Rural Nonfarm Enterprise and Its Impact on Household Livelihood in Ethiopia: Evidence from Gurage Zone

4.1. Descriptive Analysis From the total 400 questionnaires distributed, only 352 (88%) were filled appropriately and returned, whereas the remaining 12% were lost and filled inappropriately. Table 3 shows 59.77% of the respondents were male, while 40.23% were female. The prime and mature working age groups accounted for 78.18% and 16.15% of total respondents, respectively, while early working age accounted for 5.67%. The result indicated that most respondents are at their mid-year of active working age. Respondents with no education accounted for 40.51%, while 36.26% and 23.23% of respondents have attained primary and secondary education, respectively. The majority (41.64%) of the respondents were located in Cheha Woreda, while 29.46% and 28.90% were residents of Ezha and Gummer Woreda, respectively. Most (83.85%) respondents were married, while 16.15% were single-headed households. The average family size is 3.83, per household, which is less than the national average household size of 4.8. Of 352 respondents, 194 respondents were participants in NFEA, while 158 were nonparticipants. From respondents of NFEA, it is expected that none of them will be involved over the year. 87.63% of respondents participated in all seasons. On the other hand, 10.31%, 1.55%, and 0.52% of the respondents participated in autumn, winter, and during both spring and winter, respectively. Alike agriculture, NFEA is not subject to the vagaries of natural dependence, so households participated in the sector over the year. Additionally, there is less risk in NFEA compared to the agriculture sector. However, the result contrasts with WB (2009), who shows that NFEA in Ethiopia is seasonal and complementary to agriculture. This could be due to variations in household response to incentives and risk behavior in different areas of Ethiopia. RNFEA comprises a diverse set of economic activities. As shown in Table 4, most of the respondents were involved only in trade and handcraft activities, which accounted for 24.74% and 15.46%, respectively. However, 59.8% of respondents have participated in more than one set of activities. For instance, 9.79% have been involved in trade and wage work concurrently, while 9.28% have been involved in trade, handcraft activities, street trade, and vegetables. In addition, information from focus group discussants shows various arrays of traded items such as chat, livestock, vegetables, fruits, cereals, and consumer goods. Households participated in handcrafts such as masonry, tannery, pottery, blacksmiths, weaving, decoration, etc. Selling beverages, food, and beauty salons is another NFEA practiced by households. Households in the study area have experienced mixed farming systems, such as crop production and livestock rearing. Households also participated in NFEA in addition to farming to enhance farm income for various motives such as consumption, saving, purchase of agricultural inputs, or combinations of all. 39.9% of the respondents participated in NFEA for both consumption and saving motives. Respondents also participated in saving and house-building motives which accounted for 4.66% and 4.15%, respectively. Even if its share is trivial (1.04%), respondents have been involved in NFEA to cover the education spending of families. In addition, 17.10% of the respondents participated for the consumption motive while 33.16% participated in NFEA, which was driven by the combinations of all motives. In addition, information collected from key informants’ interviews indicated that households have participated in NFEA for various motives such as credit payment and purchase of farming inputs. One of the focus group discussants explained that:

“I have participated in selling vegetables and fruits at the nearest markets almost every day, and then the income obtained is used to cover family consumption expenditures such as food, clothes, and student expenses”.

