Agricultural sectors play an important role in the national economy and livelihood of rural households of Ethiopia .The main objective of this study was to identify the determinants of farm households’ livelihood strategies in Demba Gofa District of Gamo Gofa Zone of Southern Nation Nationalities and People Regional State of Ethiopia. A purposive and random sampling technique was employed to select 147 household heads, and the triangulation method of research design was employed to conduct this research. Primary data was collected using key informant interview, focus group discussion, household observations and interview schedule while secondary datawere collected from office reports, documents, books, and publications. Both descriptive and econometric data analysis techniques were applied. The multinomial logit model result for determinants of choices of livelihood strategies revealed that out of the 15 explanatory variables, the choice of agriculture plus off farm livelihood strategy was determined by agro-ecology, farmland size, livestock holding and leadership position, the choice of agriculture plus nonfarm was affected by irrigation use, credit use, distance from near market center and leadership position, and choice of agriculture plus off farm plus nonfarm was influenced by, the education level of the household head, size of farmland, frequency of extension service, irrigation use, credit use, distance from near market center and leadership position of household heads. The finding of this study revealed that most farm households who chose farm and nonfarm livelihood strategies were improved their livelihood status. The results of this study suggest that any interventions in the livelihood of farm households need to consider the asset level they own, the dynamics of livelihood system and socio-economic contexts of the area.

Ethiopia is an agrarian economy and the agricultural sector plays an important role in the national economy and livelihood of rural households. Ethiopian agriculture depends on rainfed farming which is subjectto climatic stresses. The outcome of these stress factors affects the food security, health, and wellbeing ofrural households.Since the rural poor are never the same in their characteristics and response strategies to various stresses, policy information, and implementation has to fit with these diversities. The majority of households in Ethiopia are poor, often face income fluctuation, and fail to smooth their consumption patterns due to price changes, weather related shocks, pests, death, and illness of family member, as well as livestock deaths (Woinishet, 2010).Many farm households lack opportunity and capability to engage in satisfactory livelihood activities and face food insecurityover the various months of the year.The livelihood strategy choices play an important role in determinig the livelihoods of the rural farm households.Ellis and Friedman (2005) indicate that the livelihood strategies that rural households choosedetermine their entery to and exit from poverty.

Scholarly studies on household livelihoods focus on contextualized analysis. The studies on livelihood diversification of rural households are meant to focus on the thinking for multiple entry options to various contexts and sets of livelihood strategies of households (Adugna, 2008). Since livelihood systems are related to local economies, policy has to focus on this. Such study entails the identifications of the numerous factors that determine the abilities of rural households in their choice of livelihood strategies and this approach has received attention due to the increasing threat of livelihood vulnerability sets to the poor. Thus, a thorough understanding of alternative livelihood strategies of rural households is important to design policies, support rural farm households, and empower them to manage the risk sets and enhance their resilience against the available and perceived risk sets. Such understanding is also important for informing local and contextualized policies for local development.

Many researchers have analyzed the different factors that determine the choice of livelihood strategies by rural farm households. Stifel (2010) analyzed factors such as age of household head, migrant (dummy), household structure, education, landholding, access to formal credit, community characteristics (phone, piped water and electricity access), distance to nearest city.Similarily, Berihanu(2010) analyzed the demogiraphic and socio-economic features such as sex, age, family size, educational status, asset holding, initial occupation and financial sources, and practices of saving and credit usage in relation to income generating activities. Besides commonly analyzed factors like education and gender differences, distance from urban centers plays a key role in determining access to non-farm employment (Amendola and Vecchi, 2007). All the above socio-economic and other related factors that researchers pointed out conistrain or access the different groups of rural household’s participation in a range of livelihood strategies. Different factors affect participation or choice of livelihood strategies differently. This means that the factors that favor or constrain one livelihood strategy may or may not favor or challenge the other. The livelihood strategy mix of rural households always vary and this requires periodic assessments in terms of policies implemented and new farm strategy sets that are available to rural households from time to time. The diversity, the dynamics and the set of ‘portfolios’, and the mix of assets available to rural households vary in context and is an issue of continuous study (Farrington et al., 2002).

