This paper derives a formula for the optimal forecast of a discounted sum o
f future values of a random variable, where optimal is defined in terms of
the minimized H-infinity-norm of the forecast error. This problem reflects
a preference for robustness in the presence of (unstructured) model uncerta
inty. The paper shows that revisions of a robust forecast are more sensitiv
e to new information, and discusses the relevance of this result to previou
s findings of excess sensitivity of consumption and asset prices to new inf
ormation. (C) 2001 Elsevier Science B.V, All rights reserved.