頑健な推定方法(アウトプット)
Current sample: 1965 to 1994 Current sample: 1966 to 1994 Equation 1 ============ FIRST-ORDER SERIAL CORRELATION OF THE ERROR Objective function: Exact ML (keep first obs.) Working space used: 667 STARTING VALUES C YD RHO VALUE -13.54048 0.90469 0.00000 F= 92.010 FNEW= 69.550 ISQZ= 1 STEP= 0.50000 CRIT= 38.735 F= 69.550 FNEW= 67.911 ISQZ= 0 STEP= 1.0000 CRIT= 4.3899 F= 67.911 FNEW= 67.608 ISQZ= 0 STEP= 1.0000 CRIT= 1.6846 F= 67.608 FNEW= 67.532 ISQZ= 2 STEP= 0.25000 CRIT= 1.5270 F= 67.532 FNEW= 67.272 ISQZ= 0 STEP= 1.0000 CRIT= 0.62601 F= 67.272 FNEW= 67.249 ISQZ= 0 STEP= 1.0000 CRIT= 0.49449E-01 F= 67.249 FNEW= 67.248 ISQZ= 0 STEP= 1.0000 CRIT= 0.85135E-03 F= 67.248 FNEW= 67.248 ISQZ= 0 STEP= 1.0000 CRIT= 0.36608E-06 F= 67.248 FNEW= 67.248 ISQZ= 0 STEP= 1.0000 CRIT= 0.38184E-13 F= 67.248 FNEW= 67.248 ISQZ= 1 STEP= 0.50000 CRIT= 0.77169E-27 CONVERGENCE ACHIEVED AFTER 10 ITERATIONS 23 FUNCTION EVALUATIONS. Dependent variable: CP Current sample: 1966 to 1994 Number of observations: 29 Mean of dep. var. = 173.620 Adjusted R-squared = .998226 Std. dev. of dep. var. = 56.8939 Durbin-Watson = 1.15431 Sum of squared residuals = 164.155 Rho (autocorrelation coef.) = .926127 Variance of residuals = 6.31364 Schwarz B.I.C. = 72.2991 Std. error of regression = 2.51270 Log likelihood = -67.2482 R-squared = .998353 Standard Parameter Estimate Error t-statistic P-value C .190906 10.3039 .018528 [.985] YD .850070 .042830 19.8476 [.000] RHO .926127 .063642 14.5522 [.000] Standard Errors computed from analytic second derivatives (Newton) Current sample: 1965 to 1994 Equation 2 ============ Method of estimation = Ordinary Least Squares Dependent variable: CP Current sample: 1965 to 1994 Number of observations: 30 Mean of dep. var. = 170.137 LM het. test = 1.60150 [.206] Std. dev. of dep. var. = 59.0697 Durbin-Watson = .170327 [.000,.000] Sum of squared residuals = 1039.17 Jarque-Bera test = 2.11490 [.347] Variance of residuals = 37.1132 Ramsey's RESET2 = 40.7643 [.000] Std. error of regression = 6.09206 F (zero slopes) = 2698.46 [.000] R-squared = .989730 Schwarz B.I.C. = 99.1441 Adjusted R-squared = .989363 Log likelihood = -95.7429 Estimated Standard Variable Coefficient Error t-statistic P-value C -11.4246 3.87448 -2.94867 [.006] YD .895756 .016876 53.0781 [.000] Standard Errors are heteroskedastic-consistent (HCTYPE=2). GENERALIZED METHOD OF MOMENTS ============================= WITH STARTING VALUES VIA: NONLINEAR TWO STAGE LEAST SQUARES EQUATIONS: EQ1 INSTRUMENTS: C YD OPTIONS FOR THIS ROUTINE ======================== COVOC = COVU = DEBUG = FALSE HETERO = TRUE INST = O 0001 ITEROC = FALSE ITERU = TRUE KERNEL = BARTLETT LSQSTART = TRUE MASK = MAXITW = 20 NMA = 2 OPTCOV = FALSE ROBUST = TRUE WNAME = NOTE => The model is linear in the parameters. Working space used: 699 STARTING VALUES B0 B1 VALUE 0.00000 0.00000 F= 0.96855E+06 FNEW= 0.10273E-22 ISQZ= 0 STEP= 1.0000 CRIT= 27.970 CONVERGENCE ACHIEVED AFTER 1 ITERATIONS 2 FUNCTION EVALUATIONS. Working space used: 699 F= 0.44785E-26 FNEW= 0.64797E-30 ISQZ= 0 STEP= 1.0000 CRIT= 0.13436E-24 CONVERGENCE ACHIEVED AFTER 1 ITERATIONS 4 FUNCTION EVALUATIONS. Covariance Matrix of Orthogonality Conditions 1 2 1 91.38913 2 17216.34687 3548845.60729 Number of observations = 30 E'PZ*E = .647973E-30 Standard Parameter Estimate Error t-statistic P-value B0 -11.4246 5.73981 -1.99041 [.047] B1 .895756 .024988 35.8473 [.000] Standard