A Non-Censored Binomial Model for Mean Reverting Stochastic Processes

Binomial trees are widely used for both financial and real option pricing due to their ease of use, versatility and precision. However, the classic approach developed by Cox, Ross, and Rubinstein (1979) applies only to a Geometric Brownian Motion diffusion processes, limiting the modeling choices. Nelson and Ramaswamy (1990) provided a general method to construct recombining binomial lattices which was used by Hahn and Dyer (2008) to develop a censored recombinant Mean Reverting model. These models, although more computationally complex in programming than the Cox et. al. (1979) binomial model, are fundamentally simpler than alternative approaches such as trinomial trees or simulation methods for American options. In this paper we extend the mean reverting model of Hahn and Dyer (2008) and propose a non-censored model that is more precise and has some other distinct advantages. We compare these two approaches and present the results of applying these models to evaluate a hypothetical real option.