Research Article
Impact of Smallholder Farmers Agricultural Commercialization on Rural Households Poverty
Department of Agricultural Economics and Extension, Jimma University, P.O. Box 307, Jimma, Oromia Regional State, Ethiopia
Smallholders cultivate over 96% of the total agricultural land. The average smallholder cultivates less than one hectare of arable land and consumes more than 65% of total production within the household (EEA, 2006). In many parts of the country, market participation of smallholder family farms (measured either in terms of per capita market share, the volume of farm output supplied to markets or their profit motive) is limited. Agricultural markets are fragmented and not well integrated into a wider market system which increases transaction cost and reduces farmers incentive to produce for the market. Government policy or the lack of it, has contributed to this general characteristic of the smallholder agricultural sector in Ethiopia (MoFED, 2006).
Commercial transformation of subsistence agriculture is an indispensable pathway towards economic growth and development for many agriculture dependent developing countries (Von Braun, 1994; Pingali and Rosegrant, 1995; Timmer, 1997; World Bank, 2007). Sustainable household food security and welfare also requires commercial transformation of subsistence agriculture (Pingali, 1997). Commercial agricultural production is likely to result in welfare gains through the realization of comparative advantages, economies of scale and from dynamic technological, organizational and institutional change effects that arise from the flow of ideas due to exchange-based interactions (Romer, 1993, 1994). According to a study by Gebreselassie and Sharp (2007), smallholders with high degree of market engagements have better potential of enjoying better standards of welfare. Similarly, Sharp et al. (2007) noted that enhancing the degree of commercialization of the smallholders can have more impact on reducing poverty than promotion of few large ventures.
Commercialization enhances the links between the input and output sides of agricultural markets. Commercialization entails market orientation (agricultural production decision destined for market based on market signals) and market participation (offered for sale and use of purchased inputs) (Gebremedhin and Jaleta, 2010). Empirical evidence indicates that commercialization of smallholder farms has the potential to enhance incomes and welfare outcomes and take smallholder farmers out of poverty if constraining factors such as lack of capital, basic skills (farming and commercialization), high transaction costs, lack of infrastructure, lack of information and lack of education could be eliminated (Lerman, 2004). So far the literature on commercialization of smallholders makes little study on the impact of market participation on rural poverty in the study area, especially at a household level. Therefore, this study attempts to fill the gap by conducting an empirical research on identifying, analyzing and understanding the impact of smallholders commercialization on poverty and those elements that are responsible for variation in smallholder poverty that is needed to guide policy decisions, device appropriate interventions and integrated efforts to combat poverty. Therefore, the study aimed at analyzing the impact of agricultural commercialization on smallholder farmers poverty level in south western Ethiopia. The objectives of the study are:
• | To assess the socioeconomic determinants of smallholder farmer poverty |
• | To assess the degree of smallholder farmer agricultural commercialization |
• | To analyze the impact of smallholder farmer agricultural commercialization on rural poverty |
METHODOLOGY
Description of the study area: The proposed study was conducted in two sub-districts in Jimma zone in 20012/2013 in Mana and Goma Weredas of Jimma zone, south western Ethiopia. The study area was selected from potential coffee producing area since the product is important to the national economy, grown and marketed by smallholders for generations, high policy attention and intervention. This study area also selected to represent a dominantly subsistence farming community where land degradation coupled with erratic rainfall, drought problems pose a serious threat on households food security in south western Ethiopia. Jimma zone is one of the 20 administrative zones in Oromia Regional State, is divided under 18 administrative districts with 2.5 million population from which 94% are rural inhabitants (FDRE, 2008). The zone covers a total area of 15,569 km2 that receive reliably good rains ranging from 1,200-2,800 mm per annum. Subsistence farming is the dominant form of livelihood in the area where only 15% of the population is in non-farm related jobs. The area has suitable agro-ecological potential with the lowest drought risk rating (298) in the country (Milas and El Aynaoui, 2004). Cereals (maize, teff Eragrostis tef, sorghum and barley), pulses (beans and peas), cash crops (coffee and khat Catha edulis) and root crops (Ensete ventricosum false banana and potato) are the major crops produced in the area. Different fruits and vegetables are also commonly grown where home-gardening by small holder families was observed to increase household income and food security (Kebebew et al., 2011).
