1、Chapter 3 一元线性回归模型第一节 回归分析与回归方程第1页,共21页。回归分析:1.根据经济理论或考察样本数据去设定回归方程 Y:dependent variable;:independent :random error or disturbance term(,)Yf XZNoImageNoImageNoImageNoImage,XZ第2页,共21页。A special and simple case(univariate linear regression model):这是本章研究的重点。2.参数估计(Estimation of parameter)3.Testing4.Pre
2、dicting设有样本为 ,则01Ybb X,iiY X01(1,)iiiYbbXin 第3页,共21页。模型的假设:1.2.(同方差)3.4.满足这四条件的LRM称为 经典线性回归模型(CLRM)。()0iE()0ijEij2var()i()0iiE X第4页,共21页。n由假设得 Population regression equation(function)The pity is the parameters are unknown.我们要利用样本来估计参数.如得参数估计值 ,则 称为sample regression equation(function).How to estimate
3、 them?The OLS method.01()E Ybb X01bb和01YbbX第5页,共21页。普通最小二乘法(Ordinary least squares procedure):求 使残差平方和最小:Let Then (OLSE)01bb和22201()()iiiiieY YYbbXandiiiixXXyY Y 2101iiibx yxbYb X第6页,共21页。The properties of the OLSE:1.无偏性(unbiased):2.1100(),()E bbE bb2221022var(),var()iiiXbbxnx2012cov(,)iXb bx 第7页,共2
4、1页。3.关于样本 的线性性:4.Gauss-Markov theorem:如果 是经典线性回归模型(CLRM),则其参数的OLSE 为BLUE。即,在所有线性无偏估计中,OLSE的 方差最小。iY1021,()()ii iiiijxbkYbXk Yknx01(1,)iiiYbbXin 第8页,共21页。Estimation of the variance of the random disturbance term,:We know and it is unknown.Thus,and so on are also unknown.To estimate them,we have to fi
5、rst evaluate .It is not difficult to show that is an unbiased estimator for ,22var()i2221022var(),var()iiiXbbxnx222(2)ien2第9页,共21页。Whereare the residuals.Example3.1(P39)(how to use Eviews)(iieY01()iiieYbb X第10页,共21页。模型的假设:模型的假设:5.Normality assumption:The properties of the OLSE:5.2(0,)iN222110022(,),
6、(,)iiiXbN bbN bxx第11页,共21页。nModel testing(模型的检验模型的检验):总离差分解公式:即,TSS =ESS +RSS TSS:Total sum of squares ESS:Error(residual)sum of squares RSS:Regression(explained)sum of squares222()()()iiiiY YY YY Y222iiiyey第12页,共21页。1.Goodness-of-fit testing(R2检验检验):Coefficient of determination(判定系数):In general,the
7、 larger R2,the better.2.Sample coefficient of correlation:222ESSRSS1TSSTSSiiyRy 2rR 第13页,共21页。3.Hypothesis testing We have known Let (standard error)222110022(,),(,)iiiXbN bbN bxx22(2)ien21()iS bx第14页,共21页。ThenAnd we can test the following hypothesis:Moreover,interval estimator for is 111(2)()bbtt n
8、S b0111:versus:H bcH bc(1)1b112()bt S b第15页,共21页。Forecasting(预测预测)1.Point forecastingSince we know and the sample regression equationthen given ,what about and?As(an unbiased estimator for )01Ybb X01Ybb X1nX1nY1()nEY101110111()()()nnnnnEYEb bXb bXEY 1()nEY第16页,共21页。nand (误差均匀)nNaturally,we use as a
9、point predictor for bothn and .n2.Interval forecastingn(1)Forecast interval for n Forecast error:n 1nY11()0nnE YY1nY1()nE Y1nY111nnneYY第17页,共21页。The variance of the forecast error:Therefore21121011101var()()var()var()var()2cov(,)nnnnneE ebXbXb b2212()11niXXnx22112()1(0,1)nniXXeNnx第18页,共21页。It is a pity that is unknown.Fortunately,we have Thus,Hence,a Forecast interval for is 22(2)ien211122221122(2)()()1111nnnnniieYYt nXXXXnxnx 1nY(1)第19页,共21页。n(2)Forecast interval forn Similarly,we can obtain a forecast interval for :21122()1 1nniXXYtnx1()nE Y(1)1()nE Y第20页,共21页。21122()1nniXXYtnx第21页,共21页。