Bangladesh University of Professionals (BUP)
Faculty of Business Studies (FBS)
Department of Business Administration in Finance and Banking
Course name: Advanced Business Statistics
Course code: ALD 2204
ABS_Major Tasks#02: Term Paper
Submitted to
Md. Nahid Alam
Assistant Professor
Department of Business Administration in Finance and Banking
Bangladesh University of Professionals (BUP)
Submitted by
Group: 06
Section: A
Session: 2020-21
BBA in Finance and Banking
Date of submission: 05/05/2023
1
Name of the Group Members
Name ID
Sajia Afrin Syenthi 2122151017
Hafsa Ahmed 2122151027
Tasfia Islam Ami 2122151029
Maliha Rashid 2122151051
Tasnim Mehjabin Zara 2122151113
Md Mahmudul Islam 2122151127
Md. Abu Taheir Limon 2122151135
2
Answer to the question no.- i
We conduct a Residual Diagnosis test in EViews for checking if there is any evidence of
multicollinearity in the given data. The result is -
Dependent Variable: DV
Method: Least Squares
Sample: 1976 2019
Included observations: 44
Variable Coefficient Std. Error t-Statistic Prob.
C 18.09998 0.741897 24.39689 0.0000
IV1 0.007074 0.005190 1.363053 0.1811
IV2 0.013208 0.011459 1.152634 0.2565
IV3 0.842082 0.259398 3.246293 0.0025
IV4 -0.017005 0.076412 -0.222544 0.8251
IV5 0.044816 0.008719 5.139976 0.0000
IV6 0.132360 0.065522 2.020095 0.0507
R-squared 0.981482 Mean dependent var 24.62291
Adjusted R-squared 0.978479 S.D. dependent var 0.920151
S.E. of regression 0.134987 Akaike info criterion -1.022365
Sum squared resid 0.674196 Schwarz criterion -0.738517
Log likelihood 29.49204 Hannan-Quinn criter. -0.917101
F-statistic 326.8384 Durbin-Watson stat 0.983879
Prob(F-statistic) 0.000000
We know that multicollinearity exists when independent variables are correlated and if the
correlation between two independent variables is not between -0.70 and 0.70, then there’s evidence
of multicollinearity.
From the correlation test, we get that the correlation coefficient of IV3 is 0.84 which exceeds the
range of -0.70 to 0.70.
For more precise test we conduct a VIF test. We consider the Centered VIF value and a VIF>10 is
unsatisfactory.
3
Variance Inflation Factors
Sample: 1976 2019
Included observations: 44
Coefficient Uncentered Centered
Variable Variance VIF VIF
C 0.550411 1329.094 NA
IV1 2.69E-05 7.744116 1.738577
IV2 0.000131 97.79161 13.82857
IV3 0.067287 2337.433 47.41654
IV4 0.005839 6.499977 3.746476
IV5 7.60E-05 11.71783 6.247462
IV6 0.004293 4803.254 25.40529
From the VIF test, we get that IV2, IV3 and IV6 exceeds the limit of 10.
So, we can conclude that there’s evidence of the presence of multicollinearity in data.
4
Answer to the question no.- ii
For the part of the remedial action, as multicollinearity exists in the data, we would eliminate IV3
first which has the highest VIF value among IV2, IV3 and IV6.
Eliminating IV3
Dependent Variable: DV
Method: Least Squares
Sample: 1976 2019
Included observations: 44
Variable Coefficient Std. Error t-Statistic Prob.
