DETERMINANTS OF FOOD INSECURITY AMONG FARMING HOUSEHOLDS IN KATSINA STATE , NORTH WESTERN NIGERIA : AN ORDINAL LOGIT REGRESSION APPROACH

The study identified the determinants of food insecurity among farming households in Katsina State, north western Nigeria. A cross sectional sample survey design was used to select a total of 150 small-holder farmers from 15 communities across 10 Local Government Areas of the state. A structured questionnaire, Focus Group Discussion and Key Informant Interview were used for data collection. The coping strategy index was used to determine the food security status of the household and the ordered logit regression was used to identify the determinants of food insecurity among the households. The majority (73%) were found to be food insecure. In terms of food insecurity status, 44% of the respondents were less food insecure, while 17% and 12% were moderately food insecure and severely food insecure respectively. Eating the less preferred meal, purchasing food on credit and reducing the quantity of food consumed were the major coping strategies adopted by the food insecure households. The result of the ordered logit model shows that the total quantity of cereal saved, number of income sources and dependency ratio were significant for both the moderately and severely food insecure groups at p<0.05 while access to credit was also significant for the two groups but at p<0.01. The output of other crops was significant at p=0.10 but only for the severely food insecure group. The study concluded that food insecurity was high in the study area and therefore recommended that the farming households be provided with opportunities to diversify their livelihood activities.


Introduction
Katsina state is one of the Northern states that have not been the focus of food security studies in Nigeria.According to NEST (1991), the state is faced with enormous environmental challenges, hence food insecurity is prevalent.On the other hand, previous research efforts on food security in Nigeria have failed to Hussaini Y. Ibrahim et al. 292 identify the determinants of food insecurity for households with varying food insecurity levels ranging from less, moderately to severely food insecure (Omotesho et al., 2006 ;Idrisa, Gwary and Shehu, 2008;Ayantoye et al., 2011;Olagunju et al., 2012;Obayelu, 2012;Ayoade and Adetunbi, 2013;Adepoju and Kayode, 2013;Henry-Ukoha, 2013;Agada and Igboke, 2014;Ahmed and Naphali, 2014;Edeh and Gyimah-Brempong, 2015).Rather, these studies classify households into two groups of either being food secure or food insecure and logit as well as probit regression models were used respectively to identify relevant policy variables that influence food security.However, to make policy relevant recommendations that will be instrumental in combating a herculean development challenge such as food insecurity and to improve the nutritional and health outcomes among farming households, it is pertinent to identify the relevant variables that can influence or determine food insecurity depending on the level of food insecurity of a given household.Furthermore, an understanding of factors that affect different household groups is therefore invaluable as factors that affect food security for particular households might not necessarily apply to other household groups (Abdallah et al., 2013).In addition, according to Dare et al. (2013), more detailed analyses of food security at the household level are still needed in Nigeria.

Objectives of the Study
The objectives of the study are as follows: -to determine the level of food insecurity in the study area; -to identify the coping strategies adopted when there is food insecurity and -to identify the determinants of the varying levels of food insecurity among the farming households in the study area.

Material and Methods
The study was conducted in Katsina state and a cross sectional sample survey design was adopted.A total of 15 communities from 10 LGAs in the state prone to droughts, desertification and flood were sampled for the study after a reconnaissance survey.Ten farming households were sampled per community to give a total of 150 smallholder farmers for the study.An interview schedule was used for data collection.A range of participatory tools such as focus group discussion and key informant interviews were also utilised.

