Assessing the Determinants of Food Security Status in Bangladesh: A Micro-Econometric Analysis

Food security is an intricate issue which includes diverse aspects as well as many linkages. In Bangladesh, food security is tried to be achieved by increasing the production of rice both by employing modern agricultural technology as well as by increasing the area under rice production. Despite the impressive gains in increasing domestic food grain production, problems of food and nutrition security still remain. Bangladesh is yet to achieve comprehensive food security that resolves the problems of inadequate food intake and chronic malnutrition among those who are poor and vulnerable. The main objective of this paper is to the contribution of different factors behind household food security status of 180 households in three Northern districts of Bangladesh. The study area was chosen because relatively little energy consumption data are available concerning this geographical area. The study used both primary and secondary data. Food security status of each household was assessed on the basis of the food security line using the daily calorie intake recommended by FAO. This method has proven to be efficient in measuring food security at household level. Additionally, the use of a logistic regression model identified the factors that plays crucial role in determining the food security status of the households. Results from the food security index revealed that more than 60 percent of households were with food insecurity. In addition, we found that total monthly household income, age of household head, education level of household head, household size, farm size, gender of household head, livestock ownership and quantity of cereal production had significant influence on food security status at the household level.

increasing the production of rice both by employing modern agricultural technology as well as by increasing the area under rice production. Despite the impressive gains in increasing domestic food grain production, problems of food and nutrition security still remain. Bangladesh is yet to achieve comprehensive food security that resolves the problems of inadequate food intake and chronic malnutrition among those who are poor and vulnerable. Addressing these problems satisfactorily would not only require rethinking of strategies and policies to promote food security in the country but also require decisive actions by all stakeholders-the government, the NGOs, the private sector and individual households. In fact, ensuring food security through physical availability and economic access to food would continue to be a major challenge for Bangladesh in the coming years. Food security has been emphasized on the policy agenda of the government of Bangladesh and the government has been making efforts to augment domestic production of food grain through technological innovations and investments in irrigation, infrastructure development and subsidies. Food security is an intricate issue which includes diverse aspects as well as many linkages. Thus, the objective of this study is to assess the contribution of different factors behind food security status of Northern Bangladesh.
The rest of the paper is organized as follows. Section 2 briefly reviews the most recent published articles that discuss the concept of food security in Bangladesh and in other countries in the world.
Section 3 provides the methodology used in the empirical analysis of the study including data collection, research method, and micro-econometric model. Section 4 presents the results with required discussions and Section 5 concludes the study.

Recent Studies of Food Security
First, we present a brief review of those studies that examined factors that affected food security in Bangladesh. Second, we discuss the issue in other developing countries, such as Ethiopia, Nigeria, analyzed that the regional variation of food security in terms of food availability, food access, and food utilization under natural and socio-economic circumstances people became vulnerable to low food security in Bangladesh. They suggested that development of food polices would contribute to improved food security by considering regional food security conditions, region-specific deficits, climate change, other future risks, and devised actions related to the respective components.
In another study, Hossain et al. (2013) investigated the link between internal migration and food security in Bangladesh. The results of their study showed that the number of migrants at household level had significantly positive impact on the per capita calorie intake which was found to be increased by 2.2 percent for one unit increase in number of migrants. The researchers concluded that food security outcome was highly appreciable both at the origin and destination of migration. Anik et al. (2013) analyzed the impact of farm level corruption on households' food security using survey data collected from 210 Bangladeshi rice farmers. The researchers specified a regression model by using a multi-stage sampling technique and found that an average daily calorie intake for the sample households was 2,693 kilocalorie per person, whereas the national daily average for 2005 was 2,238 kilocalorie per person. The results showed that the cost of corruption adversely affected households' calorie consumption as the marginal effect of corruption was higher for the low expenditure households relative to the high expenditure households. Furthermore, variables such as the women education level and land ownership positively contributed to solving the problem of food security in the country. Rahman and Islam (2012) explored the nutritional status and food security of farm households under different land use patterns. The study revealed that households of alternate shrimp and rice farming consumed the highest amounts of food followed by year round shrimp farming whereas the highest per capita calorie intake was observed in households of alternate rice and wheat production. Large households consumed the highest amount of food and protein followed by medium, small and marginal households, respectively. Ahmed and Garnett (2011) examined the challenges of food security and found that rice monoculture could not provide a sustainable food supply without harming the environment in the long run. The authors suggested that higher yields might be achieved by increasing inputs in the integrated farming system. Using a Cobb-Douglas production function, the researchers suggested that higher yields might be achieved by increasing inputs in the integrated farming system. Ahmed and Garnett (2011) concluded that rice and fish production should be increased to meet the demand for food in the country. Kashem and Faroque (2011) focused on the availability of food as an essential element of food security and the role of government to get food sufficiency. The researchers found that Bangladesh attained self-sufficiency in food production which marginally met the country's per capita on a daily basis in 2010-2011. Kashem and Faroque (2011  physical and economic access to food by exploring the link between landlessness and access to food in rural Bangladesh and found that landlessness of the farmers led to their insufficient purchasing power to buy adequate nutritious food for their families.

