For the two-act linear loss decision problem with normal priors, conditions are derived for which the expected value of perfect information about two independent risks is super-additive in value. Several applications show how a variety of decision problems can reduce to the canonical problem, and how the general results obtained here can be translated simply to prescriptions for specific situations.
Keisler, Jeffrey, "Additivity of information value in two-act linear loss decisions with normal priors" (2005). Management Science and Information Systems Faculty Publication Series. 38.