Confidence Intervals for Price Discovery
研究了协整向量自回归系统中永久-暂时分解的渐近和自助法置信区间，比较了不同估计量的效率，并应用于价格发现中信息份额和成分份额的推断。
ABSTRACT This paper discusses asymptotic and bootstrap confidence intervals for multivariate permanent‐transitory decompositions of cointegrated vector autoregressive I(1) systems, with a focus on price discovery. Alternative estimators of the permanent components are compared in terms of efficiency also under separable linear restrictions on the parameters, and it is found that the one based on model residuals is most efficient. Bootstrap implementations are discussed and compared. Asymptotic results are also derived for information and component shares widely used in price discovery to measure the relative importance of different markets. A Monte Carlo study and an application illustrate the findings.