Regression with time independent data

[bram tanke]

Hello,

I need to do a regression with the following variables:

Dependent variable: Number of M&A announcements per month

Independent variable: Appearance of a financial crises (dummy ==1 if month is a crisis month)

Control variables: Firm size (number) and Cross-country (dummy ==1 if M&A is a cross-country M&A)


As you can see I have a problem here. The dependent and independent variable are measured per month, the control variables are measured at a firm-level. I cannot exclude the control variables because of the omitted variable bias.

Do I need to take an average of all firm sizes within a month and use that in the regression? And also an average of the dummy cross-country? I don't know if this is econometrically right but don't see another possibility.

Thanks in advance !

Bram

[Carlo Lazzaro]

Bram:
welcome to this forum.
Iìm not clear with what you're after.
Are you dealing with a panel dataset or else?

[bram tanke]

Hi Carlo,

It is a panel dataset.

This is the regression:

Beta 1 is the independent variable (dummy)
Beta 2 and 3 are the control variables (cross border = dummy, firm size is a number)
Beta 4 and 5 are fixed effects

This is how the data look in Stata:


My supervisor told me the following; "the variables are not measured at the same frequency (look at the firm-level variables)“.

I don't know how I can fix this? Do you have any idea?


Kind regards,
Bram

[Carlo Lazzaro]

Bram:
going OLS with -i.panelid- (and -timevar-) with clustered standard errors, is simply another way to go -xtreg,fe- 8f you have a sort panel dataset.
BTW: are you dealing with a short or long panel dataset?

[bram tanke]

Carlo:

Short panel set. Some companies have multiple M&As in the dataset, some just have one.

What is the best way to measure the control variables? Do I have to aggregate them monthly because the dependent and independent variable are also measured monthly? Is this also appropriate to do with the dummy control variables?

[Carlo Lazzaro]

Bram:
thanks for clarifying.
If Stata return -repeated time values within panel- error, you cam simply set your dataset as follows:

xtset panelvar

and then go -xtreg,fe-.

[bram tanke]

Hi Carlo;

Thanks!
And statistical/econometrically; is it right to aggregate the control variables (firm size / cross border) per month?

[Carlo Lazzaro]

Bram:
could you please give an example (via -dataex-) of your variables, especially the ones that you want to aggregate? Thanks.