The Coping Strategies Index (CSI)
The coping strategies index (CSI) according to CARE and WFP (2003) was used to estimate the level of food insecurity and to identify the coping strategies used by farming households.Corbett (1988) has opined that one of the ways of identifying food insecure households is to look at the types and the number of times each coping strategy is used to manage food shortages or inadequacies.The basic idea of CSI is to combine the frequency and severity of coping strategies.The frequency of coping strategies requires the means of scoring of relative frequency which measures how many days per week a household has to rely on the various coping strategies ranking from "never" to "frequently".The mean of the scores was used as the value for a given category as presented in Table 1.The consensus ranking of coping strategies (Table 2) was determined using a focus group discussion.The method of scoring shows the severity weight of each coping strategy that a household has chosen.The mean average severity weight was estimated by finding the product of the mean intensity of severity for each of the chosen coping strategy and its consensus ranking from the FGD.Thus, the CSI score was calculated by combining both "frequency" and "severity" of coping strategies.The result of the emanating coping strategy score denotes that the higher it is, the higher the insecurity level of the household.
The ordered probit regression model was used to assess the factors that influence the food insecurity situation among the farm households based on the results of the CSI scores.The formulation of the model is such that the responses are represented by a variable Yi which denotes an average CSI score.The average CSI score presents the level of the food consumption gap and food insecurity among the rural households.In this case, the explained variable is grouped and ranked based on the level of the CSI score.The CSI score denotes that the higher its value, the higher level of food insecurity, and the other way round.The CSI score rank will be applied for all households i, and it takes the following four (4) ordered values (j = 0, 1, 2, 3).However, these observed values are assumed to derive from some unobservable latent variable Yi*, which is expressed in this equation as follows: is the hypothesised predicators of food insecurity, βs is a vector of parameters to be estimated, and is an error term which is assumed to be normally distributed (Greene, 2003).The values for the observed variable Yi are assumed to be related to the latent variable Yi* in the following manner: µ means the unknown threshold parameters.For the estimated cut-off points, µ follows the order µ1 <µ2 <µ3.Using the maximum likelihood estimate technique, the values for the βs parameters can be estimated.The marginal effects were computed to elucidate the interpretations of the ordered probit coefficients.The marginal effects or the marginal probabilities show the probabilities that would rank the CSI in any of the four level categories or groups of less food insecure, moderately food insecure, or severely food insecure and food secure due to a unit change in a particular variable (Gloy et al., 2000).
The expression below was used to estimate the relationship between the CSI ranks that indicate food insecurity and the other attributes of socioeconomic and farming characteristics.For our purpose, the specific food insecurity model can be implicitly written in the following way: CSI rank = F (X 1 , X 2 , X 3 , X 4 , X 5 , X 6 , X 7 , X 8 , X 9 , X 10 , X 11 , e) The CSI rank represents the dependent variable; it was measured as an ordered response, and reflects the food insecurity among the rural farming households.The explanatory variables are defined as follows: X 1 = total cereal saved for home subsistence (kg grain equivalent), X 2 = total output produced from legumes (kg grain equivalent), X 3 = total output produced from other crops (kg grain equivalent), X 4 = average of crop diversity (Shannon index), X 5 = number of income sources for the household head, X 6 = dependency ratio (number of persons unemployed/number of persons employed), X 7 = total number of livestock (tropical livestock unit), X 8 = age of the household head (years), X 9 = access to credit, X 10 = farm size (hectares), X 11 = household social capital index, The STATA version 11 computer program was used for all statistical analyses.

