Anova is used to see if the two groups (or more) are different from each other assuming the population they are drawn from is normally distributed. It allows between and within group comparison. You should read some material on it. It's pretty straightforward.
We use NOVA two test cross effects. Frankly I would rather have testing statistical hypothesis independently from the standard models used by econometricians who are not always trained well in probability theory and mathematical statistics. This way the analysis is not burdened by the propagated errors such as restricting the choice of optimal to linear statistics instead of choosing from the space of all estimators and solving the Fischer differential equations as Irving Fischer did a century ago