So you have a deadline to submit a project. Your cost, in terms of utility for example, for working on the project is stochastic with a known distribution. When should you start working on it? The solution to this problem is a stopping rule, which will of course depend on the way you discount future disutility.
Alberto Bisin and Kyle Hyndman look at this with three different types of discounters and come up with interesting solutions. Agents with a standard exponentially discounted utility will choose to work on their project when it is most convenient, which may be before the deadline. Agents with hyperbolic discounting will delay until the last moment. But hyperbolic discounter may not be that naïve. Sophisticated ones realize they have a procrastination problem and will end up completing their project before the deadline.
This last result is important because sophisticated hyperbolic discounters and exponential discounter end up being, potentially, observationally identical. However, looking at me and around me, there are many hyperbolic discounters who are aware of their problem, but still cannot do anything about it except complaining. I am thus not quite convinced of the empirical relevance of this result. Of course, my empirics are anecdotal, but I saw no empirical evidence at all in the paper.