Various studies conducted in Ethiopia rural context confirm that households focus on on-farm, off-farm and non-farm strategies. However, variations in contexts, institutional set ups and livelihood resources to which households have access to and control over, always affect the set of choices from the available set of strategies in the specific location and the risk sets. Therefore, participation of rural households in diverse portifolio of livelihood activities is important for withstanding the different conditions of vulnerability contexts and the requirements for the diverse entry options households choose from.

Therefore, assessing the determinants in this context is timely and worthwhile to inform policy alternatives that match to the dynamics in and contexts about the livelihood portfolios of farm households in the selected district.

The objectives of the study were:

a) To assess the main livelihood strategies adopted byrural farm households in the study area.

 b) To examine the determinants of choice of livelihood strategies in the study areas.

 Research Methodology

Description of the Study District

Demba Gofa Woreda is one of the 15 Districts of Gamo Gofa zone in SNNPRS of Ethiopia. The administrative center, Sawla town, is located at a distance of 285 km from Hawassa and 525 km from Addis Ababa. The District is located between altitudes of 1350-2600m above sea level, latitudes of 8 ^o11' 21”- 9^o 12' 24” North and longitudes of 42 ^o 19' 35”- 44 ^o 30' 26” East. The capital town of the Woreda, Sawla, which is north western part of the zonal center, Arba Minch town, located SouthWest of Hawassa. It bordered by Dawro Zone to the north, Geze Gofa to North West, Oyda to the South West, and Zala to the south and east. The total area of the Woreda is 979 km² which covers 8% of Gamo Gofa Zone. From total 97,900 hectare of land, 33,290,  23, 400,  6596, 7700, 5050, 5000 and 16,864 hectare of land covered by crop, rangeland, future unproductive land, public and private forest, bush, future productive land, and population settlement respectively, and the Woreda classified into 38 rural Kebeles (CSA, 2007).

Sample and Sampling Procedure

                                                                       In this study, both purposive and random sampling method was used. First, purposely lowland and midland areas were selected since most of households were aided by Productive Safety Net Program compared to the highland areas (Woreda Agriculture Office report).Then after, sample households were selected randomly from each of lowland (three Kebele Administrations) and midland (one Kebele Administration) agro-ecology. This study applied a simplified formula provided to determine the required sample size at 95% confidence level, degree of variability=0.5 and level of precision = 9 %, which was provided by (Mesfin Duguma, 2006). The formula is written as follows:

                        n =    ^__N______. …………… …………………………(1)

                      1+N (e) ²

Where n= Sample size, N= population size, and (e) refers to level of pricision, which is 9% (0.09). Based on this, 147 sample households were selected from Kebele diary of 2416 households (though based on the formula only 117 sample households were assumed to be selected).

Type, Source and Method of Data Collection

The data type was both qualitative and quantitative in nature, and both primary and secondary data were used. Primary data was collected through household survey, focus group discussion, and key informant interview and household observations. The primary data was collected by Enumerators who were development agent of the sample Kebele administrations. Secondary data which are also quantitative and qualitative in their form were collected from governmental organizations such as Office of agriculture, health office, finance and economic development office reports, documents, books, publications and other available source from records of NGOsoperating in the area. The field study combined key informant interviews, focus group discussions, and household survey interview. Structured and semi-structured interview schedule were developed to undertake the household survey interview.

