Serie estacionaria: ye=.5*ye(-1)+nrnd serie integrada: yi=yi(-1)+nrnd 4 30 3 2 20 1 0 10 -1 -2 0 -3 -4 -10 250 500 YE 750 1000 250 500 YI 750 1000 Serie integrada con tend.: yit=1+.5*t+u u=u(-1)+5*nrnd serie estacionaria con tend.: yet=1+.5*t+nrnd*50 600 500 500 400 400 300 300 200 200 100 100 0 0 -100 -200 250 500 YET 750 1000 -100 250 500 YIT 750 1000 Tipo de cambio y precios relativos 2.8 2.4 2.0 1.6 1.2 0.8 60 65 70 75 ITCPTAEURO 80 85 90 PRPTAEURO 95 Tipo de cambio real peseta-euro .20 .15 .10 .05 .00 -.05 -.10 60 65 70 75 80 85 90 LOG(ITCRPTAEURO) 95 Augmented Dickey-Fuller Unit Root Test on LOG(ITCRPTAEURO) Null Hypothesis: LOG(ITCRPTAEURO) has a unit root Exogenous: Constant Lag Length: 0 (Automatic based on SIC, MAXLAG=9) Augmented Dickey-Fuller test statistic Test critical values: 1% level 5% level 10% level t-Statistic Prob.* -1.970689 -3.610453 -2.938987 -2.607932 0.2980 *MacKinnon (1996) one-sided p-values. Augmented Dickey-Fuller Test Equation Dependent Variable: D(LOG(ITCRPTAEURO)) Method: Least Squares Date: 05/10/02 Time: 12:54 Sample(adjusted): 1961 1999 Included observations: 39 after adjusting endpoints Variable Coefficient Std. Error t-Statistic Prob. LOG(ITCRPTAEURO (-1)) C -0.207423 0.105254 -1.970689 0.0563 0.011454 0.009856 1.162177 0.2526 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat 0.094992 0.070532 0.046793 0.081014 65.10697 1.821167 Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion F-statistic Prob(F-statistic) -0.001164 0.048536 -3.236255 -3.150944 3.883617 0.056276 KPSS Unit Root Test on LOG(ITCRPTAEURO) Null Hypothesis: LOG(ITCRPTAEURO) is stationary Exogenous: Constant Bandwidth: 4 (Newey-West using Bartlett kernel) LM-Stat. Kwiatkowski-Phillips-Schmidt-Shin test statistic Asymptotic critical values*: 1% level 5% level 10% level 0.288537 0.739000 0.463000 0.347000 *Kwiatkowski-Phillips-Schmidt-Shin (1992, Table 1) Residual variance (no correction) HAC corrected variance (Bartlett kernel) 0.005216 0.015951 KPSS Test Equation Dependent Variable: LOG(ITCRPTAEURO) Method: Least Squares Date: 05/10/02 Time: 12:59 Sample: 1960 1999 Included observations: 40 Variable C R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Coefficient Std. Error t-Statistic Prob. 0.058177 0.011565 5.030578 0.0000 0.000000 0.000000 0.073142 0.208638 48.36317 Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion Durbin-Watson stat 0.058177 0.073142 -2.368158 -2.325937 0.429310 Índice de precios en la Unión Europea 2.4 2.0 1.6 1.2 0.8 0.4 0.0 65 70 75 80 85 LOG(IPUE) 90 95 Augmented Dickey-Fuller Unit Root Test on LOG(IPUE) Null Hypothesis: LOG(IPUE) has a unit root Exogenous: Constant, Linear Trend Lag Length: 1 (Automatic based on SIC, MAXLAG=9) Augmented Dickey-Fuller test statistic Test critical values: 1% level 5% level 10% level t-Statistic Prob.* -2.377412 -4.219126 -3.533083 -3.198312 0.3848 *MacKinnon (1996) one-sided p-values. Augmented Dickey-Fuller Test Equation Dependent Variable: D(LOG(IPUE)) Method: Least Squares Date: 05/11/02 Time: 12:44 Sample(adjusted): 1962 1999 Included observations: 38 after adjusting endpoints Variable Coefficient Std. Error t-Statistic Prob. LOG(IPUE(-1)) D(LOG(IPUE(-1))) C @TREND(1961) -0.034204 0.964289 0.003334 0.002176 0.014387 0.059364 0.006457 0.001072 -2.377412 16.24365 0.516324 2.029369 0.0232 0.0000 0.6090 0.0503 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat 0.900240 0.891438 0.010775 0.003947 120.3553 1.624514 Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion F-statistic Prob(F-statistic) 0.061610 0.032701 -6.123964 -5.951587 102.2726 0.000000 Augmented Dickey-Fuller Unit Root Test on LOG(IPUE) Null Hypothesis: LOG(IPUE) is stationary Exogenous: