The determinants of household food expenditures in Turkey: A comperative analysis

Consumption expenditures, as it is known, has been a subject affecting the whole society starting from the family and the family members who constitute the smallest building block of the society in every society. The need for food and shelter is an ongoing situation since the existence of human beings. There are many factors affecting food expenditures. These are demographic and behavioral factors such as household income, age, educational status, marital status, number of children in household, occupation, place of residence, wealth, consumer preferences, plans, optimistic and pessimistic moods. The first purpose of this study is to analyze the factors that determine the food expenditure using Household Budget Survey micro data set of the 2007, 2012 and 2016 conducted by TUIK and to see the changes over time in Turkey. The second aim is to contribute to the literature by comparing with semi-logarithmic regression and ANN models by forecasting household food expenditures and comparing the forecasting performance of both methods. According to the findings obtained from the models in the analysis, the increase in the age and income of the household head has increased food expenditures. It has been revealed that the variables that cause the increase of food expenditures are due to the increase in household size, age, education and income by years. In addition, when the estimation methods have been compared, ANN has been found to be a stronger and alternative method than semi-logarithmic modeling. Keywords: Household Food Expenditure, Semi Logarithmic Model, Artificial Neural Network, Forecasting, Turkey.