Determinants of teff commercialization among smallholder farmers: Beta regression approach

Abstract In Ethiopia, agricultural commercialization is not well developed. Smallholder farmers with a subsistence farming system dominate crop production, resulting in incompetent and less commercialized produce. As a result, producing market-oriented products can increase the well-being of smallholder farmers. Accordingly, we conducted this study to analyze determinants of teff commercialization among smallholder farmers in Hulet Eju Enese Woreda, Ethiopia. The primary data were collected from 384 randomly selected smallholder farmers to measure the level of teff commercialization and analyze determinants of teff commercialization among smallholder farmers. To address the objectives of this study, an output commercialization index, and a beta regression model were used. The findings show that about 77.2% of smallholder farmers are classified as commercial, while semicommercial farmers account for 22.8% of all observations. Furthermore, the model results revealed that the number of oxen, teff land size, farming experience in teff production, market distance, and agroecology had statistically significant effects on teff commercialization. Therefore, sources of improved traction power, land productivity, market infrastructure, experience-sharing strategies, and new varieties that can adapt to varied agroecology should be given special priority to increase smallholder farmers’ commercialization.


PUBLIC INTEREST STATEMENT
Teff is an important commercial crop in the development of the agricultural sector. It is one of the cash crops mostly cultivated by smallholder farmers. For instance, in terms of area coverage, teff takes the lions share but in terms of production and productivity it is below the other cereal crops. Teff market demand is increasing through time but it is not fully commercialized due to smallholders' production dominance and their subsistence farming systems. Accordingly, this study focused on why smallholder farmers are unable to fully commercialize teff production. The findings show that 77.2% and 22.8% of smallholder farmers are classified as commercial and semicommercial. Moreover, the number of oxen, teff-allotted acreage, farming experience in teff production, market distance, and agroecology had statistically significant effects on teff commercialization Therefore, the intervention strategies should target significant factors to improve smallholder farmers teff commercialization.