Evidence Outside Bangladesh
There are some studies which examined the determinants of household food security using different quantitative techniques. For instance, Osabohien et al. (2018) explained that the efficiency of credit facility positively contributes to agricultural sector of the agro-based economy in Nigeria by the Autoregressive Distribution Lag (ARDL) econometric approach on the time series data. The results of the study showed that commercial banks' credits and agricultural credit guaranteed scheme fund (ACFSF) increased food security by 8.12 and 0.002 percent, respectively, while the rate of population reduced food security by 0.001 percent. They recommended that population growth rate should be controlled through family planning and adequate financing of the ACFSF by the government and monitor commercial banks lending interest rates on credit facilities. Fawole and Ozkan (2016) investigated the food security status at the household level in Nigeria using daily per capita calorie consumption method. The researchers found that 54 percent of the households surveyed were food insecure implying that their daily per capita calorie intake fell below the recommended daily per capita calorie requirement and food insecurity gap of 0.0038. The main limitation to this study was the inability to collect data on the income distribution, budget share on foods and socio-economic characteristics of the households which could affect households' food security status. Endalew et al. (2015) examined the factors that affected food insecurity in Ethiopia. The authors found that poverty was not the only reason that caused food insecurity, but also population pressure, drought, shortage of farmland, poor soil fertility, and poor farming technologies had significant impacts on the food security   (2011) concluded that three factors were statistically significant with expected signs which were place of residence (urban), dependency ratio and households' level of education. Babatunde et al. (2010) investigated the nature of the relationship between income and calorie intake among farming households in rural area of Nigeria using both parametric and nonparametric techniques. The result of the research showed that the average daily per capita calorie intake in the region was 2,428 kilocalories, which was below the recommended intake level. Moreover, the findings of the study showed a positive relationship between calorie intake with income and farm size. Babatunde et al. (2010) suggested that a combination of policy strategies, including income growth, agricultural development and targeted food programs might be reduced the problems of inadequate calorie consumption among poor households.
Finally, Khan and Gill (2009) examined the determinants of three aspects of food security, i.e., food availability, accessibility, and absorption in rural areas of Pakistan using district level data. The study found that the production of wheat, rice, maize, pulses, oilseeds, poultry meat and fish at the district level had a positive effect on the availability of food. The researchers concluded that all the district except of Sindh were more probable to be food insecure in availability. In addition, electrification and adult literacy emerged as the factors having negative impact in the food accessibility. Besides, child immunization, safe drinking water, and number of hospitals had positive effects on food absorption.
In summary, most of the above articles focused on different issues of food security with diversified findings worldwide. Some studies found that total annual income, household size and level of educational, and quantity of food obtained from own production were the significant factors of food security in different countries. Most of the studies carried out in Bangladesh have analyzed the food security situation at the national level and not at the household level using production, import, and aid based data of food grains. In addition, the literature showed a few studies in the context of food security in the country for which strong quantitative tools were used in their empirical analysis. Moreover, precedent studies did not cover all part of the country, such as north western Bangladesh as it contains distinct geophysical and socioeconomic characteristics. Our study contributes substantially to the literature on the following grounds. First, we used quantitative tools in the empirical section of the paper. Second, the geographical area of our study is north western Bangladesh. Finally, our study is at the household level since the data collected at the national level were unable to identify differences in food security amongst the households as it is the case and data specific.

