This paper shows that the informativeness principle, as originally formulated by Holmstrom (1979), does not hold if the first-order approach is invalid. We introduce a "generalized informativeness principle" that takes into account non-local incentive constraints and holds generically, even...

This paper shows that the informativeness principle does not automatically extend to settings with limited liability. Even if a signal is informative about effort, it may have no value for contracting. An agent with limited liability is paid zero for certain output realizations. Thus, even if...

The informativeness principle demonstrates qualitative benefits to increasing signal precision. However, it is difficult to quantify these benefits -- and compare them against the costs of precision -- since we typically cannot solve for the optimal contract and analyze how it changes with...