and consumption can occur in the home, and policy changes may not affect household behavior until they are significant. The sole connection between production and consumption in a functioning market is the agricultural income achieved during production. Mixed markets and lack of markets occur in emerging and transitional economies, and policy changes may not affect household behavior until they are significant [20].

Several variables are thought to impact farmers' market participation decisions. Socio-economic factors, institutional factors, market factors, and external influences are all examples of such factors. Age, gender, off-farm income, level of education, years of farming, household size, farm size, and production level are all socio-economic characteristics to consider. Membership in a group, access to extension services, access to credit, infrastructure, and regulations and laws are all examples of institutional factors. Farmers market participation is also determined by market characteristics such as access to market information, output pricing, distance to the marketplace, means of transportation, and other external factors such as natural catastrophes [21].

Addressing UA output market participation necessitates multifaceted approaches and a thorough understanding of the factors contributing to or impeding the sub-sector. The existing literature highlights contributions from refs. [17, 22], and [23]. For example, Campbell et al. [23] studied commercial urban agriculture to characterize commercial urban farms and identify their primary barriers to business development and expansion and their perceptions of future opportunities using qualitative methods. Likewise, [22] investigated market participation behavior among urban okra growers using the Tobit model. However, a major flaw in the Tobit model is that the actual value and the value that would arise if the value were positive are both determined by the same underlying process (i.e., parameters). Unlike these, Bekele et al. [17] explored the factors determining the market participation of dairy cattle farmers in selected urban and peri-urban areas using descriptive and DH negative binomial count data models, but the study focused solely on dairy cattle farmers, neglecting the market participation of urban farmers in other UA products.

Several variables are thought to impact farmers’ market participation decisions. Socio-economic factors, institutional factors, market factors, and external influences are all examples of such factors. The extent and factors influencing market participation decisions in Ethiopia have subsequently been studied by several researchers [13, 21, 24,25,26,27]. Age, gender, off-farm income, level of education, years of farming, household size, farm size, and production level are all socio-economic characteristics to consider. Membership in a group, access to extension services, access to credit, infrastructure, and regulations and laws are all examples of institutional factors. Farmers' market participation is also determined by market characteristics such as access to market information, output pricing, distance to the marketplace, means of transportation, and other external factors such as natural catastrophes.

Despite the Ethiopian government’s policy choice to prioritize UA, studies on UA product market participation in Ethiopia are limited, and UA markets are fragmented and poorly integrated into wider market systems, increasing transaction costs and reducing farmers' incentive to produce for the market. As a result, there is insufficient information on the marketing process and behavior of urban farmers in Ethiopia in general, and in the study areas in particular.

4 Conceptual framework

Adoption of UA practices tends to increase agricultural productivity and hence, influence urban farmers’ tendency of market participation (Fig. 1). UA product market participation decisions are assumed to be affected by a wide range of factors. From the extensive reviews, the various factors can be grouped into the following three broad categories: socioeconomic factors, institutional factors, and market factors [12, 28,29,30,31].

Fig. 1
figure 1

Conceptual framework of the study

Full size image

The factors related to the socio-economic characteristics include livestock holding, age, sex, and family size, education, and size of the land: the institutional factors include use of credit, extension service, and membership of local organization: as well as the aspects related to the marketing characteristics include accessibility of markets, availability of market information, and distance from the market center.

5 Methodology

5.1 Description of the study areas

The study is conducted in the Dire Dawa city administration, one of the two city administrations, and the east Hararghe zone of Oromia, found in eastern Ethiopia. Dire Dawa is located 515 km away from Addis Ababa, the capital city of Ethiopia, and 55 km north of Harar [32]. Geographically, Dire Dawa lies between 9°27′'N to 9°49′N latitude and 41°38′E to 42°19′E longitude. Whereas, the east Hararghe zone capital Harar is located 510 km to the east of the Ethiopian capital, Addis Ababa. Its altitude ranges from 500 to 3400 m above sea level. Geographically, the east Hararghe zone lies between coordinates of 8° 48′28.9008″ N latitude and 41°36′4.2516″ E longitude. The east Hararghe zone encompasses several small towns; the major ones are eleven, including Harar City. However, in addition to Dire Dawa city, this study was carried out in east Hararghe zone towns namely Harar, Haramaya, Kersa, and Kulubi Demographically, the zone has a total population of 3,538,361, While 350,014 or 9.89% are urban inhabitants, the remaining 3,188,346 or 90.11% are rural inhabitants [33]. According to the same report, Dire Dawa city administration has a total population of 466,000. While 293,000 or 62.88% are urban inhabitants, the remaining 173,000, or 37.12% are rural inhabitants.