Study Area
The study is undertaken in the north western part of Bangladesh which is covered 43 percent area of Rajshahi Division known as Rajshahi, Natore and Naogaon districts. Geographically, this part of the country has been identified as the most important area for food production because of the availability of cultivable land. Nearly 40 percent of total population of Rajshahi Division lives in this area (see Figure  Figure 1

. Study Area Map
Source: Adopted from https://en.wikipedia.org

Data Sources
We used a multistage sampling technique to select the households and interviewed them. The first stage involved the selection of districts from which the participants in the survey were selected for interview.
This was done using purposive sampling techniques where the districts were grouped into northern part of Bangladesh. The second stage involved selection of sub-districts, unions and villages using the simple random sampling method. This was achieved with the help of district's recorded information.
The second stage involved a random selection of two sub-districts from each of the three selected districts and two unions from each of the six selected sub-districts, and one village was selected for the www.scholink.org/ojs/index.php/elp Economics, Law and Policy Vol. 3, No. 2, 2020 selected unions. The final stage was the selection of the households that were interviewed. We selected households using simple random sampling and collected primary data regarding their socio-economic characteristics, food production, food consumption, and other factors. The population for the study comprised both farm-and-non-farm households of northern part of the three districts of Bangladesh. In the last stage, 180 households were selected from the earlier 12 villages sampled for the study, i.e., 15 household per village.

Research Methods
This research mainly relies upon primary data collected from sample households of the northern part of Bangladesh. With a view to fulfilling the research objectives with regards to determining the food security status, identifying the factors having influence on food security and drawing a comparison among the different households belonging to different living standards and different socioeconomic characteristics, the sample was selected in such a way that it covered all the aforementioned criteria. To identify the factors influencing food security status of the households, we carried out a two-stage analysis method which was as follows. First, we constructed a Food Security Index (FSI), and second we specified a logistic regression model to estimate the food security status of households as a function of a set of independent variables.

Food Security and Head Count Indexes
This study analyzed the food security of rural households by employing a quantitative approach. To do this some statistical techniques, such as adult equivalent unit, Food Security Index (FSI), and head count index were applied. A couple of methods have intensively been used in the literature which wee the expenditure and calorie intake methods (Maxwell, 1996). In this study we chose the calorie intake method, through constructing a FSI, as this method reflects the actual food consumption pattern of households (e.g., Adeniyi & Ojo, 2013;Ahungwa et al., 2013;Babatunde et al., 2010;Iorlamen et al., 2013Iorlamen et al., , 2014and Kuwornu et al., 2013). Food security status of each household is a measure based on the food security line using the daily calorie intake recommended by FAO (2005). The average daily calorie requirement for a moderately active adult is 2,850 kilocalorie and a safe minimum daily intake should not fall below 80 percent of the above calorie requirement (i.e., 2,280 kilocalorie). This food security line is used in this study after converting all household members into adult equivalent units (e.g., May 1996, Swindale & Ohri-vichaspati, 2005.
For construction of the index, first, we took the quantity of food consumed by the households in 7 days period and then estimated the obtained calorie intake. We converted the quantities to gram and estimated the calorie content by using the nutrient composition table of commonly eaten foods in Bangladesh (GoB 2005;BIDS 1997). A household whose daily per capita calorie intake was at least 2,280 kcal was regarded as food secure household, while any number beyond that was regarded as food insecure household. We also calculated the head count index on the basis of the food security status of households (FAO, 2005;Fakiyesi, 2001;Olayemi, 1998

A Micro-Econometric Model
In addition to measuring the food security status of participants in the survey, we used a logistic regression model to identify the factors that determined their food security condition. To do this, precedent studies have used various models, such as the Cobb-Douglas production function model ( A logit model is defined as follows. Let the probability that a household is food secure can be written as where, X i is the independent variable and Y=1 means that the household is food secure. (2) which means the log of the odds ratio is not only linear in X i but also linear in the parameters. The parameters of the logit function (L i ) are estimated using the Maximum Likelihood (ML) approach. In this study the empirical relationship between the dependent and independent variables is specified in where, L i =log odds ratio in favor of a household being secure in food, OL=livestock ownership, QCP=quantity of cereal production.
Since the collected sample data in the model was cross sectional the multicollinearity problem might arise. Thus, we calculated partial correlation coefficients among the independent variables used in the model and found that the values of correlation coefficients were less than 0.5 implying that multicollinearity among variables did not appear to be a serious problem in the model. Table 1 lists the definition of variables, their measurements and expected hypotheses.   Babatunde et al., 2007Babatunde et al., , 2010Kuwornu et al., 2013;Ahungwa et al., 2013 www.scholink.org/ojs/index.php/elp Economics, Law andPolicy Vol. 3, No. 2, 2020

Results
In this section, we presented the results of the empirical analysis on the issue of food security at the household level.

Food Security Situation: Aggregate Level
Ensuring food security for all is one of the major challenges in Bangladesh today. Despite the impressive achievements in food grains production during the last few decades, food security at the households and individual levels remains a major concern for the government. Different aspects and conditions of the food security at household level are described in this section.

