A two-step simulation procedure to analyze the exercise features of American options

Abstract In this contribution we propose a two-step simulation procedure that enables to compute the exercise features of American options and analyze the properties of the optimal exercise times and exercise probabilities. The first step of the procedure is based on the calculation of an accurate approximation of the optimal exercise boundary. In particular, we use a smoothed binomial method which effectively reduces the fluctuating behavior of a discrete boundary. In the second step the boundary is used to define a stopping rule which is embodied in a Monte Carlo simulation method. A broad experimental analysis is carried out in order to test the procedure and study the behavior of the exercise features. Mathematics Subject Classification (2000): 60G40, 60J60, 65C20 Journal of Economic Literature Classification: G13