We test Hicks’ induced innovation hypothesis (IIH) in U.S. agriculture using a state-level panel dataset of input prices and quantities for the period 1960-2004. A two-stage cost minimization model where decisions of research investment allocation predates those of input allocations is set forth. Parametric relationships between expected relative price change and input quantity allocation that represent technical changes are obtained distinctively from those between current relative price change and input quantity allocation, which represent general factor substitutions. Our model allows both for time-varying, non-neutral research costs across states and time while only requiring that unobserved trends in relative costs of research are equal among states. Considerable evidence for the IIH is found for three of six input pairs.