Enhanced Monte Carlo Estimates for American Option Prices

Monte Carlo simulation has trouble with American options because the exercise decision at a given date must compare the option's immediate exercise value against its continuation value. The option value if it is not exercised is a function of its value along all possible future price paths from that point on, and each path will present further exercise decisions with the same difficulty in resolving them. The authors propose a hybrid valuation technique that bridges Monte Carlo simulation and lattice methods. Instead of simulating price paths, they simulate whole price trees. The tree emanating from each point is used to assess the option continuation value for that date and stock price. While the results are accurate, inevitably the procedure requires a large number of computations. The authors then offer a variety of techniques that substantially increase efficiency.