Introduction
Agriculture contributes 33.3% of Ethiopian gross domestic product (GDP) and 72.7% of employment (ATA, 2018;Getahun, 2020;Senbeta, 2018). This motivates the designing of polices which targeting the sector (Bachewe et al., 2015). Thus, to develop the agricultural sector, the country introduced agricultural development lead industrialization (ADLI) policy which its central aim is bringing commercialized agricultural development (Mazengia, 2016). According to ATA (2018), shifting smallholder farmers from subsistence-based production towards market-based production remains the most important aspect of agricultural transformation. As commercial transformation is a vital to improve the well-being of the farmers (Addisu et al., 2019), Ethiopia following agricultural commercialization strategy through the practice of commercialization cluster (Louhichi et al., 2019;Senbeta, 2018). Such agricultural commercialization among poor smallholder farmers is an important mechanism to reduce poverty, improve household food security and foster sustainable development goals of the country (Carlettoa et al., 2017;Gebreselassie & Sharp, 2007;Murichoa et al., 2017;Ogutu & Qaim, 2019;Tirkaso, 2013).
Furthermore, the ever increasing population at alarming rate and demand for agricultural produce requires a more commercialized pattern of production (Hailua et al., 2015). However, commercialization of agriculture in Ethiopia is still poor (Boka, 2017;Leta, 2018). In the country, smallholder agriculture has been the backbone for centuries in which large part of the commodity is provided by the smallholder farmers (Beyene, 2018). According to Getahun (2020) and Taffesse et al. (2012), smallholder farmers account 96% of the total area cultivated and contribute about 90% of the total agricultural output. But crop production is still tremendously a subsistence nature (Matouš et al., 2013). Thus, they produced low yielding output (ATA, 2018) which is incompetent to international market standards (USAID, 2015); their market participation is limited; agricultural markets are fragmented; even it is not well integrated into wider market systems both in input and output sides (Mitiku, 2014). For sustainable way continuing in such subsistence farming system may not be effective to attain development goals of the country (Pingali, 1997). Therefore, commercialization of the smallholder agriculture is an important pathway for sustainable and best economic transition to improve smallholder farmers' welfare (Beyene, 2018;Ogutu & Qaim, 2019;Senbeta, 2018).
In Ethiopia, crop production makes up 72% of the total agricultural gross domestic product (ATA, 2018) and more than 80% of the population earns their livelihood from crop production (Duguma et al., 2012;Hagos et al., 2009). But crop production is still tremendously a subsistence nature (Matouš et al., 2013). (Boka, 2017;Leta, 2018). In the country, smallholder agriculture has been the backbone for centuries in which large part of the commodity is provided by the smallholder farmers (Beyene, 2018). According to Getahun (2020) and Taffesse et al. (2012), smallholder farmers account 96% of the total area cultivated and contribute about 90% of the total agricultural output. The study by Wana and Sori (2020) reported that the average growth rate of major crops was 0.3% per annum though the population was grown by 3%. Thus, the traditional production system is not sufficient to sustain the ever-increasing population with food and energy demands (Dessale 2019). Consequently, output growth is a viable solution to feed the population growing at an alarming rate.
Specially, wheat, maize, sorghum, barley and teff are the major cereals that covers the biggest share of crop production (ATA, 2018;Louhichi et al., 2019;Taffesse et al., 2012). But crop production is still tremendously a subsistence nature (Matouš et al., 2013). Thus, they produced low yielding output (ATA, 2018) which is incompetent to international market standards (USAID, 2015). Teff is a major cash crop for both rural and urban households (Assefa et al., 2015;B. Minten et al., 2013;Gebreselassie & Sharp, 2007) which is the source of farm income and food security status (Demeke & DiMarcantonio, 2013;Habtewold & Challa, 2017). Thus, Ethiopia is the only teff producer country to have adopted it as stable crop (Lee, 2018). Due to its high and unique nutritional value (Lee, 2018) and cultural preferences (Demeke & DiMarcantonio, 2013), its demand is increasing specially in urban areas and consumed by about 70% of Ethiopian population in the form of flat bread food called injera (B. Minten et al., 2018;Tesfaye et al., 2015). It accounts the largest share of total cultivated area among cereals but in terms of total production, teff takes the second rank next to maize (CSA, 2019).
According to Cochranea and Bekele (2018), the productivity of teff is increasing through time while teff market is characterized by small scale operations with limited scale of economies (B. Minten et al., 2013;Demeke & DiMarcantonio, 2013) which leads to low level of teff commercialization (Orr et al., 2020). Study by Osmani and Hossain (2015) indicated that several demographic, socio-economic and institutional factors hinder the development of teff commercialization. To identify factors affecting teff commercialization, researchers conducted various studies in different parts of Ethiopia. Yet, previously conducted research varies with the area of their study and the methodology they used. For instance Gebreselassie and Sharp (2007), analyzed the amount of teff commercialized by smallholder farmers using two stage least square estimation (2SLS) method. On the other hand, Medhane (2014) identified and analyzed factors which determined the extent of smallholders farmers' participation in teff output market by applying ordinary least square estimation econometric model. Similarly, Addisu et al. (2019) and Leta (2018) examined determinants of commercialization of teff using tobit and Heckman two stage econometric model, respectively. For instance, Gidelew et al. (2022) examined determinants of farmers' teff commercialization level using double hurdle model. However, the double hurdle model has been used to examine market participation and volume of supply in different studies but not the level of teff commercialization. When the dependent variable is explained in the form of an index like teff commercialization, fractional outcome regression models (fractional response and beta regression models) are applicable (Baum, 2008;Ferrari, 2013). Thus, beta regression and fractional response logit/probit models are the right models to analyze determinants of teff commercialization level in particular and crop commercialization in general (Aktaş & Unlu, 2017;Hunger et al., 2011Hunger et al., , 2012Pullenayegum et al., 2010). Therefore, the econometric model applied by Gidelew et al. (2022) is not appropriate to analyze determinants of teff commercialization level. Moreover, the model specification and findings of Gidelew et al. (2022) focused on market participation decision and volume of supply. This shows that the authors did not use commercialization index as a dependent variable. Therefore, the previous studies have methodological gaps to address teff and other crops commercialization. Even though the beta regression model has been used in different discipline to analyze indexed dependent variables (Aktaş & Unlu, 2017;Hunger et al., 2011Hunger et al., , 2012Pullenayegum et al., 2010), it is rarely used in the crop commercialization studies. As a result, this study applied an output commercialization index and a beta regression model to measure smallholder farmers' level of teff commercialization and to analyze determinants of teff commercialization among smallholder farmers in Hulet Eju Enese woreda, Ethiopia.

