近年全球金融市場歷經新冠疫情、聯準會激進升息及地緣政治衝突等多重衝擊，呈現高度非線性波動，傳統線性模型難以捕捉轉折。台灣公債殖利率深受美國貨幣政策外溢影響，台美 10 年期利差於 2023 年一度突破 3.5%，預測難度大幅提升。
本研究以 2003 年至 2024 年月資料，針對台灣 2 年、5 年及 10 年期公債殖利率建構多期限預測框架，涵蓋 67 個國內外指標，採滾動視窗法進行向前 1至 24 期樣本外預測，以 RMSE、MAE 及 MAD 評估績效。共建構 17 組模型，含時間序列基準模型（隨機漫步、AR、ARMA）、Lasso 家族（LASSO、Ridge、Elastic Net、Adaptive 系列）、因子模型（主成分分析、目標因子）、非線性機器學習（套袋樹、隨機森林、XGBoost）及組合式預測方法。
實證顯示，短期（h = 1）LASSO 表現最佳，基準模型仍具競爭力；隨期距拉長（h = 3 至 h = 24），XGBoost 在短中長三個天期、各期距下均維持最佳表現，非線性機器學習優勢顯著。主成分分析法在各天期、各期距下表現均最差，脊迴歸在 Lasso 家族中表現最弱，模型組合並不必然能改善預測精度，整體表現遜於最佳單一模型。變數重要性方面，前期同天期公債殖利率為最關鍵因子，外匯存底、隔夜拆款利率與美國長短利差為跨模型核心變數。

In recent years, global financial markets have been rattled by the COVID-19 pandemic, the Fed’s aggressive rate-hiking cycle, and geopolitical conflicts, exhibiting highly nonlinear volatility that traditional linear models struggle to capture. Taiwan’s government bond yields are strongly affected by U.S. monetary policy spillovers, with the Taiwan–U.S. 10-year spread exceeding 3.5% in 2023.
Using monthly data from 2003 to 2024, this study builds a multi-horizon forecasting framework for Taiwan’s 2-, 5-, and 10-year government bond yields with 67 indicators and rolling window forecasts at horizons of 1 to 24 steps ahead (RMSE, MAE, MAD). Seventeen models span time-series benchmarks, the Lasso family,factor models, nonlinear machine learning (Bagging, Random Forest, XGBoost), and combination methods.
LASSO performs best at h = 1 while benchmarks remaincompetitive; as horizons extend (h = 3toh = 24),XGBoost consistently dominates across short-, medium-, and long-termmaturities. PCA performs worst, Ridge Regression is the weakest in the Lasso family, and model combination does not improve forecast accuracy. The lagged same maturity yield is the most critical predictor; foreign exchange reserves, the overnight interbank call loan rate, and the U.S. term spread are common core predictors. The proposed framework provides a precise quantitative basis for policy formulation and risk management in Taiwan’s bond market.

中文參考文獻：
王凱立、林卓民、王美智（2010）。總體經濟因素與資訊傳遞效果於美國與台灣債券市場動態過程之研究。管理與系統，17（4），611–636。
袁麗梅（2018）。應用機器學習預測台灣十年期公債殖利率（未出版碩士論文）。國立臺灣科技大學，台北。
徐霈宜（2025）。應用機器學習方法預測台灣十年期公債殖利率走勢（未出版碩士論文）。國立政治大學，台北。
張惟婷（2022）。用深度學習預測台灣十年期債券殖利率（未出版碩士論文）。
國立臺北商業大學，台北。
陳谷劦（2003）。公債殖利率與總體變數間關係之探討：台灣的實證研究（未出版碩士論文）。世新大學，台北。
劉有仁（2005）。台灣十年期公債殖利率與國內外總體經濟變數關聯性之研究
（未出版碩士論文）。國立成功大學，台南。
蕭伊倩（2007）。台灣公債市場殖利率與流動性影響因素之研究（未出版碩士論文）。朝陽科技大學，台中。

