New Nonparametric Methods in Financial Econometrics and Additive Models

• Activity Number: 481
• Type: Invited
• Date/Time: Wednesday, August 6, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Noether Award Committee
• Author(s): Yingying Fan*+ and Jianqing Fan and Gareth James and Jiancheng Jiang and Jinchi Lv and Peter Radchenko
• Companies: University of Southern California and Princeton University and University of Southern California and University of North Carolina at Charlotte and University of Southern California and University of Southern California
• Keywords: Nonparametric methods ; Volatility estimation ; Jump detection ; Additive models ; Functional additive regression ; Variable selection

Abstract

Nonparametric models are commonly used for flexible modeling. The first part of the talk focuses on volatility estimation and jump detection in financial econometrics. We proposed two nonparametric methods to construct time and state domain estimators for volatility estimation. A dynamic integration method was introduced to obtain a more efficient estimator. Jumps in asset price processes play an important role in pricing and managing risks of financial instruments. To test and detect jumps, we proposed a family of nonparametric tests and established their nice sampling properties. A new procedure using multiple comparisons was proposed for jump identification. The second part of the talk focuses on estimation of additive models. Two nonparametric methods, integration estimation and pooled backfitting estimation, were proposed for additive models with highly or non-highly correlated covariates, respectively. We then considered this problem in the functional data domain. We proposed functional additive regression using nonparametric estimation methods and variable selection to simultaneously select important functional predictors and estimate the corresponding additive components.