Description of the study area
The research was carried out in Ethiopia's Hulet Eju Enese woreda, which is located southeast of Bahir Dar. It is bordered by Debay Telatgen in the south, Bibugn in the west, the Abay River in the north, and Goncha Siso Enese in the east (Figure 1). It is situated between 10°5′N and 37°52′ E in terms of latitude and longitude, respectively (HEEWAO 2017). The elevation of the woreda is 2,487 meters above sea level. The research area's agro-ecologies are Woina dega (mid highland) and Kola (low land). The area's rainfall distribution varies by season from year to year (Assefa et al., 2016), with annual rainfall ranging from 900 to 1200 mm and lowest and maximum temperature records ranging from 14 to 16 and 24 to 28 °C, respectively (HEEWAO 2017).
Agriculture provides a livelihood for more than 98% of the population. They are involved in livestock and crop cultivation (Assefa et al., 2016). Teff is grown by the majority of smallholder farmers because its market price is two to three times that of other cereals (Demeke & DiMarcantonio, 2013). Farmers are encouraged to produce more teff because of the high market price (Louhichi et al., 2019). Hulet Eju Enese is also one of the top seven teff-producing districts in the east Gojam zone (Warner et al., 2015).

Methods of data collection
A multi-stage sampling technique will be applied to sample teff producer smallholder farmers. In the forst stage, Hulet Eju Enesie is selected thanks to the following reasons. The first reason is availability of teff cluster farming practice since 2017/18 production season. The next reason is the woreda has huge teff production potential in Amhara region, Ethiopia. Next to this, relevance of teff cluster farming scale up to other areas and crops to improve crop production and productivity and market access. The other reason is teff cluster farming practice was not documented before. The woreda has 29 teff producer kebeles and 25,508 teff producer smallholder farmers. The teff production potential varies from one kebele to another due to agroecology variation. In the second stage, the 29 teff producer kebeles will be stratified into two strata based on their agroecology, namely mid highland (17 kebeles) and low land (12 kebeles). In the third stage, sample kebeles will be randomly selected from each stratum. Then, representative sample respondents will be drawn from each kebele considering population proportion to sample size. Finally, systematic sampling technique will be applied to select each sample respondent.
Primary sources were used to gather qualitative and quantitative data. Face-to-face interviews, focus group discussions, and key informant interviews were used to collect primary data from selected smallholder teff producer households. The face-to-face interview was conducted using a well-prepared semi-structured questionnaire. The draft questionnaire was designed based on the inputs of focus group discussions and key informant interviews. Then after, pretest survey was undertaken to customize the draft questionnaire under a local context. Thus, a pretest survey was undertaken to check the reliability and validity of the drafted survey questionnaire. Accordingly, the draft questionnaire was revised based on the pre-test survey result. Finally, well trained, and experienced enumerators were employed to collect the required data for this study.

Measurement of teff commercialization
Commercialization deals with the shift of production from consumption to more market-oriented production system (Abafita et al., 2016). Similarly, it refers to a significant proportion of specific commodity in output market (Medhane, 2014). This indicates measurement of the farmers' teff commercialization tells us the amount which the given sample smallholder farmer commercialized. Teff output commercialization index was calculated to measure the level of smallholder farmers' participation in the output market. Finally, teff commercialization index was calculated by applying output commercialization index formula employed by previous studies (Addisu et al., 2019;Boka, 2017;Carlettoa et al., 2017;Dube & Guveya, 2016;Falola et al., 2017;Fikadu et al., 2020;Gari, 2017;Mohammeda et al., 2016).