C 16.80600 0.699861 24.01335 0.0000
IV1 0.003537 0.005676 0.623136 0.5369
IV2 0.046052 0.006019 7.651090 0.0000
IV4 0.085570 0.077817 1.099636 0.2784
IV5 0.024464 0.006778 3.609559 0.0009
IV6 0.319643 0.034741 9.200734 0.0000
R-squared 0.976207 Mean dependent var 24.62291
Adjusted R-squared 0.973077 S.D. dependent var 0.920151
S.E. of regression 0.150981 Akaike info criterion -0.817200
Sum squared resid 0.866222 Schwarz criterion -0.573901
Log likelihood 23.97839 Hannan-Quinn criter. -0.726973
F-statistic 311.8266 Durbin-Watson stat 1.389876
Prob(F-statistic) 0.000000
5
Variance Inflation Factors
Sample: 1976 2019
Included observations: 44
Coefficient Uncentered Centered
Variable Variance VIF VIF
C 0.489805 945.4323 NA
IV1 3.22E-05 7.402701 1.661928
IV2 3.62E-05 21.56588 3.049600
IV4 0.006055 5.388517 3.105850
IV5 4.59E-05 5.659596 3.017461
IV6 0.001207 1079.423 5.709266
Answer to the question no.- iii
UNIT ROOT TEST RESULTS TABLE (ADF)
Null Hypothesis: the variable has a unit root
At Level
IV1 IV2 IV4 IV5 IV6
With Constant t-Statistic -3.9629 -2.8818 -1.6146 -0.2350 0.4592
Prob. 0.0044 0.0558 0.4666 0.9259 0.9831
*** * n0 n0 n0
With Constant & Trend t-Statistic -4.9219 -4.6505 -2.7503 0.8502 -2.0438
Prob. 0.0016 0.0029 0.2228 0.9997 0.5606
*** *** n0 n0 n0
Without Constant & Trend t-Statistic -2.9869 0.2654 -0.9931 0.7664 3.0232
Prob. 0.0038 0.7584 0.2826 0.8756 0.9991
*** n0 n0 n0 n0
At First Difference
d(IV1) d(IV2) d(IV4) d(IV5) d(IV6)
With Constant t-Statistic -10.4502 -8.3672 -2.9457 -6.2786 -7.5295
Prob. 0.0000 0.0000 0.0497 0.0000 0.0000
*** *** ** *** ***
With Constant & Trend t-Statistic -11.6355 -8.2418 -2.7684 -4.6394 -7.5933
Prob. 0.0000 0.0000 0.2173 0.0035 0.0000
*** *** n0 *** ***
Without Constant & Trend t-Statistic -10.2560 -8.2262 -8.0117 -6.1054 -5.3985
Prob. 0.0000 0.0000 0.0000 0.0000 0.0000
*** *** *** *** ***
6
Notes:
a: (*)Significant at the 10%; (**)Significant at the 5%; (***) Significant at the 1% and (no) Not Significant
b: Lag Length based on SIC
c: Probability based on MacKinnon (1996) one-sided p-values.
This Result is The Out-Put of Program Has Developed By:
Dr. Imadeddin AlMosabbeh
College of Business and Economics
Qassim University-KSA
All the variables in the model must
be integrated in the same order to run either the VAR or the VECM model.
However, the ARDL bounds testing approach introduced by Pesaran et al.
(2001) does not need this requirement to be satisfied. So, the decision to
choose a suitable model relies on the result of stationarity. As the variables under consideration
are integrated at different orders, we
have no option but to use the ARDL model instead of other cointegration
models.
1. IV1 is stationary at level
2. IV2 is stationary at 1st difference
3. IV4 is stationary at 1st difference
4. IV5 is stationary at 1st difference
5. IV6 is stationary at 1st difference
7
Answer to the question no.- iv, v
Dependent Variable: DV
Method: ARDL
Sample (adjusted): 1978 2019
Included observations: 42 after adjustments
Maximum dependent lags: 4 (Automatic selection)
Model selection method: Schwarz criterion (SIC)
Dynamic regressors (4 lags, automatic): IV1 IV2 IV4 IV5 IV6
Fixed regressors: C
Number of models evalulated: 12500
Selected Model: ARDL(1, 0, 2, 0, 0, 2)
Note: final equation sample is larger than selection sample
Variable Coefficient Std. Error t-Statistic Prob.*
DV(-1) 0.693554 0.079264 8.749960 0.0000
IV1 0.005807 0.002338 2.484077 0.0186
IV2 0.000418 0.004018 0.103988 0.9178
IV2(-1) -0.002132 0.003720 -0.573095 0.5707
IV2(-2) 0.009754 0.003720 2.622081 0.0134
IV4 0.062760 0.025226 2.487934 0.0184
IV5 0.007732 0.002963 2.609950 0.0138
IV6 0.002012 0.026484 0.075953 0.9399
IV6(-1) 0.016855 0.034111 0.494124 0.6247
IV6(-2) 0.093748 0.034310 2.732401 0.0103
C 4.968300 1.410650 3.521993 0.0014
R-squared 0.997933 Mean dependent var 24.69953
Adjusted R-squared 0.997266 S.D. dependent var 0.869283
S.E. of regression 0.045455 Akaike info criterion -3.124041
Sum squared resid 0.064052 Schwarz criterion -2.668937
Log likelihood 76.60486 Hannan-Quinn criter. -2.957227
F-statistic 1496.356 Durbin-Watson stat 1.257567
Prob(F-statistic) 0.000000
*Note: p-values and any subsequent tests do not account for model
selection.
