Analysis of maize commercialization among smallholder farmers: empirical evidence from North Western Ethiopia

Commercialization in smallholder farming is very important for low income countries since it has a potential to enhance incomes and play a key role in reducing rural poverty. Even though few studies are conducted about agricultural commercialization in Ethiopia, so far there is no study conducted on maize commercialization among smallholder farmers. Therefore, the aim of this study was to analyze the factors that determine market participation and degree of commercialization among smallholder maize producers in North Western Ethiopia. The study was carried out from October 2019 to December 2020. Data were collected from 385 smallholder maize producers using systematic random sampling. Interview schedule, focus group discussion and key informant interview were used to collect the required data. Tobit model was employed to analyse both market participation and intensity of commercialization. From the analysis commercialization was significantly influenced by education level, livestock holding, frequency of extension contact, training, off/non-farm income activity, quantity of maize and lagged price. Based on the findings, smallholder maize producers should be supported regularly by extension agents in order to increase their practical skills which results enhancement of their market participation and intensity of commercialization.


Introduction
In many Asian countries agricultural commercialization of both traditional and high-value crops is mainly achieved through smallholder farmers which plays significant role in bringing rural development and poverty reduction. Investments in greater access to markets, roads and technology contributes substantially to expand agricultural commercialization (Cazzuffi et al. 2020). Agriculture is the back bone of the economies of developing countries since the sector is the means to achieve food security, gaining foreign currency from export, and bringing rural development. However, in Sub-Saharan Africa agricultural productivity has continued to decline short of expectation. Thus, to meet the increased demand, agricultural output would need to more than double by 2050 in Sub-Saharan Africa, but about one-third above from current levels is projected for the rest of the world (Etuk & Ayuk 2021).
The contribution of agriculture is crucial for the Ethiopian economy like other Sub-Saharan African countries. It is a means of livelihood for about 84% of the population and it constitutes about 33.3% of the country's GDP (NBE 2018). The majority of farmers in Ethiopia are smallholders and they are a source of 95% of the country's agricultural production (CSA 2018). Enhancing the productivity of smallholder farmers has been the primary goal of the government in order to foster the economic growth in Ethiopia.
Despite the fact that agriculture is showing a remarkable progress in Ethiopia, it is not yet developed to the expected level. The sector is still characterized by low productivity, low employment of agricultural technologies, and subsistence oriented smallholder farming (Doss et al. 2003;Shita et al. 2018). In one or another way strategies and policies designed to bring economic growth in the country such as Agriculture Development Led Industrialization and Growth and Transformation Plan has been mainly focusing on agricultural development through shifting the current smallholder subsistence based farming to commercialized agriculture (Gebreselassie 2006;MOFED 2006).
Agricultural commercialization is a process of transformation from subsistence farming system to market oriented production system (Alemu et al. 2006). Commercialization in smallholder farming is very important for low income countries since it has a potential to enhance incomes and play a key role in reducing rural poverty level (Awotide et al. 2016;Hailu et al. 2015;Osmani et al. 2014). Agricultural commercialization particularly grain crops are more subsistence than cash crops in Ethiopia due to market imperfections, lack of capital, lack of market accessibility and high transaction costs (Getahun 2020;Hagos & Geta 2016;Senbeta 2018).
In Ethiopia, grain crops are the primary source of food in addressing food security and are ranked first in area coverage of the total cultivated lands, which is 15,270,526 hectares (CSA 2018). Particularly, maize is one of the most commonly available cereals and source of staple foods in different parts of the country (Legesse et al. 2018;Matthews et al. 2015;Nigussie et al. 2001). However, over half of Ethiopia's smallholder farmers grow maize mostly for subsistence 1 3 Analysis of maize commercialization among smallholder farmers:… purposes and about three-fourth of the output is consumed at the household level (IFPRI 2010). There are socio-economic, institutional and demographic factors which contribute for the low level of agricultural commercialization by smallholders in Ethiopia (Abafita et al. 2016;Alemu et al. 2006).
Even though there are few studies conducted on agricultural commercialization in Ethiopia (Abafita et al. 2016;Abadi, 2014;Getahun 2020;Hailu et al. 2015) so far there is no study conducted maize commercialization among smallholder farmers. Therefore, this study was conducted to identify factors that determine market participation and level of commercialization among smallholder maize producers in North Western Ethiopia.

