{"title":"美联储如何决策:机器学习增强的泰勒规则","authors":"Boyu Wu, Amina Enkhbold, Asawari Sathe, Qian Wang","doi":"10.3905/jfi.2022.32.3.049","DOIUrl":null,"url":null,"abstract":"The Federal funds rate is a cornerstone of asset pricing that has a significant impact on asset valuation and portfolio performance. However, estimating it reliably can be a challenging issue given that the FOMC makes monetary policy decisions based on complex economic conditions. The authors leveraged existing literatures’ findings on factors and combined those major factor categories into the new model, which includes inflation, labor markets, financial condition, and proxy of global market, and the authors selected the optimal data series to optimize the effectiveness of detecting Fed decisions through a classification factor selection process. Also, the authors improved the regression from fixed coefficients to gradient boosting nonlinear regression approach to reflect the dynamic interconnections among all the factors and their lags through different periods. Upon conducting out-of-sample forecasting, with these selected factors and machine learning gradient boosting regression, the out-of-sample RMSE improved by 77% from traditional Taylor rule model, which offered an alternative robust solution for forecasting the Federal fund rates that can be further applied to asset pricing and investing.","PeriodicalId":53711,"journal":{"name":"Journal of Fixed Income","volume":"32 1","pages":"49 - 60"},"PeriodicalIF":0.0000,"publicationDate":"2022-12-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"How Does the Fed Make Decisions: A Machine Learning Augmented Taylor Rule\",\"authors\":\"Boyu Wu, Amina Enkhbold, Asawari Sathe, Qian Wang\",\"doi\":\"10.3905/jfi.2022.32.3.049\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"The Federal funds rate is a cornerstone of asset pricing that has a significant impact on asset valuation and portfolio performance. However, estimating it reliably can be a challenging issue given that the FOMC makes monetary policy decisions based on complex economic conditions. The authors leveraged existing literatures’ findings on factors and combined those major factor categories into the new model, which includes inflation, labor markets, financial condition, and proxy of global market, and the authors selected the optimal data series to optimize the effectiveness of detecting Fed decisions through a classification factor selection process. Also, the authors improved the regression from fixed coefficients to gradient boosting nonlinear regression approach to reflect the dynamic interconnections among all the factors and their lags through different periods. Upon conducting out-of-sample forecasting, with these selected factors and machine learning gradient boosting regression, the out-of-sample RMSE improved by 77% from traditional Taylor rule model, which offered an alternative robust solution for forecasting the Federal fund rates that can be further applied to asset pricing and investing.\",\"PeriodicalId\":53711,\"journal\":{\"name\":\"Journal of Fixed Income\",\"volume\":\"32 1\",\"pages\":\"49 - 60\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2022-12-31\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of Fixed Income\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.3905/jfi.2022.32.3.049\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Fixed Income","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.3905/jfi.2022.32.3.049","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
How Does the Fed Make Decisions: A Machine Learning Augmented Taylor Rule
The Federal funds rate is a cornerstone of asset pricing that has a significant impact on asset valuation and portfolio performance. However, estimating it reliably can be a challenging issue given that the FOMC makes monetary policy decisions based on complex economic conditions. The authors leveraged existing literatures’ findings on factors and combined those major factor categories into the new model, which includes inflation, labor markets, financial condition, and proxy of global market, and the authors selected the optimal data series to optimize the effectiveness of detecting Fed decisions through a classification factor selection process. Also, the authors improved the regression from fixed coefficients to gradient boosting nonlinear regression approach to reflect the dynamic interconnections among all the factors and their lags through different periods. Upon conducting out-of-sample forecasting, with these selected factors and machine learning gradient boosting regression, the out-of-sample RMSE improved by 77% from traditional Taylor rule model, which offered an alternative robust solution for forecasting the Federal fund rates that can be further applied to asset pricing and investing.
期刊介绍:
The Journal of Fixed Income (JFI) provides sophisticated analytical research and case studies on bond instruments of all types – investment grade, high-yield, municipals, ABSs and MBSs, and structured products like CDOs and credit derivatives. Industry experts offer detailed models and analysis on fixed income structuring, performance tracking, and risk management. JFI keeps you on the front line of fixed income practices by: •Staying current on the cutting edge of fixed income markets •Managing your bond portfolios more efficiently •Evaluating interest rate strategies and manage interest rate risk •Gaining insights into the risk profile of structured products.