Paper: arXiv 2601.00011

Authors: Jiawei Du, Yi Hong

Abstract

This study focuses on forecasting the ultimate forward rate (UFR) and developing a UFRbased bond yield prediction model using data from Chinese treasury bonds and macroeconomic variables spanning from December 2009 to December 2024. The de Kort-Vellekooptype methodology is applied to estimate the UFR, incorporating the optimal turning parameter determination technique proposed in this study, which helps mitigate anomalous fluctuations. In addition, both linear and nonlinear machine learning techniques are employed to forecast the UFR and ultra-long-term bond yields. The results indicate that nonlinear machine learning models outperform their linear counterparts in forecasting accuracy. Incorporating macroeconomic variables, particularly price index-related variables, significantly improves the accuracy of predictions. Finally, a novel UFR-based bond yield forecasting model is developed, demonstrating superior performance across different bond maturities.

Complexity vs Empirical Score

  • Math Complexity: 7.5/10
  • Empirical Rigor: 8.5/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper employs advanced mathematical methods like the Smith-Wilson framework and de Kort-Vellekoop-type estimation with parameter optimization, requiring strong mathematical understanding. It demonstrates high empirical rigor through extensive backtesting on 15 years of Chinese treasury and macroeconomic data, utilizing both linear and nonlinear machine learning models (including neural networks), SHAP interpretability analysis, and clear out-of-sample performance metrics.

Research Flowchart

  flowchart TD
  A["Research Goal: UFR & Bond Yield<br>Forecasting using ML"] --> B["Data Collection<br>Treasury Bonds & Macro Variables<br>Dec 2009 - Dec 2024"]
  
  B --> C["Methodology Application<br>1. de Kort-Vellekoom UFR Estimation<br>2. Optimal Turning Parameter<br>3. Linear & Nonlinear ML Models"]
  
  C --> D["Computational Analysis<br>Model Training &<br>UFR/Bond Yield Forecasting"]
  
  D --> E["Key Findings & Outcomes<br>• Nonlinear ML > Linear Models<br>• Macro Variables (Price Index) boost accuracy<br>• Novel UFR-based bond yield model developed"]

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper proposes a novel UFR‑based bond yield forecasting model that outperforms existing approaches across different maturities.

Source sentence
Finally, a novel UFR‑based bond yield forecasting model is developed, demonstrating superior performance across different bond maturities.

Assumes 3

  • First‑differencing is applied to UFR and ultra‑long‑term bond yields to address non‑stationarity. statistical
    Source sentence
    Given the negative impact non‑stationary series can have on the predictive performance of OLS and linear machine learning algorithms, we apply first differencing to the UFR and ultra‑long‑term treasury bond yields for our forecasting models.
  • The optimal turning parameter λ is chosen by minimizing the sum of squared errors between UFRZJW and benchmark series. model
    Source sentence
    In this paper, we propose that the optimal λ be determined by minimizing the sum of squared errors between UFRZJW and both UFRSFR and UFRSYC.
  • The ZJW improved method mitigates extreme values of f∞ that arise from the SDF method. model
    Source sentence
    One advantage of the ZJW improved method is its ability to mitigate the extreme values of f∞that arise from the SDF method.

Data

  • China Treasury Bond Yields · China Central Depository & Clearing Co., Ltd. market-prices commercial
    December 31, 2009 to December 31, 2024 · Zero‑Coupon China Treasury bonds (1,2,3,5,7,10,15,20,30,40,50‑year maturities)
    Source sentence
    This study utilizes monthly data on Zero‑Coupon China Treasury bonds provided by the China Central Depository & Clearing Co., Ltd. The dataset covers bond maturities of 1, 2, 3, 5, 7, 10, 15, 20, 30, 40, and 50 years, spanning from December 31, 2009, to December 31, 2024.
  • Macroeconomic Variables · Wind database macro commercial
    December 2009 to December 2024 · 105 variables from 13 categories (e.g., Price Index, Interest Rates, etc.)
    Source sentence
    Our selection of macroeconomic variables is similar to that of Jiang et al. (2024), utilizing 105 variables from 13 categories sourced from the Wind database.

Recognised sources: Wind

Costs

Not applicable The paper reports no trading or execution result.

Finds 5

  • Nonlinear machine learning models achieve higher forecasting accuracy than linear models. predictive-accuracy
    Source sentence
    The results indicate that nonlinear machine learning models outperform their linear counterparts in forecasting accuracy.
  • Incorporating macroeconomic variables, especially price‑index related ones, significantly improves prediction accuracy. predictive-accuracy
    Source sentence
    Incorporating macroeconomic variables, particularly price index‑related variables, significantly improves the accuracy of predictions.
  • The proposed UFR‑based bond yield forecasting model shows superior performance across bond maturities. performance
    Source sentence
    Finally, a novel UFR‑based bond yield forecasting model is developed, demonstrating superior performance across different bond maturities.
  • After first‑order differencing, both UFR and ultra‑long‑term bond yield series become stationary. statistical
    Source sentence
    After applying first‑order differencing, both series pass the unit root tests.
  • Optimal λ is obtained by minimizing the summed squared errors between UFRZJW and benchmark series. theoretical
    Source sentence
    we propose that the optimal λ be determined by minimizing the sum of squared errors between UFRZJW and both UFRSFR and UFRSYC.

