When I have a general profit - functions as follows:
Y = X * P - W * K - F
Y is a vector of components of yield, P, X and W, K are vecotrs of prices and quantities of input and output
Is it statistically right to model Y, X, P, W, K and F as formative measured LVs (in order to show the influences i.e. of X onto P simultaneously) or conflict that with problems of endogenity like using multiple regression? Are LVs like instrumental variables?
For example: X==>Y, P==>Y, X==>P in one path diagram
Every dataset has a different production function (is a different producer).