Hi sir,

I have 3 first differenced series. They do not contain unit root.

they are I(0).

can I use them to run an ECM for panel data and say this is a short-run analysis?

Note: the series at level were I(0) for two of them and one is I(1). so no way dor cointegration here.

the above question is different, as we are using the 1rst Differences as "new" series.

Thanks in advance

I have 3 first differenced series. They do not contain unit root.

they are I(0).

can I use them to run an ECM for panel data and say this is a short-run analysis?

Note: the series at level were I(0) for two of them and one is I(1). so no way dor cointegration here.

the above question is different, as we are using the 1rst Differences as "new" series.

Thanks in advance

- danon
**Posts:**17**Joined:**Thu Nov 10, 2016 3:05 pm

You can, but what does it mean? If x and y are stationary, then you can write dx=linear in lagged dx's, lagged dy's,(x{1}-b*y{1}) for any value of b and have a well-behaved estimation.

- TomDoan
**Posts:**7147**Joined:**Wed Nov 01, 2006 5:36 pm

Indeed i was trying to establish the determinants of stock price relative to long-term interest rates and industrial production index.

the panel contain 20 countries belonging to different continents.

In level, the log of stock prices index is I(0) , the log of industrial production is I(0) and long-term interest rates is I(1). Therefore I could not run cointegration test on them.

But when a use the Retuns on stck prices (ret = price at time n - prices at time n-1) and change in industrial production and changes in long erm intereste rates, THERE ARE ALL I(0)

Therefore i said to myself , ok, I can run and ECM as the series of the panel are all integrated at the same order....

Question: Am i right to do that and how to interpret ?

Sorry to disturb you again.

the panel contain 20 countries belonging to different continents.

In level, the log of stock prices index is I(0) , the log of industrial production is I(0) and long-term interest rates is I(1). Therefore I could not run cointegration test on them.

But when a use the Retuns on stck prices (ret = price at time n - prices at time n-1) and change in industrial production and changes in long erm intereste rates, THERE ARE ALL I(0)

Therefore i said to myself , ok, I can run and ECM as the series of the panel are all integrated at the same order....

Question: Am i right to do that and how to interpret ?

Sorry to disturb you again.

- danon
**Posts:**17**Joined:**Thu Nov 10, 2016 3:05 pm

In level, the log of stock prices index is I(0) , the log of industrial production is I(0)

That doesn't look right at all. I would seriously doubt that stock price, IP and interest rates are cointegrated, but it's hard to imagine that you're getting log of the price index as I(0). (IP could be trend-stationary, though that doesn't seem likely either).

- TomDoan
**Posts:**7147**Joined:**Wed Nov 01, 2006 5:36 pm

Tom,

I have a an heterogeneous panel model such as;

lnex = A+Byear+ Clnfdi

But some of my data (FDI) contains negative values. So, what is the best form to handle or deal with (transform) the negative values, given that they are not in single form?

I hope to find the long-run relationship.

Please i need your responses.

Thank you.

I have a an heterogeneous panel model such as;

lnex = A+Byear+ Clnfdi

But some of my data (FDI) contains negative values. So, what is the best form to handle or deal with (transform) the negative values, given that they are not in single form?

I hope to find the long-run relationship.

Please i need your responses.

Thank you.

- Osabuohien247
**Posts:**10**Joined:**Sun Sep 14, 2014 4:20 am

What is FDI that it would have negative values?

- TomDoan
**Posts:**7147**Joined:**Wed Nov 01, 2006 5:36 pm

TomDoan wrote:What is FDI that it would have negative values?

Thank you for your kind response.

Its FDI inward and outward flow (US dollars at current prices in millions)

- Osabuohien247
**Posts:**10**Joined:**Sun Sep 14, 2014 4:20 am

Are you saying that your variable is the net flow? (inward minus outward or vice versa)? Obviously you can't do logs of a series that could be zero or negative. If your model has the net as the explanatory variable, you may need to think of a different way to express the relationship---deflating by GDP is sometimes helpful in similar situations, though that may change the interpretation of the coefficient which wouldn't be an elasticity any longer.

- TomDoan
**Posts:**7147**Joined:**Wed Nov 01, 2006 5:36 pm

TomDoan wrote:Are you saying that your variable is the net flow? (inward minus outward or vice versa)? Obviously you can't do logs of a series that could be zero or negative. If your model has the net as the explanatory variable, you may need to think of a different way to express the relationship---deflating by GDP is sometimes helpful in similar situations, though that may change the interpretation of the coefficient which wouldn't be an elasticity any longer.

Thank you Tom, your explanation is quiet clear.

I understand clearly.

Please is there any way i can get "abdata.dta" data used for the replication of Arellano and Bond(1991) paper "Some Tests of

Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations"?

Thanks for your anticipated assistance

- Osabuohien247
**Posts:**10**Joined:**Sun Sep 14, 2014 4:20 am

Isn't that in the paper replication folder for the A-B paper in the RATS distribution?

- TomDoan
**Posts:**7147**Joined:**Wed Nov 01, 2006 5:36 pm

10 posts
• Page **1** of **1**

Users browsing this forum: No registered users and 1 guest