Does it seem reasonable to do a PCA on independent variables and factor analysis to group dependent variables and after all of this do a RA?

#1
I have a survey data, so to get rid of the multicolinearity problem, I am doing a Principal component analysis. However, I also have 2 dependent variables each consisting of 2 questions from a survey, e.g. one is internal sources of financing, the other is traditional external sources of financing. In order to get these 2 dependenat variables, I have to somehow group them first, like with a factor analysis for e.g. And afterwards, I want to do a regression analysis.
This sounds a bit complicated, using all these methods in a sample that cannot be classified as a big sample, n = 140.

Has someone a suggestion how to make things simpler, or is this completely reasonable thing to do?