Sampling size and sampling method: The study was applied a simplified equation:
(1) |
where, n is the sample size, N is the population size and e is the level of precision provided by Yamane (1967) to determine the required sample size at 95% confident level and 90% level of precision.
Data for the study was generated from a farm survey of 280 farmers selected by multistage stage sampling procedure. In the first stage, Jimma zone was purposively selected from south western Ethiopia. In the second stage, Mana and Goma Weredas were purposively selected from Jimma zone on the ground that they are the potential coffee producing areas. In third stage, five Peasant Associations (PA), or kebeles, from each Woreda were selected randomly. In fourth stage, the sample was stratified within each Peasant Associations (PA) to ensure that a representative number of females were included. Finally, a probability proportional to sample size sampling procedure was employed to select total of 280 sample farm households (28 households per PA). The number of households was obtained from the 2007 population and housing census while the households were systematically selected from the fresh list of households within the PA made during the survey.
Method of data analysis: Two types of data analysis used, namely descriptive statistics and econometric analysis.
Descriptive statistics: This method of data analysis refers to the use of ratios, percentages, means and standard deviations in the process of comparing demographic and socio-economic factors related to farming households.
Measure of poverty line: Foster et al. (1984) method was used in the assessment of poverty. The FGT measure is given:
(2) |
And:
Pα≥0 for Y<Z
Where:Pα | = | Weighted poverty index |
n | = | Total number of households |
q | = | No. of households |
Y | = | Per adult consumption expenditure of household |
Z | = | Poverty line2 when α = 0, 1 or 2, P0 = q/n |
Crop commercialization index: Following Von Braun et al. (1994), the household crop output market participation in annual crops as the proportion of the value of crop sales to total value of crop production which is refer to Crop-Output Market Participation index or Crop Commercialisation Index (CCI) specified as in Eq. 3:
(3) |
Sik | = | Quantity of output k sold by household i evaluated at an average community levelprice |
Qik | = | Total quantity of output k produced by household i |
A value of zero for the CCI signifies total subsistence, while a CCI value approaching 100 indicates higher degrees of commercialisation i.e., a greater percentage of marketed crop production.
Logit model: The binary logit model was used to analyse the impact of smallholders agricultural commercialization on the probability of being poor. In this model the dependent variable is Household Poverty Status (HPS) that is dichotomous taking a value of 1 if the household is poor; 0 otherwise. The information which identifies the poor from non-poor, was obtained by comparing poverty line. A household below this threshold is said to be poor (Zi = 1), otherwise non-poor (Zi = 0). The cumulative logistic probability model is econometrically specified in Eq. 4 (Gujarati, 1995):
(4) |
I | = | Individual i = 1, 2, .., n |
Li | = | Log of the odds ratio which is not only linear in Xi but also linear in the parameters |
Pi | = | Probability that an individual is being poor |
1-Pi | = | Probability that a household will not be non-poor |
αo | = | Intercept or constant term, that implies the combined impact of these fixed factors on household poverty |
α1, ,α8 | = | Coefficients of continuous explanatory variables (X1, ,X9) |
β1 β6 | = | Coefficient of explanatory dummy variable and εi is error term |
Table 1: | List of variables, codes and variable definitions |
After specification of the model, the parameter of the model was estimated by Maximum Likelihood Function (MLE) using STATA software package version 12. The model is based on the hypotheses mentioned in Table 1 (STATA, 2008).
Descriptive results: Descriptive statistics of variables used in the regression analysis are given in Table 2. The result of the survey revealed that the head count ratio or incidence of poverty is 0.43. This implies that 43% of the sampled households were poor or not able to meet the daily recommended caloric requirement. The average crop output market participation of rural households was 39%, indicating moderate market participation. The study showed that about 36% of sample households were female headed. Classification of household head as literate and illiterate exhibited that 47% of household heads are literate. It was hypothesized that as the level of education increases, the probabilities of being poor decrease.
