Panel Cointegration

[mezeadi]

Hi, I am really new to econometrics with eviews so i'm sorry if my questions are silly and have obvious answers.
First I am trying to run panel unit root and panel cointegration test on eviews 6. I have been able to confirm the presence of unit root as the results show high p values at levels and low p values at first difference confirming the null of unit root in all my variables. However when i proceed to test for cointegration the seven Pendroni statics mostly show high p values, and I am unable to interpret the Kao and also the Fisher (johansen) test. I know that cointegration is definately as I am using the very same data set used previously to establish that. My questions are:
1. What can i do to my data set to better my result?

2. Are there certain things i ought to do differently in running the cointegration test other than the syntax which i have used i.e. "coint RGDP PRIV" and select the pedroni or any other test type?

3. Is it possible to run the same command above including all my ancillary variable which also have unit root? i.e. "coint RGDP PRIV GFCF OPEN CPI". However in this case I am even more confused with the test results.

4. Can a few missing observation in my data set distort my result?

5. Does it matter if i get better results when i dont include intercept and time trent in my coint test result even though one of my primary variable RGDP (real GDP per capita) clearly shows a deterministic trend for all the countries from their plots.

6. Intrestingly, When i do the cointegration test by changing the position of the variables i.e. "coint PRIV RGDP" more of the pendroni test stats become significant enuogh for me to reject the null of no cointegration. What does mean then? that Priv(private credit) is the dependent variable right? and any causality is from PRIV to RGDP? In that case something is wrong because it is well established from the paper i am replicating using the same Variables and DAta set that it is the other way round.

7. you may view these result and give an opinion.

Pedroni Residual Cointegration Test
Series: RGDP PRIV
Date: 08/14/10 Time: 14:56
Sample: 1970 2008
Included observations: 585
Cross-sections included: 15
Null Hypothesis: No cointegration
Trend assumption: No deterministic trend
Automatic lag length selection based on SIC with lags from 6 to 9
Newey-West automatic bandwidth selection and Bartlett kernel

Alternative hypothesis: common AR coefs. (within-dimension)
Weighted
Statistic Prob. Statistic Prob.
Panel v-Statistic -0.876052 0.8095 -0.825505 0.7955
Panel rho-Statistic 2.369772 0.9911 2.092733 0.9818
Panel PP-Statistic 2.439077 0.9926 1.824248 0.9659
Panel ADF-Statistic 1.918379 0.9725 1.019751 0.8461

Alternative hypothesis: individual AR coefs. (between-dimension)

Statistic Prob.
Group rho-Statistic 3.381017 0.9996
Group PP-Statistic 3.179659 0.9993
Group ADF-Statistic 2.351580 0.9907


Cross section specific results

Phillips-Peron results (non-parametric)

Cross ID AR(1) Variance HAC Bandwidth Obs
Chile 0.918 78679.77 166639.1 3.00 34
Egypt, Arab Rep. 0.996 3917.671 10075.07 4.00 38
India 0.932 396.7040 774.1924 4.00 38
Jamaica 0.926 29978.35 38937.69 1.00 38
Kenya 0.696 190.3928 327.9324 2.00 38
Malaysia 1.012 87754.92 103729.3 2.00 38
Mexico 0.991 52353.33 52353.33 0.00 38
Morocco 0.772 5480.730 6944.347 3.00 38
Pakistan 0.924 1179.070 1083.723 5.00 38
Philippines 1.042 1045.993 2511.623 4.00 38
South Africa 1.000 6613.645 12570.27 3.00 38
Sri Lanka 0.861 6017.810 8439.900 1.00 38
Thailand 1.053 20572.80 33931.88 2.00 38
Turkey 0.863 67528.11 67528.11 0.00 27
Venezuela, RB 0.872 69221.69 94114.19 2.00 38

Augmented Dickey-Fuller results (parametric)

Cross ID AR(1) Variance Lag Max lag Obs
Chile 0.892 49645.13 1 8 33
Egypt, 0.900 2302.554 1 9 37
India 0.932 396.7040 0 9 38
Jamaica 0.876 27212.47 1 9 37
Kenya 0.743 129.1915 2 9 36
Malaysia 1.012 87754.92 0 9 38
Mexico 0.959 45780.28 1 9 37
Morocco 0.667 4507.504 2 9 36
Pakistan 0.924 1179.070 0 9 38
Philippines0.962 706.3100 1 9 37
South Afr 0.918 5041.032 1 9 37
Sri Lanka 0.755 4684.824 1 9 37
Thailand 0.942 15218.71 1 9 37
Turkey 0.863 67528.11 0 6 27
Venezuela 0.847 65147.84 1 9 37
Pedroni Residual Cointegration Test
Series: PRIV RGDP
Date: 08/14/10 Time: 14:59
Sample: 1970 2008
Included observations: 585
Cross-sections included: 15
Null Hypothesis: No cointegration
Trend assumption: No deterministic trend
Automatic lag length selection based on SIC with lags from 6 to 9
Newey-West automatic bandwidth selection and Bartlett kernel