As indicated at the beginning of this section, 158 of the respondents did not participate in NFEA. Respondents were asked why they did not participate. Consequently, 30% of the respondents did not participate because income from agriculture was sufficient and 29.38% responded that due to the absence of a market and infrastructure. In addition, lack of government support and market access accounted for 12.50% of the factors that contributed to the nonparticipation of households in NFEA. In comparison, 8.13% responded that there is a lack of infrastructure, although agricultural income is adequate. Moreover, 32% have not participated due to a combination of the above factors. The key informant interviewee was also asked about and responded that households have not participated in NFEA due to a lack of government support and infrastructure. Further, it explained that government support in the form of credit and technical advice is minimal and infrastructures such as electricity, market linkage, and transport are another challenge. Lack of interest in engaging is another factor that hinders participation. Land and livestock are the two most vital inputs in farming activity. The average landholding in the study area is 0.615 hectares, less than the national holding because Gurage zone is the most dense area. The maximum landholding is 1.5 hectares, whereas the minimum is zero. Livestock is a vital asset in the household economy, as well as farming or a means of livelihood. The main livestock assets include oxen, sheep, goats and pack animals. The average livestock per household is 3.584, and the maximum and minimum are 9 and zero, respectively. As shown in Table 5, 81.59% of the respondents have no access to improved seeds, while 18.41% have access to improved seeds. Moreover, 44.76% of respondents have access to fertilizer, while 55.24% have no access to fertilizer. About credit, 83% of respondents have no access to credit, while 17% have access to credit. Therefore, the results suggested that households have limited access to improved seeds, fertilizer, and credit. In addition, 13.88% of respondents have access to electricity, while the majority (86.12%) have no access to electricity. Besides, 40.79% of the respondents have access to transport services, while 59.21% have not. Concerning the market, 36.26% of the respondents have access to the market, while 63.74% have no access to the market. Similarly, 44.48% of respondents have access to extension services, whereas the rest (55.52%) have no access. Only 23.86% of respondents have access to training, however, 76.14% have no access to training. Regarding remittances, 62.04% of respondents reported not receiving any, while 37.96% stated they had received remittances from either overseas or domestic sources. Regarding social capital, 38.81% of respondents were members of associations, while 61.81% were not members. The participation rate of households in NFEA is 54.96%, while 45.04% were nonparticipants. The average distance to the nearest city is 6.15 km, with the maximum distance being 14 km and the minimum distance being 0.5 km. 4.2. Empirical Analysis of Participation in Nonfarm Economic Activities In the logistic regression model, the assumptions of the classical linear regression model (CLRM), such as linearity, normality, homoscedasticity, and autocorrelation, need not be satisfied [32]. However, independent variables should not be collinear. To confirm this, a multicollinearity test was conducted using both correlation analysis and the Variance Inflation Factor (VIF). Although there is no clear cutoff value, ref. [33,34,35] exhibited that a VIF less than 10 suggests that the relationship between predictor variables is weak and the smaller the VIF, the lesser multicollinearity problems. The results showed that the VIF for the predictor variable “age” is close to 1, indicating a low level of association between the variables. The mean VIF is 1.56, indicating that there is no multicollinearity problem among the variables. The overall evaluation of the logit model is checked by using various techniques. One is the likelihood ratio (LR) test, which shows how well the independent variables jointly affect the dependent variable. The null hypothesis stated that independent variables do not jointly affect dependent variables. As shown in Table 6, LR with degree of freedom 20, LR (20) = 226.88 is the likelihood ratio chi-square statistic, and probability (prob) > LR = 0.0000 is the p-value. Thus, independent variables fit well in the model and have a high effect on jointly predicting the dependent variable. Another measure of goodness of fit is the Pearson or Hosmer-Lemeshow goodness of fit test, and the null hypothesis assumes that the model fits the data well. Hosmer-Lemeshow test statistics (19.03) are insignificant at (prob. > 0.05), so we accept the null hypothesis, suggesting that the model well fits the data. Various variables, such as demographic features of households, economic and infrastructure, and government-related variables, are included in the regression model. Accordingly, the logit model estimate, odd ratio, and marginal effect are reported in Table 7, along with their robust standard errors. Six variables—access to electricity, remittances, fertilizer, improved seeds, age (25–54 years), and distance—are statistically insignificant, while the remaining 14 variables