Food Expenditure Pattern in the Sampled Districts
In general food access depends upon the price food, household income and its distribution within households, and household size. In addition, the sources of income and their reliability for a steady flow and reliable amounts are important to both individuals and households for ensuring food. Table 2 shows the actual food expenditure scenarios in Bangladesh and selected districts in 2016. It is observed that the required food expenditure of household per month at the national level was 5,346.29 (Tk) with a food expenditure gap of 524.47 (Tk). Moreover, the average required and actual food expenditure of households in Natore was higher than that of in the other two regions (Table 2). When access gap is compared amongst the three regions the same pattern was observed (Table 2).

Per Capita Food Intake Per Day
In order to achieve food security, the food ingested must be safe and enough to meet the physiological requirements of each individual. The daily per capita food intake is essential to ensure nutritional requirement of body. Figure  vegetables, 2 percent of sugar, and another 2 percent of fruits. Cereals, largely rice, are the main food in Bangladesh. Nearly two-thirds of the daily diet of Bangladeshi people consists of rice, some vegetables, a little amount of pulses and small quantities of fish. Milk, dairy products and meat are consumed occasionally and in a very small amount (Islam & Buysse, 2012;Kabir et al., 2018). The consumption of protein and micronutrient-rich foods like fish, meat, eggs, milk, dairy products, fats and oil is often low at the rural poor areas in Bangladesh (Lalita et al., 2007).

Figure 2. Minimum Required Food Intake
Source: Government of Bangladesh 2005. Figure 3 depicts the percentages of food consumption according to food items by the households per capita on a daily basis in the study area. It is revealed that the staple food and vegetables, respectively, included 50 percent and 20 percent of total consumption of food. Moreover, households received most of the calories from the consumption of rice and vegetables. The next important food item is wheat followed by potato, fish, pulse, milk and milk product, egg, sugar, oil, fruits and meat.

Figure 3. Actual Food Consumption
Source: Sample data.

Results from Indices: Micro-Level
To assess whether a household in the sample observation was food secured or not, we calculated the food security index suggested by Fakiyesi (2001). We also compared the food security status of household in different districts which is explained as follows.

Head Count Index
We classified respondents in the survey into food secure and insecure on the basis of recommended daily calorie intake of 2,280 kilocalorie suggested by FAO (2005). Figure 4 shows the headcount index calculated for the participants in the survey. It is observed that the ratio of all sample food unsecured households was 0.61 indicating 61 percent (110 households) of the total participants in the survey were food insecure. However, this scenario was different for the three districts in the study. The headcount index of food secure for the Naogaon region (0.47) was higher than the other two regions. Table 4 shows that the most food insecure district was Natore followed by Rajshahi.

Figure 4. Headcount Index for the Study Area
Source: Sample data.

Food Security Status at the Household Level
We calculated a food consumption index (FSI) that enabled us to understand whether an individual was secured in the availability of food. Table 3 presents the mean value of FSI for a participant in our survey measured in kilocalorie (kcal). Table 3 shows that in case of all households in the survey the mean value of FSI was higher in Naogaon district than that of the other two districts. Furthermore, respondents in the Rajshahi district were more food secure when compared to the other regions in the survey.  Table 4 presents the empirical results of the logit model obtained from collecting data from 180 participants in the survey. We conducted our analysis in three phases to explore the influence of the key factors on food security status. First, we combined all the collected data and estimated the parameters of the regression. Second, we separated the sample observations into three districts where the data were collected and estimated the parameters of the regression separately.

Estimation Results
Finally, we compared the estimation results of the three regions to find any significant variations amongst the districts. Table 4 shows the comparative analysis of food security determinants in all three districts. It is observed that four out of eight variables included in the regression model were statistically significant at the 0.01 level. These variables were total monthly income, quantity of cereal production, educational status of household, and farm size in all study areas.  In the following we briefly provide the explanation for each of the statistically significant coefficients.

Total Monthly Income
An increase in the household income helped it afford more food for the family which made the family secured in food. It was found that the coefficient of the variable "monthly income" (0.0006) was positive and statistically significant at the 0.01 level. The coefficient of the variable monthly income was positive and statistically significant with 99 percent in the Natore and Rajshahi districts implying that the households with higher income level had the higher probability of being secured in the consumption of food but this coefficient was not statistically significant in the region of Naogaon. This result indicated that the higher the income of a household was the higher the probability that the household would be secured in the consumption of food. This result was consistent with Adeniyi and Ojo (2013)

Age of Household Head
The coefficient of the variable "age" had a negative sign and statistically significant at the 0.1 level only in the Naogaon district reflecting that household head with higher age was less likely to be food secure compared to the other regions. This might stem from the fact that the respondents with less than

Education Level of Household Head
An educated household is more informative about the consumption of food. The result of the study showed that the coefficient of education was positive and statistically significant with 99 ercent confident only in the Natore district implying that respondents in the Natore district were more conscious about food consumption than the other two districts. This result suggested those households with relatively higher level of education strived better than those with lower level of education (see, e.g., Adeniyi & Ojo, 2013;Ramakrishna & Demeke, 2002).