Determinants of commercialization
Commercialization was measured using commercialization index which is computed as a ratio of quantity of maize supplied to market by a particular household in the specified year to the quantity of maize produced by the same household in the same year. The commercialization index measured in this way was used to measure the intensity of maize commercialization (Abafita et al. 2016;Makombe et al. 2017). Market participation is a dummy variable which was measured whether or not maize producer household supplied to market. Overall, the probability of smallholder maize producers to sell their maize in the output markets was 76.3%. All variables that were found to influence the participation decision and degree of commercialization do not have similar contribution for influencing the participation decision and degree of commercialization. Hence, using a decomposition procedure suggested by (Moffitt & Mcdonald 1980), the marginal effect results of Tobit model was used to assess the effects of changes in the explanatory variables into participation decision and intensity. The marginal effects were computed for the dependent variable conditional on the censoring and on the unconditional expected value of the dependent variable. Therefore, the effects of each significant explanatory variables which affects smallholder maize producers' probability of maize market participation and degree of commercialization is discussed as follows based on the marginal effect results presented in Table 2.

Education level
The variable education level is a continuous variable measured using a grade of formal schooling which had positively influence the probability of market participation and degree of commercialization of maize at 1% level of significance. This indicates that household who were more educated had better market participation and high degree of commercialization. The positive relationship could be due to the fact that educated people can more easily contribute to the generation of new technologies and more readily utilize those technologies (Derso et al. 2016). Furthermore, educated people manage their fields properly and then this activity results have pushes 1 3 to get good production and productivity of the land. Educated members are expected to have more exposure to the external environment and familiar with their duties and rights they have in different social activities and need to actively participate in economic and democratic right to take right decision. Education increases human capital which enhances the farmer's ability to adopt new agricultural technology which in turn leads to high degree of commercialization. The results of the econometric model indicated that, an increase in the education level of households by one grade increase the probability of farmer's market participation and the expected level of commercialization of maize producing farmers by 2.2% and 0.01 units, respectively. Moreover, the education level of households increases by one grade the unconditional expected value of maize commercialization increases by 0.01 units. This result is in line with the findings of Awotide et al. (2016) which is analyzed by Heckman two stage model and confirmed that level of education has positive and statistically significant effect on market participation of farmers in rice marketing.

Livestock holding
This variable was a continuous variable measured in Tropical Livestock Unit (TLU) was found to have negatively and statistically significant at 10% level of significance on the probability of smallholder maize producer market participation as well as degree of commercialization. The negative relationship could imply that as the households' have more livestock endowment, their market participation and degree of commercialization decreases. The possible reason might be to purchase farm inputs which can enhance production and productivity like fertilizer, improved seed, pesticides and insecticides, farmers directly sell their livestock and store their maize output. The marginal effect of this variable revealed that, as the number of livestock increases by one TLU from the mean, the probability of farmer's market participation and the expected level of commercialization of maize producing farmers decreased by 1% and 0.04 units, respectively keeping other variables constant at their mean value. Moreover, as the number of livestock increases by one TLU from the mean, the unconditional expected value of maize commercialization decreases by 0.06 units. This finding contradicts with the findings of Abafita et al. (2016). In their study on smallholder cereal farmers' commercialization in Ethiopia by using Heckman two stage models, Ox that is a proxy for total livestock holding had positive effect on probability of participation on cereal marketing.

