Inflation indexed swap pricing risk: is the market standard approach for seasonality modeling still reliable?

GZ - Psychologie · Springer Nature · 2026-08-18 · DOI: 10.1007/s10203-026-00599-0

Аннотация

Inflation modeling has become of great interest again in financial research. After years in which investors have grown accustomed to a world of low (if not negative) interest rates, inflation has been rising sharply and unexpectedly over the past few years. The paper aims to make a comparison between two models for determining future values of inflation: a traditional seasonality modeling methodology and a highly nonlinear approach, based on a Long Short-Term Memory architecture. Results show that the LSTM model, based on the deep learning methodology, depicts a more accurate behavior of the projection of future inflation rates compared to the more traditional approach, since the LSTM-based model is also able to generate a more accurate within-tenor inflation path in a generic point in time, while keeping the projections anchored to market-implied inflation rates values.

Читать у издателя →

Похожие статьи