A question about Carrion-i-Silvestre, et al. (2005) pankpss test results

Hi everyone,
I am trying interpret Carrion-i-Silvestre, et al. (2005) pankpss test results using gauss 21 carrionlib, pdlib libraries.

I have two issues:

1) What is the bootstrap parameter set to? How can i see that?
2) Where the PANKPSS test critical values are reported? How can i see them?

Kindest regards.

Here are the test results reported by Gauss for my dataset:

K_BIC 5.0000000
K_BIC 1.0000000
K_BIC 1.0000000
K_BIC 1.0000000
K_BIC 1.0000000
K_BIC 1.0000000
K_BIC 1.0000000
K_BIC 1.0000000
K_BIC 1.0000000
K_BIC 1.0000000
************************************
Results for the PANKPSS test
************************************
Stationarity test with structural breaks (homogeneous): 1.0743847 with p-val: 0.14132515
Stationarity test with structural breaks (heterogeneous): 2.3701649 with p-val: 0.0088900766
Maximum number of breaks allowed 5.0000000
Matrix of individual tests
0.020552716 0.0000000 0.0000000 0.0000000
0.11726387 0.0000000 0.0000000 0.0000000
0.034238388 1.0000000 1.0000000 1.0000000
0.041515483 0.0000000 0.0000000 0.0000000
0.12090409 0.0000000 1.0000000 0.0000000
0.10958435 0.0000000 0.0000000 0.0000000
0.15801470 0.0000000 1.0000000 0.0000000
0.075400641 0.0000000 0.0000000 0.0000000
0.061361775 0.0000000 0.0000000 0.0000000
0.19587219 0.0000000 1.0000000 0.0000000
Number of observations 205.00000
Estimated breaking points
54.000000 42.000000 42.000000 42.000000 42.000000
0.0000000 89.000000 89.000000 89.000000 76.000000
0.0000000 0.0000000 122.00000 126.00000 114.00000
0.0000000 0.0000000 0.0000000 160.00000 144.00000
0.0000000 0.0000000 0.0000000 0.0000000 175.00000
43.000000 43.000000 43.000000 43.000000 43.000000
0.0000000 171.00000 83.000000 83.000000 76.000000
0.0000000 0.0000000 171.00000 118.00000 109.00000
0.0000000 0.0000000 0.0000000 171.00000 140.00000
0.0000000 0.0000000 0.0000000 0.0000000 171.00000
80.000000 81.000000 81.000000 33.000000 33.000000
0.0000000 117.00000 111.00000 81.000000 81.000000
0.0000000 0.0000000 161.00000 111.00000 111.00000
0.0000000 0.0000000 0.0000000 161.00000 141.00000
0.0000000 0.0000000 0.0000000 0.0000000 171.00000
37.000000 80.000000 37.000000 37.000000 37.000000
0.0000000 112.00000 80.000000 86.000000 77.000000
0.0000000 0.0000000 112.00000 138.00000 107.00000
0.0000000 0.0000000 0.0000000 175.00000 138.00000
0.0000000 0.0000000 0.0000000 0.0000000 175.00000
78.000000 40.000000 40.000000 40.000000 40.000000
0.0000000 76.000000 78.000000 78.000000 78.000000
0.0000000 0.0000000 113.00000 116.00000 113.00000
0.0000000 0.0000000 0.0000000 161.00000 143.00000
0.0000000 0.0000000 0.0000000 0.0000000 174.00000
118.00000 83.000000 83.000000 39.000000 39.000000
0.0000000 141.00000 141.00000 83.000000 81.000000
0.0000000 0.0000000 174.00000 141.00000 111.00000
0.0000000 0.0000000 0.0000000 174.00000 141.00000
0.0000000 0.0000000 0.0000000 0.0000000 174.00000
175.00000 32.000000 32.000000 32.000000 32.000000
0.0000000 175.00000 128.00000 113.00000 79.000000
0.0000000 0.0000000 175.00000 144.00000 113.00000
0.0000000 0.0000000 0.0000000 175.00000 144.00000
0.0000000 0.0000000 0.0000000 0.0000000 175.00000
158.00000 48.000000 48.000000 48.000000 48.000000
0.0000000 87.000000 87.000000 87.000000 79.000000
0.0000000 0.0000000 158.00000 128.00000 113.00000
0.0000000 0.0000000 0.0000000 158.00000 145.00000
0.0000000 0.0000000 0.0000000 0.0000000 175.00000
81.000000 37.000000 37.000000 37.000000 37.000000
0.0000000 81.000000 81.000000 81.000000 70.000000
0.0000000 0.0000000 130.00000 126.00000 103.00000
0.0000000 0.0000000 0.0000000 158.00000 133.00000
0.0000000 0.0000000 0.0000000 0.0000000 165.00000
69.000000 39.000000 39.000000 61.000000 30.000000
0.0000000 69.000000 69.000000 91.000000 61.000000
0.0000000 0.0000000 166.00000 132.00000 91.000000
0.0000000 0.0000000 0.0000000 166.00000 132.00000
0.0000000 0.0000000 0.0000000 0.0000000 166.00000

2 Answers

0

Hello,

Thank you for your question. In regards to the critical values, the GAUSS pankpss test provides the p-values for the test statistic but does not provide critical values.

However, the test statistic is distributed standard normal. You could calculate these critical values using the standard normal distribution. This results in critical values of 1.281 for 10%, 1.644 for 5%, and 2.323 for 1%.

I am not sure what Bootstrap parameter you are referring to. If you mean the kernel used for estimating the LR-variance, this is controlled with the kernel input.

You have three options for this input:

kernel[1] = 0 for Sul, Phillips and Choi (2003) with the Bartlett kernel
kernel[1] = 1 for Sul, Phillips and Choi (2003) with quadratic spectral kernel
kernel[1] = 2 for Kurozumi (2002) proposal with Bartlett kernel;

link

Eric

105

0

Thank you so much for your answering. I actually meant the critical values referenced for individual tests. I think the program gives the critical values separately for each country, as reported in the article below (Table 2, Table 3). I'm trying to achieve this sort of results. Bootstrap also points to the number of cycles used to obtain these critical values. Could you please guide me in this direction. Thank you.

Bulent Guloglu, Serdar Ispir & Deniz Okat (2011): Testing the validity of quasi PPP hypothesis: evidence
from a recent panel unit root test with structural breaks, Applied Economics Letters, DOI:10.1080/13504851.2011.564124 .

Quicj link for the article:

https://www.dropbox.com/s/7e491n0k1db62zn/Guloglu%20etal%282011%29-PPP%20multiple%20breaks.pdf?dl=0

link

Ahmet

0