Bai and Carrion-i-Silvestre (2009) Panel unit root tests

Hello, I want to perform a panel data analysis. I want to obtain the results shown below as an example with Bai and Carrion-i-Silvestre (2009) - Panel unit root test. But even with these codes I use, I can't get this result. Which library and code should I use?

new;
cls;

library carrionlib;

// Load data
test_data = loadd(__FILE_DIR $+ "brics.xlsx", "lco2");

// Time periods
bigt = 29;

ncross = rows(test_data)/bigT;

// Create wide panel data
lco2_wide = reshape(test_data, ncross, bigT)';

// Declare control structurea
struct breakControl bCtl;

// Number of breaks
m = 3;

// Model
model = 4|m|1|1|2;

// Set the number of factors
k = 2;

// Number of maximum factors to allow
// and estimation method
kmax = 3|1;

// AR degress
p_ar = 0;

// Datevec
datevec = 0;

{ Z_test, test_n, test_chi, Z_test_sim, test_n_sim, test_chi_sim, fhat } = panelbreak(lco2_wide, model, p_ar, kmax, datevec);

print "Z test: " Z_test;
print "Pval (normal): " test_n;
print "Pval (Chi-square): " test_chi;

print;

print "Simplified tests";
print "Z test: " Z_test_sim;
print "Pval (normal): " test_n_sim;
print "Pval (Chi-square): " test_chi_sim;</blockquote>

1 Answer

0

accepted

Thank you for your question. I was able to locate the publication where the sample results you provided are located.

Are you hoping to replicate this table? The examples in the carrionlib do not correspond to this paper and will not replicate this table. You would need to contact the authors directly for more information about the data being used.

More generally, the panelbreak procedure returns six different test results:

Variable	Description
z_test	Test statistic based on the moments.
test_n	Test statistic based on p-values using the N(0,1) distribution.
test_chi	Test statistic based on p-values using the Chi-squared(2N) distribution.
z_test_sim	Simplified test statistic based on the moments.
test_n_sim	Simplified test statistic based on p-values using the N(0,1) distribution.
test_chi_sim	Simplified test statistic based on p-values using the Chi-squared(2N) distribution.
fhat	Matrix with the estimated common factors.

The sample code you provided will print each of these statistics to the GAUSS Command Window. Alternatively, you can open these from the Symbol Window on the GAUSS Data Editor tab.

link

Eric

105