Econometrics.ppt

* Econometrics Ch1 The nature and scope of Econometrics Y: dependent var. effect 果 X1, …Xk: independent var. cause 因 Ch2-Ch5:Review of statistic t X2 F N.D Ch6-7:simple regression 不教 Ch8: multiple regression 多元迴歸 * appendix 1A: p.15 Econ data on the world wide web Ch8 multiple regression 8.1 the three-variable linear regression model Y:dependent var. or explained var. x2 and x3: independent var. or explanatory var. ut: the stochastic disturbance term 不確定的誤差項，殘差項，x2, x3不能100% affect Y : intercept term 截距項 、 : partial regression coeff. Ch8 multiple regression 8.2 Assumption of reg. model A 8.1 Linear in parameter in Eq 8.1 Y 與x2, x3為線性關係 A 8.2 x2, x3 are uncorrelated with u A 8.3 E ui 0 用二手資料 A 8.4 Var ui the variance of ui is constant Homokedasticity 變異數齊一 A 8.5 Cov ui, uj 0 No autocorrelation exists between ui, uj 無自我相關 A 8.6 No exact collinear between x2, x3 A 8.7 for hypothesis testing , Ch8 multiple regression 8.3 estimation of parameters of multiple reg. population regression fun. PRF : sample regression fun. SRF : Where e: residual term bi: the estimate of Ch8 multiple regression Method of Ordinary Least Square OLS 一般最小平方法 Choose b1, b2, b3 to min where et: : Residual sum of square RSS Min RSS F.O.C : ~ normal equation 三個方程式求三個未知數，見p.215 Ch8 multiple regression Properties of : p.217 If A1~A4 are satisfied , bOLS will be BLUE Best Linear Unbiased Estimation Linear: b is a linear combination of sample observation. 線性組合 Unbiased: E 真的 值，母體真正的 不偏 Best: efficient , 用OLS估計比其他方法做的還要佳,變異數最小 Ch8 multiple regression 8.4 Goodness of fit 配適度 找出一條線最能代表觀察值 R2 multiple coeff. of determination the proportion of the total variation in Y. :explained by x2 andx3 or regression line jointly. Ch8 multiple regression TSS ESS+RSS TSS: total sum of squares ESS: explained sum of squares RSS: residual sum of squares R2 ESS/TSS 衡量配適度 用 ESS/TSS的比例來看 若RSS 0 TSS ESS R2 1 RSS ESS Ch8 multiple regression 8.6 Hypo. Testing 假設檢定