Urban agriculture in the study areas takes place under varying socio-economic conditions with various production systems involving a diversity of interdependent activities. The agro-climatic condition of the study location enables the production of a range of agricultural products including cereal crops like sorghum and maize; vegetables like potatoes, tomatoes, onions, salad, green peppers, and cabbage; fruit trees like oranges, mangoes, bananas, lemons, and papayas; and perennial crops like coffee and khat (Catha edulis) [34]. The areas are also convenient for livestock rearing, such as milk cows, beef cattle fattening, goats, sheep, poultry production, and beekeeping activities. The dominant UA products produced include vegetables such as cabbage, onion, lettuce, green peppers, tomatoes, and potatoes; the grain of corn; livestock products such as egg, milk, butter, and cheese; and live animals such as goats, sheep, beef cattle, and chicken into the market.

5.2 Sampling technique

A multi-stage sampling procedure was employed to draw the target sample households. First, the east Haraghe zone and Dire Dawa city administration were selected purposively based on the representativeness of eastern Ethiopia and the study’s purpose [35]. In the second stage, Dire Dawa city and four towns from the east Haraghe zone of Oromiya were selected based on the existence of UA practices (major UA-practicing towns in the zone) and discussions held at the zone’s office of agriculture. In the third stage, Kulubi, Kersa, and Haramaya towns as a whole, four kebeles from Harar and four kebeles from Dire Dawa city, were randomly selected for the study. Finally, sample respondents were randomly selected and interviewed from each town/city, proportionate to the size [35].

5.3 Sample size determination

Sample size determination is affected by several factors, among which are the rationale of the study, population size, and tolerable sampling errors. Three decisive factors need to be specified to determine the appropriate sample size for this study; these include the level of precision, the level of confidence, and the intensity of variability in the attributes being measured [36].

The [37] formula was used to determine the total sample size. In order to get the most conservative sample size in the absence of information on the standard deviation of the population, the estimated proportion of households participating in UA was assumed to be 0.5. Thus, the required sample size was determined as follows:

$$n = \frac{{z^{2} pq}}{{e^{2} }}\,\, = \frac{{(1.96)^{2} \times 0.5\times0.5}}{{0.05^{2} }}\,\, \approx 385$$
(1)

where n is the required sample size, Z is the inverse of the standard cumulative distribution that corresponds to the level of confidence, e is the desired level of precision, q = 1−p, and p is the estimated proportion of an attribute that is present in the population. The value of Z is found from the statistical table which contains the area under the normal curve of 95% confidence level [38]. However, out of the total 385 sample households surveyed 115 were non-adopters while the remaining were UA practices adopters. Therefore, the 270 UA practice adopters were the focus of this study’s investigation.

5.4 Data types, sources, and data collection methods

This study is based on primary data collected from 270 UA farmers. Primary data were collected through face-to-face personal interviews using structured questionnaires. The questionnaires were developed based on previous research outputs on determinants of market participation decisions [17, 22, 23, 39]. A mixed type of research method was employed in this study as the sole approach of research method, either qualitative or quantitative, may not be enough to grasp the major constraints in UA practices. Hence, it is recommended to supplement quantitative methods with qualitative methods [40]. Accordingly, FGDs by considering sex, educational background, and age at each kebele and Key Informant Interviews (KII) were conducted to enrich and triangulate the data from the questionnaires. KII was conducted with experts from zonal and district offices of agriculture and town/city development agents.

5.5 Methods of data analysis

The decision to participate can be viewed as a two-stage decision-making process in the model. Households make a discrete decision on whether or not to participate in the urban agricultural product market in the first stage. In the second stage, households make a continuous decision on their intensity of market participation contingent on their decision. The term "intensity of market participation" refers to the output value (UA products) sold by an urban farmer in the output market, which is measured in birr. The models that are most frequently employed in the two-step method are the Heckman, Tobit, and DH Models [41, 42]. As opposed to the Tobit model, the DH model does not presuppose that the process determining participation and intensity of participation be the same [43]. As a result, the model offers a valuable framework for examining independently the impact of certain factors on participation and the quantity of produce sold. Nonetheless, the DH Model was selected for this study instead of the Heckman and Tobit models because, first, there was no sample selection problem in the data, and second, the Tobit model's assumption that participation and sells decisions are made simultaneously means that factors influencing both decisions are the same [44, 45].

