Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 40 views

    Tutorial 11

    Uploaded by

    HENG QUAN PAN
    Based on the output provided, conduct the following tests and report your conclusions: a. Test for heteroskedasticity at 5% significance level. b. Test for autocorrelation at 10% significance level. c. Test for normality of the residuals at 1% significance level.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd

    Tutorial 11

    Uploaded by

    HENG QUAN PAN
    40 views6 pages
    Based on the output provided, conduct the following tests and report your conclusions: a. Test for heteroskedasticity at 5% significance level. b. Test for autocorrelation at 10% significance level. c. Test for normality of the residuals at 1% significance level.

    Copyright

    © © All Rights Reserved

    Available Formats

    DOCX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Based on the output provided, conduct the following tests and report your conclusions: a. Test for heteroskedasticity at 5% significance level. b. Test for autocorrelation at 10% significance level. c. Test for normality of the residuals at 1% significance level.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    40 views6 pages

    Tutorial 11

    Uploaded by

    HENG QUAN PAN
    Based on the output provided, conduct the following tests and report your conclusions: a. Test for heteroskedasticity at 5% significance level. b. Test for autocorrelation at 10% significance level. c. Test for normality of the residuals at 1% significance level.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    of 6

    Question 1

    Table 4.2

    Y X1
    100 3540
    120 3500
    130 3420
    140 3410
    152 3360
    162 3351
    178 3250
    185 3210
    195 3150
    200 3100
    250 3050
    254 3024
    260 3018
    270 3008
    285 2988

    a. Given the data in Table 4.2, Regress lny = β1 + β2 lnx1 + β3 lnx2 + β4 lnx3 + β5 lnx4.
    Dependent Variable: LNY
    Method: Leas t Squares
    Date: 04/14/22 Time: 13:44
    Sample: 1 15
    Included observations: 15

    Variable Coefficient Std. Error t-Statistic Prob.

    C -37.43388 39.52267 -0.947150 0.3659


    LNX1 -0.307061 1.840905 -0.166799 0.8709
    LNX2 0.756649 0.400973 1.887030 0.0885
    LNX3 0.042310 0.493233 0.085781 0.9333
    LNX4 4.173017 2.366343 1.763487 0.1083

    R-squared 0.985122 Mean dependent var 5.210477


    Adjusted R-squared 0.979171 S.D. dependent var 0.324007
    S.E. of regres sion 0.046762 Akaike info criterion -3.026309
    Sum squared resid 0.021866 Schwarz criterion -2.790292
    Log likelihood 27.69732 Hannan-Quinn criter. -3.028823
    F-statistic 165.5343 Durbin-Watson stat 2.471554
    Prob(F-statistic) 0.000000

    lny= -37.4339- 0.3071 lnx1+ 0.7566 lnx2+ 0.0423 lnx3+ 4.1730 lnx4

    b. Test at 10% significance level if the errors are series uncorrelated.


    Breusch-Godfrey Serial Correlation LM Test:
    Null hypothes is: No serial correlation at up to 2 lags

    F-statistic 1.110182 Prob. F(2,8) 0.3754


    Obs*R-squared 3.258735 Prob. Chi-Square(2) 0.1961

    Test Equation:
    Dependent Variable: RESID
    Method: Leas t Squares
    Date: 04/14/22 Time: 13:48
    Sample: 1 15
    Included observations: 15
    Pres ample missing value lagged residuals set to zero.

    Variable Coefficient Std. Error t-Statistic Prob.

    C -14.62019 40.52313 -0.360786 0.7276


    LNX1 0.954037 1.931466 0.493944 0.6346
    LNX2 0.336897 0.456550 0.737918 0.4816
    LNX3 0.007406 0.497371 0.014890 0.9885
    LNX4 0.426233 2.378051 0.179236 0.8622
    RESID(-1) -0.547654 0.380611 -1.438881 0.1881
    RESID(-2) -0.342911 0.359228 -0.954577 0.3677

    R-squared 0.217249 Mean dependent var -3.79E-15


    Adjusted R-squared -0.369814 S.D. dependent var 0.039521
    S.E. of regres sion 0.046255 Akaike info criterion -3.004583
    Sum squared resid 0.017116 Schwarz criterion -2.674160
    Log likelihood 29.53437 Hannan-Quinn criter. -3.008103
    F-statistic 0.370061 Durbin-Watson stat 2.203376
    Prob(F-statistic) 0.878800

    H0: No autocorrelation
    H1: Autocorrelation
    P= 0.1961 (>0.1), not significant at 10% significance level.
    Fail to reject H0, there is no autocorrelation in the model.

    c. Test at 5% significance level if the errors are heteroscedasticity.


    Heteroskedasticity Test: White
    Null hypothes is: Homoskedasticity

    F-statistic 1.539391 Prob. F(6,8) 0.2791


    Obs*R-squared 8.037968 Prob. Chi-Square(6) 0.2353
    Scaled explained SS 2.893494 Prob. Chi-Square(6) 0.8221

    Test Equation:
    Dependent Variable: RESID^2
    Method: Leas t Squares
    Date: 04/14/22 Time: 13:52
    Sample: 1 15
    Included observations: 15
    Collinear test regressors dropped from specification

    Variable Coefficient Std. Error t-Statistic Prob.

