# st: interpretation of FE model with interaction of 2 dummies with continuous variable

 From P K To statalist@hsphsun2.harvard.edu Subject st: interpretation of FE model with interaction of 2 dummies with continuous variable Date Thu, 16 Jul 2009 01:50:27 +0000 (GMT)

```Hi,

I am estimating a fixed-effects panel model with an interaciton between the
following variables:

strategic_change => coded as a dummy (1 if change happens in year t, 0
otherwise)
ceo_change => coded as a dummy (1 if change happens in year t-1, 0
otherwise)
duration_change => continuous variable (duration clock, measured in year
t-1)

The FE regression is:
xtreg   performance    strategic_change
ceo_change   duration_change   strategic_change*ceo_change
strategic_change*duration_change
ceo_change*duration_change    strategic_change*ceo_change*duration_change,
fe

Results are shown below. My question is how to interpret the interaction terms, especially the last term where I interact two dummy variables (strategic change and CEO change) with a continuous change (duration change).

Does STATA pick the case that each dummy is 1 as a baseline category, or can I conclude that while the combined effect of strategic change and CEO change is positive and significant (coefficient = 12.66), the combined effect of strategic change, CEO change and change duration is negative and significant (coeff. = -4.91). Does this imply that when strategic change =1 and CEO change =1 and change duration is high/increasing, this has a negative impact on performance?

Thanks a lot,
Pat

Results:

Variables  Performance (corrected for collinearity)
Strategic change 2.01
(2.06)
Duration change  (dropped)
(dropped)
CEO change -2.8
(4.16)
Strategic change* change duration -0.47
(0.9)
Strategic change* CEO change 12.66**
(4.88)
Prior CEO change*change duration -0.74
(1.8)
Strategic change* change duration*CEO change -4.91**
(1.73)
_cons                           72.04*
(40.33)

No. of Obs. 430
F statistic               2.46***
R-Squared 0.09

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
```