The DH Model, according to Burke [46], is helpful because it allows a subset of the data to pile up at a particular value without causing a bias in estimating the determinants of the continuous dependent variable in the second stage, allowing to obtain all of the data in the remaining sample for the participants. Prior to data analysis, diagnostic tests such as multicollinearity, sample selection bias, normality tests, and an appropriateness test between Tobit and the DH Model were carried out to ensure the validity of the DH Model.

The DH Model is based on a two-step estimating process that was first developed by Cragg [47]. The first stage (participation decision) was analyzed by the Probit model. The participation decision can be formulated as:

$$Y_{i}^{*} = \beta X_{i} + \varepsilon_{i} \,\,\,\,\,\,and\,\,\,Y_{i} = \left\{ {\begin{array}{*{20}c} {1\,\,if\,\,Y_{i}^{*} > 0} \\ {0\,\,if\,\,Y_{i}^{*} \le 0} \\ \end{array} } \right.$$
(2)

where: \(Y_{i}\) is the observed dependent variable for decision in urban agricultural product market participation; \(Y_{i}^{*}\) is the latent (unobserved) variable describing the smallholder’s decision to participate in urban agricultural product market; \(X_{i}\) is a vector of the independent variable on participation; \(\beta\) is a parameter of the model to be estimated; and \(\varepsilon_{i}\) is a random error.

The next stage involves an outcome equation, in which the intensity of urban agricultural product market participation is determined by using a truncated model. A truncated regression is applied to analyze determinants of the intensity of urban agricultural products sold [48]. The extent of participation is indicated by:

$$Z_{i}^{*} = \alpha X_{i} + u_{i} \,\,\,\,\,\,and\,\,\,Z_{i} = \left\{ {\begin{array}{*{20}c} {Z_{i}^{*} } & {if\,Z_{i}^{*} > 0\,\,and\,\,Y_{i} = 1} \\ 0 & {\,otherwise} \\ \end{array} } \right.$$
(3)

where: \(Z_{i}\) is the observed dependent variable for and its intensity of urban agricultural product market participation decision (actual value of urban agricultural products sold, in birr); \(Z_{i}^{*}\) is the latent (unobserved) variables describing the household’s in-quantity sales decision; \(X_{i}\) is a vector of independent variable on the quantity of sell’s decision; \(\alpha\) is the parameter of the model to be estimated; and \(u_{i}\) is a random error for quantity sells decision.

The log-likelihood function from the Cragg type DH Model is the sum of the log-likelihood from a probit model and a truncated regression. Hence, DH Model is given as follows:

$$Log{\mkern 1mu} L = \sum\nolimits_{0} {\ln \left[ {1 - \Phi \left( {Z_{i} ^{\prime } \alpha } \right){\mkern 1mu} \left( {\frac{{X_{i} ^{\prime } \beta }}{\sigma }} \right)} \right]} + \sum\limits_{ + } {\ln \left[ {\Phi \left( {Z_{i} ^{\prime } \alpha } \right)\frac{1}{\sigma }\varphi \left( {\frac{{Y_{i} - X_{i} ^{\prime } \beta }}{\sigma }} \right)} \right]}$$
(4)

where: \(\Phi\) and \(\varphi\) are the standard normal probability function (PDF) and cumulative density functions (CDF), respectively.

6 Results and discussion

6.1 Socio-demographic characteristics

The summary of the general disparities in each variable between UA output market participants and non-participants is illustrated in Table 1 and Table 2. The findings reveal that UA output market participants can be differentiated by demographic characteristics such as age, education, sex, family size, and dependency ratio. Participants are, on average, younger, have more formal years of schooling, more likely to be female-headed, and have a low dependency ratio. The mean age of household heads for the UA output market participation was 41.16 years for participants and 44.09 years for non-participants. The mean difference test result for age is statistically significant at a 10% level. Another factor is the level of education achieved by the heads of households, who are typically the decision-makers. The mean education level of participants and non-participants was 11.25 and 9.97 years of schooling, respectively. The mean difference test result for education is statistically significant at a 5% level.

