邹至庄经济研究院博士生论文研讨会第77期
论文题目：Re-mining the Forward Rates
报告人：王道炜
导师：洪永淼教授
摘要：
The bond market plays a pivotal role in financing, risk management, decision-making references, and economic growth. As a core indicator, the Treasury bond forward rate serves to predict economic trends, guide investment and policy formulation, and enhance the bond market's function in stabilizing the economy. Consequently, estimating forward rates holds significant importance. Previous studies typically first estimated the yield or discount function and then derived forward rates using their equivalence formulas. This paper proposes the Flexibly-fused Method (FLUME), which directly estimates forward rates, filling a critical research gap in this field. FLUME improves the penalty term of the traditional Fused LASSO method by allowing the penalty mechanism to cease when the estimated coefficients exhibit local polynomial characteristics, thereby overcoming the restriction that coefficients must be equal locally to avoid penalization. This innovation enables more flexible fusion-penalized estimation of coefficients. Applying this method, we re-mine the term structure of U.S. Treasury bonds. Compared with the results from Fama and Bliss (1987), Gürkaynak, Sack and Wright (2007), Liu and Wu (2021), and Filipović et al. (2025), our estimates achieve a smaller Mean Square Yield Error (MSYE). Furthermore, based on these results, we investigate the prediction of Treasury zero-coupon bond excess returns. Empirical findings reveal that time-varying effects are crucial factors to consider when forecasting excess returns. Additionally, the FLUME dataset demonstrates stronger predictive explanatory power compared to datasets used in other studies.
地点：D235
腾讯会议：589-607-327
时间：5月6日，周二中午，12:30-14:00
语言：中文