Errors computed from heteroscedastic-consistent matrix (Robust-White) (also robust to autocorrelation: NMA= 2, Kernel=Bartlett) Equation: EQ1 Dependent variable: CP Mean of dep. var. = 170.137 Std. dev. of dep. var. = 59.0697 Sum of squared residuals = 1039.17 Variance of residuals = 37.1132 Std. error of regression = 6.09206 R-squared = .989730 Adjusted R-squared = .989363 Durbin-Watson = .170327 [.000,.000] GENERALIZED METHOD OF MOMENTS ============================= WITH STARTING VALUES VIA: NONLINEAR TWO STAGE LEAST SQUARES EQUATIONS: EQ1 INSTRUMENTS: C YD OPTIONS FOR THIS ROUTINE ======================== COVOC = COVU = DEBUG = FALSE HETERO = TRUE INST = O 0001 ITEROC = FALSE ITERU = TRUE KERNEL = LSQSTART = TRUE MASK = MAXITW = 20 NMA = 0 OPTCOV = FALSE ROBUST = TRUE WNAME = NOTE => The model is linear in the parameters. Working space used: 699 STARTING VALUES B0 B1 VALUE -11.42457 0.89576 F= 0.10341E-26 FNEW= 0.10341E-26 ISQZ= 0 STEP= 0.00000 CRIT= 0.27865E-28 CONVERGENCE ACHIEVED AFTER 1 ITERATIONS 1 FUNCTION EVALUATIONS. Working space used: 699 STARTING VALUES B0 B1 VALUE -11.42457 0.89576 F= 0.13580E-29 FNEW= 0.61006E-30 ISQZ= 0 STEP= 1.0000 CRIT= 0.40741E-28 CONVERGENCE ACHIEVED AFTER 1 ITERATIONS 3 FUNCTION EVALUATIONS. Covariance Matrix of Orthogonality Conditions 1 2 1 34.63902 2 6463.54285 1327954.52336 Number of observations = 30 E'PZ*E = .610061E-30 Standard Parameter Estimate Error t-statistic P-value B0 -11.4246 3.64928 -3.13063 [.002] B1 .895756 .015874 56.4302 [.000] Standard Errors computed from heteroscedastic-consistent matrix (Robust-White) Equation: EQ1 Dependent variable: CP Mean of dep. var. = 170.137 Std. dev. of dep. var. = 59.0697 Sum of squared residuals = 1039.17 Variance of residuals = 37.1132 Std. error of regression = 6.09206 R-squared = .989730 Adjusted R-squared = .989363 Durbin-Watson = .170327 [.000,.000] Equation 3 ============ Method of estimation = Ordinary Least Squares Dependent variable: CP Current sample: 1965 to 1994 Number of observations: 30 Mean of dep. var. = 170.137 LM het. test = 1.60150 [.206] Std. dev. of dep. var. = 59.0697 Durbin-Watson = .170327 [.000,.000] Sum of squared residuals = 1039.17 Jarque-Bera test = 2.11490 [.347] Variance of residuals = 37.1132 Ramsey's RESET2 = 40.7643 [.000] Std. error of regression = 6.09206 F (zero slopes) = 2698.46 [.000] R-squared = .989730 Schwarz B.I.C. = 99.1441 Adjusted R-squared = .989363 Log likelihood = -95.7429 Estimated Standard Variable Coefficient Error t-statistic P-value C -11.4246 3.64928 -3.13063 [.004] YD .895756 .015874 56.4302 [.000] Standard Errors are heteroskedastic-consistent (HCTYPE=0). Equation 4 ============ Method of estimation = Ordinary Least Squares Dependent variable: CP Current sample: 1965 to 1994 Number of observations: 30 Mean of dep. var. = 170.137 LM het. test = 1.60150 [.206] Std. dev. of dep. var. = 59.0697 Durbin-Watson = .170327 [.000,.000] Sum of squared residuals = 1039.17 Jarque-Bera test = 2.11490 [.347] Variance of residuals = 37.1132 Ramsey's RESET2 = 40.7643 [.000] Std. error of regression = 6.09206 F (zero slopes) = 2698.46 [.000] R-squared = .989730 Schwarz B.I.C. = 99.1441 Adjusted R-squared = .989363 Log likelihood = -95.7429 Estimated Standard Variable Coefficient Error t-statistic P-value C -11.4246 3.77737 -3.02448 [.005] YD .895756 .016431 54.5167 [.000] Standard Errors are heteroskedastic-consistent (HCTYPE=1).