Alternative hypothesis: common AR coefs. (within-dimension)
Weighted
Statistic Prob. Statistic Prob.
Panel v-Statistic 0.727573 0.2334 1.670357 0.0474
Panel rho-Statistic 2.699482 0.9965 1.688395 0.9543
Panel PP-Statistic 3.116951 0.9991 1.300650 0.9033
Panel ADF-Statistic 2.193477 0.9859 -0.879577 0.1895

Alternative hypothesis: individual AR coefs. (between-dimension)

Statistic Prob.
Group rho-Statistic 2.615581 0.9955
Group PP-Statistic 1.900027 0.9713
Group ADF-Statistic -3.169011 0.0008


Cross section specific results

Phillips-Peron results (non-parametric)

Cross ID AR(1) Variance HAC Bandwidth Obs
Chile 0.843 0.002211 0.004617 3.00 34
Egypt, Arab Rep. 0.952 0.000686 0.001756 4.00 38
India 0.925 0.000155 0.000323 4.00 38
Jamaica 0.759 0.000704 0.000598 6.00 38
Kenya 0.881 0.000506 0.000506 0.00 38
Malaysia 0.929 0.010513 0.013454 2.00 38
Mexico 0.873 0.001087 0.001087 0.00 38
Morocco 0.782 0.002148 0.002757 2.00 38
Pakistan 0.764 0.000215 0.000289 3.00 38
Philippines 0.938 0.001363 0.003812 4.00 38
South Africa 1.034 0.005470 0.007374 1.00 38
Sri Lanka 0.764 0.000647 0.000983 2.00 38
Thailand 0.978 0.009086 0.016563 2.00 38
Turkey 0.842 0.000598 0.000598 0.00 27
Venezuela, RB 0.938 0.002846 0.004935 4.00 38

Augmented Dickey-Fuller results (parametric)

Cross ID AR(1) Variance Lag Max lag Obs
Chile 0.844 0.001367 1 8 33
Egypt, 0.758 0.000319 3 9 35
India 0.531 5.60E-05 8 9 30
Jamaica 0.571 0.000381 1 9 37
Kenya 0.881 0.000506 0 9 38
Malaysia 0.929 0.010513 0 9 38
Mexico 0.804 0.000903 1 9 37
Morocco 0.656 0.001774 2 9 36
Pakistan 0.621 0.000156 1 9 37
Philippine 0.848 0.000475 1 9 37
South Afr 1.006 0.004889 1 9 37
Sri Lanka 0.633 0.000460 1 9 37
Thailand 0.886 0.005739 1 9 37
Turkey 0.842 0.000598 0 6 27
Venezuela 0.938 0.002846 0 9 38

but when i dont use time trend or intecept and still PRIV as dependent variable
then:
Pedroni Residual Cointegration Test
Series: PRIV RGDP
Date: 08/14/10 Time: 15:01
Sample: 1970 2008
Included observations: 585
Cross-sections included: 15
Null Hypothesis: No cointegration
Trend assumption: No deterministic intercept or trend
Automatic lag length selection based on HQIC with lags from 6 to 9
Newey-West automatic bandwidth selection and Bartlett kernel

Alternative hypothesis: common AR coefs. (within-dimension)
Weighted
Statistic Prob. Statistic Prob.
Panel v-Statistic 3.980828 0.0000 3.195715 0.0007
Panel rho-Statistic 0.544709 0.7070 -0.100555 0.4600
Panel PP-Statistic 0.161579 0.5642 -1.103364 0.1349
Panel ADF-Statistic -1.152914 0.1245 -2.727807 0.0032

Alternative hypothesis: individual AR coefs. (between-dimension)

Statistic Prob.
Group rho-Statistic 2.201855 0.9862
Group PP-Statistic -0.499727 0.3086
Group ADF-Statistic -4.055857 0.0000


Cross section specific results

Phillips-Peron results (non-parametric)

