时间序列中向量自回归模型的拟合优度检验

本文提出并研究了一些诊断检验工具,用于一般参数型的时间序列向量自回归模型的拟合优度检验．该检验在零假设下渐近服从卡方分布,并能侦察到以参数速度收敛到零假设模型的备择模型．检验涉及到权函数,因此可以灵活地选择权函数以提高检验功效,尤其是在可能的偏离方向已知情形．如果备择不是方向型的,而只知道其属于某一个模型类中,此时可构造一个渐近分布自由的极大极小（maximin）检验．对于饱和备择,基于得分型思想给出了构造万能（omnibus）检验的可行性构想．本文对提出的检验从理论上进行了功效研究．另外,为提高检验在小样本情形的功效,本文把非参数Monte Carlo检验方法推广到相依数据情形．最后,通过模拟研究和实际数据分析进一步表明检验的有用性．     

多元时间序列GARCH型模型的诊断检验

多元时间序列GARCH型模型已被证实在理论和实际中具有重要作用.该文对这一类模型的拟合优度提出了一组得分型检验统计量.这些检验在零假设模型下渐近服从卡方分布,计算简单,临界值容易得到.检验对备择模型比较敏感,能侦察到以1/n~(1/2)的速度收敛到零假设的备择模型.对于可能的多个备择,构造了渐近分布自由的Maximin检验;而对于饱和备择情形,基于得分型检验的思想提出了一个构造Omnibus检验的可能性.值得指出的是构造的这组检验能检测到零假设模型的条件协差阵的每一部分可能的偏离,从而当模型被错误指定时,该检验能提供相关信息进行模型修正.模拟结果表明该文的检验表现理想.     

一类上界型拟合优度检验统计量的精确分布

对于简单假设的拟合优度检验,Zhang (2002)构造出一类上界型检验.取不同的参数$\lambda$和不同的权函数$q(t)$,这类检验包含了Kolmogorov-Smirov检验, Berk and Jones(1979)检验等已有的上界型检验.文献中仅对极少数$\lambda$和$q(t)$所对应的检验给出了零假设下的精确分布.然而, 针对不同的问题, ``好'的检验是不同的,因此有必要对任意给定的$\lambda$和$q(t)$情况, 讨论该类检验.本文对任意给定的$\lambda$和$q(t)\equiv 1$情况,导出了相应上界型检验统计量在零假设下的精确分布. 当样本容量$n$较大时,精确分布的计算时间较长, 本文还通过模拟比较得到了在不同样本量下,应采用的计算方法. 最后, 给出一个实际例子对前述方法加以简单说明.     

函数型二次模型的假设检验

本文研究了函数型二次回归中二次参数函数的显著性检验问题。采用函数型主成分分析将预测变量函数进行投影降维,利用零模型和全模型的残差平方和构造F型检验统计量。在一定的正则条件下证明了检验统计量在原假设下渐近于F分布,在备择假设下检验统计量依概率趋于无穷,从而表明该检验方法是相合的。进一步证明了在一定收敛速度的局部备择假设下,检验统计量渐近于非中心F分布。最后通过数值模拟研究了该检验方法在有限样本下的表现,并给出了一个实际例子进一步验证所提方法的有效性。     

预测回归模型序列相关性的Jackknife经验似然比检验

预测回归模型是计量经济学中的重要模型,而相关的序列检验问题在文献中尚未被提及.考虑到序列相关性检验在模型实证中的重要性,文章基于Jackknife经验似然法,构造了该模型的序列相关检验统计量,并在一定条件下推导了零假设成立时检验统计量的渐近分布,分析了它在局部备择假设下的功效情况,最后通过随机模拟和实际数据例子验证了该检验方法的有限样本性质.     

一类多元重复测量模型参数的似然比检验及其功效分析

在多元重复测量试验模型下,当受试对象观测矩阵的协方差矩阵∑为等方差等协方差结构时,给出了参数的似然比检验统计量.给出该检验在原假设下的渐近零分布和在备择假设下的渐近非零分布,并就检验的功效进行了分析.     

PLM中误差方差未知时的GLR

本文检验部分线性回归模型(PLM)中,误差的方差未知时,函数部分是否是线性函数,在备择假设下,先用局部多项式方法估计出函数部分,再估计参数部分.计算出了零假设下广义似然比(GLR)检验统计量的表达式,给出了它的渐近分布,并对结果进行了模拟.     

重复测量试验模型参数似然比检验及其功效分析

本文给出了在重复测量试验模型下, 当受试对象观测向量的协方差矩阵$\Sigma$为复合对称阵时,参数的似然比检验统计量; 给出该检验在原假设下的渐近零分布和在备择假设下的渐近非零分布;并就其功效进行了分析.     

含有测量误差的Panel数据模型的统计推断

本文考虑含有测量误差的Panel数据模型在约束条件下的参数估计和检验问题.首先,给出未知参数向量的纠正最小二乘估计,在一些正则条件下研究其渐近性质.其次,对线性约束条件的假设检验问题,利用原假设和备择假设下纠正残差平方和的差构造检验统计量,并证明该检验统计量的渐近分布.最后,通过数值模拟研究验证所提估计和检验方法的有效性.     

复合零假设下的修正Berk-Jones检验

首先给出带估计参数的上界型和积分型两种修正Berk-Jones拟合优度检验统计量,接着在复合零假设下研究了这两种检验统计量的渐近分布.所得结论可以为进一步讨论非参数似然比拟合优度检验提供较好的理论基础.     

