How to take an exam if you must: Decision under Deadline

This paper uses the example of an exam to model multi-dimensional search under a deadline. When the dimension is two, an order-invariance property allows simple characterization of the optimal search policy. Behavior is shown to be highly sensitive to changes in the deadline, and a wide variety of policies can be rationalized (as being optimal) as the length of the deadline increases. This is contrasted with behavior under the traditional case of geometric discounting, in which a similar sensitivity to changes in the discount factor cannot hold. For dimensions higher than two, the invariance principle does not hold; this increases complexity of the problem of finding the optimal search policy.