Censored(11000)估计 参数估计结果、似然函数值都 Dependent Variable:CONS 与OLS估计差异较大。为什么 Method:ML-Censored Normal(TOBIT) 似然函数值大于OLS估计？ Date:11/29/04Time:17:25 Sample:1 31 View Procs Objects Print Name Freeze Estimate Forecast Stats Resids Included observations:31 Right censoring (value)series:11000 Dependent Variable:CONS Convergence achieved after 8 iterations Method:Least Squares Date:11/1905Time:09:36 Covariance matrix computed using secc Sample:131 Coefficient Included observations:31 Q 25.62933 Variable Coefficient Std.Error t-Statistic Prob. INCOM 0.775212 283.3025 273.2348 1.036847 0.3084 Error INCOM 0.740774 0.031782 23.30833 0.0000 SCALE:C(3) 396.7539 R-squared 0.949325 Mean dependent var 6427.886 R-squared 0.949968 Adjusted R-squared 0.947578 S.D.dependent var 1746.959 Adjusted R-squared 0.946394 S.E.of regression 399.9813 Akaike info criterion 14.88305 S.E.of regression 404.4725 Sum squared resid 4639566 Schwarz criterion 14.97557 Sum squared resid 4580745. Log likelihood -228.6873 F-statistic 543.2782 Log likelihood -215.7708 Durbin-Watson stat 1.241862 Prob(F-statistic) 0.000000 Avg.log likelihood -6.960348 Left censored obs 0 Right censored obs 2 Uncensored obs 29 Total obs 31Censored(11000) 估计 Dependent Variable: CONS Method: ML - Censored Normal (TOBIT) Date: 11/29/04 Time: 17:25 Sample: 1 31 Included observations: 31 Right censoring (value) series: 11000 Convergence achieved after 8 iterations Covariance matrix computed using second derivatives Coefficient Std. Error z-Statistic Prob. C 25.62933 306.6661 0.083574 0.9334 INCOM 0.775212 0.036891 21.01348 0.0000 Error Distribution SCALE:C(3) 396.7539 52.22918 7.596403 0.0000 R-squared 0.949968 Mean dependent var 6427.886 Adjusted R-squared 0.946394 S.D. dependent var 1746.959 S.E. of regression 404.4725 Akaike info criterion 14.11425 Sum squared resid 4580745. Schwarz criterion 14.25302 Log likelihood -215.7708 Hannan-Quinn criter. 14.15948 Avg. log likelihood -6.960348 Left censored obs 0 Right censored obs 2 Uncensored obs 29 Total obs 31 参数估计结果、似然函数值都 与OLS估计差异较大。为什么 似然函数值大于OLS估计？