Cross ID AR(1) Variance HAC Bandwidth Obs
Chile 0.847 0.002256 0.004733 3.00 34
Egypt, 0.943 0.000614 0.001619 4.00 38
India 0.950 0.000158 0.000301 4.00 38
Jamaica 0.876 0.000743 0.001089 3.00 38
Kenya 0.881 0.000406 0.000406 0.00 38
Malaysia 0.937 0.010795 0.013731 2.00 38
Mexico 0.868 0.001030 0.001446 2.00 38
Morocco 0.934 0.001803 0.002604 2.00 38
Pakistan 0.921 0.000253 0.000205 6.00 38
Philippine 0.949 0.001367 0.003843 4.00 38
South Afr 1.014 0.005809 0.007780 1.00 38
Sri Lanka 0.808 0.000680 0.000960 1.00 38
Thailand 0.986 0.009314 0.016786 2.00 38
Turkey 0.818 0.000581 0.000581 0.00 27
Venezuela, RB 0.949 0.002524 0.003952 3.00 38

Augmented Dickey-Fuller results (parametric)

Cross ID AR(1) Variance Lag Max lag Obs
Chile 0.846 0.001395 1 8 33
Egypt, 0.826 0.000294 3 9 35
India 0.774 8.03E-05 8 9 30
Jamaica 0.813 0.000568 1 9 37
Kenya 0.881 0.000406 0 9 38
Malaysia 0.821 0.008937 3 9 35
Mexico 0.832 0.000871 1 9 37
Morocco 0.855 0.001599 2 9 36
Pakistan 0.894 0.000220 1 9 37
Philippines0.799 0.000424 3 9 35
South Afr 0.992 0.005230 1 9 37
Sri Lanka 0.691 0.000507 1 9 37
Thailand 0.888 0.005994 1 9 37
Turkey 0.818 0.000581 0 6 27
Venezuela 0.949 0.002524 0 9 38

With RGDP as dependent variable and no intecept and time trend i get:

Pedroni Residual Cointegration Test
Series: RGDP PRIV
Date: 08/14/10 Time: 15:05
Sample: 1970 2008
Included observations: 585
Cross-sections included: 15
Null Hypothesis: No cointegration
Trend assumption: No deterministic intercept or trend
Automatic lag length selection based on HQIC with lags from 6 to 9
Newey-West automatic bandwidth selection and Bartlett kernel

Alternative hypothesis: common AR coefs. (within-dimension)
Weighted
Statistic Prob. Statistic Prob.
Panel v-Statistic -2.053214 0.9800 -1.816872 0.9654
Panel rho-Statistic -0.015270 0.4939 0.032182 0.5128
Panel PP-Statistic -1.269953 0.1021 -0.960535 0.1684
Panel ADF-Statistic -1.579260 0.0571 -1.821344 0.0343

Alternative hypothesis: individual AR coefs. (between-dimension)

Statistic Prob.
Group rho-Statistic 2.394793 0.9917
Group PP-Statistic -0.183500 0.4272
Group ADF-Statistic -3.369877 0.0004


Cross section specific results

Phillips-Peron results (non-parametric)

Cross ID AR(1) Variance HAC Bandwidth Obs
Chile 0.846 108744.9 226569.9 3.00 34
Egypt, 0.953 7765.020 20356.60 4.00 38
India 0.949 342.8971 654.3158 4.00 38
Jamaica 0.880 142478.2 218935.7 2.00 38
Kenya 0.886 937.8972 937.8972 0.00 38
Malaysia 0.962 124905.3 155918.6 2.00 38
Mexico 0.905 559757.7 882673.7 3.00 38
Morocco 0.916 18286.32 25990.92 2.00 38
Pakistan 0.938 962.7561 923.3603 5.00 38
Philippine 0.968 10011.43 28116.14 4.00 38
South Afr 0.988 62419.45 90248.81 2.00 38
Sri Lanka 0.825 7079.066 9979.353 1.00 38
Thailand 1.021 31495.69 55130.16 2.00 38
Turkey 0.813 277120.0 277120.0 0.00 27
Venezuela 0.957 508974.4 794447.4 3.00 38

Augmented Dickey-Fuller results (parametric)

Cross ID AR(1) Variance Lag Max lag Obs
Chile 0.846 68009.43 1 8 33
Egypt, 0.904 4288.202 1 9 37
India 0.760 166.1040 8 9 30
Jamaica 0.820 110093.3 1 9 37
Kenya 0.886 937.8972 0 9 38
Malaysia 0.827 104550.9 3 9 35
Mexico 0.874 482890.1 1 9 37
Morocco 0.852 16286.64 2 9 36
Pakistan 0.909 843.7002 1 9 37
Philippine 0.809 3251.103 3 9 35
South Afr 0.976 55700.92 1 9 37
Sri Lanka 0.704 5336.032 1 9 37
Thailand 0.903 21240.35 1 9 37
Turkey 0.813 277120.0 0 6 27
Venezuela 0.957 508974.4 0 9 38