are significant. Household Size: Households with larger family sizes are more likely to participate in NFEA than households with less family size. Citrus paribus, as household size increases by one person, the odd ratio in favour of participation in NFEA increases by a factor of 2.27. Because households with larger family sizes have high expenditures thus, the probability of participation in income-generating activities to diversify and increase income sources is higher than relying on agriculture income only. There may also be surplus labour in the household that increases the probability of participation. The result is consistent with the findings of [16,18,19,36,37] that showed that households with large family sizes are more likely to participate in NFEA than households with less family size. Sex of household head: It is expected that females are more likely to participate in non-farm employment activities (NFEA) than males. Consistently, the result indicated that citrus paribus, the odd ratio in favour of participation in NFEA decreases by a factor of 0.535 as a household is male. This is because males participate in farm activity, which is assumed to be more productive. In contrast, females participate in NFEA, which is assumed to be non-productive and requires less energy. The result is in line with the findings of [13,16,18,37,38] that women are more likely to participate in nonfarm economic activities than males. Marital Status of household: A single-headed household is more likely to engage in NFEA than a married household. Citrus paribus, the odd ratio in favour of participation in NFEA decreases by 0.077 as the household head is married. Single-headed households are mostly assumed younger and have no access to land and other assets; hence, they are obligated to participate in nonfarm activities. The result aligns with the findings of [38], who demonstrated that married households are less likely to participate in NFEA compared to single-headed households. Age of household: It is expected that as the age of a household increases, the probability of participating in non-farm employment activities (NFEA) decreases, as older households typically have less incentive to engage in this sector. As expected, mature working-age households are less likely to participate in NFEA than other age groups. Citrus paribus, the odd ratio in favour of participation in NFEA decreased by 0.096 for the mature working age group. The finding is consistent with the findings of [18,37,38] that young-age households have higher incentives to participate in NFEA than old-age households. Education level of household: Households in the upper secondary and primary education are more likely to participate in NFEA than those without education. Other things were constant, as such the odd ratio in favour of participation in NFEA increased by 2.39 and 2.355 as households attained primary and secondary education, respectively. Education increases the awareness of individuals to explore the opportunities obtained from NFEA, and the result is analogous to the findings of [35,37,38]. Cooperative membership: Households that are members of cooperatives are more likely to participate in NFEA than those who are not members of any association or cooperative. All else being equal, the odds of participating in NFEA increase by a factor of 5.409 for households that are members of cooperatives. Being a member of associations increases the social networks and sharing of experiences about starting a business, which in turn increases participation in NFEA. The result is consistent with the findings of [18,19,38,39,40]. Being a member of a cooperative lessens financial constraints through informal financial institutions such as Equib and Edir ⁴; it increases entrepreneurial skill, social capital, and bargaining power in the selling and buying process and other shared actions [18,38]. An association can be transformed into market information, business opportunities, resources, skills, and knowledge sharing about the importance of NFEA [19,40]. Access to extension service and training: Access to extension services increases the probability of household participation in NFEA. Citrus paribus, the odd ratio in favour of participation in NFEA increases by 2.855 as households have access to extension services. Rural extension workers with development agents increase households’ awareness about the importance of NFEA in enhancing income in addition to agriculture production support. Similarly, households with access to training are more likely to participate in NFEA than those without access to training. Citrus paribus, the odd ratio in favor of participation in NFEA increases by 2.207 as households have access to training. Training increases awareness about the diversification of income sources such as by being involved and rural entrepreneurship. The finding is consistent with [3,38], who state that training provides individuals with special and business skills to participate in NFEA. Access to market and transport: Households with market access are more likely to engage in NFEA than those without market access. Citrus paribus, the odd ratio in favour of participation in NFEA rises by 12.099 as households have access to the market. The finding is consistent with the findings of [18,39,41]. Similarly, households with access to transportation are more likely to engage