摘要:This paper uses recently developed panel cointegration techniques and error correction modeling to test the hypothesis of induced innovation in a state-level panel of U.S. agricultural production. These methods provide greatly improved power compared to conventional approaches. A 5-step testing procedure is used to test the induced innovation hypothesis based on a 2-stage CES production function with aggregate output and four input categories. For both pairs of inputs, the test fails to support the hypothesis of induced innovation at one or more of the critical steps.