As for the Kao and Fisher test am clueless as to how to begin to interprete.
Heres an example of each:
Kao with intercept but no trend
Kao Residual Cointegration Test
Series: RGDP PRIV
Date: 08/14/10 Time: 15:10
Sample: 1970 2008
Included observations: 585
Null Hypothesis: No cointegration
Trend assumption: No deterministic trend
Automatic lag length selection based on SIC with a max lag of 6
Newey-West automatic bandwidth selection and Bartlett kernel

t-Statistic Prob.
ADF -0.077956 0.4689

Residual variance 16177.33
HAC variance 29794.83




Augmented Dickey-Fuller Test Equation
Dependent Variable: D(RESID)
Method: Least Squares
Date: 08/14/10 Time: 15:10
Sample (adjusted): 1974 2008
Included observations: 510 after adjustments

Variable Coefficient Std. Error t-Statistic Prob.

RESID(-1) -0.049469 0.014230 -3.476349 0.0006
D(RESID(-1)) 0.348364 0.043823 7.949299 0.0000
D(RESID(-2)) 0.001672 0.045357 0.036866 0.9706
D(RESID(-3)) 0.160472 0.044304 3.622028 0.0003

R-squared 0.117530 Mean dependent var 23.37165
Adjusted R-squared 0.112298 S.D. dependent var 146.8915
S.E. of regression 138.3982 Akaike info criterion 12.70596
Sum squared resid 9691952. Schwarz criterion 12.73917
Log likelihood -3236.020 Hannan-Quinn criter. 12.71898
Durbin-Watson stat 2.053277

And the Fisher with intercept and no trend

Johansen Fisher Panel Cointegration Test
Series: RGDP PRIV
Date: 08/14/10 Time: 15:11
Sample: 1970 2008
Included observations: 585
Trend assumption: Linear deterministic trend
Lags interval (in first differences): 1 1

Unrestricted Cointegration Rank Test (Trace and Maximum Eigenvalue)

Hypothesized Fisher Stat.* Fisher Stat.*
No. of CE(s) (from trace test) Prob. (from max-eigen test) Prob.

None 68.54 0.0001 59.64 0.0010
At most 1 47.75 0.0210 47.75 0.0210

* Probabilities are computed using asymptotic Chi-square distribution.

Individual cross section results

Trace Test Max-Eign Test
Cross Section Statistics Prob.** Statistics Prob.**

Hypothesis of no cointegration
Chile 16.7708 0.0320 15.9644 0.0267
Egypt, 7.0738 0.5691 6.3969 0.5628
India 17.2305 0.0271 14.5855 0.0445
Jamaica 18.8444 0.0150 17.7350 0.0136
Kenya 11.5201 0.1814 9.2428 0.2665
Malaysia 5.0655 0.8018 4.3451 0.8213
Mexico 9.7849 0.2976 9.3932 0.2549
Morocco 14.0476 0.0816 10.9337 0.1575
Pakistan 14.7362 0.0648 14.7355 0.0421
Philippine 14.3750 0.0732 14.0966 0.0531
South Af 4.1265 0.8931 3.8132 0.8785
Sri Lanka 27.5463 0.0005 16.4972 0.0218
Thailand 12.2134 0.1470 11.9872 0.1111
Turkey 4.0337 0.9008 2.6421 0.9677
Venezuela 11.5518 0.1797 8.7814 0.3048
Hypothesis of at most 1 cointegration relationship
Chile 0.8064 0.3692 0.8064 0.3692
Egypt, 0.6769 0.4106 0.6769 0.4106
India 2.6450 0.1039 2.6450 0.1039
Jamaica 1.1094 0.2922 1.1094 0.2922
Kenya 2.2772 0.1313 2.2772 0.1313
Malaysia 0.7203 0.3960 0.7203 0.3960
Mexico 0.3917 0.5314 0.3917 0.5314
Morocco 3.1139 0.0776 3.1139 0.0776
Pakistan 0.0007 0.9802 0.0007 0.9802
Philippines 0.2784 0.5977 0.2784 0.5977
South Afr 0.3133 0.5756 0.3133 0.5756
Sri Lanka 11.0491 0.0009 11.0491 0.0009
Thailand 0.2262 0.6343 0.2262 0.6343
Turkey 1.3916 0.2381 1.3916 0.2381
Venezuela 2.7704 0.0960 2.7704 0.0960

**MacKinnon-Haug-Michelis (1999) p-values
I am so sorry for the very lenghty post I could not make an attachment. I eagerly await you reply!
Thanks