Teff commercialization index ¼
value of teff sold in the market value of total teff produced (1)

Statistical analysis
Statistical software (SATA version 14) was utilized to perform descriptive and inferential statistics. Percentage, frequency, minimum, maximum, mean, and standard deviation were employed as descriptive statistics. Furthermore, the chi-square test was used to determine the relationship between categorical variables and teff commercialization, while the independent sample t-test was used to determine the mean difference between continuous variables and teff commercialization.

Econometric model
Researchers applied a variety of econometric models to examine crop commercialization as a function of individual characteristics.  Gidelew et al., 2022;Hailua et al., 2015;Kabiti et al., 2016;Leta, 2018;Mamo et al., 2017;Mohammeda et al., 2016). Multiple linear regression, for example, assumes that the error term is homoscedastic and normally distributed, however this assumption may not work for bounded dependent variables (Hunger et al., 2011;Pullenayegum et al., 2010). Moreover, the obit model is not appropriate for analyzing the teff commercialization index because teff commercialization index scores cannot exceed the fully commercialization level (Gallani et al., 2015). As a result, these econometric models are inappropriate when the dependent variable is restricted to an interval [0, 1] (Baum, 2008;Ferrari & Cribari-Neto, 2004;Gallani et al., 2015). Moreover, the double hurdle model has been used in previous studies to examine market participation and volume of supply. However, no studies have used this approach to examine the level of commercialization.
As a result, if the dependent variable is explained in the form of index, fractional outcome regression models (fractional response models and beta regression) are applicable (Baum, 2008;Ferrari, 2013). The beta regression model and fractional response probit models capture the nonlinear relationship between the fractional outcome variable and exogenous variables. Furthermore, the model was used to avoid model misspecification and dubious statistical validity. For example, unlike the beta regression model, fractional response probit model is appropriate when the outcome variable includes zero, values between zero and one, and itself i.e. 0 � y � 1 (Ferrari, 2013;Papke & Wooldridge, 2008). The outcome variable for this study is the Teff commercialization index score, which ranges from 0.26 to 0.85. Even though the beta regression model has been used in a different discipline to address these issues (Aktaş & Unlu, 2017;Hunger et al., 2011Hunger et al., , 2012Pullenayegum et al., 2010), it is rarely used in relation to crop commercialization index scores. Consequently, the beta regression model was used in this study to examine the relationship between the teff commercialization index and an exogenous variable. The outcome variable is subject to a beta distribution with the density function shown below (Ferrari & Cribari-Neto, 2004;Ferrari, 2013;Hunger et al., 2011). f ðy; μ; ΦÞ ¼ ΓΦ ΓðμΦÞΓðð1 À μÞΦÞ y μΦÀ 1 ð1 À yÞ ð1À μÞΦÀ 1 ; 0<y<1 (2) Where μ denotes expected value of Y i.e., E Y ð Þ¼ μ The parameter Φ fulfills the definition of a precision parameter because the greater the value of Φ, the smaller the variance of the dependent variable. More specifically, Only the mean parameter μ of the beta distribution is expressed as a function of covariates in the classical beta regression model; whereas, the precision parameter Φ is treated as a nuisance and E y i jX i ð Þ ¼ μ i (Hunger et al., 2011).
Where xi denotes the vector of covariates, β 0 is the slope of the constant term and β i refers to the vector of regression coefficients.

