### PPURoot (Perron 1997 unit root test)

Posted:

**Mon May 07, 2012 4:20 pm**This thread is for the PPURoot add-in that performs the Perron 1997 Unit Root test for data with structural breaks.

Page **1** of **2**

Posted: **Mon May 07, 2012 4:20 pm**

This thread is for the PPURoot add-in that performs the Perron 1997 Unit Root test for data with structural breaks.

Posted: **Tue Aug 28, 2012 2:57 am**

Hello,

I have a question about the Perron 1997 unit root test. How do you choose the perfect lag for it? I am using 12 because I have monthly information.

I am working with 177 monthly observations and 5 variables using trend and constant. I know it has structural breaks and I am trying to identify them. However, I am not sure how to identify them.

Could you please help me?

Blanca

bvguizar

Posts: 7

Joined: Sat Aug 28, 2010 2:56 pm

I have a question about the Perron 1997 unit root test. How do you choose the perfect lag for it? I am using 12 because I have monthly information.

I am working with 177 monthly observations and 5 variables using trend and constant. I know it has structural breaks and I am trying to identify them. However, I am not sure how to identify them.

Could you please help me?

Blanca

bvguizar

Posts: 7

Joined: Sat Aug 28, 2010 2:56 pm

Posted: **Sun Sep 09, 2012 4:14 pm**

Hello Blanca

The value of the lag that you give the program at first, is the maximum lag, and usually the maximum lag is given by the frequency at which data are recorded, 12 for annual data, 4 for quarterly data, the program only needs maximum lag (4 or 12) and internally determines the optimal lag.

As for your second question, the program generates the possible breakpoint and this where it says "Chosen break point: xxxxx. Then you must be careful to ensure that the break date generating program makes economic sense, which you determine through the knowledge that you have about the economic phenomenon that describe your data. In a way, the test determines the possible break date, is most striking economic phenomenon, during the historical horizon under study.

Hope this helps ...

Prof. Ruben Ibarra

The value of the lag that you give the program at first, is the maximum lag, and usually the maximum lag is given by the frequency at which data are recorded, 12 for annual data, 4 for quarterly data, the program only needs maximum lag (4 or 12) and internally determines the optimal lag.

As for your second question, the program generates the possible breakpoint and this where it says "Chosen break point: xxxxx. Then you must be careful to ensure that the break date generating program makes economic sense, which you determine through the knowledge that you have about the economic phenomenon that describe your data. In a way, the test determines the possible break date, is most striking economic phenomenon, during the historical horizon under study.

Hope this helps ...

Prof. Ruben Ibarra

Posted: **Mon Sep 10, 2012 2:02 pm**

Thank you Professor.

I read a paper about unit root tests

Glynn, J, Perera, N and Verma, R, Unit Root Tests and Structural Breaks: A Survey with Applications, Revista de Métodos Cuantitativos para la Economía y la Empresa = Journal of Quantitative Methods for Economics and Business Administration, 3(1), 2007, 63-79.

And I found that this test and the Zivot and Andrews are used for finding breaks. I ran the Z-A test and I used a dummy but still get insignificant results. Do you know which one has more power or it might help me more?

Thank you Professor.

Blanca

I read a paper about unit root tests

Glynn, J, Perera, N and Verma, R, Unit Root Tests and Structural Breaks: A Survey with Applications, Revista de Métodos Cuantitativos para la Economía y la Empresa = Journal of Quantitative Methods for Economics and Business Administration, 3(1), 2007, 63-79.

And I found that this test and the Zivot and Andrews are used for finding breaks. I ran the Z-A test and I used a dummy but still get insignificant results. Do you know which one has more power or it might help me more?

Thank you Professor.

Blanca

Posted: **Wed Sep 12, 2012 9:13 pm**

Hello Blanca

I think the Perron-test is much more complete than the ZA-test, as it allows both under the null and the alternative hypothesis, the series Xt is subject to a break, while the ZA-test, under the null hypothesis, only allows a Nonstationary series Xt, without structural breaks. I recommend you read the paper: Perron, Further Evidence on Breaking Trend Functions in Macroeconomic Variables", Journal of Econometrics, 80, 1997, which will allow you to understand the program.