Household Size
The coefficient of the household size variable was negative (-0.69) and statistically significant at the 0.1 level only in the Naogaon district. This indicated that as the household size increased, the probability of food security decreased in this district than that of those respondents who lived at the Rajshahi and Natore districts. In other words, large size households were more likely to be food insecure than small size households. The result of our study was in line with the findings of Adeniyi

Farm Size
The result of the study showed that a household with more cultivable land could produce more crops which ensure food for the family. The coefficient of farm size (0.289) was positive and statistically significant with 95 percent confidence.

Gender of Household Head
The result of the research showed that the gender of household head had a positive impact on the household food security. The coefficient of gender of household head was statistically significant at the 0.1 level only in the Natore district. This result showed that a household with male headed was more likely to be secured in food compared to the one with female headed. Precedent studies concluded the same result as the finding of this study (e.g., Babatunde et al., 2010). However, this conclusion should be interpreted cautiously because the coefficients of the gender of household head were not statistically significant at the aggregate level and other districts.

Livestock Ownership
The coefficient of the livestock ownership was not statistically significant for the food security status.
This finding was not in line with theoretical expectation outlined in Table 1. One reason that could justify this result is that most of the households participated in the survey did not own sufficient livestock population.

Quantity of Own Cereal Production
The result of the regression model showed that the coefficient of the quantity of own cereal production (0.134) was positive and statistically significant with 99 percent confidence in the region. This implies that those households who produced more cereal crops were more likely to have security in food through increasing either their purchasing power food self-consumption.
The finding of our study was also consistent with the results of Ahungwa et al. (2013), Babatunde et al. (2010) and Kuwornu et al. (2013).

Conclusions and Policy Implications
The main objective of this study was to assess the contribution of different factors behind food security status of Northern Bangladesh. To achieve our research objectives we constructed a food security index and used a binary logit model. Finding from the food security index revealed that households in the study areas were nearly food-secure as the mean of the FSI value was slightly below one. Moreover, using the index along with the food security line based on calorie intakes we found that 39 percent of the respondents were food-secure. The estimation results of the econometric model showed that some factors such as monthly household income, level of education of household head, farm size, and quantity of own cereal production had positive effect on the food security of the households. Any changes in these factors would change the status of food security. However, the findings of the research varied across the districts (i.e., Natore, Rajshahi and Naogaon). It was found that four statistically significant variables which were age of household head, household size, quantity of owned cereal production and farm size for Naogaon whereas total monthly income, educational status of household, gender of household head and farm size were statistically significant for Natore. In the Rajshahi district, only three variables namely total monthly income, quantity of owned cereal production and farm size were statistically significant. Moreover, food security situation at national/aggregate level and household level in the study area are completely different and influencing factors the food security situation in this area are also different. Because, at the national level, food security situation has improved in context of availability side, although Bangladesh remains a food insecure country in terms of access to food and utilization of food. Considering income and food expenditure of household it is found that the households lacked to achieve the ability of food expenditure which is necessary to meet food requirement for the members of their households.
Based on the major findings of this study some policy recommendations might be devised. We recommended the government should expand alternative income generating activities for the people who lived in the rural regions. We also suggested that the household heads and other members who engaged in farming activities should be educated or trained through programmes like training and farmers' field schools which would have potential to enhance their capabilities, which, in turn, would improve households' productivities. Educated households will be more conscious about their dietary composition for better health. Furthermore, we recommended that farmers should have easy access to reliable and quality inputs such as seeds, fertilizers and credit support to increase production. These suggestions will help the marginal farmers enhance their farm production practicing. Finally, farming households should take advantage of the food security by increasing their farmland holding through government owned land distribution. However, globalization is connect with achieving food security because globalization refers to the way in which developments in one country or area or region can rapidly come to have significant consequences for the security and well-being of communities in quite distant countries or areas or regions of the world. As a result, it has a positive influence on the globally household food security status, especially in the developing countries like Bangladesh where food related crises are frequently occurring.