Level of teff commercialization
The mean teff commercialization index of commercial and semi-commercial teff producer farmers was 0.6 and 0.39, respectively. The average commercialization indices of commercial farmers were higher than the average of the whole observation (0.55) unlike semi-commercial farmers. According to Goitom (2009), andPingali (1997), smallholder farmers were divided into two categories based on their teff commercialization index score: semi-commercial 1 and commercial. 2 About 77.2% and 22.8% of smallholder farmers were categorized under commercial and semicommercial groups, respectively (Table 1). As compared to the previous research finding, the number of smallholder farmers fall under commercial category (77.2%) was greater than 30.1%, 40.5% and 49% which studied by Gari (2017), Addisu et al. (2019) and Medhane (2014), respectively. This means that the level of teff commercialization is higher than other areas.

Descriptive statistics result of categorical variables
Male-headed households accounted for 89.4% of smallholder farmers, while female-headed households accounted for 10.6%. This implies that the vast majority of teff producer smallholder farmers were headed by male. Furthermore, approximately 69.9% and 19.5% of male-headed households produced commercial and semi-commercial teff, respectively. In contrast, approximately 7.3% and 3.3% of female-headed smallholder farmers produced commercial and semicommercial teff, respectively. Similarly, 52.0% of smallholder farmers were literate, while 47.9% were illiterate. At the 5% level of significance, the chi square test results show that educational status has a statistically significant association with teff commercialization (Table 2).
Nearly 82.1% of smallholder farmers used extension services, while 17.9% of respondents did not use extension services. Similarly, 40.7% of smallholder farmers received credit services. On the other hand, approximately 59.4% of smallholder farmers did not receive credit. Farmers who have access to credit can reduce their financial constraints and purchase inputs more easily than those who do not. The reason for this is that credit enables farmers who are unable to purchase production inputs in cash on time.
In the study area smallholder teff producers have used both improved and local teff seed (Table 2). Even if the availability of sufficient improved seed is a challenge in most part of Ethiopia (ATA, 2018), about 57.7% of smallholders used improved teff seed whereas, 42.3% of respondents used only local seed. The result of the chi-square test revealed that seed type and teff commercialization had statistically significant association at 5% level of significance.
Availability of improved agricultural input help to increase production, productivity and market size (Fikadu et al., 2020) (2015) and Fikadu et al. (2020), among the inputs necessary for teff production, seed is the major one which increases teff marketed surplus. In Ethiopia, the use of traditional seed is still predominant (T. S.  though the amount of improved seed application has been growing over time (Louhichi et al., 2019). Study by Kabeta et al. (2019), Hailua et al. (2015) and Demelash (2017) stated that application of improved seed brings variation in yield of teff. Moreover, Aktaş and Unlu (2017) and Boka (2017) also stated that the adoption of improved varieties can improve the level of smallholders' amount of output which actually commercialized. The findings of previous studies confirms that majority of improved seed teff user are categorized under commercial production relative to farmers who used local seed (Addisu et al., 2019;Fikadu et al., 2020;Hauenstein, 2015;Louhichi et al., 2019).
Moreover, 55.3% of smallholder teff producers were found in mid highland agro-ecology, while 44.7% were found in low land agro-ecology. When compared to low land agro-ecology, the majority of smallholder teff producers in the mid highland were classified as commercial producers (Table 2). Households in the mid highland benefits largely by growing marketed crops (Boka, 2017). This implies that mid highland agro-ecology has more potential in production of cereals like teff than low land agro-ecology (Louhichi et al., 2019). According to Tura et al. (2016), the probability of market participation of mid highland smallholder farmers are higher than highland and low land agro ecologies. As chi square result indicated teff commercialization and type of agro-ecology had statistically significant association at 5% level of significance.

