【主讲】宋晓军 (北京大学)
【主题】Testing Symmetry of a Nonparametric Conditional Distribution
【时间】2015年6月18日 (周四) 15：30-17：00
【地点】上海财经大学经济学院楼701室
【语言】英文
【摘要】This article proposes tests of symmetry of conditional distributions around a nonparametriclocation function able to detect general non-parametric alternatives. The test is developed ina general serial dependence context, where innovations may exhibit an unknown higher orderserial dependence structure. The test statistic is a functional of the joint empirical distributionof non-parametric residuals and explanatory variables, which can detect non-parametric alternativesconverging to the null at the parametric rate root-n. Critical values are estimated withthe assistance of a bootstrap technique easy to implement, and the validity of the resulting testis formally justified under the regularity conditions. A Monte Carlo study examines the finitesample properties of the test. We also investigate whether losses are more likely than gainsgiven the available information in stock markets using our testing approach.