Methods of Data Analysis

Analyses of both qualitative and quantitative data were carried out. Data collected through structured interview schedule were processed and coded using SPSS software for further analysis. Quantitative categorical type of data was analyzed using chi-square test. But narrative type of analysis was used to analyze qualitative type of data, and to enrich and illustrate a qualitative conclusion. After computing the descriptive statistics, a multinomial logistic regression was used (Green, 2003).The dependent variable of the study, livelihood strategy, is unordered, categorical variable with four alternatives. The dependent variables for the determinants of rural farm households’ choice of livelihood strategies were specified as;

                            Y=1, if the choice lies in On-farm alone

                            Y=2, if the choice lies in On-farm and Off-farm

                            Y=3, if the choice lies in On-farm and Non-farm

                            Y=4, if the choice lies in combination of On-farm, Non-farm, and Off-farm

Different variables were expected to affect livelihood strategies of rural farm households in the study area. These variables are dummy, categorical, and continuous in nature. The continuous variables are age, education level, family size, livestock number, dependency ratio, land size, distance to market center, and annual income from cash crops. The dummy, and or categorical variables include: sex, agro-ecology, credit use, participation in cooperatives, and participation in leadership, and access to irrigation use.

Results And Discussion

Livelihood Strategies Adopted By Households

Based on wealth criteria and local perceptions of the society, discussions conducted with the focus groups revealed the existence of socio-economic inequality within the society. The wealth category of sample respondents were distributed as poor (34.7%), middle (40.1%), and well-off (25.2%).

The survey result showed that 11.6, 10.2, 52.4 and 25.9 percent of sample respondents derived their livelihood from on-farm alone, on-farm and off-farm, on-farm and no-farm, and on-farm, off-farm and on-farm activities respectively. More than half of sample household derive their livelihood from on-farm + non-farm. This indicate that agriculture alone is not chosen as favorable livelihood strategy rather livelihood diversification strategy. The result of the sample survey also depicted that, most of sample households diversify their livelihood particularly above 90% of the poor diversify their livelihoods due to shortage of farm land (compared to the middle and better-off households). The difference in livelihood strategies choice implies the existence of significant barriers in selection of the most remunerative livelihood strategy (Yishak, 2014). The statistical analysis also indicated that, there is significant difference across the wealth category of the sample households to engage in diverse household livelihood strategies at less than 1% probability level.

                            Table 1: Choice of Livelihood Strategies by Wealth Category

Factors Affecting Farm Household Livelihood Strategies

Factors that (either positively or negatively) affect rural households’ choice of on-farm and off-farm livelihood strategy over on-farm alone (since it’s a reference category) were agro-ecology, farmland size, livestock holding and leadership position in the village. Whereas choice of on-farm and non-farm livelihood strategy were affected by irrigation use, credit use, distance from near market center and leadership position in the locality, but farm households engagement in the combination of the three livelihood strategies (on-farm, off-farm and non-farm) affected by the education level of the household head, size of farmland, extension contact, irrigation use, credit use, distance from near market center and leadership position of household head in the village.

It has to be considered that the multinomial logit estimates are reported for three of the four categories of livelihood strategies choice. Meaning that the first alternative (i.e. choice of on-farm alone) was used as a reference group (a benchmark to alternative category which is the choice of the other three). The credible implication and odds ratio of the significant explanatory variables on the choice of households’ livelihood strategies are presented as follows:

Education Level Household Head (EDU): The multinomial model result indicates that the variable education had negatively and significantly influenced the household choices of on-farm + off-farm + non-farm livelihood strategy at less than 5% probability level. The result of the regression also revealed that, if other factors held constant, as education level of the household increases by one year, the odds-ratio in favor of the probability of the household to choose on-farm with off-farm and non-farm livelihood strategy decrease by a factor 0.715. This finding indicates that those farm household heads with high educational level are less likely diversify livelihood strategies into  on-farm + off-farm + non-farm livelihood activities than those do not. This is most probably due to the inclusion of off-farm activities in this livelihood strategy since they think the activity as low remunerative, and less social value has given, and thus, the better educated farmers who have better skill, experience and knowledge less likely to participate in it. On other hand, a better educated household more likely to choice farming alone since they believe that technology based farming as preferable and can improve household wealth status than off-farm activities. Less availability of the work in the village, low social acceptance and seasonability of the off-farm activities are some of the causes that make the better educated households less likely to choice it. The other thing that is similar with what has discussed is that most of households who participate in this livelihood strategy also participate in less remunerable non-farm activities. Rural farm households who participate in this livelihood strategy were middle income categories while the educated farmers were more empowered with relatively strong social connection. Better educated farmers analyze the situations where they are living and well predict that what will have to be done rather than simply accepting what others say in general since they go with relatively adequate information that they are not as such negligence in comparison to less educated ones regarding to their technology acceptance and use. More educated farmers easily communicate others and have better access to claim as well as to have better linkage with someone who has assets such as irrigation farmland in his area. This result finding is similar with Bhatta et al., 2002 and Destaw (2003) but opposite with the findings of Berhanu (2007), Adugna (2008), Gebru and Beyene (2012), and Yishak et al., 2014.