Descriptive statistics result of continuous variables
Commercial and semi-commercial smallholder farmers had an average household size of 4.8 and 4.2 man days, with standard deviations of 1.6 and 1.7, respectively. Besides, commercial, and semi-commercial smallholder farmers in teff production had an average of 38.9 and 28.2 years of farming experience, respectively. The independent sample t-test result showed that the average farm experience of commercial and semi-commercial smallholder farmers was statistically significant at the 1% level of significance, implying that more experienced smallholder farmers can sell more teff to the market (Table 3).
Smallholder farmers allocated 1.39 hectares of land for teff production on average, with a standard deviation of 0.23. Similarly, commercial, and semi-commercial farmers set aside 1.43 and 1.29 hectares of land, respectively, for teff production. Oxen provides draft power and is a major input in the teff production process. On average, commercial smallholder farmers owned 2.37 oxen, which was more than the 0.67 oxen owned by semi-commercial smallholder farmers. At the 1% level of significance, the average land size and oxen ownership among commercial and semi-commercial smallholder farmers were statistically significant.
Furthermore, the distance between the homesteads of smallholder farmers and the nearest market center has taken about 2.2 and 5.5 h, respectively, for commercial and semi-commercial farmers. At the 1% level of significance, the t-test result showed that the average distance from the nearest market differed statistically between commercial and semi-commercial farmers. This implies that smallholder farmers near the central market were more commercialized than farmers further away from the central market. Commercial and semi-commercial farmers, on the other hand, spent approximately 6395.76 Birr and 4652.04 Birr, respectively, on fertilizer. This implies that more commercialized farmers used more fertilizer to produce teff because the cost of fertilizer was higher for commercialized farmers than for semi-commercialized farmers. Studies by Lee (2018) and Louhichi et al. (2019) revealed teff takes the second rank next to maize in terms of large amount of fertilizer consumption. Moreover, the t-test result indicated that the cost of fertilizer between commercial and semi-commercial smallholder farmers was statistically significant at 5% level of significance.

Econometric model result
Five of the twelve explanatory variables included in the beta regression model had a statistically significant effect on teff commercialization. The variables are the number of oxen owned, the amount of land allocated to teff production, farm experience with teff production, agro-ecology, and market distance. Furthermore, the chi-square value indicated that the variables included in the model are significantly different from zero at the 1% level of significance. Table 4 shows the coefficients, magnitude, statistical test, and significant level of each explanatory variable.

Discussion
A smallholders' decision to enter and make use of market is influenced by various factors (Boka, 2017;Gebreselassie & Sharp, 2007;Hailua et al., 2015). The model result shows that teff commercialization of smallholder farmers is influenced by demographic, socio-economic and institutional factors. As expected, the number of oxen owned had a positive and significant effect on teff commercialization at 1% level of significance. This implies that an increase in the number of oxen increases commercialization level of smallholder farmers. Availability of traction/draft power is the most important factor that induces farmers to be market oriented which enhance agricultural commercialization (Addisu et al., 2019;Dube & Guveya, 2016;Gebremedhin & Hoekstra, 2007;Kim et al., 2016). This implies that producers who owned oxen are more likely to till in time and can produce the required amount which can increase the amount of teff sold to the market. The model result shows that a unit increase in number of oxen owned by smallholder farmers resulted in a 5.8% increase in teff commercialization. This finding is in line with previous studies (Goshu, 2016;Leta, 2018;Medhane, 2014) who found possession of oxen had positive influence on smallholder farmers' participation decision and level of teff commercialization.
Similarly, land allocated for teff production and teff commercialization had a positive and significant relationship at 1% level of significance. This implies that one hectare increases in size of land allocated for teff results in 6.6% increase in teff commercialization. This is because more land size owners can produce more output which increases the frequency and volume of teff supplied to the market (Boka, 2017;Carlettoa et al., 2017). Furthermore, the study by Bekele and Alemu (2015) revealed that an increase in size of land allocated for cereal crops results in an increase in percentage of sales in the output market. Similarly, the findings of previous studies confirmed that land size allocated for teff production increases the quantity of produce and level of teff commercialization (Addisu et al., 2019; Kabeta et al., 2019;Leta, 2018;Medhane, 2014;Tura et al., 2016) Moreover, farming experience in teff production had positive and significant influence on teff commercialization at 5% level of significance. More experienced farmers have good know-how on land preparation, input selection, improved agricultural technology, means of teff production and seasonality of teff production. A unit increase in farming experience in teff production increases teff commercialization by 4.30%. The result of this study is congruent with the findings of previous studies (Endalew et al., 2020;Kabeta et al., 2019;Kabiti et al., 2016) who found experienced farmers have better knowledge about farming, weather forecasting ability and use of improved production technology which increases productivity and amount supplied to the market.
Similarly, agro-ecology and teff commercialization had a positive and significant relationship at 5% level of significance. This revealed that mid high land agro-ecology increases teff commercialization by 1.50% as compared to low land agro-ecology. Thus, smallholder farmers in mid highland agro-ecology produce and supply more amount of teff because mid highland is the most suitable type of agro-ecology in which teff easily adapts. As discussed by Boka (2017), smallholder farmers in the mid highland agro-ecology benefits largely by growing crops that can be marketed because the land receives important soil nutrients coming from the highland by floods Likewise, Louhichi et al. (2019) reported that mid high land agro-ecology is the best suited to rain fed crops like teff. This finding is consistent with the study conducted by Tura et al. (2016) and Boka (2017) who revealed that mid highland agro-ecology also increases market participations of smallholder farmers.
In the opposite side, market distance had negative and significant effect on teff commercialization at 1% level of significance. Moreover, a unit increase in market distance from homestead decreases teff commercialization by 1.3%. The farmers who live close to the central market have an opportunity to sell more amount of teff than farmers who live far from the central market because long distance increase transportation cost of teff production and marketing. Therefore, the high cost of transportation discourages smallholder farmers from participating in teff production and marketing because it has a direct effect on the smallholder farmer's profit. The study by Bekele and Alemu (2015) revealed that distance from the main market adds more weight to the opposition of heavy transaction cost on most of subsistence farmers. This finding is in line with previous studies (Boka, 2017;Hailua et al., 2015;Kabeta et al., 2019;Kabiti et al., 2016;Mamo et al., 2017;Mazengia, 2016;Medhane, 2014;Mihretie, 2020) whose finding stated the closer the smallholder farmers to the output market, the higher is commercialization.

