Fundamentally, the benefit of postponing commitments to flights is that additional information about the eventual hotel prices becomes known. Thus, the benefit of postponing commitment is computed by sampling possible future price vectors and determining, on average, how much better the agent could do if it bought a different flight instead of the one in question. If it is optimal to buy the flight in all future scenarios, then there is no value in delaying the commitment and the flight is purchased immediately. However, if there are many scenarios in which the flight is not the best one to get, the purchase is more likely to be delayed.
The algorithm for determining the benefit of postponing commitment is similar to that for determining the marginal value of hotel rooms. It is detailed, with explanation, in Table 8.