Error Correction Model Using Panel Data

[lnp3]

Dear list,

I am trying to program an ERROR CORRECTION MODEL estimation using a RANDOM EFFECTS & FIXED EFFECTS PANEL DATA framework. I was wondering if someone might have offered some advise that I could use for my project. I am doing a study on bank profitability using a panel of 62 banks over 17 year period (1994-2011).

I have noticed that all error corrections models using panel data first do the 1) panel unit root tests, 2) Panel co-integration tests (Pedroni/Engle-Granger based for variables 7 or less), 3) final estimation, which includes co-integrating regression (or error correction model, in other words).

1) Since there is no built in framework for panel co-integrating regression in E-views, how can I calculate or write the syntax for the **error correction term**? (EC). In other words, how can I add (EC) as one of my explanatory variables to the regression equation?

2) Is it OK to do EC model estimation without doing co-integration? Is co-integration necessary component of EC modeling?

I really appreciate your response to these questions.

Best Regards,
Mine Doyran
:o