8
Long run bounds test,
ARDL Long Run Form and Bounds Test
Dependent Variable: D(DV)
Selected Model: ARDL(1, 0, 2, 0, 0, 2)
Case 2: Restricted Constant and No Trend
Sample: 1976 2019
Included observations: 42
Conditional Error Correction Regression
Variable Coefficient Std. Error t-Statistic Prob.
C 4.968300 1.410650 3.521993 0.0014
DV(-1)* -0.306446 0.079264 -3.866163 0.0005
IV1** 0.005807 0.002338 2.484077 0.0186
IV2(-1) 0.008040 0.005151 1.560817 0.1287
IV4** 0.062760 0.025226 2.487934 0.0184
IV5** 0.007732 0.002963 2.609950 0.0138
IV6(-1) 0.112614 0.026511 4.247835 0.0002
D(IV2) 0.000418 0.004018 0.103988 0.9178
D(IV2(-1)) -0.009754 0.003720 -2.622081 0.0134
D(IV6) 0.002012 0.026484 0.075953 0.9399
D(IV6(-1)) -0.093748 0.034310 -2.732401 0.0103
* p-value incompatible with t-Bounds distribution.
** Variable interpreted as Z = Z(-1) + D(Z).
Levels Equation
Case 2: Restricted Constant and No Trend
Variable Coefficient Std. Error t-Statistic Prob.
IV1 0.018950 0.009228 2.053443 0.0485
IV2 0.026236 0.011939 2.197543 0.0356
IV4 0.204798 0.083776 2.444593 0.0204
IV5 0.025232 0.007131 3.538194 0.0013
IV6 0.367485 0.050396 7.291989 0.0000
C 16.21263 0.908674 17.84207 0.0000
EC = DV - (0.0190*IV1 + 0.0262*IV2 + 0.2048*IV4 + 0.0252*IV5 + 0.3675
*IV6 + 16.2126 )
F-Bounds Test Null Hypothesis: No levels relationship
Test Statistic Value Signif. I(0) I(1)
Asymptotic: n=1000
F-statistic 21.66474 10% 2.08 3
k 5 5% 2.39 3.38
2.5% 2.7 3.73
1% 3.06 4.15
Actual Sample Size 42 Finite Sample: n=45
10% 2.276 3.297
5% 2.694 3.829
1% 3.674 5.019
Finite Sample: n=40
10% 2.306 3.353
5% 2.734 3.92
1% 3.657 5.256
9
F-statistic> upper bound (at 1%level of significance)
So, reject null. which means there is level of relationship. The null hypothesis is rejected in case
of calculated F statistic being greater than the upper bound critical value. If the calculated F statistic
falls below the lower bound critical value, then the null hypothesis cannot be rejected. If the F
statistic stays between the lower and upper bound critical values, then we cannot reach any
conclusion.
Answer to the question no.- vi
Normality,
8
Series: Residuals
7 Sample 1978 2019
Observations 42
6
Mean -1.78e-15
5
Median 0.003465
4 Maximum 0.074057
Minimum -0.113867
3 Std. Dev. 0.039525
Skewness -0.461764
2 Kurtosis 3.486011
1
Jarque-Bera 1.905942
0 Probability 0.385594
-0.12 -0.10 -0.08 -0.06 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08
Null hypothesis, Ho; normally distributed
Alternate hypothesis, Hi; not normally distributed
P-value >o
So, do not reject null.
