Frequency domain estimation of temporally aggregated Gaussian cointegrated systems
Frequency domain estimation of temporally aggregated Gaussian cointegrated systems
复制标题
时间聚合高斯协整系统的频域估计
DOI：
10.1016/j.jeconom.2006.03.005
复制
发表时间：
2007
影响因子：
6.3
通讯作者：
J. R. McCrorie
中科院分区：
文献类型：
--
作者：
Marcus J. Chambers;J. R. McCrorie
本文考虑了连续时间建立的全参数高斯协积分系统的长期平衡系数和控制短期动态的参数的联合估计问题。该模型允许由随机微分方程组产生平稳扰动，并允许变量为库存和流量的混合。我们推导出了连续时间模型的三角误差修正形式的精确离散模拟的精确形式，作为利用离散时间数据进行未知参数频域估计的基础。我们正式地建立了这种估计量的相合阶和渐近抽样性质。协整参数的估计以样本量收敛于混合正态分布的速率收敛，而短期参数的估计以样本量的平方根收敛于极限正态分布的速率收敛。
This paper considers joint estimation of long run equilibrium coefficients and parameters governing the short run dynamics of a fully parametric Gaussian cointegrated system formulated in continuous time. The model allows the stationary disturbances to be generated by a stochastic differential equation system and for the variables to be a mixture of stocks and flows. We derive a precise form for the exact discrete analogue of the continuous time model in triangular error correction form, which acts as the basis for frequency domain estimation of the unknown parameters using discrete time data. We formally establish the order of consistency and the asymptotic sampling properties of such an estimator. The estimator of the cointegrating parameters is shown to converge at the rate of the sample size to a mixed normal distribution, while that of the short run parameters converges at the rate of the square root of the sample size to a limiting normal distribution.