英文參考文獻：
Abad, P., & Novales, A. (2005). An error correction factor model of term structure slopes in international swap markets. Journal of International Financial Markets, Institutions and Money, 15(3), 229–254.
Ang, A., & Piazzesi, M. (2003). A no-arbitrage vector autoregression of term structure dynamics with macroeconomic and latent variables. Journal of Monetary Economics, 50(4), 745–787.
Bai, J., & Ng, S. (2008). Forecasting economic time series using targeted predictors.Journal of Econometrics, 146(2), 304–317.
Breiman, L. (1996). Bagging predictors. Machine Learning, 24(2), 123–140.
Breiman, L. (2001). Random forests. Machine Learning, 45(1), 5–32.
Bianchi, D., Büchner, M., & Tamoni, A. (2021). Bond risk premiums with machine learning. The Review of Financial Studies, 34(2), 1046–1089.
Breiman, L., Friedman, J. H., Olshen, R. A., & Stone, C. J. (1984). Classification and Regression Trees. Belmont, CA: Wadsworth International Group.
Chen, T., & Guestrin, C. (2016). XGBoost: A scalable tree boosting system.
Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 785–794.
De Pooter, M., Ravazzolo, F., & van Dijk, D. J. C. (2010). Term structure forecasting using macro factors and forecast combination (Working Paper). SSRN.
Diebold, F. X., & Li, C. (2006). Forecasting the term structure of government bond yields. Journal of Econometrics, 130(2), 337–364.
Duffee, G. R. (2002). Term premia and interest rate forecasts in affine models. The Journal of Finance, 57(1), 405–443.
Gu, S., Kelly, B., & Xiu, D. (2020). Empirical asset pricing via machine learning. The Review of Financial Studies, 33(5), 2223–2273.
Gupta, K. L., & Moazzami, B. (1991). Recent interest rate behavior: The error
correction modeling approach. Journal of Macroeconomics, 13(3), 479–493.
Hoerl, A. E., & Kennard, R. W. (1970). Ridge regression: Biased estimation for nonorthogonal problems. Technometrics, 12(1), 55–67.
Hastie, T., Tibshirani, R., & Friedman, J. (2001). The Elements of Statistical Learning:Data Mining, Inference, and Prediction. New York: Springer.
Matsumura, M., Moreira, A., & Vicente, J. (2011). Forecasting the yield curve with linear factor models. International Review of Financial Analysis, 20(5), 237–243.
Moench, E. (2008). Forecasting the yield curve in a data-rich environment: A no arbitrage factor-augmented VAR approach. Journal of Econometrics, 146(1), 26–43.
Nelson, C. R., & Siegel, A. F. (1987). Parsimonious modeling of yield curves. The Journal of Business, 60(4), 473–489.
Oosterlaken, J. (2020). Predicting the US Treasury yields using machine learning techniques (Master’s thesis). Erasmus University Rotterdam, Rotterdam.
Svensson, L. E. O. (1994). Estimating and interpreting forward interest rates: Sweden 1992–1994 (Working Paper No. 4871). National Bureau of Economic Research.
Stock, J. H., & Watson, M. W. (2002). Macroeconomic forecasting with diffusion indexes. Journal of Business and Economic Statistics, 20(2), 147–162.
Tibshirani, R. (1996). Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society: Series B (Methodological), 58(1), 267–288.
Vereda, L., Lopes, H., & Fukuda, R. (2008). Estimating VAR models for the term structure of interest rates. Insurance: Mathematics and Economics, 42(2), 548–559.
Zou, H. (2006). The adaptive lasso and its oracle properties. Journal of the American Statistical Association, 101(476), 1418–1429.
Zou, H., & Hastie, T. (2005). Regularization and variable selection via the elastic net. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 67(2), 301–320.
Zou, H., & Zhang, H. H. (2009). On the adaptive elastic-net with a diverging number of parameters. The Annals of Statistics, 37(4), 1733–1751.