Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
Suppose a researcher observes a finite number of preference comparisons between pairs (x,t) of rewards x delivered at time t. We provide necessary and sufficient conditions under which such a dataset can be rationalised with the most common specifications of the discounted utility model. The distinctive feature of our characterisation is that it refers to notions of stochastic dominance among empirical distributions defined over the observed rewards and time delays.