The panel data regression concept in consumption modelling

Abstract

Panel data research constitutes a new methodological approach to studies covering the area of food consumption. The modelling procedure described in this article was carried out taking into account the cross-section of individual types of households and product categories. Product categories included: bread and cereals, cakes and bakery products, meat, fish, milk, yoghurts and dairy drinks, cheese, oils and other vegetable fats, animal fats, fruits, vegetables, confectionery products and juices. Surveys focusing on households budgets in the period of 2003-2019 and provided by the Central Statistical Office of Poland (CSO) were the source of information used for the panel data analysis applied in this work. The author has undertaken the task of building all the presented models for panel data, namely fixed effects, random effects and pooled regression. It turned out that the correct models are those with fixed individual effects. In conclusion, it was established that panel models are a useful tool for modelling the consumption of food products. When analysing the constructed models, it is possible to observe significant differences in the consumption of the examined products between the correspondent quintile groups. This study confirms the tendency in food consumption in Poland, observed in 2003-2015.