FINANCING CONSTRAINTS AND FIRM HETEROGENEITY IN INVESTMENT BEHAVIOR - AN APPLICATION OF NONNESTED TESTS

Estimating an extended q model against Canadian panel data it is demonstrated that the usual practice of conditioning tests of financing constraints on the a priori sorting of firms into financially constrained and unconstrained groups may lead to spurious inference about the existence of credit rationing. The sample sorting bias is minimized by empirically determining the best sorting criterion. The results reveal that while liquidity is the most important and robust determinant of corporate investment, the data do not support critical predictions of either the underinvestment or the over-investment theories.