Family size which measure number of individual members of a household, is a variable used by many empirical studies to see how it affects poverty status of households. The mean family size in AE of the household was 7.14. The result indicated that the mean age poor household was 45.39 years (Table 3).
Survey result showed the mean cultivated land size of households was 0.92 ha. The annual total crop production of households was 1941.3 kg. The average livestock owned by the sample respondents were 7.16 TLU (In Appendix 1 conversion factors are mentioned that are used to calcalate the Tsopical Livestock Units (TLU)). The average value of annual crop produced per household is Ethiopian birr (ETB) 5534.40, of which ETB 2002.70 worth of produce was sold. Households in the study area get about ETB 1165.30 income from non-farm and off-farm employment and remittances. On the other hand, the mean annual consumption expenditure per AE for sample households was ETB 1064.7. Similarly, the mean annual food and non-food consumption expenditures per AE were ETB 893.54, respectively. Almost 78% of households consumption expenditure was spent on food.
Table 2: | Descriptive statistics for dummy variables used in econometrics |
Survey result (2013) |
Table 3: | Descriptive statistics for continuous variables used in econometrics |
Survey result (2013) |
Table 4: | Logit estimation results for the impact of smallholder commercialization on rural poverty |
Survey result (2013), *,**, ***Significance at 10, 5 and 1%, level, respectively |
Results of econometric analysis: Table 4 shows the result of the logit model on the impact of smallholder agricultural commercialization on rural poverty. The results of the existence of serious problem of multicollinearity among the hypothesized explanatory variables showed that values of Variance Inflation Factor (VIF) for each of the continuous variables were found to be less than ten hence, there is no a multicollinearity problem among all the hypothesized continuous variables included in the model (Appendix 2). The result of Contingency coefficient (C) revealed that there was no a serious problem of association among discrete explanatory variables as the contingency coefficients did not exceed 0.75. Therefore, all the hypothesized dummy variables were included in the logistic regression model (Appendix 3).
Out of the fourteen independent variables hypothesized to have influence on rural households poverty, seven variables were found to be statistically significant. The maximum likelihood estimates of the logit model showed that sex of household head, age of household (year), education of household head, household family size in AE, livestock holding (TLU), annual farm income per AE (birr), access to credit previous year and distance from settlement center to nearest market place (km) were found to be the important determinants identified to influence rural households poverty in the study area (Table 4).
The result revealed that agricultural commercialization (crop output market participation) does not affect the rural households poverty due to the non-significance of the agricultural commercialization in the logit model used, however, it has a negative relationship to poverty. This means that the higher the smallholder farmers commercialization, the lower the probability of being poor (Table 4).
The survey result showed that male-headed households and negatively related to the probability of being poor. The possible explanation for the negative relationship indicates that male headed households less likely to be poor than female headed households. This may be due to the fact that male headed households usually have higher potential of crop production efficiency advantages access to market information and incomes than the female-headed households. The odds-ratio of 0.209 indicates that, if other factors are kept constant, the odds in favor of being poor decrease by a factor of 0.209 than female headed households. This result indicates that male-headed households were less likely to poor than female-headed households.
The sign of the coefficient of change in age of the household head showed a negative relationship with poverty and is significant at 5% probability level. Keeping other factor unchanged, the odds ratio in favor of poverty decrease by a factor of 0.97 when age of the household head increases by one year.
The household family size in AE was significant at 1% probability level and has positive association with the household poverty. The positive relationship indicates that the odds ratio in favor of the probability of being poor increases with an increase in the family size in AE (In Appendix 4 conversion factors are mentioned that are used to Compute Consumption Unit (AE). The odds ratio of 1.21 for family size in AE implies that, other things being constant, the probability of being poor increases by a factor of 1.21 as family size increases by one adult equivalent. This is in agreement with the hypothesis that the family size is likely to play a role in determining the state of rural household poverty. This clearly shows the importance of controlling population growth in the area (Mitiku et al., 2012a, b).
