sidenav header background
sidenav background

[10月15日] 计量金融与大数据分析workshop

发布日期:2021-10-11 09:28    来源:

时间:20211015日(周五)10:00 AM -- 11:30 AM


题目:Semiparametric Conditional Factor Models: Estimation and Inference (with Nikolai Roussanov and Xiaoliang Wang) (半参数条件因子模型:估计和推断)





报告摘This paper introduces a simple and tractable sieve estimation for semiparametric conditional factor models, and develops a simple bootstrap procedure for inference on nullity of intercept function and linearity of intercept and factor loadings functions. We establish large-$N$-asymptotic properties of the estimators and the tests without requiring large $T$. We additionally provide two consistent estimators for the number of factors.  The results enable us to estimate conditional (dynamic) behavior of a large set of individual assets from a number of characteristics exhibiting nonlinearity without the need to pre-specify factors, while allowing us to disentangle the alpha versus beta explanations. We apply the methods to explain the cross-sectional differences of individual stock returns in the US market, and find strong evidence of nonzero pricing error and nonlinearity in both alpha and beta functions.