## 经济代写|计量经济学代写Econometrics代考|ECON2300

2022年10月7日
couryes™为您提供可以保分的包课服务

couryes-lab™ 为您的留学生涯保驾护航 在代写计量经济学Econometrics方面已经树立了自己的口碑, 保证靠谱, 高质且原创的统计Statistics代写服务。我们的专家在代写计量经济学Econometrics代写方面经验极为丰富，各种代写计量经济学Econometrics相关的作业也就用不着说。

• Statistical Inference 统计推断
• Statistical Computing 统计计算
• (Generalized) Linear Models 广义线性模型
• Statistical Machine Learning 统计机器学习
• Longitudinal Data Analysis 纵向数据分析
• Foundations of Data Science 数据科学基础

## 经济代写|计量经济学代写Econometrics代考|Maximum Likelihood and GNLS

A second approach that is widely used in place of feasible GLS when $\boldsymbol{\Omega}$ is assumed to equal $\Omega(\alpha)$ with $\alpha$ unknown is the method of maximum likelihood. To use it we must make some assumption about the distribution of the error terms (in practice, almost always an assumption of normality). This allows us to write down the appropriate loglikelihood function as a function of the $q$-vector $\boldsymbol{\alpha}$ and the $k$-vector $\boldsymbol{\beta}$.

Consider the class of models
$$\boldsymbol{y}=\boldsymbol{x}(\boldsymbol{\beta})+\boldsymbol{u}, \quad \boldsymbol{u} \sim N(\mathbf{0}, \boldsymbol{\Omega}(\boldsymbol{\alpha})) .$$
By modifying the loglikelihood function (9.03) slightly, we find that the loglikelihood function corresponding to (9.31) is
\begin{aligned} \ell^n(\boldsymbol{y}, \boldsymbol{\beta}, \boldsymbol{\alpha})=&-\frac{n}{2} \log (2 \pi)-\frac{1}{2} \log |\boldsymbol{\Omega}(\boldsymbol{\alpha})| \ &-\frac{1}{2}(\boldsymbol{y}-\boldsymbol{x}(\boldsymbol{\beta}))^{\top} \boldsymbol{\Omega}^{-1}(\boldsymbol{\alpha})(\boldsymbol{y}-\boldsymbol{x}(\boldsymbol{\beta})) . \end{aligned}
There will be two sets of first-order conditions, one for $\boldsymbol{\alpha}$ and one for $\boldsymbol{\beta}$. The latter will be similar to the first-order conditions (9.05) for GNLS:
$$\boldsymbol{X}^{\top}(\hat{\boldsymbol{\beta}}) \boldsymbol{\Omega}^{-1}(\hat{\boldsymbol{\alpha}})(\boldsymbol{y}-\boldsymbol{x}(\hat{\boldsymbol{\beta}}))=\mathbf{0} .$$
The former will be rather complicated and will depend on precisely how $\Omega$ is related to $\boldsymbol{\alpha}$. For a more detailed treatment, see Magnus (1978).

In Section 8.10, we saw that the information matrix for $\boldsymbol{\beta}$ and $\sigma$ in a nonlinear regression model with covariance matrix $\sigma^2 \mathbf{I}$ is block-diagonal between $\boldsymbol{\beta}$ and $\sigma$. An analogous result turns out to be true for the model (9.31) as well: The information matrix is block-diagonal between $\boldsymbol{\beta}$ and $\boldsymbol{\alpha}$. This means that, asymptotically, the vectors $n^{1 / 2}\left(\hat{\boldsymbol{\beta}}-\boldsymbol{\beta}_0\right)$ and $n^{1 / 2}\left(\hat{\boldsymbol{\alpha}}-\boldsymbol{\alpha}_0\right)$ are independent. Thus the fact that $\hat{\boldsymbol{\alpha}}$ is estimated jointly with $\hat{\boldsymbol{\beta}}$ can be ignored, and $\hat{\boldsymbol{\beta}}$ will have the same properties asymptotically as the GNLS estimator $\hat{\boldsymbol{\beta}}$ and the feasible GNLS estimator $\breve{\boldsymbol{\beta}}$.