in NFEA than those without transportation access. Citrus paribus, the odd ratio in favour of participation in nonfarm economic activities, increases by 2.679 as households have access to transport. Access to transport connects output and input markets, which, in turn, facilitates transactions. Access to credit: Households with access to credit are more likely to participate in NFEA than those without access to credit. Citrus paribus, the odd ratio in favour of participation in NFEA increases by a factor of 2.603 for households with access to credit because it helps to finance inputs. Therefore, the finding is consistent with [18,36], who indicated that households with access to credit are more likely to participate in nonfarm economic activities. Livestock and landholding: Households who have more livestock units are less likely to participate in NFEA than those who have less or no livestock holding. Citrus paribus, the odd ratio in favour of participation in NFEA declined by a factor of 0.814 as livestock holding of households increased by one unit. Similarly, households who have more land are less likely to participate in NFEA than those who have less or no landholding. Citrus paribus, the odd ratio in favour of participation in NFEA decreases by 0.024 as landholding increases by one hectare. The finding concerning landholding is consistent with [16,41], who showed that small land-holding households are more likely to participate in NFEA than those with more land. 4.3. Nonfarm Economic Activities and Household Livelihood In addition to crop and livestock, off-farm and nonfarm income is a vital source of income for households in Meskan and Mareko Woredas of the Gurage zone [42]. As shown in Table 8, the household average yearly income from farm and nonfarm income was 38,188.16 ETB, while the maximum and minimum income were 98,000 and 20,000 ETB, respectively. Similarly, the average expenditure from farm and nonfarm income was 35,228.5 ETB, and the minimum and maximum expenditure were 21,000 and 82,000 ETB, respectively. Average expenditure is less than income, which shows the existence of income left from expenditure, either saved or invested. However, only marketed expenditure and income are measured, while home consumption expenditure is not taken into account. The t-test is used to compare the mean income and expenditure difference between treatment and control groups and is presented in Table 9. T-test is employed to check whether the mean of treatment and control group are equal or not. The null hypothesis assumes that the mean income and expenditure of the treatment and control groups are equal. the null hypothesis is rejected, which indicates the presence of a difference in the mean income of participants and nonparticipants. Therefore, the mean income of nonparticipants is less than that of participants by 10,085.06 ETB. This is due to the additional income from NFEA, with farm income contributing to the difference for treatment and control groups. Concerning expenditure, the null hypothesis is rejected, signifying that the mean expenditure of participants is higher than nonparticipants by 7682.68 ETB. Those who participated in nonfarm income-generating activities have higher incomes, leading to higher expenditures. According to [43], before the propensity score, first, an outcome analysis must be undertaken. Outcome analysis is vital to measure the effect of participation if there are no propensity scores to control selection biases related to group assignment. For income analysis, all variables discussed in logistic regression were included as covariates. As shown in Table 10, participation in NFEA has a positive effect on income, which is consistent with [36,44]. Other covariates such as land, age (55–64 years) and (25–54 years), education (grade 9–12), and access to electricity positively influence income, but others are insignificant. Variables that are assumed to affect household expenditure are included as covariates. Demographic characteristics and economic variables are included, as shown in Table 11. Participation in NFEA positively affects expenditure and other covariates such as sex and education influence expenditure. After outcome analysis, the next step of PSM ⁵ is a balance test of covariates. The balancing test shows where the distribution of observed characteristics of treatment and control groups overlap. Using the propensity score algorithm, the balancing assumption is satisfied for the covariates of income and expenditure. The next step is to estimate the effects of participation in NFEA on livelihood outcomes using nearest-neighborhood, stratification, radius, and kernel matching methods. As presented in Table 12, the output of the four PSM techniques revealed participation in NFEA has a positive impact on household income. Nearest-neighbour matching indicated ATT is 11,423 ETB per year. In the same way, the stratification (interval) matching revealed ATT is 8308.1 ETB per year. The radius (calliper) matching also indicated ATT is 10,688.5 ETB per year and using the kernel (local linear) matching the ATT is 8438.4 ETB per year. Therefore, an estimate of ATT using the four matching exhibited households participating in NFEA have on average, more income than nonparticipants. However, the ATT estimated by nearest neighbours is higher than that of another matching. In contrast, ATT estimated by