Coefficient Std. Error T-Ratio P-Value: Running OLS On Data
Coefficient Std. Error T-Ratio P-Value: Running OLS On Data
Coefficient Std. Error T-Ratio P-Value: Running OLS On Data
Now checking Autocorrelation :- For this data were defined as time series in gretl option.
Hypothesis:-
In the above output we have Durbin Watson calculated value 1.100797 (d=1.100797), we have to
check this value in critical region. The upper and lower value of Durbin Watson critical region value
will be checked at 5% level of significance. We don’t have Dl and DU points for 88 th observation
therefore at 90th observation the points are taken.
Dl=1.566
DU=1.751
4-DU= 4-1.751=2.249
4-DL=4-1.566=2.434
After checking Durbin Watson calculated value in critical region we have concluded that d calculated
value fall in rejection area therefore our null hypothesis will be rejected and we have problem of
autocorrelation. Or p-value = 1.68134e-006 (0.000000168134) less than significance level 0.05 so we
reject our null hypothesis and we have problem of autocorrelation.
Hypothesis:
---------------------------------------------------------
Interprtation
2 2
χ calculated value is (n− p )R =(88-2)x 0.205498=17.67283
Now χ 2 tabulated
χ 2 (0.05)(2)=5.9914.
Our Chi square calculated value is greater than (17.67283> 5.9914) chi square tabulated value so
we may reject the null hypothesis and we have problem of autocorrelation.
Remedy to Autocorrelation
Interpretation:-
In the above result it is shown that at third iteration our value of Roh became constant and it changes
no more therefore we conclude that there is no more autocorrelation problem. The value of roh at 3 rd
iteration is 0.48 and at 4th iteration it is also 0.48 this indicate no autocorrelation.
HetroSkedasity
Detection Test
1. Gold Felf quadnt Test. (This test is for one indepent variable but for the sake of understanding
this is applied)
Fcal=RSS2/RSS1
Ftab=(a)(v1,v2)
n−c −2 k
2
v1=v2=
Hypothesis:-
RSS1= 0.008544
Step 2 RSS2
RSS2 = 0.000010
Interpretaion
Fcal=RSS2/Rss1= 0.000010/0.008544= 0.000117041
Ftab=(a)(v1,v2)
n−c −2 k
2
v1=v2=
v1=v2= (88-22-2(5))/2
v1=v2=28
Fcal value is less than Ftab i.e. (0.000010 < 1.87) which means we may accept our null hypothesis and
conclude that we have no problem of heteroskedasticity.
Hypothesis:-
H0: There is no problem of heteroskedasticity
H1: There is problem of heteroskedasticity.
Interpretation:-
χ2
In the above results ESS value is 80.2249 to divide this value by 2 we will get calculated
value i.e. 80.2249/2= 40.11247
2
χ
tabulated=(0.05)(5-1)
2
χ
tabulated=(0.05)(4)=9.48773
2 2
χ χ
Conclusion:- calculated value is greater than tabulated value i.e. 40.11247 > 9.48773
we may reject our null hypothesis and we have problem of heteroskedasticity.
Hypothesis:-
H0: There is no problem of heteroskedasticity
H1: There is problem of heteroskedasticity.
Interpretation
In the above table we have unadjusted R2 = 0.201673 we have to multiply this value with n i.e.
88 to get chi square calculated value.
2
χ calculated= 0.201673 x 88 = 17.747224.
2
χ tabulated=(0.05)(14)=23.6848
2 2
Conclusion:- As χ calculated value is greater than χ tabulated (17.747224 >23.6848) we
will reject our null hypothesis and we have problem of heteroskedasticity.
Remedy to hetroskedasticity
White corrected Test
Interpretation:-
After running robust test the option of heteroskedasticity was not avaible in test tab which means we
have no more heteroskedasticity in our model.
Multicolinearity:-
Detection Test
1.Correlation Matrix
Interpretation:-
We have three methods to check whether we have multicolineartiy among the independt variables.
First Method:-
Check correation between independent variables and comparing it with 0.6. If the value of correlation is
greater than 0.6 so we have munlticolinearty among independent variables.
Correlation value of CA and SZ is 0.5224 < 0.6 so there is no multicolineartiy among them
Correlation value of lgdt and SZ is 0.2144< 0.6 so there is no multicolineartiy among them
Correlation value of lvrg and sz is 0.3270< 0.6 so there is no multicolineartiy among them
Correlation value of CA and Lqdt is 0.0826> 0.6 so there is multicolineartiy among them
Correlation value of CA and lvrg is 0.7441> 0.6 so there is multicolineartiy among them
Correlation value of lvrg and lqdt is 0120< 0.6 so there is no multicolineartiy among them
Method 2.
Comparing the correlation value of two independent variables with the correlation value of one
dependent and one independent variable (X1,X2 with Y,X1 and X1,X2 with YX2).
Correlation value of CA and SZ is 0.5224 while the correlation value of CA and ROA is 0.0091 (0.5224 <
0.00910), Further correlation value of SZ and ROA is 0.0891as (0.5224<0.0891) so we have no
multicolinearity problem.
Method 3.
VIF method
Interpretation:
Value of VIF of all independent variables is less than 10 So we have no multicolineartiy problem among
the independent variables.
Diagnostic Tes
Structural break.
Chow Test:-
Hypothesis:-
Interpretation :-
In the above table we have F calculated value is 4.79397 and its p value is 0.0007 which is less than 0.05
so we have to reject our null hypothesis and we concluded that we have structure break problem in our
data.
2. MWD TEST (Makinnon, White, Davidson Test).
Hypothesis
Ho: Linear Model is good.
H1: Log linear Model is good.
Linear Model
Interpretation:-
In the above results coefficient value of Z is -0.0175839 and its p value is 0.46 which is greater
than level of significance 0.05 which means our null hypothesis will be accepted which is linear
model is good.
Ramsey Test:-
-------------------------------------------------------
Interpreation:-
F calculated value is 3.84 and its p value is 0.053 which is greater than level of significance 0.05 which
means we may accept our null hypothesis and we have no specification problem.
Cube Only
---------------------------------------------------------
---------------------------------------------------------
Interpreatiaon