Short-Term Inflation Projection Based on Marker Models

Abstract

The article proposes a hybrid model for inflation projection that combines econometric and neural network models. At the same time, both factor variables and market markers of consumer price index growth are used as explanatory variables. It has been shown that such an approach makes it possible both to maintain the theoretical fullness of the model and to ensure high accuracy of calculations, which is unattainable when using only one type of model toolkit.