Comparing difference of two independent variables in a panel regression

#1
Hi Guys,

currently writing on my thesis on the effect of executives' optimism on cash holdings. I am measuring the degree of optimism through insider trades and would like to differenciate between male and female optimism. Furthermore, i would like to test empircally the hypothesis that male's optimism has a more negative impact on the cash balance than female's optimism.
And that's the problem: Let me give you an example:


Obs 1:
Cash balance: 200
Tobin's Q: 5.2
Dividend payout: 66
Male optimism measure: 0.88
Female optimism measure: -0.24

Obs 1:
Cash balance: 117
Tobin's Q: 9.1
Dividend payout: 89
Male optimism measure: 0.10
Female optimism measure: 0.53



Now, i would like to test my hypothesis that the coefficient of female optimism has a smaller negative influence on the cash balance than male optimism. Also, i would like to test its significance.

Creating an interaction term is imo not possible because every year has values for males and females and thus cannot be identified as one or another.

Comparing regression coefficients of two seperate regressions, once with incorporating male and once with incorporating female optimism is also not a good idea i think.


I would be very pleased if you could help me with this problem!

Manuel
 

maartenbuis

TS Contributor
#2
Something like:

Code:
reg cash femopt maleopt dividend tobinsq
test femopt = maleopt
You need to addept this to the variables names that you used in your dataset and the regression type model you think is appropriate for your data, but the general strategy is the same: add both variables to your model and use the test command to compare the coefficients.