Checked by others

Supported by

  • Retail Demand Forecasting: A Comparative Study for Multivariate Time Series (2023)
    On the question: Do macroeconomic variables improve forecast accuracy in time‑series models?
    Compare the findings

    This paper: Incorporating macroeconomic variables, especially price‑index related ones, significantly improves prediction accuracy.

    That paper: Incorporating macroeconomic variables yields modest improvements in forecast accuracy for most machine learning models, with LightGBM benefiting the most.

    Both studies find that adding macro variables boosts accuracy; A reports modest gains (LightGBM most), B reports significant gains, especially from price‑index variables.

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.

Cited References (55)

  1. Core and ‘crust’: consumer prices and the term structure of interest rates (2020)
  2. Does lending relationship help or alleviate the transmission of liquidity shocks? Evidence from a liquidity crunch in China (2022)
  3. On constructing a market consistent economic scenario generator (2011)
  4. Expectations hypotheses tests (2001)
  5. Bond risk premiums with machine learning (2021)
  6. Exponentials, polynomials, and Fourier series: More yield curve modelling at the Bank of Canada (2002)
  7. Yield spreads and interest rate movements: A bird’s eye view (1991)
  8. Long-term government bond yields and macroeconomic fundamentals: Evidence for Greece during the crisis-era (2014)
and 47 more
  1. Extrapolating Long-Maturity Bond Yields for Financial Risk Measurement (2021)
  2. Approximately normal tests for equal predictive accuracy in nested models (2007)
  3. An intertemporal general equilibrium model of asset prices (1985)
  4. Giving flexibility to the Nelson-Siegel class of term structure models (2011)
  5. Modeling and forecasting the yield curve by an extended Nelson‐Siegel class of models: A quantile autoregression approach (2013)
  6. Yield curves from different bond data sets (2020)
  7. Yield curve data choice and potential moral hazard: An empirical exercise on pricing callable bonds (2022)
  8. Comparing predictive accuracy (2002)
  9. The macroeconomy and the yield curve: a dynamic latent factor approach (2006)
  10. Forecasting the term structure of Treasury bond yields (2006)
  11. Yield curve modeling and forecasting: the dynamic Nelson-Siegel approach (2012)
  12. Market-Consistent Economic Scenario Generation System for Treasury Bond Yields with the Smith-Wilson Method (2025)
  13. Term premia and interest rate forecasts in affine models (2002)
  14. The information in long-maturity forward rates (1987)
  15. China’s official rates and bond yields (2010)
  16. Why are excess returns on China’s treasury bonds so predictable? The role of the monetary system (2012)
  17. Interest rate term structure modeling using free-knot splines (2006)
  18. A dynamic Nelson–Siegel model with forward-looking macroeconomic factors: Forecasting performance and economic implications (2019)
  19. Technical Documentation of the Methodology to Derive EIOPA’s Risk-Free Interest Rate Term Structures (2022)
  20. Note on the Smith–Wilson interest rate curve (2016)
  21. Bond pricing and the term structure of interest rates: A new methodology for contingent claims valuation (1992)
  22. Estimating and forecasting the yield curve using a Markov switching dynamic Nelson and Siegel model (2015)
  23. Term structure movements and pricing interest rate contingent claims (1986)
  24. Bond return predictability: Macro factors and machine learning methods (2024)
  25. An analysis of the solvency ii regulatory frameworks Smith-Wilson model for the term structure of risk-free interest rates (2018)
  26. Risk premiums in dynamic term structure models with unspanned macro risks (2014)
  27. Analyzing the term structure of interest rates using the dynamic Nelson–Siegel model with time-varying parameters (2010)
  28. Term structure extrapolation and asymptotic forward rates (2016)
  29. Issues with the Smith–Wilson method (2016)
  30. How can an economic scenario generation model cope with abrupt changes in financial markets? (2021)
  31. Extrapolation methods of the term structure of interest rates under Solvency II (2017)
  32. Interest Rate Modeling: A MATLAB Implementation (2007)
  33. Parsimonious modeling of yield curves (1987)
  34. Out-of-sample forecast tests robust to the choice of window size (2012)
  35. Fitting Yield Curves with Long-Term Constraints (2001)
  36. Estimating and interpreting forward interest rates: Sweden 1992-1994 (1994)
  37. Interest rate term structure and the Chinese fiscal policy: a mixed frequency term structure approach (2023)
  38. A Technical Note on the Smith-Wilson Method (2010)
  39. The relationship between yield curve components and equity sectorial indices: evidence from China (2021)
  40. A preferred‐habitat model of the term structure of interest rates (2021)
  41. A comparative analysis of parsimonious yield curve models with focus on the Nelson-Siegel, Svensson, and Bliss versions (2022)
  42. The sovereign yield curve and the macroeconomy in China (2015)
  43. Forecasting the term structures of Treasury and corporate yields using dynamic Nelson-Siegel models (2011)
  44. Predicting Chinese bond risk premium with machine learning (2024)
  45. An empirical study of the asset price channel of monetary policy transmission in China (2017)
  46. The time-varying bond risk premia in China (2022)
  47. Construct Smith-Wilson risk-free interest rate curves with endogenous and positive ultimate forward rates (2024)

0 of 55 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.