In contrast to the views put forward in a recent paper, the following points are made concerning multicollinearity in regression models: firstly, multicollinearity is almost invariably a problem of degree rather than kind; and, secondly, the search for an appropriate functional form for a regression equation should not be confused with the treatment of multicollinearity. Finally, some comments are made on the role of prior detrending in both linear and nonlinear models.
No takes yet. Share an insight, caveat, or question.
Andrew Harvey (1977) studied this question.
Synapse has enriched 5 related papers. Consider them for comparative context: