Predicting Changes in Household Consumption Using Neural Networks

Anastasia D. Petaykina – Junior Researcher of the Russian Presidential Academy of National Economy and Public Administration (Moscow, Russia). Е-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.

The article is devoted to research of influence of income changes on consumption of the households in Russia – in particular, the task is to develop and train a neural network capable of forecasting changes in consumption based on data on income changes, individual characteristics of households, as well as the factor of the existence of liquidity constraints. The relevance of the use of a neural network is explained by the assumption of a non-linear relationship between consumption and factors that can explain its change. The study is conducted using data broken down by individual households available in the RLMS HSE database for the period from 2006 to 2019. According to the results of the study, it was found that the use of neural networks improves the predictive properties of the model compared to the use of linear regression, which is evidence in favor of the existence of non-linear relationships between the indicators.

The article was prepared within the framework of the research work of the state assignment of RANEPA.

Key words: households’ consumption, neural networks, multilayer perceptron, RLMS.

JEL-codes: C23, C45, C53, E21.