Frequency of extension contact
It is obvious that agricultural extension services play a vital role in motivating farmers towards accepting and implementing improved agricultural technologies and agronomic practices. However, the result of this study shows that frequency of extension contact negatively and significantly influence the probability of maize market participation and degree of commercialization at 5% level of significance. This might be smallholder maize producers who have frequent contact with development agent could not get practical information on new technologies and agronomic practices which might boost their production and productivity of maize. Instead development agents out of their profession, might spent their time with farmers talking about politics and other issues which is not directly relevant to enhance farmers' production and productivity. The marginal effect shows as the smallholder maize producers increase extension contact by one time in a year from the mean, the probability of farmer's market participation and the expected level of commercialization of maize producing farmers decreased by 0.2% and 0.01 units, respectively keeping other variables constant at their mean value. Moreover, as the smallholder maize producers increase extension contact by one time in a year from the mean, the unconditional expected value of maize commercialization decreases by 0.01 units. Negative but statistically significant effect of extension service on market participation and commercialization level had been reported in some other African countries such as in rural Nigeria (Awotide et al. 2016) and in Ghana (Martey et al. 2012).

Training
Training was found to have positive and statistically significant influence on both the probability of maize market participation and degree of commercialization at 5% significance level. Trainings on application of new agricultural technologies, agronomic practices, harvest and post-harvest loss minimization and other related trainings could build smallholder farmers' production capacity. Ultimately, it increases the likelihood of maize market participation and degree of commercialization for producers. The marginal effect revealed that smallholder maize producer who got agricultural training had 12.5% more probability for market participation and 0.05 unit more expected level of maize commercialization compared to those who didn't took training keeping other variables constant at their mean value. Moreover, smallholder maize producers who got agricultural training had 0.08 unit more unconditional expected value of maize commercialization compared to producers who didn't took training. In line with our finding a study conducted in the northern part of Ethiopia, Tigray Region, confirmed that training on crop marketing has a positive and significant effect on intensity of crop commercialization (Hailu et al. 2015).

Off/non-farm income activity
This variable was measured in terms of whether or not respondents get additional income from off/non-farm income beyond their own agricultural activity. Off/nonfarm income activity had positively and statistically significant influence at 1% level of significance on the probability of market participation and degree of commercialization. The positive relationship could be because of farmers who have got additional income from off/non-farm activities might not face financial shortage to purchase farm inputs to increase their maize production and productivity which ultimately increases their market participation and degree of commercialization. This variable could be interpreted as farmers who were engaged in off/non-farm activities had 9.4% and 0.04 unit more probability of market participation and expected level of maize commercialization respectively compared to farmers who didn't get any income from off/non-farm income activities keeping all other variables constant at their mean value. Furthermore, farmers who have extra income from off/non-farm 1 3 income activities beyond their farm activities had 0.06 unit more unconditional expected value of maize commercialization compared their counter parts. This result was in line with the findings of Hailu et al. (2015) which states that off-farm income is the driving force of increased crop commercialization. In addition, Matthews et al. (2015) confirmed that the direct effect of off/non-farm income in enabling smallholder farmers to be technical efficient in maize farming in Ethiopia. This might increase the production level and market participation of the farmers. Contradicting to this result, off/non-farm income had shown significant negative influence on farmers' market participation was reported by Awotide et al. (2016) and commercialization level by (Martey et al. 2012).

Quantity of maize produced (in quintal)
this variable was found to have positive and statistically significant influence on the probability of participation in maize marketing as well as degree of commercialization at 5% level of significance. As the evidence obtained from sample respondents, maize producers who produced more had better chance to participate in maize marketing and supply high amount of maize in to the market. The results of the econometric model indicated that, as the quantity of maize production increased by one quintal from the mean, the probability of farmer's market participation and the expected level of commercialization of maize producing farmers could increase by 5.8% and 0.02 units, respectively keeping other variables constant at their mean value. Furthermore, as the quantity of maize production is increased by one quintal from the mean, the unconditional expected value of maize commercialization increases by 0.03units. A previous study conducted in Ethiopia has shown a significant positive effect of value of crop produced on the probability of market participation and the level of commercialization by smallholder cereal farmers (Abafita et al. 2016). In addition, the study conducted in rural Nigeria confirmed our result and it indicates that the positive and statistically significant effect of yield of rice on farmers' rice market participation and welfare maximization (Awotide et al. 2016).