Agro-Ecology (AGRO): This variable was found to be positively and significantly influenced diversification of the livelihood strategies into farm plus off-farm at less than 10% probability level. This result demonstrates that the magnitude of diversifying the livelihood into farming with off-farming increases as we go from low land to mid land. Unfair land distribution and grazing land, in addition to marginalized land and infertile soil with steep slope land structure, forces people who live in midland agro-ecology to engage in alternative livelihood strategies like off-farm activities. Consequence of this, all the agricultural production is low, and the poor households in the area involve in low remunerative works such as collecting fire wood, making charcoal as well as mat and selling house construction materials which comes from the near forest resources to cope up with the food insecurity situations. Moreover, lack of infrastructure in the area also constrains the delivery of many new technology as well as public services which improves the way of living of the poor households. Opposite to this, in low land areas the better topography which is convenient for farming, and land distribution is relatively better than the meddlers’. This fair land distribution and plain topography in lowland allow farmers to produce crops and rear various types of livestock so that they stick mostly with farming. Thus, the probability of diversifying livelihood into farming with off-farming activity increases by a factor of 60.736 for those farmers who are living in midland area. This model result finding is similar with the previous finding by (Yishak et al., 2014).

Size of Farmland Owned (LANDSIZE): As hypothesized, the area of land owned by the farm household has a significant and negative correlation with the likelihood of choosing on-farm + off-farm, and agriculture + off-farm + non-farm livelihood strategies at less than 10% probability level. The results of this study suggest that rural households with more land and hence gain more crop output either through intensification or not rather than diversifying from agriculture since they draw incentives of land productivity. This implies the chances of choosing agriculture in the context of having large land size decreases the probability of diversifying to on-farm + off farm, and to the three activities (on-farm + off-farm, non-farm activities) by  a factor of 0.067 and 0.3 respectively. A unit increase in land size results in a more decrease in probability choosing farm + off-farm livelihood strategies than the probability of choosing farm + off-farm + non-farm. On the other hand the probability of diversifying livelihoods decreases by increasing land size as farmers with more land supposed to stay on farm since land stimulates farming. As with agriculture, the effects of diversification on environmental resource management are mixed and context-specific. The growth of non-farm income sources might be expected to reduce the need for landless rural dwellers to carry out extractive practices in local environments for survival. On the other hand, for settled agriculturalists non-farm earning opportunities can result in neglect of labour-intensive conservation practices if labour availability is reduced (Ellis, 2008). This result is in line with that of Adugna (2008) and Yishak, 2014. The implication is that access to farm land is the most critical issue for farm households since those who with promising farmland do not need to involve in off-farm activities even if there is other challenges.

Livestock Holding in Total Livestock Unit (LVSTK): This variable was expected to influence the choice of livelihood strategies by the households negatively. In line with prior expectation, livestock holding (in TLU) negatively influence households’ choice of farming + off-farm livelihood strategy at less than 5% probability level.This result suggests that a household having larger size of livestock are less likely to diversify the livelihood strategies into non-farm and/or off-farm activities compared to those who own small number. Majority of farm households depend on livestock production for their farm income. The negative association between livelihood diversification and livestock ownership indicates that herd size creates better opportunity to earn more income from livestock production. The income generated from livestock helps households to fulfill family requirement including food. Moreover, increased livestock ownership has also a great importance in improving soil fertility and hence this increases field crop production. As a result, households who can get the required amount of food from livestock may not engage in another income generating activities unless and otherwise aimed to increase their asset holding. On other hand, the result shows that the households with lower livestock holding would be obliged to diversify livelihoods into farm and off-farm in order to meet needs.  According to the study, keeping other variables constant, the likelihood of diversifying the livelihoods into farm + off-farm activities decrease by a factor 0.511 while with one additional unit increase of livestock number (in TLU).This finding is congruent with many others; findings by Adugna (2008) and (Yishak et al., 2014).