Constant, Linear Trend Bandwidth: 5 (Newey-West using Bartlett kernel) LM-Stat. Kwiatkowski-Phillips-Schmidt-Shin test statistic Asymptotic critical values*: 1% level 5% level 10% level 0.136746 0.216000 0.146000 0.119000 *Kwiatkowski-Phillips-Schmidt-Shin (1992, Table 1) Residual variance (no correction) HAC corrected variance (Bartlett kernel) 0.018632 0.085851 KPSS Test Equation Dependent Variable: LOG(IPUE) Method: Least Squares Date: 05/11/02 Time: 12:44 Sample: 1961 1999 Included observations: 39 Variable Coefficient Std. Error t-Statistic Prob. C @TREND(1961) -0.049068 0.071544 0.044031 0.001994 -1.114402 35.88226 0.2723 0.0000 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat 0.972066 0.971311 0.140139 0.726638 22.32773 0.059611 Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion F-statistic Prob(F-statistic) 1.310272 0.827367 -1.042447 -0.957137 1287.536 0.000000 TEST DE COINTEGRACIÓN PRECIOS RELATIVOS-TIPO DE CAMBIO PESETA FRENTE A EURO Ecuación estática Dependent Variable: LOG(ITCPTAEURO) Method: Least Squares Date: 05/11/02 Time: 12:46 Sample: 1960 1999 Included observations: 40 Variable Coefficient Std. Error t-Statistic Prob. C LOG(PRPTAEURO) -0.095699 1.075237 0.020236 0.034003 -4.729237 31.62207 0.0000 0.0000 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat 0.963390 0.962426 0.069741 0.184825 50.78701 0.487241 Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion F-statistic Prob(F-statistic) 0.440840 0.359787 -2.439351 -2.354907 999.9553 0.000000 Ajuste modelo autorregresivo aplicado a los residuos Dependent Variable: D(RES) Method: Least Squares Date: 05/Time: 12:55 Sample(adjusted): 1962 1999 Included observations: 38 after adjusting endpoints Variable Coefficient Std. Error t-Statistic Prob. RES(-1) -0.290528 0.109016 -2.665006 0.0113 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood 0.160421 0.160421 0.045015 0.074974 64.41619 Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion Durbin-Watson stat -0.001318 0.049127 -3.337694 -3.294600 1.759691 TEST DE COINTEGRACIÓN A PARTIR DE LA ESTIMACIÓN DEL MODELO DE MECANISMO DE CORRECCIÓN DE ERROR Estimación por mcnl del modelo de mecanismo de corrección de error Dependent Variable: DLOG(ITCPTAEURO) Method: Least Squares Date: 05/10/02 Time: 12:42 Sample: 1961 1999 Included observations: 39 Convergence achieved after 4 iterations DLOG(ITCPTAEURO) =C(1)*DLOG(PRPTAEURO) +C(2) *(LOG(ITCPTAEURO(-1)) +C(3)+C(4)*(LOG(PRPTAEURO(-1)))) C(1) C(2) C(3) C(4) R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Coefficient Std. Error t-Statistic Prob. 0.676916 -0.294986 0.097588 -1.136331 0.335606 0.111976 0.056251 0.083537 2.016998 -2.634361 1.734872 -13.60274 0.0514 0.0125 0.0916 0.0000 0.305598 0.246077 0.045562 0.072656 67.23033 Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion Durbin-Watson stat 0.024752 0.052473 -3.242581 -3.071959 1.809688 Estimación por mco del modelo de mecanismo de corrección de error aplicando el método en dos etapas Dependent Variable: DLOG(ITCPTAEURO) Method: Least Squares Date: 05/11/02 Time: 13:08 Sample(adjusted): 1962 1999 Included observations: 38 after adjusting endpoints DLOG(ITCPTAEURO) =C(1)*DLOG(PRPTAEURO) +C(2)*RES(-1) C(1) C(2) R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Coefficient Std. Error t-Statistic Prob. 0.847287 -0.312570 0.223838 0.111090 3.785263 -2.813658 0.0006 0.0079 0.302119 0.282733 0.044992 0.072874 64.95580 Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion Durbin-Watson stat 0.025127 0.053125 -3.313463 -3.227274 1.734923 En procesos cointegrados la estimación de la ecuación estática ofrece un estimador consistente del efecto multiplicador a largo plazo Estimación de la ecuación estática