Lagged price
which was measured in Ethiopian birr had positive and statistically significant relationship with probability of maize market participation and degree of commercialization at 1% level of significance. This was due to the fact that lagged price of maize was high. High price level of the output in the previous year could motivate smallholder maize farmers to produce more in the form of allocating more land and use of appropriate agricultural technologies and this could increase their market participation and degree of commercialization. The results of the econometric model indicated that, as lagged price increases by one Ethiopian birr per kilogram of maize from the mean, the probability of farmer's market participation and the expected level of commercialization of maize producing farmers increased by 8.4% and 0.04units, respectively keeping other factors constant at their mean value. Furthermore, as lagged price increases by one Ethiopian birr per kilogram of maize from the mean, the unconditional expected value of maize commercialization increases 1 3 Analysis of maize commercialization among smallholder farmers:… by 0.05units. In line with this result, in Ghana the significant effect of unit of price output on intensity of commercialization was documented in the study by Martey et al. (2012).

Description of study areas
This study was conducted in North Western Ethiopia in two districts namely Dembia and Takusa which represents different agro ecological zones. In both districts, mixed farming systems (livestock rearing and crop productions) is widely adopted agricultural practices. The crop production systems are highly relied on rain with rare irrigation technology. Based on information from the zonal agriculture department, the two districts are potential in maize production compared to other districts of the Zone. Figure 1 shows map of the study areas.

Sampling technique and sample size
A combination of different sampling procedures was used to select the units of analysis. The sampling frame of the study was the list of households in the selected kebeles, 1 in the respective districts of Dembia and Takusa districts, who are engaged in maize production. Households are the unit of analysis for this study and a multistage sampling procedure was employed to select respondents. In the first stage, Dembia and Takusa districts were selected purposively based on their maize production potential. In the second stage from the selected districts, ten maize producing kebelles were selected using simple random sampling technique. In the third stage, households in the selected kebelles were stratified in to maize producers and nonproducers. Finally, 385 smallholder maize farmers were selected using systematic random sampling technique by taking in to account the proportion of number of maize producers in each kebelle in the corresponding districts.
To obtain a representative sample size, for cross-sectional household survey the study employed the sample size determination formula given by Kothari (2004): where; n = Sample size; Z = confidence level (α = 0.05, hence, Z = 1.96); p = proportion of the population containing the major interest, q = 1-p and e = allowable error.

Data source and data collection method
Both primary and secondary sources of data were used. Before the data collection, the questionnaire was pre-tested on selected farmers to evaluate the appropriateness of the design, clarity, and relevance of the questions. Having the appropriate modification on the questionnaire based on the pre-tested result three enumerators from each districts were recruited and trained about the content of the questionnaire and interviewing process. Primary data were collected using interview schedule, focus group discussion and key informant interview whereas secondary data were also collected to supplement the primary data from published and unpublished sources.