Table 1 Summary statistics of continuous variables
Full size table
Table 2 Summary statistics of discrete variables
Full size table

As indicated in Table 1, the farming experience of the households influences market participation. The mean farming experience of the household heads was 12.17 years for market participants and 10.33 years for non-participants. Market participants and non-participants had average distances to the nearest market of 2.22 and 2.39 km, respectively. This proximity to the market might assist in providing better access to information and, thus, to market opportunities. The mean difference test result for farming experience is statistically significant at a 1% level. Additionally, in terms of occupation, 79% of market participants were civil servants. According to the results of the chi-square test, a significant difference between participants and non-participants was found for all variables at a 10% level.

Looking at the family size of the sampled households, the market participants accounted for about 5.35 persons, while the non-participant households accounted for about 5.76 persons. However, market participants showed a slightly lower mean dependency ratio (0.28) than non-participants (0.42), which implies that market participants supported fewer persons who were either young or very old compared to non-participant households. The t-test results of family size and dependency ratio are significant at a 1% significance level (Table 1).

Market-oriented urban farmers appear to be better off in terms of land size and tropical livestock units (TLU) (Table 1). The average land size owned by the market participant households was 0.27 hectare, which is higher than the average land size owned by the non-participants (0.21). Larger land holdings are predicted to provide people with a competitive advantage in terms of UA output and market participation, as urban farmers can only devote more of their land to UA practices if they have sufficient land. Having livestock as a wealth indicator shows a household's capacity to engage in other UA practices, such as cultivating fruits, vegetables, and crops. The mean TLU was 1.25 for market participants and 0.89 for the non-participants. This suggests that compared to non-participants, market participants owned significantly more livestock. The t-test results show that land size and TLU are significant at a 10% and 1% significance levels, respectively.

Concerning the sex of the household heads and market participation, 29% of UA output market participants were female-headed households, whereas 46% were non-participants (Table 2). The chi-square test result of sex concerning market participation shows that there is a strong association between the two at a 1% level. This implies that there is a disparity in market participation concerning the sex of household heads. Female-headed household’s participated less than male-headed household’s maybe because they had less extra to sell relative to their male counterparts.

Market participants have more access to institutional services such as UA-related training and market information and are also members of community groups than non-participants. Those household heads participating in the UA output market are relatively those who got trained in UA-related practices (87%), have access to market information (95%), and belong to community groups (33%). The chi-square test result indicated a strong association between participants and non-participants at a 1% level for all variables (Table 2).

The average number of extension contacts provided to market participants and non-participants was 1.15 and 0.69, respectively. The result indicates extension services help improve the market participation of urban farmers. The test result for extension service shows a strong association between the two at a 1% level (Table 1). Market participants (30%) had access to credit services, while 28% of non-market participants had access to credit. The chi-square test result shows a statistically insignificant association between access to credit and market participation (Table 2).

6.2 The market for urban agriculture products

The effectiveness of market participation in improving food security and household welfare depends on the efficiency of the market. The descriptive statistics show that 42% of the urban farmers sold their output in a market. In terms of total production, the UA output produced by urban farmers was valued at 36,693.81 birrFootnote 1 on average, while the average value sold by urban farmers was 27,053.02 birr annually. This shows, that UA has a significant portion of the food supply and pays special care to urban diets, which include vegetables, fruits, crops, fresh milk, and especially chicken products. Most urban farmers produce modest quantities of relatively fresh UA products, sell them in urban markets at lucrative prices, and earn a higher income. In this way, UA significantly contributes to a greater diversity of foods in urban markets, as well as to employment, livelihoods, and food security. However, based on the information obtained during the KII and FGD, the local market for UA output in the study towns/cities is undeveloped.

Households in the study area sell their UA output through various channels, including the local market, the farm gate, retailers, and hotels. As can be observed from Fig. 2, the majority of the sample households (74.05%) sold their UA output at the farm gate. About 60.31% of farmers sold their UA output at the local market. Around 30% of the sample households sold their output to retailers and the rest (16.03%) supplied their UA output to hotels. The point of sell had implications for farm profitability and consequently household welfare.

Fig. 2
figure 2

Source: Own estimation results based on survey data, 2022

Urban agriculture products point of sell.