Household food security status
The food security status of the farming households is as presented in Table 3.The majority (73%) of the households are food insecure.In terms of the level of food insecurity, most of the food insecure households are less food secure followed by the moderately and severely food insecure households respectively.This obviously shows that the food security situation in the area is perilous.Previous studies on food insecurity among farming households in Nigeria have also pointed out that the food security situation among farming households in Nigeria is precarious.Ayoade and Adetunbi (2013) observed that food insecurity was about 65% among farming households in south western Nigeria.Adepoju and Kayode (2013) reported that almost half (49.4%) of rural farming households in the Nigeria were food insecure during the post-planting period.The high level of food insecurity among farming households is quite alarming despite the fact that the bulk of agricultural production activities take place in these areas.Food insecurity coping strategies The food insecurity coping strategies employed by farming households are presented in Table 4.The result shows that eating of less preferred meals ranked first followed by purchasing food on credit, reducing or rationing quantity of food consumed and skipping of meals in that order.This finding implies that households faced with food insecurity do not rely on a single coping strategy during periods of food shortages.Idrisa, Gwary and Shehu (2008) also observed in a study conducted in north eastern Nigeria that several coping strategies ranging from eating once a day, allowing children to eat first and buying food on credit were employed to cope with food insecurity.The finding also agrees with that of Agada and Igboke (2014) in a study on coping strategies during periods of food shortages.They observed that the commonly adopted coping strategies by ethnic groups in north central Nigeria were reliance on less preferred food and limiting food portions at meal times rather than on skipping of meal by eating once a day.On the other hand, the least utilised coping strategies include: sending household members to eat elsewhere, begging for food and gathering of wild fruits.These strategies obviously imply some form of shame to a household, hence seldom used in the area.In terms of the relative usage of food insecurity coping strategies, Table 5 shows that most of the farming households never send their members to eat elsewhere or to beg for food.This result corroborates the findings in Table 4.It further shows that households frequently use the skipping of meals, eating less preferred food and reducing the quantity of food consumed in order to bear with the incidence of food insecurity in their households.Determinants of food insecurity.
The results of the ordered probit regression model are presented in Table 6.The overall model is highly significant (p<0.01) with a lower log likelihood of about -160.73422.The estimated cut-off points (µ) satisfy the conditions that µ1 < µ2 < µ3.This implies that these categories are ranked in an ordered way (Knight et al., 2005).The first cut-off point (Y=0 for "food secure group") was used as a mark for the purpose of comparison.The table also shows the significant explanatory variables that influence the household food insecurity.The total amount of cereal saved, number of income sources, amount of credit and dependency ratio are the significant explanatory variables that influence the household food insecurity p<0.05.However, extent of crop diversity, number of livestock kept, farm size, social capital, outputs of other crops, age, and total legume saved were not significant.
The marginal effects for the probabilities of farm household food insecurity situation as "less food insecure," "moderately food insecure," and "severely food insecure" as opposed to "food secure" are also presented in Table 6.None of the variables has a significant effect at p=0.10 on food insecurity for the less food insecure ranks, but the marginal effects for these variables are all negative.The negative sign of these variables indicates that an increase will cause the level of food insecurity for the less food insecure ranks to decrease.On the other hand, the total quantity of cereal saved, number of income sources and dependency ratio were all significant for both the moderately and severely food insecure groups at p<0.05 while access to credit was also significant for the two groups, but at p<0.01.The finding for the amount of cereal saved agrees with the finding of Babatunde, Omotesho and Sholotan (2007).Cereal crops such as millet, maize and sorghum are major staple crops in north western Nigeria and the quantity a household is able to save is an indication of food availability in the household.The number of income sources implies the availability of off-farm employment opportunities.A study by Babatunde and Matin (2010) showed that off-farm income contributes to higher access to food and household welfare.The result for access to credit is also in line with that of Olagunju et al. (2012) and Ayantoye, Yusuf ,Omonona, and Amao (2011).If farming households have a more secure access to credit, they will be able to purchase improved inputs and livestock which can be sold and also consumed in the household.
The output of other crops was significant at p=0.10, but only for the severely food insecure group.An increase in the output of other crops especially cash crops such as vegetables and cotton will lead to a boost in household income.The income acquired from selling cash crops would eventually assist and support the accessibility to food for the severely food insecure group.This finding underscores the need for packaging advisory service activities aimed at enhancing cash crop production among the severely food insecure groups.The result for the dependency ratio was not expected and it contradicts previous studies on determinants of food security (Ojogbo, 2010;Beyene and Muche, 2010;Sultana and Kiani, 2011;Adepoju and Kayode, 2013).The practice of utilising under-aged children on the farm could be a likely explanation for this result.It is very common in northern Nigeria to see very young children helping out on their parent's farms.These children obviously contribute to available farm labour and subsequently farm output.If the marginal effects are considered, it is very clear that an increase in total quantity of cereal saved, number of income sources, access to credit and dependency ratio will have a greater impact on reducing food insecurity in the moderately food insecure category than on any other group.This finding portraits the importance of a livelihood diversification program, a post-harvest handling training and a farm credit scheme in reducing the level of food insecurity among the moderately food insecure groups in north western Nigeria.

Table 1 .
Numerical value of relative frequency of coping strategies.

Table 2 .
Consensus ranking of coping strategies.

Table 3 .
Level of food insecurity among farming households.

Table 4 .
Food insecurity coping strategies utilised by households.

Table 5 .
Relative usage of food insecurity coping strategies.

Table 6 .
Determinants of food insecurity among farming households.