Goodness-of-fit tests based on the empirical characteristic function

The paper is devoted to the supremum-type multivariate goodness-of-fit tests based on the empirical characteristic function. Particular attention is devoted to the composite hypothesis of normality and Gaussian distribution mixture model. An analytical way to approximate the null asymptotic distributions of the considered test statistics is discussed applying the theory of large excursions of differentiable Gaussian random fields. The produced comparative Monte Carlo power study shows that the considered tests are powerful competitors to the existing classical criteria, clearly dominating in verification of the goodness-of-fit hypotheses against the specific types of alternatives.     

Multivariate linear rank statistics for profile analysis

For some general multivariate linear models, linear rank statistics are used in conjunction with Roy's Union-Intersection Principle to develop some tests for inference on the parameter (vector) when they are subject to certain linear constraints. More powerful tests are designed by incorporating the a priori information on these constraints. Profile analysis is an important application of this type of hypothesis testing problem; it consists of a set of hypothesis testing problem for the p responses q-sample model, where it is a priori assumed that the response-sample interactions are null.     

线性模型中M检验原假设分布的随机加权逼近

在线性模型中M-方法可以用于线性假设检验, 其中M检验、Wald检验和Rao的计分型检验是最常用的检验准则. 但是在计算这些检验的临界值时都涉及到未知参数的估计. 在本文中我们利用随机加权的方法来逼近这些检验的原假设分布. 结果表明在原假设和局部对立假设之下随机加权统计量的渐近分布与原检验统计量在原假设之下的渐近分布相同. 因此我们不需要对冗余参数进行估计,利用随机加权的方法就可以得到这些检验的临界值. 而且在局部对立假设之下可以实现对功效的计算. 当取不同的误差分布和不同的随机权时, 我们对本文的方法进行了蒙特卡洛模拟. 结果表明用随机加权方法来逼近原假设分布是非常精确的.     

Test for a specified signal when the noise covariance matrix is unknown

In the univariate case it is well known that the one sided t test is uniformly most powerful for the null hypothesis against all one sided alternatives. Such a property does not easily extend to the multivariate case. In this paper, a test derived for the hypothesis that the mean of a vector random variable is zero against specified alternatives, when the covariance matrix is unknown. This test depends on the given alternatives and is more powerful than Hotelling's T². The results are derived both for real and complex vector observations and under normal and spherical distributions. The properties of the proposed tests are investigated in detail when a single alternative is specified.     

On consistent testing for serial correlation of unknown form in vector time series models

Multivariate autoregressive models with exogenous variables (VARX) are often used in econometric applications. Many properties of the basic statistics for this class of models rely on the assumption of independent errors. Using results of Hong (Econometrica 64 (1996) 837), we propose a new test statistic for checking the hypothesis of non-correlation or independence in the Gaussian case. The test statistic is obtained by comparing the spectral density of the errors under the null hypothesis of independence with a kernel-based spectral density estimator. The asymptotic distribution of the statistic is derived under the null hypothesis. This test generalizes the portmanteau test of Hosking (J. Amer. Statist. Assoc. 75 (1980) 602). The consistency of the test is established for a general class of static regression models with autocorrelated errors. Its asymptotic slope is derived and the asymptotic relative efficiency within the class of possible kernels is also investigated. Finally, the level and power of the resulting tests are also studied by simulation.     

Invariant Tests for Covariance Structures in Multivariate Linear Model

The null hypothesis that the error vectors in a multivariate linear model are independent is tested against the alternative hypothesis that they are dependent in some specified manner. This dependence is assumed to be due to common random components or autocorrelation over time. The testing problem is solved by classical invariance arguments under multinormality. The most powerful invariant test usually depends on the particular alternative and may even lack a closed form expression. Then the locally best test is derived. The power is maximized at the null hypothesis in the direction of some alternative. In most applications the direction where the maximization is performed does not enter the test. Then the locally uniformly best test exists. Several applications are outlined.     

Asymptotic power properties of the Cramér-von Mises test under contiguous alternatives

The asymptotic power of the Cramér-von Mises test (CvM test) when parameters are estimated from the data is studied under certain local (contiguous) alternatives. The notion of (asymptotic) direction and distance from the null hypothesis of alternatives is introduced, and it is shown that there exist directions with maximum, minimum, and arbitrary intermediate power. For each direction, there exists a natural asymptotic testing problem with an uniformly most powerful test that is compared with the CvM test. For that, the notion of asymptotic local efficiency (ALE) of the CvM test is introduced. finally, the influence of more information on the (unknown) parameter is studied for three tests of the CvM-type for independence. it is shown that for certain directions, a better knowledge of the parameter may increase the power, and for other ones decrease it. These properties are analogous to that of the X²-test in similar situations.     

Bahadur efficiency and robustness of studentized score tests

We derive the exact Bahadur slopes of studentized score tests for a simple null hypothesis in a one-parameter family of distributions. The Student's t-test is included as a special case for which a recent result of Rukhin (1993, Sankhy Ser. A, 55, 159–163) was improved upon. It is shown that locally optimal Bahadur efficiency for one-sample location models with a known or estimated scale parameter is attained within the class of studentized score tests. The studentized test has an asymptotic null distribution free of the scale parameter, and the optimality of likelihood scores does not depend on the existence of a moment generating function. We also consider the influence function and breakdown point of such tests as part of our robustness investigation. The influence of any studentized score test is bounded from above, indicating certain degree of robustness of validity, but a bounded score function is needed to cap the influence from below and to ensure a high power breakdown point. We find that the standard Huber-type score tests are not only locally minimax in Bahadur efficiency, but also very competitive in global efficiency at a variety of location models.     

浅谈假设检验中原假设和备择假设的建立

建立原假设和备择假设是进行假设检验的第一步.如何设立假设检验中的原假设和备择假设,是学生和初学者普遍困扰的问题.而根据奈曼—皮尔逊准则,原假设和备择假设的地位不对等.本文对如何建立假设检验中的原假设和备择假设进行了讨论,给出了几个建立原假设和备择假设的原则.