This study was carried out in the South-West Nigeria. South-West Nigeria is one of the six geopolitical zones of Nigeria. It has six states (Ondo, Ogun, Ekiti, Osun, Lagos and Oyo State). It falls on latitude 6⁰ to the North and latitude 4⁰ to the south. It is marked by longitude 4⁰ to the west and 6⁰ to the east. The geographical location of South-West Nigeria covers about 114,271 kilometers square that is approximately 12% of Nigeria’s total land mass and the vegetation is typically rainforest. The total population is 27,581,992 out of which more than 96% is Yoruba [8]. The major occupation of people residing in this geopolitical zone include farming, trading, food vendor, fish smoking, hair dressing, sewing, carpentry, marketing as well as food processing. Agriculture provides employment for the majority of people with respect to commercial and local livestock production, food crops such as maize, cassava, yam, soybeans, plantain, and melon; tree crops such as cocoa, citrus, and oil palm, cola nut and leafy vegetables such as amaranthus, celosia, waterleaf etc. South-West Nigeria is bounded in the North by Kogi and Kwara States, Edo and Delta States in the East, Atlantic Ocean in the south and by Republic of Benin in the west.

The study adopted a multi stage sampling procedure. The first stage involves purposive selection of six local governments from the twenty local governments in Ogun state namely; Ido, Ipokia, Odeda, Olorunsogo, Remo North and Surulere LGAs. The second stage involves random selection of thirteen (13) villages from the six local governments. The selection of the villages was done based on the proportion to size of the local government area; and is represented in the parenthesis as follows: in Ido (Igbonna and Odebode), Ipokia (Idemose, Ifoyintedo and Mogbara), Odeda (Ojebiyi, Orileilugun), Olorunsogo (Igbeti, Tesi-Garuba), Remo North (Gbasemo, Ilara and Jowoje) and lastly Surulere (Pooro). The third stage constitutes random selection of 30 farming households from each of the selected villages, giving a total number of 325 respondents. However, 291 questionnaires were correctly filled and returned out of the 325 administered. The returned questionnaires were used for analysis.

The descriptive statistics was used to analyze the socio-economic data of the respondents. The assessment of income level inequality across the rural farming households in the study area on gender basis was analyzed using Lorenz curve and Gini coefficient while poverty status of the respondents by gender was estimated using the method provided in another study [15]. In a quest to determine the socio-economic status of respondents that determines the poverty status, the logistic regression model was utilized.

The probability that a farmer will fall below the poverty line was defined as function of the socioeconomic characteristics and institutional factors using logit model. Hence the logistic probability model is econometrically defined below;

(1)

Where R_i is the probability that the farmer falls below the poverty line or not. γ and β₁ and are parameters to be estimated. Logit model could be written in terms of the odds and logs of odds which allow quick understanding of the interpretation of the coefficients [16]. The odds ratio simply means the probability (R_i) that a farmer is poor to the probability (1-R_i) that a farmer is not poor.

(2)

Hence;

(3)

The natural log of equation (3), will give us:

(4)

If the stochastic term e_i is taken into consideration, then the logit model can be re-written as:

(5)

Equation (3) was estimated using the maximum likelihood method. This procedure does not require assumptions of normality or homoskedasticity of errors in predictor variables.

Note: The definition of the variables imbedded in the logistic regression is presented in Table 1.

A number of poverty measurements were adopted in literature [13, 17, 18]. Observably, the p-alpha (P_α) class of poverty measure, is the most popular because α is a policy parameter that can be varied to approximately reflect the poverty “aversion” and also the P_α class of poverty indices is subgroup decomposable [13]. Thus, this study adopted the standard FGT (1984) to generate the poverty profile of the selected farming households. FGT takes the form:

(6)

Where Z = the relative asset poverty line

n= number of the cassava farmers below the poverty line

N = Total number of cassava farmers sampled

Y_i= estimated per capita household income of the ^th household

Z-Y_i = poverty gap of the ^th household

poverty gap ratio

α= poverty aversion parameter, with values: 0, 1, 2

α= 0, equation (6) gives the poverty headcount

α=1, equation (6) gives the poverty depth

α=2, equation (6) gives the poverty severity index.

The inequality in income can be estimated by using Gini-coefficient. If incomes are ordered such that T₁≤T₂≤T₃≤T₄≤T_n, then the Gini-coefficient can be calculated based on the procedure adopted by Morduch and colleagues [19].

The result of Gini coefficient showed that income inequality value for the population was 0.1168. By implication, there is 11.68% income inequality among the rural households in the Southwest rural Nigeria. However, decomposing the inequality status by gender, we found that income inequality was lower (0.4692) among the male respondents than the female counterparts (0.4994), but these may not be very significant. This means that elimination of gender inequality will not reduce total income inequality significantly in the study area. On the contrary, it was observed that the relative contribution to overall inequality was higher among male respondents with a value of 0.7592 while that of the female counterpart was at the other low extreme with a value of 0.0114. This finding is congruent with the study of [13].

The estimated poverty line based on two-third of the annual mean per capita income in rural households showed that the mean annual per capita income among the rural households was ₦22793.78 ($114.54) while the poverty line was ₦15271.83 ($76.74). Without doubt, poverty is a rural phenomenon, 58.89% of the rural households were found to be living below poverty line in the south western region of Nigeria. However, more than half (53.71%) of the male population have incidence of poverty while 47.22% accounted for female population living below poverty line. This is also graphically represented below (Fig. 1). IFAD, 2009 re-ported that over 75 percent of rural dwellers engage in farming activities as a source of livelihood. Therefore, this finding is consistent with the previous studies [13, 20].

