Keynes and the logic of econometric method

This paper analyzes the controversy between Keynes and Tinbergen on econometric testing of business cycle theories. In his writings, Keynes repeatedly emphasizes a logical objection to Tinbergen s work. It is clarified what exactly this logical objection is, and why it matters for a statistical analysis of investment. Keynes arguments can be traced back to his Treatise on Probability, where the principle of limited independent variety is introduced as the basic requirement for probabilistic inference. This requirement is not satisfied in case of investment, where expectations are complex determinants. Multiple correlation, sometimes thought to take care of required ceteris paribus clauses, does not help to counter Keynes critique