Title: The value of turning-point detection for optimal investment
Authors: Michail Chronopoulos - City, University of London (United Kingdom) [presenting]
Lars Sendstad - NTNU Norway (Norway)
Abstract: To capture the dynamic evolution of economic indicators and its impact on option pricing, we develop a regime-switching, real options framework for investment under uncertainty that facilitates time-varying transition probabilities. Considering a private firm with a perpetual option to invest, we use machine-learning techniques to forecast the evolution of transition probabilities and analyse how they affect the value of an investment opportunity. Results indicate that: (a) ignoring the dynamic evolution of transition probabilities can result in severe valuation errors; and (b) when the probability of a regime switch is low, the option value is greater in the good (bad) regime under time-varying than under fixed transition probabilities.