Dependent Variable: LOG(ITCPTAEURO) Method: Least Squares Date: 05/11/02 Time: 12:55 Sample: 1960 1999 Included observations: 40 Variable Coefficient Std. Error t-Statistic Prob. C LOG(PRPTAEURO) -0.095699 1.075237 0.020236 0.034003 -4.729237 31.62207 0.0000 0.0000 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat 0.963390 0.962426 0.069741 0.184825 50.78701 0.487241 Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion F-statistic Prob(F-statistic) 0.440840 0.359787 -2.439351 -2.354907 999.9553 0.000000 Estimación de un modelo dinámico general Dependent Variable: LOG(ITCPTAEURO) Method: Least Squares Date: 05/11/02 Time: 13:15 Sample(adjusted): 1961 1999 Included observations: 39 after adjusting endpoints Variable Coefficient Std. Error t-Statistic Prob. C LOG(PRPTAEURO) LOG(ITCPTAEURO(1)) LOG(PRPTAEURO(1)) -0.028787 0.676916 0.705014 0.018054 0.335606 0.111976 -1.594480 2.016998 6.296105 0.1198 0.0514 0.0000 -0.341715 0.383095 -0.891984 0.3785 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat 0.985017 0.983732 0.045562 0.072656 67.23033 1.809688 Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion F-statistic Prob(F-statistic) 0.452144 0.357222 -3.242581 -3.071959 766.9749 0.000000 Elasticidad a largo plazo del tipo de cambio con respecto a los precios relativos: Elasticidad=(0.676916-0.341715)/(1-0.705014)=1.1363 En procesos cointegrados el sesgo de simultaneidad desparece Estimación por MCO Dependent Variable: LOG(ITCPTAEURO) Method: Least Squares Date: 05/11/02 Time: 13:23 Sample(adjusted): 1961 1999 Included observations: 39 after adjusting endpoints Convergence achieved after 7 iterations Variable Coefficient Std. Error t-Statistic Prob. C LOG(PRPTAEURO) AR(1) -0.132820 1.132369 0.734801 0.061817 0.093551 0.111475 -2.148612 12.10427 6.591608 0.0385 0.0000 0.0000 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Inverted AR Roots 0.984164 0.983284 0.046185 0.076792 66.15072 1.847350 Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion F-statistic Prob(F-statistic) 0.452144 0.357222 -3.238498 -3.110532 1118.629 0.000000 .73 Estimación por Variables Instrumentales Dependent Variable: LOG(ITCPTAEURO) Method: Two-Stage Least Squares Date: 05/10/02 Time: 13:57 Sample(adjusted): 1962 1999 Included observations: 38 after adjusting endpoints Convergence achieved after 5 iterations Instrument list: C LOG(IPUE) LOG(IPUE(-1)) LOG(IPUE(-2)) Lagged dependent variable & regressors added to instrument list Variable Coefficient Std. Error t-Statistic Prob. C LOG(PRPTAEURO) AR(1) -0.176157 1.196844 0.714983 0.068137 0.101561 0.113647 -2.585338 11.78446 6.291232 0.0141 0.0000 0.0000 R-squared Adjusted R-squared S.E. of regression F-statistic Prob(F-statistic) Inverted AR Roots 0.984432 0.983542 0.045474 1107.236 0.000000 .71 Mean dependent var S.D. dependent var Sum squared resid Durbin-Watson stat 0.463766 0.354467 0.072376 1.859968 EFECTOS DE LA COINTEGRACIÓN SOBRE EL SESGO DE SIMULTANEIDAD. UN EJERCICIO DE SIMULACIÓN Se parte del modelo: CO=0.5R+e R=CO+I Cuya forma reducida viene dada por: R=2·I+2·e CO=I+2·e Postulándose para “e” una normal (0,1). Adicionalmente, para el supuesto de estacionariedad se postula para IEST una normal (0,1), mientras que la INT se genera a través de: IINT=IINT(-1)+0.2+nrnd Este planteamiento permite definir dos muestras artificiales de consumo y de renta. En la primera ambas variables son estacionarias, mientras que en la segunda se está ante dos procesos integrados y cointegrados, cuya representación gráfica es: 600 500 400 300 200 100 0 250 500 RINT 750 COINT 1000 Estimación de la función de consumo en procesos estacionarios MCO Dependent Variable: COEST Method: Least Squares Date: 05/20/02 