Model uncertainty and the pricing of American options (Q503400)

The authors study American options under the conditions of model uncertainty. This uncertainty implies some flexibility in option pricing. The paper quantifies the potential value of this flexibility by identifying the supremum of the price of an American option when the class of all models consistent with a family of European call prices is considered. To solve the problem, a duality between pricing and hedging problems is established. The main theoretical result of the paper is that in the presence of a finite set of European call options, the supremum of the value of an American claim is equal to the cost of the cheapest superreplicating strategy. The proof demonstrates that the search for the cheapest strategy within a particular sub-family of replicating strategies (an upper bound on the price), and the search for the model which places the highest value on the American option within a particular sub-family of models, are the primal and dual of the same finite linear program, and have the same optimal value. Hence the cheapest superreplicating strategy lies in the chosen sub-family, and the model which gives the highest value to the American option is from the given sub-family of models. The methodology provides a viable method of computing the bound in practice.