第九讲 联立方程计量模型.ppt

用间接最小二乘法估计消费方程 C简化式模型估计结果 Dependent Variable: CC Method: Least Squares Date: 11/27/06 Time: 18:24 Sample(adjusted): 1979 2002 Included observations: 24 after adjusting endpoints CC=C(1)+C(2)*CC(-1)+C(3)*G Coefficient Std. Error t-Statistic Prob. C(1) 1135.937 440.3420 2.579670 0.0175 C(2) 0.619782 0.273112 2.269331 0.0339 C(3) 1.239898 0.751061 1.650861 0.1136 R-squared 0.993521 Mean dependent var 17685.46 Adjusted R-squared 0.992904 S.D. dependent var 16106.62 S.E. of regression 1356.820 Akaike info criterion 17.38014 Sum squared resid Schwarz criterion 17.52740 Log likelihood -205.5617 Durbin-Watson stat 0.661227 Y简化式模型估计结果 Dependent Variable: Y Method: Least Squares Date: 11/27/06 Time: 18:25 Sample(adjusted): 1979 2002 Included observations: 24 after adjusting endpoints Y=C(1)+C(2)*CC(-1)+C(3)*G Coefficient Std. Error t-Statistic Prob. C(1) 2014.368 1036.735 1.942992 0.0655 C(2) 0.682750 0.643012 1.061800 0.3004 C(3) 4.511084 1.768289 2.551101 0.0186 R-squared 0.992382 Mean dependent var 37485.38 Adjusted R-squared 0.991656 S.D. dependent var 34971.77 S.E. of regression 3194.479 Akaike info criterion 19.09270 Sum squared resid 2.14E+08 Schwarz criterion 19.23996 Log likelihood -226.1125 Durbin-Watson stat 0.729172 用两阶段最小二乘法估计消费方程 代替原消费方程中的Yt，应用OLS估计 比较上述消费方程的3种估计结果，证明这3种方法对于恰好识别的结构方程是等价的。估计量的差别只是很小的计算误差。 第2阶段估计结果 Dependent Variable: CC Method: Least Squares Date: 11/27/06 Time: 17:57 Sample(adjusted): 1979 2002 Included observations: 24 after adjusting endpoints CC=C(1)+C(2)*YC+C(3)*CC(-1) Coefficient Std. Error t-Statistic Prob. C(1) 582.2761 438.0324 1.329299 0.1980 C(2) 0.274856 0.166492 1.650861 0.1136 C(3) 0.432124 0.386593 1.117775 0.2763 R-squared 0.993521 Mean dependent var 17685.46 Adjusted R-squared 0.992904 S.D. dependent var 16106.62 S.E. of regression 1356.820 Akaike info criterion 17.38014 Sum squared resid Schwarz criterion 17.52740 Log likelihood -205.5617