Method of data analysis
Two types of analysis, namely: descriptive and econometric analyses were used for analyzing the data. Descriptive method of data analysis refers to the use of ratios, percentages and means in the process of examining and describing household characteristics. To analyse the factors that determine market participation and intensity of commercialization Tobit model was employed. Intensity of commercialization was measured as the ratio of the percentage of marketed output to total production. It is necessary to show n = Z 2 pq e 2 = 1.96 2 × 0.5 × 0.5 0.05 2 = 385 1 3 Analysis of maize commercialization among smallholder farmers:… the decision of smallholder maize farmers' market participation in order to estimate the degree of commercialization. The dependent variable, decision to commercialize and intensity of commercialization, in this case has an upper limit of one in all cases and a lower limit of zero. The rationale for this is to match farmers' decision to fit the Tobit model that cannot take dependent variables greater than one or a negative.
It is assumed that both the decision to market participation and intensity of commercialization are influenced by the same variables that increase the probability of market participation and also increase the intensity of commercialization (Moffitt & Mcdonald 1980).
The Tobit or censored normal regression model assumes that the observed dependent variables Yi for observations i = 1… n must satisfy: where Y * i represents the latent variable generated by the classical linear regression model. The Tobit model used to estimate the factors that influence the intensity of commercialization is specified as follows: The Tobit model does not correspond directly to changes brought about by changes in the independent parameters but rather the direction of the effect. The marginal effect of the changes in an explanatory variable on the intensity of maize commercialization is given as follows (Greene 2003): From the above, the empirical Tobit model estimated for the factors likely to affect the intensity of commercialization for maize (Y maize ) is given as: where X i represents a vector of exogenous explanatory variables, i represents the estimated maximum likelihood parameters; i represents the captured random influence on the relationship which is assumed to be normally distributed with mean zero and variance.
Intensity of commercialization measured as the ratio of the gross marketed output to the total production at household level. Explanatory variables included in the analysis that are assumed to determine the level of commercialization are described in Table 1.

Socio-economic characteristics of respondents
Sample respondents were composed of both female and male household heads. Among the total sample respondents, about 5% were female headed households and about 95% were male headed households. In terms of participation in maize market, 76% of household heads were participant while 24% of the households were not participant. The average age of the respondent households were 46 years with a range of 25 years minimum and maximum of 76 years. The average quantity of maize annually produced was 15.27 quintal while the minimum was one Quintal and 54quintal was the maximum. The average amount of maize supplied to market was 3.52 quintal and the minimum was supplied nothing and the maximum was 35 quintal. The extent of maize commercialization was 23%. The average livestock holding of the respondents in terms of tropical livestock holding was about 6 with minimum of having zero and maximum of about 26 TLU.

Estimation results of econometric model
In this study, we employed Tobit model to estimate both determinants of smallholder maize producers' market participation and intensity of commercialization Analysis of maize commercialization among smallholder farmers:… simultaneously. Accordingly, the model result showed that seven variables were found to be significantly creating variation on the probability of smallholder maize producers' market participation and degree of commercialization. These socio-economic variables are; education level, livestock holding, frequency of extension contact, training, off/non-farm income activity, quantity of maize, lagged price. The change in probability of market participation, degree of commercialization and unconditional expected value of commercialization for the explanatory variables is presented in Table 2.

Conclusions
The general objective of this study was investigation of maize commercialization among smallholder farmers with a specific objectives of analyzing factors determining market participation and intensity of commercialization among smallholder maize producers. Tobit model was employed to investigate both maize market participation and intensity of participation for smallholder maize producers. The marginal effect of the Tobit model indicated that education level of household head, attending training, getting income from off/non-farm income activities, quantity of maize produced and lagged price had positive and statistically significant influence on both the probability of smallholder maize producers market participation and intensity of market participation. However, total livestock holding and frequency of extension contact were found to have negatively and statistically significant effect on both probability of maize producers market participation and intensity of commercialization. The findings of this study revealed that about three-quarters of respondents involved in maize commercialization which is promising but needs collaborative effect to promote commercialization more than this figure. Based on the findings, it is recommended that socio-economic characteristics of respondents should be taken in to account for designing new policies and interventions regarding agricultural commercialization in the study area. Furthermore, extension agents should provide practical and professional advices to farm households in order to enhance their production which in turn increase their probability of maize market participation and intensity of commercialization. Finally, to capture the full understanding of commercialization, future researchers need to replicate our analysis using paned data.
Data availability All authors declare that the datasets used in this manuscript are fully available upon request from the corresponding author.

Declarations
Competing interests The authors declare that they have no any competing interests in this manuscript.