Regards

Prof. Rubén Ibarra

I think the Perron-test is much more complete than the ZA-test, as it allows both under the null and the alternative hypothesis, the series Xt is subject to a break, while the ZA-test, under the null hypothesis, only allows a Nonstationary series Xt, without structural breaks. I recommend you read the paper: Perron, Further Evidence on Breaking Trend Functions in Macroeconomic Variables", Journal of Econometrics, 80, 1997, which will allow you to understand the program.

Regards

Prof. Rubén Ibarra

Posted: **Thu Sep 13, 2012 12:37 pm**

Dear Professor you mean to use the adds in of the eviews:

PPURoot*

This add-in, written by Prof. Ruben Ibarra, performs the Perron (1997) unit root test with a break in the trend function at an unknown time.

Is this the same test known as Bai and Perron?

Thank you so much professor

PPURoot*

This add-in, written by Prof. Ruben Ibarra, performs the Perron (1997) unit root test with a break in the trend function at an unknown time.

Is this the same test known as Bai and Perron?

Thank you so much professor

Posted: **Fri Sep 14, 2012 6:40 pm**

Hi

Blanca not the same. The test PPUroot was proposed by Professor P. Perron to overcome some weaknesses of Zivot-Andrews test.

Regards

Blanca not the same. The test PPUroot was proposed by Professor P. Perron to overcome some weaknesses of Zivot-Andrews test.

Regards

Posted: **Mon Sep 17, 2012 5:25 am**

Hi Dr,

Do you know if the Lumsdain and Papell (1997) test can be run in eviews? Or if this one presents better features agains the Perron (1997) test. Thank you

Do you know if the Lumsdain and Papell (1997) test can be run in eviews? Or if this one presents better features agains the Perron (1997) test. Thank you

Posted: **Wed Sep 19, 2012 11:57 am**

Hi Blanca

Sorry, but I do not know about this test...

Sorry, but I do not know about this test...

Posted: **Sat Sep 29, 2012 8:00 pm**

-------------------

Posted: **Sat Oct 13, 2012 4:53 pm**

Hello Allen

1, - The non-rejection of the null hypothesis means that the series is non-stationary with structural break in 6/25/2007.

2. - Indeed must pass after applying the test on the first difference, but now the data-generating process, should be only break in the intercept. This is because to differentiate a step dummy (permanent effect); you get an impulse dummy (time effect).

3. - The maximum lag you choose, big enough (depending of the size and frequency of the data) to minimize the possibility of correlated residues.

Regards

Prof. RI

1, - The non-rejection of the null hypothesis means that the series is non-stationary with structural break in 6/25/2007.

2. - Indeed must pass after applying the test on the first difference, but now the data-generating process, should be only break in the intercept. This is because to differentiate a step dummy (permanent effect); you get an impulse dummy (time effect).

3. - The maximum lag you choose, big enough (depending of the size and frequency of the data) to minimize the possibility of correlated residues.

Regards

Prof. RI

Posted: **Tue Nov 06, 2012 11:50 pm**

wonder that the addin is test on 'log(price)' or 'price' series? thx

Posted: **Thu Mar 28, 2013 1:32 am**

Hi,

I woould like to know if the Perron (1997) structutal break unit root test implements the Additive Outlier Model or the Innovative Outlier Model ? Since in the options there is no option given regarding choice of either of the two models, I wish know which of the model has been implemented under the PPUroot Test

I woould like to know if the Perron (1997) structutal break unit root test implements the Additive Outlier Model or the Innovative Outlier Model ? Since in the options there is no option given regarding choice of either of the two models, I wish know which of the model has been implemented under the PPUroot Test

Posted: **Fri Apr 12, 2013 9:41 am**

Hello Sidtech

In the attachment, you will find the information you're asking

Regards

Prof. RI

In the attachment, you will find the information you're asking

Regards

Prof. RI

Posted: **Mon Oct 14, 2013 3:29 pm**

When using ppuroot, am I the only one who always is getting the same critical values regardless of the sample size (-5.23 för 5% in the Perron unit root test output)?

Best regards

Best regards