We propose a quasi-analytical approach for valuing American-style options under Gaussian and double exponential jumps à la Merton (1976) and Kou (2002). Our approach is based on dynamic programming coupled with finite elements. Finally, we perform a numerical experiments that show convergence and efficiency. We also address the estimation problem and report an empirical investigation based on Home Depot. Jump-diffusion models outperform pure-diffusion models.
Published October 2012 , 18 pages
This cahier was revised in February 2015