Fig. (1). FGT Curves for poverty Incidence.

For instance, it has been reported that male-headed households that engaged in farming have lower welfare while female-headed households engaged in farming have higher welfare [21]. In relation to depth and severity of poverty, male-headed households were well represented as they took a lead above the female headed households (Table 2).

Logit regression was used in this study to assess the determinants of poverty among rural households in the study area. This method was adopted in line with other studies by [13, 22-25]. From the maximum likelihood estimates of the Logistic regression (Table 3); the results show that the model (regression line) fits the data reasonably. The result of the analysis revealed a chi square value of 74.18, 22.36 and 73.59 which was significant at 1% (p<0.01). The Pseudo R² was 0.16, 0.45 and 0.14, and the log-likelihood was -195.25, -13.77 and -220.62 suggesting strong explanatory power of the model for male-headed households, female-headed households and pooled rural household heads respectively. This also indicates that variation in poverty status is explained by the (maximum likelihood) estimates of the specified explanatory variables, suggesting that the model as specified, explained significant non-zero variations in factors influencing poverty among the rural households in the study area. Interestingly, the model shows a wide variation in the drivers of poverty across the situations considered (for male-headed households, female-headed households and pooled rural household heads). Collectively, educational years, household size, farming experience, market distance, extension access, credit access and Member of social group are the explanatory variables determining poverty among the rural dwellers in the southwest Nigeria.

It has been observed that improved educational attainment has a direct effect on enabling households accessing, assessing and conceptualizing relevant information on technology adoption, improved farming practices, accessing lucrative rural market opportunities and profitable enterprise combination, thus, education remains important in poverty alleviation and income distribution [26]. The coefficient of education measured in actual years of schooling was found to be significant with a negative relationship with poverty. However, only the female-headed households experienced poverty reduction influenced by education. By implication, probability of being poor was reduced by 39.9 percent through an increase in the years of educational attainment. Unsurprisingly, this study is consistence with other studies, which concluded that increase in the educational attainment of the household head have an important impact on reducing the probability that a household is poor [27]. Furthermore, a study conducted in Kenya concluded that lack of education is a factor that accounts for a higher probability of being poor [28].

Household size is an important factor, which can play a role in bringing down the incidences of poverty by reducing the probability of remaining in the poor household category. The increasing family size implies a larger number of dependents on fewer earners and this might lead to fewer earning and lesser per capita consumption [29]. The co-efficient of household size was significant and positively influenced poverty status for male-headed households and the total sample collectively. Results obtained revealed that the likelihood event of being poor was higher with male-headed households that have large households compared to the total sample (Table 3). The larger the household size, the poorer the household is likely to be, because more of the household members will likely be children who are unproductive and yet take a big proportion of household income in terms of school fees, medical bills, food and clothing. Therefore, a unit increase in the size of the rice farming household by an individual increases the probability that the farm households will be poor by 21.75%. Similar finding was obtained in other studies [24, 25, 30, 31].

The regression coefficient for years of farming experience of the household head is 0.024, which is negative significant for the total households sampled for the study while gender disparity in relation to farming experience has no influence. By implication, additional year of farming experience obtained by the household heads will result in 6 percent decrease in poverty incidence. This could be attributed to the fact that as the years of farming experience increase, the propensity to understand productive agricultural practices increase which could enhance productivity which at the long run increases household income and reduces poverty incidence. Family labour force of the agricultural holding in the context of our study refers to persons who carry out farm work on the holding and are classified either as a holder or the members of the sole holder’s family. The coefficient of family labor was found to be significant and negative. By implication, family labor reduces the likelihood of being poor in the study area but was relevant only to the female-headed households. The probable reason for this might be the ability of proper coordination the women possess through constructive persuasion and appealing promises after work done by the household members. Specifically, family labor has the probability of reducing poverty incidence by 34.1 percent.

With respect to the distance to the market, farmers in any households which is located far away from where the resources are being purchased and likewise the receiving markets of the farm products are expected to live below the poverty line due to high transactions costs such as transportation costs, loader cost etc., that is placed as an additional cost on farm incomes and reduces the possible take home of the farmers [32]. The findings of our study were consistence with this assertion as female-headed households were found to have probabilities of being poorer by 9.9 percent if there is a kilometer increase in distance to markets. Contact with extension services provided more access information on market opportunities and interventions, improved agricultural techniques, improved inputs, subsidized farm inputs and other production incentives. These would have a positive relationship on farmers’ outputs and their income-generating activities, thereby increasing livelihood diversifications and reducing their poverty level [33]. Poverty incidence will be probably reduced by 26.2 percent and 25.1 percent for male-headed households and the total sample respectively if there is access to extension services. Supportively, the availability of extension services and membership of extension related organizations improved farmers’ productivity and profitability and hence poverty reduction [33].

The coefficient of access to credit on reducing the likelihood of a household living below poverty line does not come with a surprise and significant at 1% for both male-headed households and pooled sample. The significant effects of credit access in elevating out and alleviating the households from poverty incidence is accounted by the cushion effects brought about by the funds, which can enhance the purchasing power of the households and possibly give opportunity for livelihood diversification through non-farm activities.

The coefficient of membership of social group is negative and significant at 5 percent for both male-headed households and pooled sample. This result lends credibility to the apriori expectation because membership of social group especially farmers association, has many competitive advantages in terms of risk reduction and uncertainty because of effective access to relevant information and to low or no interest credit sources and other important less subsidized inputs needed to enhance productivity process [34, 35]. Against this backdrop, it could be said that the incidence of poverty is higher among rural household heads that do not belong to any social group [10].