Time: 11:59 Sample: 2 1000 Included observations: 999 Variable REST R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Coefficient 0.754308 0.895309 0.895309 0.725629 525.4837 -1096.623 Std. Error t-Statistic 0.008160 92.43848 Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion Durbin-Watson stat Prob. 0.0000 -0.072973 2.242635 2.197443 2.202355 1.983166 Variables Instrumentales Dependent Variable: COEST Method: Two-Stage Least Squares Date: 05/20/02 Time: 12:02 Sample: 2 1000 Included observations: 999 Instrument list: IEST Variable REST R-squared Adjusted R-squared S.E. of regression Durbin-Watson stat Coefficient Std. Error t-Statistic Prob. 0.484019 0.017134 28.24893 0.0000 0.780216 0.780216 1.051373 2.001543 Mean dependent var S.D. dependent var Sum squared resid -0.072973 2.242635 1103.174 Comentario Al estimar la función de consumo por MCO se estima un multiplicador de la inversión sobre la renta de 4.1, frente a un valor poblacional de 2, mientras que al aplicar VI el multiplicador estimado, de 1.9, prácticamente coincide con el poblacional. De hecho, el estimador por VI es consistente, lo que significa que tiende al parámetro poblacional. Estimación de la función de consumo con procesos cointegrados MCO Dependent Variable: COINT Method: Least Squares Date: 05/20/02 Time: 12:06 Sample: 2 1000 Included observations: 999 Variable RINT R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Coefficient 0.500048 0.999792 0.999792 1.019189 1036.669 -1436.008 Std. Error t-Statistic 0.000107 4661.543 Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion Durbin-Watson stat Prob. 0.0000 132.6493 70.74434 2.876893 2.881804 2.002130 Variables Instrumentales Dependent Variable: COINT Method: Two-Stage Least Squares Date: 05/20/02 Time: 12:08 Sample: 2 1000 Included observations: 999 Instrument list: IINT Variable RINT R-squared Adjusted R-squared S.E. of regression Durbin-Watson stat Coefficient Std. Error t-Statistic Prob. 0.500025 0.000107 4661.116 0.0000 0.999792 0.999792 1.019213 2.002223 Mean dependent var S.D. dependent var Sum squared resid 132.6493 70.74434 1036.717 Comentario En procesos integrados y cointegrados el sesgo de simultaneidad desaparece, lo que se traduce en que tanto el estimador MCO como por variables instrumentales tiendan al mismo parámetro poblacional. El problema de la regresión espuria 1000 800 600 400 200 0 -200 250 500 Y 750 X Proceso de generación de datos: Y=Y(-1)+1+5*nrnd X=X(-1)+0.5+2.5*nrnd Regresión estimada: Dependent Variable: Y Method: Least Squares Date: 05/12/02 Time: 10:32 Sample: 2 1000 Included observations: 999 Variable C X R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Coefficient Std. Error t-Statistic Prob. 62.91992 1.784421 3.128963 0.011211 20.10887 159.1660 0.0000 0.0000 0.962136 0.962098 50.11421 2503900. -5326.909 0.019054 Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion F-statistic Prob(F-statistic) 492.2687 257.4113 10.66849 10.67831 25333.80 0.000000 1000 Johansen Cointegration Test Date: 05/10/02 Time: 13:28 Sample(adjusted): 1963 1999 Included observations: 37 after adjusting endpoints Trend assumption: Linear deterministic trend Series: LOG(ITCPTAEURO) LOG(PRPTAEURO) Lags interval (in first differences): 1 to 2 Unrestricted Cointegration Rank Test Hypothesized No. of CE(s) Eigenvalue Trace Statistic 5 Percent Critical Value 1 Percent Critical Value None At most 1 0.271454 0.072489 14.50234 2.784277 15.41 3.76 20.04 6.65 *(**) denotes rejection of the hypothesis at the 5%(1%) level Trace test indicates no cointegration at both 5% and 1% levels Hypothesized No. of CE(s) Eigenvalue Max-Eigen Statistic 5 Percent Critical Value 1 Percent Critical Value None At most 1 