Which analysis pre/post analysis

[Erik Nevland]

Dear experts,

For my thesis I have to do an analysis in Stata but unfortunately do I have problems with the statistics/analysis part. I have a panel data set of the year 2007-2016 and I want to do a pre-post analysis. In the end of 2010 there has been a change in a certain framework and I want to measure if the companies in the years 2007-2010 are doing more risk reporting im comparison with the years 2011-2016. The dataset contains different company(id) which is unbalanced, each year I have the total words of risk reporting , so some companies there are missing one or two years. My question is which method do I need to use. A regression with dummy variables which the pre period = 0 and the post period is 1? Or something else?

Thank you for your help.

[Clyde Schechter]

My question is which method do I need to use. A regression with dummy variables which the pre period = 0 and the post period is 1?

Well, that would be the starting point for your analysis. But you will also need to incorporate other variables in the analysis. After all, lots of things happened in the world in the end of 2010, so you can't just attribute everything that changed between the pre- and post- periods to the change in framework. Ideally you would have similar data for a group of companies that were not subjected to the change in framework. If you do not (and perhaps the change was universal so such data does not exist anywhere), then you at least need to include other variables that might account for risk reporting and adjust for their effects. It's not entirely clear to me what discipline you are working in (finance? economics? organizational behavior?) but, in any case, it is out of my area of expertise. So I am not in a position to advise you more specifically what those other variables might be, nor how you might obtain suitable data for them.

[Erik Nevland]

Yes I got some other variables. Thanks for info!

[Carlo Lazzaro]

Erik:
what is your supervisor's take on the whole matter?