Panel Data reading

[Pattigi]

Hi,

I downloaded a rather large Excel file from COMPUSTAT, with panel data from a large number of companies.

I succeeded loading these data in RATSDATA (Trial version 8.3), however I have a few problems when reading data :

Problem 1 : The panel is unbalanced, which means the individuals have a different amount of data. Also, the column with the calendar date is not recognized. I am able to create a macro in Excel where I could insert blank lines to even the number of observations thus creating a balanced panel with a lot of missing values. Would that impact the results of the upcoming regressions?

Problem 2: I did a sample test with data from only two individuals, and I tried to load the panel data in a RATS file, however when I did a simple regression I couldn't load data from the second individual . The message was that there was an "illegal date"

I guess my questions here are basically how to construct a good panel in RATS.

Really appreciate any help

[TomDoan]

It's probably too big to post (or it may be contrary to Compustat rules). However, can you show roughly how it's formatted?

Although RATSDATA has some tools for re-organizing panel data, most of the instructions for that are in RATS---PFORM in particular.

[Pattigi]

I attached an Excel File of what the final file should look like (I put only part of the data). As I said in my first post, the first column is the company key (individual).

The frequency of data is quarterly, from 1991 Q1 to 2013 Q1. I specified the year and the quarter in separate columns, as well as the calendar date.

I have the ability to balance the panel with a macro that could insert blank lines so that every individual would have the same number of lines (with blanks at the appropriate quarters) if needed.

My objective would be to form a panel so I could regress INVESTMENT over SIZE, Q and CASHFLOW. I would also add another independent variable from another file (quarterly data as well).

Thanks for the help