Full size image

Furthermore, the study indicated that, despite the economic and environmental benefits of UA practices, undeveloped markets for UA products, due to low prices, knowledge gaps, and so on, constitute impediments to the expansion of the UA output market in the study area.

6.3 Empirical results of urban agriculture output market participation

This section presents the DH Model results on factors influencing the status and intensity of the market participation of urban farmers. The study used a series of model specification tests to determine which model best identifies the factors influencing urban farmers' decision to engage in the UA output market and the level of participation. The findings of the diagnostic test indicated that the Inverse Mills ratio was insignificant indicating no selectivity bias in the model. Hence, the Heckman model was found inappropriate for the data set. The Tobit and DH models were evaluated for suitability using the conventional log-likelihood ratio test. The test statistic for log-likelihood (Γ = 496.859) was higher than the chi-square value (24.996) at 15 degrees of freedom in favor of the DH model. As is typical with regression analyses, a diagnostic test result for multicollinearity, revealed that it is not a serious problem (Mean VIF of 1.26). The Wald chi-square value for the decision to participate in the market is 94.06, showing that explanatory variables jointly explained the likelihood of participating in the UA output market. This value is significant at the 1% significance level. The result of Skewness and Kurtosis for the truncated part (Joint Prob > χ2 = 0.8353) revealed that residuals were normally distributed.

Table 3 presents the DH model's findings on the variables affecting market participation and the intensity of participation. The decision of urban farmers to participate in the UA output market is positively and significantly influenced by farming experience, family size, education, participation in UA-related training, and market information, while it is negatively and significantly influenced by the household head's age, occupation, and dependence ratio. According to the second stage of the DH (log-truncated) Model, years of schooling, land size, participation in UA-related training, and market information had a positive and significant influence on market participation intensity.

Table 3 Determinants of status and intensity of UA output market participation
Full size table

The age of the household head negatively and significantly influenced the participation of urban farmers in the output market at a 5% probability level. As the age of the household head increases a year, the likelihood of UA output market participation reduces by 0.9%, indicating that older urban farmers are less likely to participate in a market than younger counterparts. This is because old household heads might reduce trust in implementing UA practices, which need financial input and intensive management. Additionally, the associated less likely market participation may be due to the incapacity of old-aged households to acquire information on inputs and marketing. A similar argument was made by Andaregie et al. [12] and Megerssa et al. [13].

The household head years of schooling have positively and significantly affected both market participation decision and the intensity of UA output market participation at a 5% and 1% significance level, respectively. The positive and significant association indicates that as the household head years of schooling (education) increases, the likelihood of UA output market participation and intensity supply also increases. This finding indicated that as the head of the household level of education rises by one year, the amount of UA output supplied to the market increases by 0.06%, according to this study. This finding revealed that educated heads of households were better suppliers in the market because enlightened urban farmers have better experience and understanding, making it easier to grasp market information. Additionally, educated households may collect not just marketing data, but also information on agricultural input availability, allowing them to boost production and sell a large volume of produce. This result is consistent with the findings of Modibedi et al. [49] and Megerssa et al. [13].

The coefficient of family size positively and significantly influences participation in the UA output market at a 1% level. For every additional family member, the likelihood that a household will participate in the UA output market increases by 6.5%. As a result, the probability of practicing UA increases with the size of the family since larger families are assumed to have more workers, have resource constraints in meeting basic needs, and seek alternative income sources like UA. Having a workforce in the house typically motivates people to try out various agricultural practices, such as vegetable and livestock rearing. Therefore, it seems reasonable that families having more family members can reduce their cost of production as UA practices are labor-intensive, particularly livestock rearing, and hence participate in the output market. A similar result was reported by Guta and Irge [50] and Githungunyi [51].

The study indicates land size positively and significantly affected the intensity of market participation at a 5% level, implying that households with larger land sizes supply more to the market. An increase in the land size by one hectare increases the intensity of market participation by 0.49%. The result shows that households with a larger land size are more likely to diversify their production into UA practices and have a higher possibility of producing more output, thus selling the surpluses. In fact, according to IFPRI, most agricultural production increases in Africa have been due to area expansion and not productivity growth [52]. Municipal authorities should, therefore, develop ways to persuade households to acquire land for UA practices, resulting in market participation and food security. The findings are congruent with those of Shitaye [53] and Kassa et al. [26], who found that land size played a role in enhancing output and market participation.