Extension Contact: As expected, this variable was found to have negative and significant influence of livelihood diversification into a combination of on-farm with off-farm and non-farm activities at less than 10% probability level. This result implies that households having more frequency of extension contact are less likely to diversify their livelihood strategies into farm + off-farm + non-farm activities. From the model result, other things being constant, the probability of a household diversifying into the three combined strategy (farm with off-farm and non-farm) decreased by a factor of 1.68, for those farmers with more frequency of extension contact. The objectives of extension is to change farmers outlook towards their difficulties which assists them adapt better solution to their livelihoods. Thus, the information obtained and the knowledge and skill gained from extension organization may influence farmers’ skill and decision making on seeking diversification. The frequent extension contact received will increase the tendency of household to participate in farm activities. This may be also explained by the factors that the message that farmer gain from extension agents help them to initiate to participate in farming activities.

Irrigation Use (IRGTN):Away from what was expected, the model result indicates that the variable, irrigation water use has found to be positively and significantly influence households’ choice of livelihood diversification in to farm + non-farm, and farm + off-farm + non-farm livelihood strategies at less than 1% and 5% probability level. This implies that households with access to irrigation water have more likely to diversify their livelihood in to farm plus non-farm and off-farm activities than those who have no irrigation water access.  There are two probable reasons. The first one is that irrigators gain enough income from production of food crops (maize, potatoes and others) during agricultural off-season/dry season, and cash crops such as vegetables (onion, tomatoes and etc) during agricultural season. Secondly, households with access to improved irrigation infrastructure have strong social networks and better relations (since they have high social capital that they can access cultivated irrigation land through different means easily regardless of the size of land). In line with this, things should be considered is that in the study area there is irrigation users’ cooperatives which is very important for information sharing and as a result households participate in another diversification options in order to improve their livelihoods. Having access to the improved irrigation infrastructure was a crucial factor that provided many households with the opportunity to diversify their livelihood activities and thereby increase their levels of income (Sellamuttu et al., 2013). Others also advocate that improved access to irrigation indirectly results in household livelihood strategy diversification through improved infrastructure such as road networks and transport access as well. Households allocate their half time for farming and the rest half time participate in different livelihood activities. In opposite to the model result, (Kitchaicharoen et al., 2008) presented that having better access to irrigation water allows farmers to use their available land for crop production to increase their income and this can enable farmers to adopt new technologies as a result leading to increased income from farming. Similarly, Christine (2004) confirms that households with access to irrigation water likely to be in a good position to produce enough food, even a surplus. As surplus is sold, families are able to pay for school fees, pay for health services and prepare their farms well for the following season. It is also noted that households with access to irrigated water are able to prepare farms for early planting. This contributes to good performance of their plots and therefore high production. But this contradicts the model result and the condition which presented in the study area.

