Forecasting Yield Curves with Survey Information

Abstract

In recent years, affine term structure models have provided alternatives to the expectations hypothesis and have become very popular in the finance literature. In particular, the widely accepted dynamic Nelson–Siegel model employs ingenious measures of the level, slope, and curvature of the yield curve that captured the attention of Francis and Hua. They supplement the dynamic Nelson-Siegel model with the Federal Reserve’s Survey of Professional Forecasters data. Because these data utilize information from dozens of professional forecasters who study numerous macroeconomic variables, the authors wanted to see if this information-rich supplementary data could be used to improve the interest rate forecasting models for out-of-sample forecasts for Treasury bond maturities ranging from three months to 10 years that extend from three months to one year into the future.

TOPICS: VAR and use of alternative risk measures of trading risk, factor-based models, statistical methods