The above argument does not require that the error terms $u_t$ actually be normally distributed. All that we require is that the vectors $n^{1 / 2}\left(\hat{\boldsymbol{\beta}}-\boldsymbol{\beta}_0\right)$ and $n^{1 / 2}\left(\hat{\boldsymbol{\alpha}}-\boldsymbol{\alpha}_0\right)$ be asymptotically independent and $O(1)$ under whatever DGP actually generated the data. It can be shown that this is in fact the case under fairly general conditions, similar to the conditions detailed in Chapter 5 for least squares to be consistent and asymptotically normal; see White (1982) and Gouriéroux, Monfort, and Trognon (1984) for fundamental results in this area. As we saw in Section 8.1, when the method of maximum likelihood is applied to a data set for which the DGP was not in fact a special case of the model being estimated, the resulting estimator is called a quasi-ML, or QML, estimator. In practice, of course, almost all the ML estimators we use are actually QML estimators, since some of the assumptions of our models are almost always wrong.

## 经济代写|计量经济学代写Econometrics代考|GLS Estimation of Multivariate Regression Models

In practice, multivariate regression models are usually estimated either by feasible GLS or by maximum likelihood, assuming normality. Except in very rare circumstances, it makes no sense to assume that $u_{t i}$ is independent of $u_{t j}$ for $i \neq j$, as we have already seen in the case of both seemingly unrelated regressions and demand systems. Depending on whether we intend to use ML or feasible GNLS, we may or may not want to assume that the vector of error terms $\boldsymbol{U}t$ is normally distributed. We will in either case make the assumption that $$\boldsymbol{U}_t \sim \operatorname{IID}(\mathbf{0}, \boldsymbol{\Sigma}),$$ where $\boldsymbol{\Sigma}$ is a (usually unknown) $m \times m$ covariance matrix, sometimes referred to as the contemporaneous covariance matrix. Thus we are assuming that $u{t i}$ is correlated with $u_{t j}$ but not with $u_{s j}$ for $s \neq t$. This is of course a strong assumption, which should be tested; we will discuss one test that may sometimes he appropriate helow. Under these assumptions, the generalized sum of squared residuals for the model (9.43) is
$$\sum_{t=1}^n\left(\boldsymbol{Y}_t-\boldsymbol{\xi}_t(\boldsymbol{\beta})\right) \boldsymbol{\Sigma}^{-1}\left(\boldsymbol{Y}_t-\boldsymbol{\xi}_t(\boldsymbol{\beta})\right)^{\top} .$$
Let us suppose initially that $\boldsymbol{\Sigma}$ is known. Then $\boldsymbol{\Sigma}$ may be used to transform the multivariate model (9.40) into a univariate one. Suppose that $\psi$ is an $m \times m$ matrix (usually triangular) such that
$$\boldsymbol{\psi} \boldsymbol{\psi}^{\top}=\boldsymbol{\Sigma}^{-1} .$$
If we postmultiply each term in (9.43) by $\boldsymbol{\psi}$, we obtain the regression
$$\boldsymbol{Y}_t \boldsymbol{\psi}=\boldsymbol{\xi}_t(\boldsymbol{\beta}) \boldsymbol{\psi}+\boldsymbol{U}_t \boldsymbol{\psi} .$$
The $1 \times m$ error vector $\boldsymbol{U}_t \boldsymbol{\psi}$ has covariance matrix
$$E\left(\boldsymbol{\psi}^{\top} \boldsymbol{U}_t^{\top} \boldsymbol{U}_t \boldsymbol{\psi}\right)=\boldsymbol{\psi}^{\top} \boldsymbol{\Sigma} \boldsymbol{\psi}=\mathbf{I}_m .$$
As written, (9.47) has only one observation, and all terms are $1 \times m$ véctors. In order to run this regression, we must somehow convert these $1 \times m$ vectors into $n m \times 1$ vectors for all observations together. There is more than one way to do this.

# 计量经济学代考

## 经济代写|计量经济学代写Econometrics代考|最大似然和GNLS

$$\boldsymbol{y}=\boldsymbol{x}(\boldsymbol{\beta})+\boldsymbol{u}, \quad \boldsymbol{u} \sim N(\mathbf{0}, \boldsymbol{\Omega}(\boldsymbol{\alpha})) .$$

\begin{aligned} \ell^n(\boldsymbol{y}, \boldsymbol{\beta}, \boldsymbol{\alpha})=&-\frac{n}{2} \log (2 \pi)-\frac{1}{2} \log |\boldsymbol{\Omega}(\boldsymbol{\alpha})| \ &-\frac{1}{2}(\boldsymbol{y}-\boldsymbol{x}(\boldsymbol{\beta}))^{\top} \boldsymbol{\Omega}^{-1}(\boldsymbol{\alpha})(\boldsymbol{y}-\boldsymbol{x}(\boldsymbol{\beta})) . \end{aligned}

$$\boldsymbol{X}^{\top}(\hat{\boldsymbol{\beta}}) \boldsymbol{\Omega}^{-1}(\hat{\boldsymbol{\alpha}})(\boldsymbol{y}-\boldsymbol{x}(\hat{\boldsymbol{\beta}}))=\mathbf{0} .$$

## 经济代写|计量经济学代写Econometrics代考|多元回归模型的GLS估计

$$\sum_{t=1}^n\left(\boldsymbol{Y}_t-\boldsymbol{\xi}_t(\boldsymbol{\beta})\right) \boldsymbol{\Sigma}^{-1}\left(\boldsymbol{Y}_t-\boldsymbol{\xi}_t(\boldsymbol{\beta})\right)^{\top} .$$让我们最初假设 $\boldsymbol{\Sigma}$ 是已知的。然后 $\boldsymbol{\Sigma}$ 可以用来将多变量模型(9.40)转换为单变量模型。假设 $\psi$ 是一个 $m \times m$ 矩阵(通常为三角形)，使
$$\boldsymbol{\psi} \boldsymbol{\psi}^{\top}=\boldsymbol{\Sigma}^{-1} .$$

$$\boldsymbol{Y}_t \boldsymbol{\psi}=\boldsymbol{\xi}_t(\boldsymbol{\beta}) \boldsymbol{\psi}+\boldsymbol{U}_t \boldsymbol{\psi} .$$
The $1 \times m$ 误差矢量 $\boldsymbol{U}_t \boldsymbol{\psi}$ 协方差矩阵是否
$$E\left(\boldsymbol{\psi}^{\top} \boldsymbol{U}_t^{\top} \boldsymbol{U}_t \boldsymbol{\psi}\right)=\boldsymbol{\psi}^{\top} \boldsymbol{\Sigma} \boldsymbol{\psi}=\mathbf{I}_m .$$

## 有限元方法代写

tatistics-lab作为专业的留学生服务机构，多年来已为美国、英国、加拿大、澳洲等留学热门地的学生提供专业的学术服务，包括但不限于Essay代写，Assignment代写，Dissertation代写，Report代写，小组作业代写，Proposal代写，Paper代写，Presentation代写，计算机作业代写，论文修改和润色，网课代做，exam代考等等。写作范围涵盖高中，本科，研究生等海外留学全阶段，辐射金融，经济学，会计学，审计学，管理学等全球99%专业科目。写作团队既有专业英语母语作者，也有海外名校硕博留学生，每位写作老师都拥有过硬的语言能力，专业的学科背景和学术写作经验。我们承诺100%原创，100%专业，100%准时，100%满意。

## MATLAB代写

MATLAB 是一种用于技术计算的高性能语言。它将计算、可视化和编程集成在一个易于使用的环境中，其中问题和解决方案以熟悉的数学符号表示。典型用途包括：数学和计算算法开发建模、仿真和原型制作数据分析、探索和可视化科学和工程图形应用程序开发，包括图形用户界面构建MATLAB 是一个交互式系统，其基本数据元素是一个不需要维度的数组。这使您可以解决许多技术计算问题，尤其是那些具有矩阵和向量公式的问题，而只需用 C 或 Fortran 等标量非交互式语言编写程序所需的时间的一小部分。MATLAB 名称代表矩阵实验室。MATLAB 最初的编写目的是提供对由 LINPACK 和 EISPACK 项目开发的矩阵软件的轻松访问，这两个项目共同代表了矩阵计算软件的最新技术。MATLAB 经过多年的发展，得到了许多用户的投入。在大学环境中，它是数学、工程和科学入门和高级课程的标准教学工具。在工业领域，MATLAB 是高效研究、开发和分析的首选工具。MATLAB 具有一系列称为工具箱的特定于应用程序的解决方案。对于大多数 MATLAB 用户来说非常重要，工具箱允许您学习应用专业技术。工具箱是 MATLAB 函数（M 文件）的综合集合，可扩展 MATLAB 环境以解决特定类别的问题。可用工具箱的领域包括信号处理、控制系统、神经网络、模糊逻辑、小波、仿真等。