Loading...
|
Please use this identifier to cite or link to this item:
https://nccur.lib.nccu.edu.tw/handle/140.119/157809
|
| Title: | 運用混合頻率方法預測高頻金融變數
Mixed-Frequency Forecasting of High-Frequency Financial Variables |
| Authors: | 賴真萍
Lai, Chen-Ping |
| Contributors: | 翁久幸
Weng, Chiu-Hsing
賴真萍
Lai, Chen-Ping |
| Keywords: | 混合頻率資料預測
RR-MIDAS 模型
十年期國債殖利率預測
股票報酬率預測
Mixed-Frequency Forecasting
RR-MIDAS Model
10-Year Treasury Yield Forecasting
Stock Returns Forecasting |
| Date: | 2025 |
| Issue Date: | 2025-07-01 15:03:33 (UTC+8) |
| Abstract: | 在金融市場中，預測高頻宏觀經濟指標與股價報酬率一直是熱門且具挑戰性的研究議題。部分學者主張，僅利用高頻資料本身的歷史資訊即可提供足夠的預測訊息，無須納入額外的低頻變數。然而，為提升預測準確性，亦有研究傾向引入低頻宏觀經濟指標作為預測變數，其中低頻宏觀經濟指標包括國內生產毛額GDP、消費者物價指數CPI等。
然而，使用低頻宏觀經濟指標來預測股票報酬率或十年期國債殖利率等高頻資料，仍面臨諸多挑戰，主因在於兩者的時間頻率不一致，若處理不當，可能導致預測精準度下降。因此，本研究旨在探討Reverse Restricted MIDAS（RR-MIDAS）模型在納入低頻變數後，是否能有效提升預測效能，優於僅以高頻歷史資料作為輸入的線性模型。
本研究的實證分析分別以台灣與美國的GDP與CPI作為低頻宏觀經濟指標，預測兩個高頻目標變數：台灣十年期國債殖利率及美國蘋果公司股票持有一季的收盤價報酬率。比較模型包括：僅使用高頻歷史資料作為輸入的線性模型、Reverse Restricted MIDAS（RR-MIDAS）模型、Reverse Unrestricted MIDAS（RU-MIDAS）模型，以及線性插值模型。最終，實證結果顯示，RR-MIDAS 模型在預測表現上優於其他三種模型，為預測準確度最高之模型。
In financial markets, forecasting high-frequency macroeconomic indicators and stock returns has been a popular but challenging research topic. Some scholars argue that only using the high-frequency historical data can provide sufficient predictive information, eliminating the need to incorporate additional low-frequency variables. However, in order to improve forecasting accuracy, many studies tend to include low-frequency macroeconomic indicators as explanatory variables, such as Gross Domestic Product (GDP) and the Consumer Price Index (CPI).
Nevertheless, using low-frequency macroeconomic indicators to forecast high-frequency variables, such as stock returns or the 10-Year treasury yield, still faces many challenges due to the mismatch in data frequency. If not properly addressed, this mismatch may reduce accuracy in prediction. Therefore, our study aims to explore whether the Reverse Restricted MIDAS (RR-MIDAS) model, which incorporates low-frequency variables, can improve forecasting performance compared to linear models only using the high-frequency historical data to predict.
The empirical analysis in our study utilizes GDP and CPI data from Taiwan and the United States as low-frequency macroeconomic indicators to forecast two high-frequency target variables: the 10-year treasury yield in Taiwan, and the quarterly holding period stock return of Apple Inc. in the U.S. The models compared include: a linear model that only use high-frequency historical data as input, the Reverse Restricted MIDAS (RR-MIDAS) model, the Reverse Unrestricted MIDAS (RU-MIDAS) model, and a linear interpolation model. In conclusion, the empirical results show that the RR-MIDAS model outperforms the other three models, achieving the highest forecasting accuracy. |
| Reference: | [1] Nai-Fu Chen, Richard Roll, and Stephen A. Ross. Economic Forces and the Stock Market. Journal of Business, 59(3):383–403, 1986.
[2] Charles L. Evans and David A. Marshall. Economic determinants of the nominal treasury yield curve. Journal of Monetary Economics, 54(7):1986–2003, 2007.
[3] Claudia Foroni, Pierre Guérin, and Massimiliano Marcellino. Using low frequency information for predicting high frequency variables. International Journal of Forecasting, 34(4):774–787, 2018.
[4] Claudia Foroni, Massimiliano Marcellino, and Christian Schumacher. Unrestricted mixed data sampling (MIDAS): MIDAS regressions with unrestricted lag polynomials. Journal of the Royal Statistical Society Series A: Statistics in Society, 178(1):57–82, 2015.
[5] Claudia Foroni, Francesco Ravazzolo, and Luca Rossini. Are low frequency macroeconomic variables important for high frequency electricity prices? Economic Modelling, 120, 2023.
[6] Eric Ghysels, Virmantas Kvedaras, and Vaidotas Zemlys. Mixed Frequency Data Sampling Regression Models: The R Package midasr. Journal of Statistical Software, 72(4), 2016.
[7] Eric Ghysels, Pedro Santa-Clara, and Rossen Valkanov. The MIDAS Touch: Mixed Data Sampling Regression Models. 2004.
[8] Feng Ma, Yu Li, Li Liu, and Yaojie Zhang. Are low-frequency data really uninformative? A forecasting combination perspective. North American Journal of Economics and Finance, 44:92–108, 2018.
[9] Jorge Nocedal and Stephen J. Wright. Numerical Optimization. Springer, 2nd edition, 2006.
[10] Ivo Welch and Amit Goyal. A Comprehensive Look at The Empirical Performance of Equity Premium Prediction. The Review of Financial Studies, 21(4):1455–1508,2008.
[11] Qifa Xu, Xingxuan Zhuo, Cuixia Jiang, Fang Sun, and Xue Huang. Reverse restricted MIDAS model with application to US interest rate forecasts. Communications in Statistics - Simulation and Computation, 50(2):462–482, 2019.
[12] Vaidotas Zemlys-Balevicius and Virmantas Kvedaras. Package ‘midasr’. CRAN, 2025. https://cran.r-project.org/web/packages/midasr/midasr.pdf. |
| Description: | 碩士
國立政治大學
統計學系
112354031 |
| Source URI: | http://thesis.lib.nccu.edu.tw/record/#G0112354031 |
| Data Type: | thesis |
| Appears in Collections: | [統計學系] 學位論文
|
Files in This Item:
| File |
Description |
Size | Format | |
|---|
| 403101.pdf | | 1079Kb | Adobe PDF | 0 | View/Open |
|
All items in 政大典藏 are protected by copyright, with all rights reserved.
|
