Using Artificial Intelligence to Monitor Inflation

The article proposes a new digital technology of inflation targeting based on daily monitoring of inflation processes in the country. The purpose of the study is to identify potential opportunities for implementation of neural networks in collecting and processing large volumes of data by statistical services. It will lead to improvements in the efficiency and accuracy of statistics provided, which will be shown on the example of the CPI. Thisresearch is based on the analysis of the use of artificial intelligence technologies applied in foreign countries in the financial market. The novelty of the study lies in the development of a neural network structure that will allow targeting both demand-pull inflation and supply-pull inflation simultaneously. It is assumed that the core of this technology will be a modular controlled recurrent neural network, the architecture of which was developed to solve this specific problem. Taking as input and processing in hidden layers daily values of macro- and microeconomic variables that can affect inflation dynamics, the neural network will calculate as output the daily value of the CPI. If this value turns out to be higher than the target inflation level set by the Bank of Russia, then the neural network, during backpropagation of the error, will issue recommendations for adjusting some of the neural network weighting coefficients and the values of the macro- and microeconomic variables themselves to reduce the inflation level. The paper concludes that only a comprehensive approach to combating such a complex economic phenomenon as inflation, which consists of taking into account not only demand-pull inflation but also supply-pull inflation, will bring tangible results.