structural breaks in all regression coefficients


I've tested my model for breaks in any of the regression coefficients, for breaks in each regression coefficient independently, but now I have to test for breaks in all the regression coefficients at once. Does anyone know How to achieve this?

Before I regressed my standard model while including the interaction break terms, and concluded with an F test on the interaction terms to test for breaks.

Y=a+B1*X1+B2*X2+B3*X3+B4*X4+B5*X5 + G*a*DUMMY + G1*X1*DUMMY + G2*X2*DUMMY + G3* X3 * DUMMY + G4*X4*DUMMY + G5* X5 *DUMMY

with an F test I tested my null hypothesis that G=G1=G2=G3=G4=G5=0
If I could reject my hypothesis, it means that any of the G's is not equal to 0.

Thus, how do I test that all of my G's are not equal to 0?

I hope someone can be of help!

Thanks in advance,

Andrea Vergouwen
The above question relates to my master thesis. It is expected from me to divide my time series in two regimes (volatile and less volatily) and check if my regression model fits better for one of the two. Now, after exploring the Markov Switching Regression approach, which is ideal for this kind of analysis, it seems STATA does not provide this kind of regression. After using CHOW testing, and the QLR ratio, which both give me breaks in many weeks of my regression coefficients, does anyone have any idea how to make a break in my sample, so that i can test two regimes?
If you think you know where there is a structural break in your model, then create a dummy variable to that effect, where D = 1 after the break and D = 0 before. Then interact the dummy variable with each of your other variables. If the dummy and the interactions are jointly significant, then you probably have a structural break around the point you specified. If not, then try some other break specifications. If you still don't get statistical significance, you probably don't have a break.

But be warned that unless you have a really good specification, the structural break you pick up on might actually be due to some omitted variable.