Market-consistent economic scenario generation system for treasury bond yields with the Smith-Wilson method

This study presents a novel model, named the SW-VAR model, to tackle the challenge of generating market-consistent economic scenarios for bond yields. The proposed methodology integrates the Smith-Wilson method, principal component analysis (PCA), and first-order vector autoregressive (VAR) models to generate bond yield scenarios driven by both historical data and market views. Underpinned by a new estimation method for the ultimate forward rate (UFR), we implement this methodology using a collection of treasury bond yields in China. As a result, our model achieves an average relative absolute error of less than 3 % compared to the target views. This study makes a pioneering contribution to generating market-consistent economic scenarios for bond yields using the SW-VAR model and provides a new risk management tool for yield movements and bond portfolios. All of these efforts extend the applicability of the Smith-Wilson method in the industry.