An Independent Dynamic Latent Factor Approach to Yield Curve Modeling

Sammanfattning: Understanding the yield curve characteristics and dynamics is important for many tasks such as pricing financial assets, portfolio allocation, managing financial risk, and conducting monetary policy. Therefore, it is important to use models that are interpretable, fits well, and make useful forecasts. In this paper, I introduce a dynamic yield curve model with latent independent factors based on Independent Component Analysis, which is a statistical method used successfully in other fields than finance. I find that one can interpret the factors as level, slope, and curvature of the yield curve. I also find that the ICA-based model fits the yield curve well and produce good forecasts. In particular, it shows significantly better out-of-sample forecasts for the short-term maturities than the commonly used dynamic Nelson-Siegel model. I find that the factors correlate with macroeconomic variables such as monetary policy instrument, real economic activity, and inflation. Finally, I find that the curvature factor seems to be more important than the previous literature state.