Farming experience appears to have a significant and positive impact on UA output market participation at the 10% significance level. Revealing an increase in the farming experience of household heads by a year is more likely to increase the probability of participating in the UA output market by 1.2%. Accordingly, market participation would increase with the number of years the households have been participating in farming. Better farming performance ideally comes with experience, skills, and knowledge [54]. The results of this study are consistent with [55] finding that farmers’ experience in production is positively and statistically significant in influencing market participation.

Training in any UA-related activities affects both the decision to participate in the UA output market and the level of participation. It is one of the extension activities where households can learn practical skills and in-depth knowledge about contemporary technology. Attending UA-related training improves farmers' comprehension, which enhances production. Household heads who obtained UA-related training can produce more and participate more in the market. UA practices and intensity are positively and significantly impacted by training at the 1% and 5% significance levels, respectively. Acquiring training in any UA practice enhances market participation likelihood by 66.2% and intensity supply by 0.45%, indicating that better information availability is more likely to increase UA output and better market participation. This finding is in harmony with the findings of Dey and Singh [56] and Bannor et al. [57].

Market information influences UA output market participation positively and significantly at a 1% probability level and intensity of market participation at a 5% level. Urban farmers who have market information are more likely to participate in the UA output market and offer their products in response to a changing market situation. Market information is a crucial tool in marketing since it educates urban farmers about market circumstances. Access to market information increases the probability of UA output market participation by 77.1% and marketed supply by 0.62%. Therefore, urban farmers who have price information before marketing tend to sell more of their products than those without. This shows that increased access to information is a vital component of efforts to increase participation in the UA output market. This agrees with the conclusions of Abate and Addis [11] and Megerssa et al. [13]. Goetz [58] further pointed out that higher market information considerably increases the likelihood of market participation for potential selling households.

The variable dependency ratio negatively impacts the UA output market participation, and it suggests that each additional dependent in the household decreases the likelihood of household market participation by 35.2% and is significant at a 1% significance level. A higher dependency ratio indicates that the dependents do not contribute to farm labor, reducing excess production and thus the household's capacity to earn additional income through market activity. Urban farmers with a higher proportion of dependents consume more agricultural produce, limiting marketable surplus. This study agrees with the findings of Lighton [59] and Abrha et al. [60].

Finally, the coefficient of occupation has a detrimental impact on the likelihood of UA output market participation and is statistically significant at a 5% significance level. Being a civil servant has negatively and significantly impacted the probability of UA output market participation decision. The result showed that being a civil servant lowered the probability of market participation by 29.5%. This could be because civil servants might not have enough time to produce more surpluses for the market. This result is consistent with the findings of [22].

7 Conclusion and recommendation

The study sought to determine factors influencing UA output market participation decisions and the extent of participation among smallholder urban farmers in eastern Ethiopia, using DH modeling, which considers possible interdependence among their actions as individuals. The descriptive statistics result revealed that of the total sample households, 131 (48.52%) participated in the UA output market, while the remaining 139 (51.48%) were non-participants.

The first stage of DH Model results shows that factors such as education, family size, participation in UA-related training, farming experience, and market information had a positive and significant influence on UA output market participation decisions however, variables like age, occupation, and dependency ratio had negative and significance impact on UA output market participation decision of urban farmers.

The second stage result revealed that factors such as education (measured in years of schooling), land size, UA-related training, and market information had positive and significant impacts on the intensity of the marketed supply of UA output market participation. In general, the main impediments influencing UA output market participation and the intensity of marketed supply in the studied areas are a lack of land, limited access to information, a lack of improved livestock breeds, and inadequate training arrangements.

Policy intervention efforts geared towards improving household participation in output markets must primarily focus on organizing and providing practical UA-related training, providing market-related information, and working on the productivity of livestock breeds, as these factors affect market participation. Appropriate information, education, and training for urban farmers are necessary to ensure their understanding and contribution to good nutrition and health, and hence market participation. A continual threat to UA is the loss of agricultural land in urban and peri-urban areas. Therefore, a land policy that opens up access to marginal land in urban areas for UA as a means of empowering poor urban farmers may be required to ease land constraints and encourage market participation. Besides, since the dependency ratio affects market participation negatively and significantly, the government should give attention to population policy and improve households' awareness of risk prevention rather than fearing risk to promote UA practice and reduce urban food insecurity.