Valuing American Options under ARMA Processes

Motivated by the empirical findings that asset returns are autocorrelated, this paper provides the pricing algorithm for American options on the stocks, the returns of which depend on an autoregressive moving average (ARMA) process, by incorporating with the least squares Monte Carlo approach of Longstaff and Schwartz (2001) and the local risk-neutralization principle of Duan (1995). Based on the results of numerical analyses, the ARMA effect has significant impacts on values of American options. Specifically, the AR effect is more significant than the MA effect.