stratification matching is smaller than other techniques. The result is consistent with [23] who revealed that participation in NFEA has a positive effect on the well-being of households in rural Nigeria. Ref. [45] also showed participation in nonfarm employment increased household income in the upper east and west regions of Ghana. As presented in Table 13, the estimation output of the four-matching showed participation in NFEA has a positive impact on household expenditure. The nearest neighbor matching indicated ATT is 9075.07 ETB per year. In the same way, stratification matching revealed ATT is 9207.1 ETB per year. The radius matching also indicated ATT is 7956.5 ETB per year. Finally, kernel matching showed ATT is 9095.70 ETB per year. Thus, an estimate of ATT in the four matchings exhibited households participating in NFEA have, on average, more expenditure than nonparticipants but ATT estimated by stratification is higher than others. ATT is estimated by using a radius that is smaller than other techniques. The result is like the findings of [24], who indicated participation in NFEA increases expenditure and reduces the poverty status of households in Pakistan. Moreover, ref. [23] showed participation in NFEA has a positive effect on well-being by increasing consumption expenditure. In addition, ref. [22] exhibited that NFEA diversification has a positive impact on the consumption expenditure of farm households. If the result of the different matching methods is comparable, then the estimated ATT is robust. In this study, the findings of the different matching are quite consistent for both income and expenditure, such that ATT is robust. Another method is applying direct nearest neighbors matching instead of estimating propensity score [30]. If both methods give comparable results, then ATT is assumed to be more reliable and robust. Thus, the nearest neighbour matching result estimated before running the propensity score is presented in Table 14. The ATT of both income and expenditure is statistically significant suggesting that participation in nonfarm economic activities has a positive effect on household livelihood outcomes. Thus, the estimated result by propensity score and nearest-neighbors matching methods are similar, which indicates that the ATT is robust. 4.4. Triangulation with Qualitative Information The narration of key informant interviews and focus group discussions is used to triangulate the survey findings. The key informant interviewee and focus group discussants revealed that NFEA is practiced in their localities and Woredas. Moreover, the key informants indicated that almost 50% of households participated in NFEA. The result is nearly consistent with the survey, which showed that 54.957% participated in NFEA. Key informant interviewees and focus group discussants also showed various array of NFEA are practiced in the study areas such as trade, street and pity trade, handcraft activity, stonework, grain milling, wood and metal work, beverages and food, charcoal work, beauty salon, daily wage work, tailoring, fruit and vegetable selling and so on. The key informant interviewee and FGD indicated household participation in NFEA is driven by consumption smoothening, saving, purchasing agriculture inputs, paying credit, and financing student expenditure. Therefore, the findings from key informant interviews and focus group discussions (FGD) are consistent with the results from the survey. Household participation in NFEA is inhibited by a lack of credit access, government support, and infrastructure access and a lack of interest and motivation.

“One of the FGD stated when we need credit from finance sectors to start our business, we cannot access it because the interest rate is high or requested collateral. Therefore, we are discriminated against and forgotten from the financial markets. Other FGD indicated that the government officials did not give us continuous support. Further officials come when they need information which seems for report purposes or come at the start of business and then after, there is no continuous support and follow-up. However, we need continuous technical and material support, and training.”

The key informant interviewee revealed participation in NFEA raises household income and improves livelihood and well-being. Moreover, it creates employment opportunities and raises consumption of locally produced goods which in turn encourage local producers. Similarly, the FGD indicated participation in NFEA boosts income, satisfies local demand for local products, creates employment opportunities, and serves as a base to start micro and small enterprises. Moreover, the key informant interviewee and FGD revealed that most households have a positive attitude towards NFEA. Finally, key informant interviewees and FGD were asked about the support they need from the government to upturn NFEA participation. The key informant interviewees responded they need to create market opportunities and chains with input and output traders, availing credit with a reasonable interest rate, life skills and entrepreneurship training, electrification, and transportation. In addition, the FGD revealed they need technical and financial support and entrepreneurship training within the context of local nonfarm activities. Moreover, households required additional land to run NFEA, credit access, and infrastructure.