markets ensure that farmers are either insured against losses, can borrow around them, or can receive help from the government to maintain livelihoods during bad times” (p.135). 4. Methodology The methodology builds on the framework of Lovendal and Knowles (2006) and the study by Karfakis, et al. (2010). Consistent with the literature, food insecurity will be measured in terms of (1) food expenditure and (2) malnutrition. Correspondingly, two datasets are employed, namely: the 2009 Family Income and Expenditure Survey (FIES) and the 2007-2010 pooled Community Based Monitoring System (CBMS) data of 16 provinces. The model specification and estimation methodology will depend on the available variables and measure of food security available from each of the datasets. The following subsection presents the matching of variables across different surveys with the variables used by Karfakis, et al (2010) in their study. Data mapping Identifying the variables for the model as executed in the FAO study is almost straightforward for the national survey of the Philippines which is the FIES as well as for the CBMS data of local government units (LGUs). Table 4 shows mapping of the variables (dependent, explanatory and instruments) used in the FAO study with FIES and CBMS data. Table 4. Data and variable mapping and initial variables Metrics or substitute metrics Model variable Dependent variable Food security Agricultural productivity FAO model FIES data CBMS data foodexp: malnutrition Value of food consumed per adult equivalent With Per capita malnourished expenditure children 0-5 on food years old Value of agri production per acre cropinc: Explanatory variables Income from crop farming Age of HH head Characteristics of the head of the household Years of education of HH head age_yr: Age of HH head educlvl : CBMS-FAO | 18