Credit Use (CREDIT): Unlike what is expected, the model result indicated that credit use found to be negative and significant influence on livelihood strategy choice of farm+ non-farm, and farm + off-farm + non-farm livelihood strategies at less than 10 % and 5% probability level respectively. This implies that farm households who have access to credit less likely to diversify in to off-and non-farm activities instead they participate in agricultural activities.  The probable reason is that farm households’ fear for taking risks if they face failure, or challenge while in their participation in off-farm and non-farm activities. On other words, credit use allows farmers to follow agricultural intensification by accessing farm inputs (DAP, UREA and improved seed varieties) which in turn improves productivity. Because of soil infertility, including poor farmers in the study area purchase farm inputs to increase crop production and secure their food security despite of unreasonable price of these inputs.  The finding of the model result also depicts that, other things being constant, the chance of diversifying the livelihood into farm +non-farm, and combination of the three (farm + off-farm + non-farm) livelihood strategy drops by a factor of 0.188 and 0.096 respectively when credit use increase by a households. This is consistent with Berhanu (2007), Khan (2007) and Adugna (2008) finding that confirm the fact that access and use of credit is very important asset in rural livelihoods not only to finance agricultural inputs activities, but also to protect loss of crucial livelihood assets such as cattle due to seasonal food shortage, illness or death. Thus, access and credit use is important for rural households to increase agricultural production than for livelihood diversification. In contrary to this, the finding was inconsistent with Stifel (2010), Gebrehiwot and Fekadu (2012), whose conclusion proves that access to credit plays a pivot role in strengthening smallholders’ potential to diversify their livelihood strategies and cope with stresses like drought.

Distance from Market Center (DSMKT): In line with what is expected, the variable, found to have negatively and significantly (at less than 10% probability level) influenced on the choice of farm +off-farm, and farm + off-farm + non-farm livelihood strategies. This means proximity to market center is another constraint that can affect farm households’ engagement into off-farm and non-farm income generating activities, which is determined by the existing infrastructural facilities to link the urban and rural people like access to road networks or transport. The interpretation of the odds-ratio for the distance from market centre indicates that, other things being constant, the probability of the respondents to choose on-farm + off-farm, and on-farm + off-farm + non-farm livelihood strategies together decreases by a factor of  0.598 and 0.568 respectively as the distance increases by one kilometer. This is because households nearby to market center gets several key advantages such as access to different information, terms of exchange for assets, save their substantial time, much lower transport costs and better and more remunerative non-farm, and off-farm activities. Also due to these all advantages households are more empowered than their counterparts and thus, their bargaining power is better and they do not face ambiguity when making decision in their way. Moreover, households near market center can cope up with harshly conditions relatively during food shortage times; they respond to shocks early by their own than waiting for food aid.  By equipping communities with the ability to manage and respond to shocks in the early stages of a crisis, strategic livelihoods interventions allow for more timely and appropriate responses to disasters than is possible with typical emergency relief assistance. In relation to this early protection and promotion of people’s livelihoods significantly reduces the need for massive food aid operations when malnutrition and disease reach acute levels. Farmers usually sell their produce at the village market place or nearby town markets. Proper functioning of markets are the basis for efficient resource allocation in the free market economy. On other hand, indirectly remoteness affects the choice set of livelihood strategies available to households by affecting transaction costs and by determining the degree of access to markets and to market information. Households living in a community which is away from market center are more likely to allocate labor to farming activities that are geared toward home consumption (Stifel, 2008). This finding was consistent with the previous finding which was done by Gebrehiwot and Fekadu (2012).

Leadership Position (LDRSP): this variable was found to be positive and significant influence on livelihood diversification into farm + non-farm, farm plus off-farm plus non-farm livelihood strategy at less than 5% probability level. This implies that those farmers who have participated in social leadership in local area are more likely to diversify livelihood into off-farm and non-farm activities than those farmers who have no leadership role in their village. The possible reason is that household head’s participation in local leadership and this can help to have more access for information, share more experience with others in social environment, creating more social network with outside societies. Survey respondents who play a leadership role were 98 %   and 91.2 % more likely to diversify in to on-farm + non-farm, and the three (on-farm + off-farm + non-farm) activities. This multinomial logistic regression result was confirmed by key Informants and Focus Group Discussants. For example, one of key informants, Ato Mengesha Gorfo, head of kebele administration, with better-off households, said that his household  involve both in off-farm activities (such as a daily wage laborer when work is available in the area like school construction)  and non-farm activities such kiosks/shop keeping, petty trading etc. in addition to farming activities. But, even most of others who are poor and with weak social relation in his/her environment do not want to participate in off-farm activities unless and otherwise they obliged to do so due to challenging conditions such as droughts and food shortages. This is why because traditional attitudes/cultural beliefs influence middle income gainers not to participate in off-farm activities that they see the activities as only for specific isolated groups.