Conclusion
The study was undertaken to measure the level of commercialization and analyze the determinants of teff commercialization by applying output commercialization index and beta regression model, respectively. The result of the study shows majority of smallholder farmers were commercialized teff producers. The average commercial index of teff production was about 54.23%. This implies that the majority of teff production is for commercial purposes. Moreover, about 45.77% gap was observed to fully commercialize teff production. Therefore, special priority should be given to shifting semi-commercial teff production to commercial farming.
As the model result revealed, socioeconomic and institutional factors had statistically significant impact on teff commercialization of smallholder farmers. For example, the number of oxen owned had a positive effect on teff commercialization. This implies that an increase in number of oxen will increase teff commercialization because oxen are sources of traction power to till the land at the required time and frequency. Therefore, it is important to provide improved sources of traction power for smallholder farmers to improve their production and productivity which leads to an increase in their market participation decision and intensity. Similarly, land allocated for teff production positively affected its level of commercialization. Land is the most important but scarce resource in agricultural production. Therefore, an intervention regarding how to improve the productivity of available land will bring viable results in teff production and commercialization. Additionally, routine research should be undertaken to increase land productivity to increase teff production, productivity, and commercialization. Additionally, policy makers should strive to design and implement strategies to improve land productivity which would target smallholder farmers' teff commercialization. Moreover, increasing the year in teff production increases the level of teff commercialization. Thus, encouraging the smallholders to stay in teff production enables to get more experienced farmers who can produce market-oriented products which is highly demanded by the consumer. Likewise, smallholder farmers in the mid highland area were commercialized more amount of teff than smallholders in the low land agro-ecology. However, to increase teff commercialization, policy makers should design strategies which target hybridization of improved teff seed varieties which can adapt the low land type of agro-ecology. Consequently, market distance had a negative and significant effect on the level of teff commercialization. It is obvious that as the market center is far from the production site, smallholder farmers are discouraged from participating in teff production. Therefore, it is very important to focus on road infrastructure and expansion of various new market centers to increase the participation of smallholder farmers in the output market.