0.271454 0.072489 11.71807 2.784277 14.07 3.76 18.63 6.65 *(**) denotes rejection of the hypothesis at the 5%(1%) level Max-eigenvalue test indicates no cointegration at both 5% and 1% levels Unrestricted Cointegrating Coefficients (normalized by b'*S11*b=I): LOG(ITCPTA LOG(PRPTAE EURO) URO) -18.59780 20.58009 4.195710 -1.098271 Unrestricted Adjustment Coefficients (alpha): D(LOG(ITCPT AEURO)) D(LOG(PRPT AEURO)) 0.025228 -0.000704 0.001517 -0.004477 1 Cointegrating Equation(s): Log likelihood 164.8164 Normalized cointegrating coefficients (std.err. in parentheses) LOG(ITCPTA LOG(PRPTAE EURO) URO) 1.000000 -1.106587 (0.05471) Adjustment coefficients (std.err. in parentheses) D(LOG(ITCPT -0.469178 AEURO)) (0.13833) D(LOG(PRPT -0.028216 AEURO)) (0.05616) Estimación por el mecanismo de corrección de error vectorial Date: 05/10/02 Time: 13:30 Sample(adjusted): 1963 1999 Included observations: 37 after adjusting endpoints Standard errors in ( ) & t-statistics in [ ] Cointegrating Eq: LOG(ITCPTAEURO(-1)) LOG(PRPTAEURO(-1)) C CointEq1 1.000000 -1.106587 (0.05471) [-20.2281] 0.119668 Error Correction: D(LOG(ITCP TAEURO)) D(LOG(PRPTAEURO)) CointEq1 -0.469178 (0.13833) [-3.39183] -0.028216 (0.05616) [-0.50244] D(LOG(ITCPTAEURO(-1))) 0.276323 (0.16272) [ 1.69819] 0.086354 (0.06606) [ 1.30722] D(LOG(ITCPTAEURO(-2))) 0.058341 (0.16491) [ 0.35378] -0.058405 (0.06695) [-0.87239] D(LOG(PRPTAEURO(-1))) 0.570310 (0.44831) [ 1.27214] 0.696958 (0.18201) [ 3.82933] D(LOG(PRPTAEURO(-2))) -0.983345 (0.43317) [-2.27011] -0.228455 (0.17586) [-1.29908] C 0.028117 (0.01224) [ 2.29652] 0.012267 (0.00497) [ 2.46801] 0.364284 0.261749 0.063452 0.045242 3.552777 65.31434 -3.206181 -2.944951 0.026938 0.052655 0.445402 0.355951 0.010458 0.018367 4.979267 98.66788 -5.009075 -4.747845 0.024925 0.022887 R-squared Adj. R-squared Sum sq. resids S.E. equation F-statistic Log likelihood Akaike AIC Schwarz SC Mean dependent S.D. dependent Determinant Residual Covariance Log Likelihood Log Likelihood (d.f. adjusted) Akaike Information Criteria Schwarz Criteria 6.60E-07 164.8164 158.2699 -7.798375 -7.188838 Estimación por el mecanismo de corrección de error vectorial con restricciones sobre la relación de cointegración Vector Error Correction Estimates Date: 05/10/02 Time: 13:36 Sample(adjusted): 1963 1999 Included observations: 37 after adjusting endpoints Standard errors in ( ) & t-statistics in [ ] Cointegration Restrictions: A(2,1)=0 Convergence achieved after 4 iterations. Not all cointegrating vectors are identified LR test for binding restrictions (rank = 1): Chi-square(1) 0.236937 Probability 0.626427 Cointegrating Eq: CointEq1 LOG(ITCPTAEURO(-1)) -18.26837 LOG(PRPTAEURO(-1)) 20.46708 C -2.315684 Error Correction: D(LOG(ITCPTAEURO)) CointEq1 0.024443 (0.00726) [ 3.36687] D(LOG(ITCPTAEURO(-1))) D(LOG(ITCPTAEURO(-2))) D(LOG(PRPTAEURO(-1))) D(LOG(PRPTAEURO(-2))) C R-squared Adj. R-squared Sum sq. resids S.E. equation F-statistic Log likelihood Akaike AIC Schwarz SC Mean dependent S.D. dependent Determinant Residual Covariance Log Likelihood Log Likelihood (d.f. adjusted) Akaike Information Criteria Schwarz Criteria 0.264712 (0.16190) [ 1.63501] 0.044903 (0.16390) [ 0.27396] 0.609137 (0.44650) [ 1.36425] -0.977940 (0.43358) [-2.25551] 0.027656 (0.01223) [ 2.26092] 0.362168 0.259291 0.063664 0.045317 3.520421 65.25286 -3.202857 -2.941628 0.026938 0.052655 D(LOG(PRPTAEURO)) 0.000000 (0.00000) [ NA ] 0.083392 (0.06572) [ 1.26887] -0.061512 (0.06653) [-0.92452] 0.703440 (0.18125) [ 3.88109] -0.222260 (0.17600) [-1.26282] 0.012109 (0.00497) [ 2.43866] 0.443692 0.353965 0.010491 0.018396 4.944903 98.61093 -5.005996 -4.744766 0.024925 0.022887 6.61E-07 164.6979 158.2445 -7.797003 -7.187466