Dynamic Models for Estimating the U.S. Treasury Yield Curve for the Costa Rica Central Bank

This research seeks to design and implement a quantitative financial model for the Treasury bond’s yield curve estimation, using a mathematical formulation that estimates, based on updated financial indicators, the trend that the interest rates are expected to have in a short-term projection. Three models were created, starting with a simple Vector Autoregression (VAR) model, an intermediate complexity Dynamic Nelson-Siegel (DNS) model, and ending with a high complexity model that incorporates macroeconomic data to the DNS model, with the purpose of being used by the Costa Rica Central Bank as a new tool to adjust its investment strategies with less uncertainty.