Data is normally distributed.
10
Autocorrelation,
Breusch-Godfrey Serial Correlation LM Test:
F-statistic 2.959268 Prob. F(2,29) 0.0677
Obs*R-squared 7.118813 Prob. Chi-Square(2) 0.0285
Test Equation:
Dependent Variable: RESID
Method: ARDL
Date: 05/04/23 Time: 10:47
Sample: 1978 2019
Included observations: 42
Presample missing value lagged residuals set to zero.
Variable Coefficient Std. Error t-Statistic Prob.
DV(-1) -0.019077 0.079266 -0.240663 0.8115
IV1 0.000827 0.002229 0.370956 0.7134
IV2 0.000335 0.003836 0.087435 0.9309
IV2(-1) 0.000520 0.003539 0.146988 0.8842
IV2(-2) 0.000575 0.003524 0.163145 0.8715
IV4 -0.007187 0.024189 -0.297126 0.7685
IV5 0.000495 0.002863 0.173004 0.8638
IV6 0.011626 0.025429 0.457189 0.6509
IV6(-1) -0.011848 0.032821 -0.361007 0.7207
IV6(-2) 0.007475 0.032688 0.228687 0.8207
C 0.283364 1.393321 0.203373 0.8403
RESID(-1) 0.448989 0.193897 2.315608 0.0279
RESID(-2) -0.255575 0.192033 -1.330893 0.1936
R-squared 0.169496 Mean dependent var -1.78E-15
Adjusted R-squared -0.174161 S.D. dependent var 0.039525
S.E. of regression 0.042829 Akaike info criterion -3.214525
Sum squared resid 0.053196 Schwarz criterion -2.676675
Log likelihood 80.50502 Hannan-Quinn criter. -3.017382
F-statistic 0.493211 Durbin-Watson stat 1.972726
Prob(F-statistic) 0.901968
11
Ho , there is no correlation
Hi, there is correlation
P-value >5%
Do not reject null. there is no serial correlation.
Hed test,
Heteroskedasticity Test: Breusch-Pagan-Godfrey
F-statistic 2.026940 Prob. F(10,31) 0.0646
Obs*R-squared 16.60474 Prob. Chi-Square(10) 0.0836
Scaled explained SS 11.24423 Prob. Chi-Square(10) 0.3388
Test Equation:
Dependent Variable: RESID^2
Method: Least Squares
Date: 05/04/23 Time: 10:47
Sample: 1978 2019
Included observations: 42
Variable Coefficient Std. Error t-Statistic Prob.
C 0.083385 0.067541 1.234585 0.2263
DV(-1) -0.006424 0.003795 -1.692657 0.1005
IV1 0.000178 0.000112 1.589506 0.1221
IV2 0.000231 0.000192 1.200211 0.2391
IV2(-1) -6.01E-06 0.000178 -0.033726 0.9733
IV2(-2) 1.18E-05 0.000178 0.066410 0.9475
IV4 -0.001446 0.001208 -1.196888 0.2404
IV5 -7.01E-05 0.000142 -0.494411 0.6245
IV6 0.004159 0.001268 3.279544 0.0026
IV6(-1) -0.001660 0.001633 -1.016182 0.3174
IV6(-2) 0.000827 0.001643 0.503187 0.6184
12
R-squared 0.395351 Mean dependent var 0.001525
Adjusted R-squared 0.200303 S.D. dependent var 0.002434
S.E. of regression 0.002176 Akaike info criterion -9.202196
Sum squared resid 0.000147 Schwarz criterion -8.747092
Log likelihood 204.2461 Hannan-Quinn criter. -9.035382
F-statistic 2.026940 Durbin-Watson stat 2.105567
Prob(F-statistic) 0.064650
p-value >5%
There is homoskedasticity
The results were obtained by using the Independent Variables provided in the dataset and running
the appropriate commands in the Stata Application to achieve the correlation matrix, regression
model and VIF tests.
13
