Conditional means of time series processes and time series processes for conditional means

• 成果类型: Article
• 署名作者: Fiorentini, G; Sentana, E
• 署名单位: Universitat d'Alacant
• 刊物名称: INTERNATIONAL ECONOMIC REVIEW
• ISSN/ISSBN: 0020-6598
• DOI: 10.2307/2527354
• 发表日期: 1998
• 页码: 1101-1118
• 关键词: arch models; persistence; returns; prices

摘要

We study the process for the conditional mean of vector linear processes, which nest many models of interest. We also consider the joint process for a variable and its mean conditional on a multivariate information set. We compare the persistence of shocks to stationary variables and their means using impulse response functions. An empirical application suggests that U.S. real stock returns are close to white noise, while expected returns follow an AR(1) with high autocorrelation. We also find that unexpected variations in expected returns immediately produce large negative observed returns, thereafter compensated by slowly diminishing increments on expected returns.