Hidden testing and selective disclosure of evidence

A decision maker faces a choice to withdraw or to retain a product but is uncertain about its safety. An agent can gather information through sequential testing. Players agree on the optimal choice under certainty, but the decision maker has a higher safety standard than the agent. We compare the case where testing is hidden and the agent can choose whether to disclose his findings to the case where testing is observable. The agent can exploit the additional discretion under hidden testing to his advantage if and only if the decision maker is sufficiently inclined to retain the product. Hidden testing then yields a Pareto improvement over observable testing if the conflict between players is larger than some threshold, but leaves the decision maker worse off and the agent better off if the conflict is smaller than this threshold.