AN OPTION VALUATION FRAMEWORK BASED ON ARITHMETIC BROWNIAN MOTION: JUSTIFICATION AND IMPLEMENTATION ISSUES
Date | 01 September 2017 |
Author | Joshua A. Brooks,Robert Brooks |
Published date | 01 September 2017 |
DOI | http://doi.org/10.1111/jfir.12129 |
AN OPTION VALUATION FRAMEWORK BASED ON ARITHMETIC
BROWNIAN MOTION: JUSTIFICATION AND IMPLEMENTATION ISSUES
Robert Brooks
University of Alabama
Joshua A. Brooks
Columbus State University
Abstract
We examine arithmetic Brownian motion as an alternative framework for option
valuation and related tasks. After reexamining empirical evidence, we compare and
contrast option valuation based on one of the simplest forms of geometric Brownian
motion with arithmetic Brownian motion. We identify an enhanced way to handle
negative stock prices within arithmetic Brownian motion that is consistent with
empirical observation. We review numerous strengths and weaknesses of both
approaches. The arithmetic Brownian motion framework allows for the aggregation of
any number of correlated factors for risk analysis.
JEL Classification: G13
I. Introduction
The Black–Scholes–Merton option valuation model (BSMOVM) is deeply embedded in
the heart of modern financial analysis. It is also well known, however, that the BSMOVM
is deeply flawed. The Black–Scholes–Merton (BSM) framework has been extended to
incorporate stochastic volatility, jumps, volatility surfaces, and so forth. Although many
of these extensions have been useful for specific applications, core problems remain,
many of which are rarely identified, much less addressed. For the purpose of this
introduction, we identify two problems with the BSMOVM related to the underlying
lognormal distribution assumption. First, there is zero probability of the stock price being
zero in the future. Second, a simple portfolio of stocks follows no known statistical
distribution. Hence, within the BSMOVM, internally consistent portfolio statistics
cannot be computed, which impairs risk management practice.
1
Thus, the lognormal
distribution suffers because it is not additive and the underlying values are strictly
positive. To address these issues, we propose an alternative framework based on
The authors gratefully acknowledge the helpful comments of Don Chance, John Deeble, Matthew Lambert,
Abbi Ondocsin, Pavel Teterin, Kate Upton, and Kaylee Brooks.
1
An important solution for this aggregation issue and lack of closed-form distributions is the use of the
nonparametric models presented in Stutzer (1996). The maximum entropy approach used is significantly sample-
period dependent and does not have internally consistent portfolio characteristics unless each possible portfolio is
separately computed.
The Journal of Financial Research Vol. XL, No. 3 Pages 401–427 Fall 2017
401
© 2017 The Southern Finance Association and the Southwestern Finance Association
RAWLS COLLEGE OF BUSINESS, TEXAS TECH UNIVERSITY
PUBLISHED FOR THE SOUTHERN AND SOUTHWESTERN
FINANCE ASSOCIATIONS BY WILEY-BLACKWELL PUBLISHING
arithmetic Brownian motion (ABM). To illustrate this framework, we use the most
simple and most canonically recognizable form, which we term the arithmetic Brownian
motion option valuation model (ABMOVM).
The ABMOVM presented here, based on the normal distribution, can easily
address these two problems. All the extensions (such as stochastic volatility, jumps,
volatility surfaces, etc.) can also be easily applied within the ABMOVM. Ultimately, the
appropriate model to deploy in practice depends on the user’s objectives, the underlying
instruments involved, and the empirical support. Given the utility of a nonzero
probability of default and the many applications of a known portfolio distribution, our
purpose is to justify the ABM framework as a reasonable approach to consider.
Additionally, we explore selected implementation issues.
Historically, the shift in financial research from ABM to geometric Brownian
motion (GBM) occurred with Samuelson (1965). Samuelson, based on earlier work by
Osborne (1959), Sprenkle (1961), Alexander (1961), Boness (1964), and many others,
introduced GBM for solving financial derivatives valuation problems. GBM was
introduced, in part, to “correct”prior errors by authors relying on ABM. Samuelson
credits Bachelier (1900) with discovering “the mathematical theory of Brownian motion
five years before Einstein’s classic 1905 paper”(p. 13). Two prior errors identified by
Samuelson include stocks possessing limited liability and warrant values exceeding the
underlying instrument’s value. It is important to note that although stocks do possess
limited liability characteristics, they still have some probability of becoming worthless.
As we address later, both are not weaknesses of ABM, but rather an incomplete
application of ABM to this finance problem.
Additionally, we reexamine ABM by focusing on the empirical evidence related
to the distribution of underlying instruments. Because finance is a social science, every
mathematical framework has both strengths and weaknesses. As Derman (2011) notes in
the context of financial theories, “Only imperfect models remain. The movements of
stock prices are more like the movements of humans than molecules. It is irresponsible to
pretend otherwise”(p. 187). Because modeling human movements remains elusive, it is
important to have multiple, pertinent frameworks available for the array of tasks facing
the finance profession.
Finally, we compare and contrast the s imple BSM-based forms of the
orthodox GBM with the ancient AB M. Our purpose is not to advocate ABM- based
tools over GBM-based tools, but rather to as sert that the ABM should be considered as
a viable framework for analys is in many and varied contexts . As option valuation
models become more complex and nu merous, the risk of combinator ial explosion
becomes more plausible. Corres pondingly, many of the most recent mode ls of option
valuation require calibrat ion within each market where they ar e deployed. Each of
these issues is further exacer bated when a risk manager attem pts to gain insight into
the aggregate position of an entire firm. Our approach gives a simple, closed- form,
option valuation framework that allows a level of aggregation never before seen in this
research space. There have been extraordinary advances in representing the finest
details of an underlying ins trument’s behavior in many of the ne west option valuation
models (e.g., Garleanu, Peders en, and Poteshman 2009; Corsi, Fus ari, and La Vecchia
2013; Fulop, Li, and Yu 2014). As ne w markets come online, expandi ng the toolbox to
402 The Journal of Financial Research
To continue reading
Request your trial