Likewise, literate households are negatively and significantly related to the probability of being poor. The possible explanation for the negative relationship indicates that literate households have better skills, better access to information and ability to process information than illiterate households.
Likewise, credit is negatively and significantly related to the probability of being poor. The negative relationship is explained by the fact that credit helps to improve the ability of farmers at critical times of the year to buy inputs and encourage farmers to adopt new technology. The model result confirms that credit is statistically significant at 5% probability level with the expected sign. The credits used for agricultural inputs improve their productivity and increase the farm income and wealth status of the farmers and those farmers with better wealth status than the others. The odds-ratio of 0.475 indicates that, if other factors are kept constant, the odds in favor of being poor decrease by a factor of 0.475 for a farmer who gets access to credit than those farmers who do not have access to credit. This result indicates that those farmers who had access to credit were less likely to poor than those who had no access to credit.
Similarly, annual farm income per AE is negatively and significantly related to the probability of being poor. The negative relationship is explained by the fact that those farmers who have better access to different types of farm income are less likely to become poor than those households who have little access. The odds ratio in favor of poverty decreases by a factor of 1 as the farm income in AE increases by one birr, keeping other factor constant.
Distance from settlement center to nearest market place (km) is positively and significantly related to the probability of being poor. The positive relationship is explained by the fact that households that have proximity to market and other public infrastructure may create opportunities of more income by providing off/non-farm employment and access to transportation facilities, market information. In addition, further distance to nearest market detract farmers from crop inputs and outputs market participation and also increasing marketing costs. Other things are held constant, for a unit increase in km the odds ratio in favor of being poor increase by a factor of 1.023 as distance from settlement center to nearest market place increases in one kilometer.
The study revealed that 43% of the households were not able to meet the daily recommended caloric requirement or below poverty line. The result showed that the impact of smallholder agricultural commercialization on rural poverty revealed that agricultural commercialization does not affect the rural households poverty in the study area due to the non-significance of the crop commercialization index, however, it has a negative relationship with poverty or probability of being poor. The negative relationship shows that agricultural commercialization reduces poverty level of the households. Further, the study has shown as the major factors affecting poverty of rural households are sex of household head, age of household, education of household head, family size, livestock holding, farm income, access to credit and distance from settlement center to nearest market place. Based on the findings and conclusion of the study, the following policy recommendations are forwarded:
• | Proper attention should be given to limit the increasing population through proper awareness creation about practicing family planning activities by using integrated health and education services |
• | Farm households with larger livestock holdings are less likely poor than farmers with less livestock holdings. Therefore, farmers should be encouraged to engage in livestock husbandry through providing with improved livestock production technologies (health service, improved breeds and feeds) to increase production and productivity of the sector, this will ultimately reduce poverty. |
I would like to thank the Jimma University for granting me with all the necessary financial and materials support for this study. I would like to extend my thanks to the enumerators, key informant and the community of Jimma zone who spent many hours in responding to my questions. Finally my thanks go to all individuals who gave me their valuable advice and encouragement during the research.
Appendix 1: | Conversion factors used to calculate Tropical Livestock Units (TLU) |
Storck et al. (1991) |
Appendix 2: | Variance Inflation Factor (VIF) for continuous variables |
Appendix 3: | Contingency coefficient (C) value of dummy variables |
Appendix 4: | Conversion factor used to compute consumption unit (AE) |
Storck et al. (1991) |
1Sokoni (2007:3) defined commercialization of smallholder production as a process involving the transformation from production for household subsistence to production for the market. Hazell et al. (2007:4) found that most definitions refer to agricultural commercialization as the degree of participation in the output markets with the focus very much on cash incomes
2In this study poverty line was estimated based on the cost of 2,200 kcal per day per adult food consumption with an allowance for essential nonfood items. The food poverty, non-food poverty and total poverty lines used were 2692, 2805 and 5622 birr at local average prices, respectively applied to real per adult household consumption expenditure in order to calculate head count, poverty gap and squared poverty gap indices (MoFED, 2012).