The data set consists of different variables from different banks during 6 years. I want to understand how these variables correlate with market share. The problem is that market share is a zero-sum game. If market share for one bank increases then the market share for another bank automatically decreases. How should I combat this? I have tried logit transformation for fixed effects and vector autoregression but it is not great.
1
-
$\begingroup$ Compositional data analysis is one way to handle this type of problem. This question seems closely related to yours. $\endgroup$– EdMCommented Nov 28 at 14:58
Add a comment
|