计量经济学实验报告三.doc

1．序列相关实验：下表是1965年-1994年美国人均真实工资Y与人均产出指数X的数据，通过建立模型进行分析： 年份 Y X 1965 69.3 58.6 1966 71.8 61 1967 73.7 62.3 1968 76.5 64.5 1969 77.6 64.9 1970 79 66.2 1971 80.5 68.8 1972 82.9 71 1973 84.7 73.1 1974 83.7 72.2 1975 84.5 74.8 1976 87 77.2 1977 88.1 78.4 1978 89.7 79.5 1979 90 79.7 1980 89.7 79.8 1981 89.8 81.4 1982 91.1 81.2 1983 91.2 84 1984 91.5 86.4 1985 92.8 88.1 1986 95.9 90.7 1987 96.3 91.3 1988 97.3 92.4 1989 95.8 93.3 1990 96.4 94.5 1991 97.4 95.9 1992 100 100 1993 99.9 100.1 1994 99.7 101.4 用普通最小二乘法估计模型参数 Dependent Variable: Y Method: Least Squares Date: 05/25/14 Time: 08:53 Sample: 1965 1994 Included observations: 30 Variable Coefficient Std. Error t-Statistic Prob. C 33.42956 2.041553 16.37457 0.0000 X 0.680115 0.025092 27.10504 0.0000 R-squared 0.963288 Mean dependent var 88.12667 Adjusted R-squared 0.961976 S.D. dependent var 8.691372 S.E. of regression 1.694788 Akaike info criterion 3.957333 Sum squared resid 80.42455 Schwarz criterion 4.050746 Log likelihood -57.35999 F-statistic 734.6830 Durbin-Watson stat 0.297779 Prob(F-statistic) 0.000000 Y=33.42956+0.680115x 用图形法判断模型是否存在一阶自相关 如图表明Ut存在一阶自相关，且大部分散点落在 一，三象限，则判 为正相关 用DW值检验模型是否存在一阶自相关 对样本容量为30，一个解释变量的模型，在显著性水平5%下，查DW统计表可知 DL=1.35 Du=1.49，模型中DWDL,所以模型中存在自相关 使用迭代法估计模型参数，并证明在5%显著性水平下迭代后的模型随机扰动项不存在一阶自相关。 Dependent Variable: Y Method: Least Squares Date: 05/25/14 Time: 09:04 Sample(adjusted): 1966 1994 Included observations: 29 after adjusting endpoints Convergence achieved after 4 iterations Variable Coefficient Std. Error t-Statistic Prob. C 46.35435 6.393929 7.249745 0.0000 X 0.534129 0.070251 7.603108 0.0000 AR(1) 0.782789 0.076701 10.20570 0.0000 R-squared 0.991395 Mean dependent var 88.77586 Adjusted R-squared 0.990733 S.D. dependent var 8.071097 S.E. of regression 0.776976 Akaike info criterion 2.430883 Sum squared resid 15.69600 Schwarz criterion 2.572328 Log likelihood -32.24781 F-statistic