     Table 2: Multinomial Logit Model Parameter for On-farm + Off-farm Livelihood Strategies

The reference category is: On-farm alone. Source: computed from own survey data, 2015.

**, and * stand for significant at 5%, and 10% respectively.

          Table 3: Multinomial Logit Model Parameter for On-farm + Non-farm Livelihood Strategies

The reference category is: On-farm alone. Source: computed from own survey data, 2015.

***, and * stand for significant at 1%, and 10% respectively.

On-farm + off-farm + non-farm

χ²                              

Source: Own Survey Result, 2015

      Odds-ratio

    Table 4: Multinomial Logit Model Parameter for On-farm + Off-farm + Non-farm Livelihood Strategies

The reference category is: On-farm alone. Source: computed from own survey data, 2015. **, and * stand for significant at 5% and 10% respectively.

Conclusions And Recommendations

                     The study characterized sample households as poor (34.7 %), middle (40.1 %) and better-off (25.2 %) based on land size, livestock owned, and crop output produced. The study show that a significant number of poor households were resource poor; they lack productive assets. In the study area, unlike the poor, better off households have better resources that help them to participate in livelihood activities they prefer. The asset level owned determines the adoption of households’ livelihood strategies. The key findings of the study revealed that, 11.6 %, 10.2%, 52.45% and 25.9% of households pursued on-farm alone, on-farm and off-farm, on-farm and non-farm, and combination of on-farm, off-farm and non-farm livelihood strategies at 1% probability level respectively. Shortage of farmland, lack of farm inputs and household income, poor quality of soil, and erratic rainfall is the case which repels households away from farming activities. To overcome such kind of obstacles poor households pursue on-farm plus off-farm, or combination of the three livelihood activities. This study has confirmed the importance of education, agro-ecology, farmland size, number of livestock owned in (TLU), annual cash crop income gained, irrigation use, credit use, distance from near market, and leadership role as important determinants of farm household livelihood strategy choice in the study area. While the log of odds of choice of livelihood diversification either in to off-farm/non-farm, or both decreases with increasing formal educational attainment of household head, farmland size, number of livestock owned (measured in TLU), annual cash crop income gained, credit use and distance from near market, but increases with increasing households’ participation in irrigation use, leadership role, and agro-ecology (more peoples living in lowland areas).

                     Land size (<0.1) and livestock ownership (< 0.05) were found to have negative association with on-farm plus off-farm livelihood strategy, but agro-ecology (<0.10) was found to be significant and positively influence household’s choice of on-farm plus off-farm livelihood strategy. Meanwhile, credit use and distance to near market significantly and negatively determine choice of on-farm plus non-farm activities at < 0.1 probability level. Irrigation use and participation in leadership, on the other hand, positively affect the same strategy at < 0.01 probability level. In the case of combination of the three activities (on-farm plus off-farm plus non-farm), education of household head (<0.05), land size (<0.10), extension contact (<0.10), credit use (<0.10), and distance to near market (< 0.1) were found significant and affect choice of this livelihood strategy negatively. On other hand, irrigation use and leadership participation positively affect the choice of on-farm plus off-farm plus non-farm livelihood strategies.

Generally, people’s options are heavily constrained by their circumstances, in particularly the place and society, into which they are living, and the asset level they own. Finally, policy makers and researchers must device a policy which addresses households’ social, psychological, attitudinal and vocational problems specifically taking under consideration of changing conditions and livelihood strategies over time first. Then, farm livelihoods can be improved through government’s policy intervention that can bring better access to resources and services in ways conducive to the area.

Acknowledgement

Authors would like to express their appreciation to various Offices of the Demba Gofa district (Demba Gofa Woreda Administration, Agriculture office, Finance and Economic Development Office) for their assistance during the course of the study and data collection. We finally appreciate numerous respondents for their patience during the interview time.