    C 1.400952 0.861188 1.626767 0.1424


    LNX1^2 -1.66E-05 0.148904 -0.000111 0.9999
    LNX1*LNX2 0.043566 0.157235 0.277077 0.7887
    LNX1*LNX3 -0.054305 0.390559 -0.139045 0.8929
    LNX1*LNX4 -0.013661 0.012214 -1.118414 0.2958
    LNX2^2 -0.022381 0.074171 -0.301748 0.7705
    LNX3^2 0.027403 0.194310 0.141027 0.8913

    R-squared 0.535865 Mean dependent var 0.001458


    Adjusted R-squared 0.187763 S.D. dependent var 0.001920
    S.E. of regres sion 0.001731 Akaike info criterion -9.575708
    Sum squared resid 2.40E-05 Schwarz criterion -9.245285
    Log likelihood 78.81781 Hannan-Quinn criter. -9.579228
    F-statistic 1.539391 Durbin-Watson stat 2.430886
    Prob(F-statistic) 0.279134

    H0: No heteroscedasticity (Homoscedasticity)


    H1: Autocorrelation
    P = 0.2353 (>0.05), not significant at 5% significance level.
    Fail to reject H0, there is no heteroscedasticity in the model.

    d. Test at 1% significance level if the errors are normally distributed.


    7
    Seri es: Residual s
    6 Sample 1 15
    Obse rvations 15
    5
    Mean -3.79e-15
    4 Median -0.005671
    Maximum 0.067277
    3 Minimum -0.075208
    Std. Dev. 0.039521
    2
    Skewness -0.026242
    1 Kurtosis 2.619902

    0 Jarque-Bera 0.092018
    -0.10 -0.05 0.00 0.05 Probability 0.955033

    H0: Normal distributed


    H1: Not normal distributed
    P = 0.9550 (>0.01), not significant at 1% significance level.
    Fail to reject H0, the model is normal distributed.

    e. Test whether the model specification is correct and stable at 1% significance level.
    Ramsey RESET Test
    Equation: UNTITLED
    Omitted Variables: Squares of fitted values
    Specification: LNY C LNX1 LNX2 LNX3 LNX4

    Value df Probability
    t-statistic 1.484871 9 0.1717
    F-statistic 2.204841 (1, 9) 0.1717
    Likelihood ratio 3.286820 1 0.0698

    F-test summary:
    Sum of Sq. df Mean Squares
    Test SSR 0.004303 1 0.004303
    Restricted SSR 0.021866 10 0.002187
    Unrestricted SSR 0.017564 9 0.001952

    LR test summary:
    Value
    Restricted LogL 27.69732
    Unrestricted LogL 29.34073

    Unrestricted Test Equation:


    Dependent Variable: LNY
    Method: Least Squares
    Date: 04/14/22 Time: 13:59
    Sample: 1 15
    Included observations: 15

    Variable Coefficient Std. Error t-Statistic Prob.

    C -310.0642 187.3634 -1.654882 0.1323


    LNX1 -3.952843 3.008812 -1.313755 0.2214
    LNX2 4.778317 2.734791 1.747233 0.1145
    LNX3 -0.040192 0.469261 -0.085650 0.9336
    LNX4 35.57970 21.26893 1.672848 0.1287
    FITTED^2 -0.717872 0.483458 -1.484871 0.1717

    R-squared 0.988050 Mean dependent var 5.210477


    Adjusted R-squared 0.981411 S.D. dependent var 0.324007
    S.E. of regression 0.044176 Akaike info criterion -3.112097
    Sum squared resid 0.017564 Schwarz criterion -2.828877
    Log likelihood 29.34073 Hannan-Quinn criter. -3.115114
    F-statistic 148.8238 Durbin-Watson stat 2.395544
    Prob(F-statistic) 0.000000

    H0: No mis-specified
    H1: Mis-specified
    P = 0.0698 (>0.01), not significant at 1% significance level.
    Failed to reject H0, there is no mis-specified in the model.
    Question 2

    Consider the following EViews’ output for an auxiliary regression:

    Test Equation:
    Dependent Variable: RESID^2
    Method: Least Squares
    Sample: 1980 2016
    Included observations: 37

    Variable Coefficient Std. Error t-Statistic Prob.

    C -0.026123 0.053191 -0.491115 0.6273


    LL^2 0.014859 0.013959 1.064436 0.2966
    LL*LR -0.024856 0.038021 -0.653762 0.5188
    LL*LCO -0.020506 0.011025 -1.859926 0.0738
    LL 0.014214 0.042722 0.332715 0.7419
    LR^2 0.021081 0.041863 0.503568 0.6186
    LR*LCO 0.007467 0.016840 0.443384 0.6610
    LR -0.021494 0.058899 -0.364924 0.7180
    LCO^2 -0.007545 0.004057 -1.859575 0.0739
    LCO 0.041021 0.027378 1.498294 0.1457

    R-squared 0.352525     Mean dependent var 0.000963


    Adjusted R-squared 0.136700     S.D. dependent var 0.001042
    S.E. of regression 0.000968     Akaike info criterion -10.81697
    Sum squared resid 2.53E-05     Schwarz criterion -10.38159
    Log likelihood 210.1140     Hannan-Quinn criter. -10.66348
    F-statistic 1.633385     Durbin-Watson stat 2.176753
    Prob(F-statistic) 0.155642

    Based on the White’s General Heteroscedasticity test, identify and explain whether
    there is a heteroscedasticity problem at 5% significance level.

    H0: No heteroscedasticity (Homoscedasticity)


    H1: Autocorrelation

    Chi-square= n x R2= 37 x 0.3525= 13.034


    X2 df= X2 9= 16.919

    Fail to reject H0 since the Chi-square statistic is smaller than the Chi-square critical value.
    There is no heteroscedasticity in the model.

    You might also like