# Real Option Methodology for Risk Assessment in Asset Management

Status: Concept
Timeframe: Near-term
Potential Funding:

Background:

A number of approaches are commonly used to manage risk, including conducting visual inspections of existing infrastructure, using design standards with conservative safety factors for new infrastructure, and applying best practices for minimizing risks of project cost and schedule overruns. Research is needed to determine how to build on existing practices to better assess the risks to transportation assets, better quantify consequences of different risks, and better prioritize investments explicitly acknowledging uncertainty in future events.

R = Σ pi × Ci

Risk (R) is generally quantified with the equation above. It is essentially a value for the expected outcome returns of a decision weighted by the probability (p) of each consequence (C) of event i. How do we calculate R if neither p nor C is certain? Do current methods address this effectively?

Investment decisions are widely made using discounted cash flows (DCF). It is assumed, that given a certain decision made in Year 0, Costs and Benefits can be assumed for a number of years to come, i.e. C is known and p is assumed 1 for all i. If the project is considered risky, the discount rate is increased accordingly. However, defining the future in- and outflow of cash with such deterministic certainty is unrealistic. Not only is the consequence (C) uncertain, but also their occurrence. This is because infrastructure is often affected by stochastically occurring events.

We can ignore the uncertainty by using expected values. Imposing an assumed expected value will nevertheless almost certainly lead to arriving at a wrong risk estimation (see figure X). This is called the “flaw of averages” (Savage, 2012). The error due to the “flaw of averages” exponentiates when systems are non-linear because outputs using expected inputs do not equal expected outputs. Ultimately, it can be said that ignoring the uncertainty and the consequent existence of a distribution instead of a deterministic expected value, is a fallacy.

Another approach to compensate for increased risk is overdesigning infrastructure. This reduces the probability of failure to negligent values, but may lead to infrastructure being overly expensive or redundant. This also ignores the fact that infrastructure owners are not passive, but actively observe their condition and relevant external factors and trends that affect the condition level of the infrastructure. Based on this, the fundamental assumptions of DCF do not seem appropriate.

Objectives:

The Real Option method allows infrastructure owners to evaluate the advantage of options that an infrastructure manager has over time. As time passes, a manager will have the ability to intervene as as an object may deteriorate at a faster rate than expected. Likewise, a manager may postpone a planned intervention if the condition is better than expected. In addition to the option to defer, a manager may have the option to expand or contract the infrastructure or the infrastructure network, as well as to shut it down temporarily, abandon it, grow it or switch it (de Neufville and Scholtes, 2011).

The options provide an owner with the flexibility adapting the infrastructure to uncertain future needs. Owners, thus, neither under-, nor overinvest and consequently minimize the risks of their decisions. The external factors affecting risk include weather events, condition development, system demands, funding and other critical variables. The methodology proposes a way to systematically analyse and define these uncertainties and make predictions taking the defined uncertainty fully into consideration.

Real option valuation is known using binomial lattices (a form of decision trees) and/or Brownian motion random walk algorithms. Infrastructure life-time net benefits can also be calculated by simulating the uncertainty using continuous Monte Carlo simulations. Using different stakeholders’ costs of different design alternatives and management strategies, the costs can be calculated over a large sample of potential futures. The methodology is able to address multiple levels of risk and weight them as necessary and thus make multi-objective, cross-asset investment decisions under uncertainty to best support the national goals identified in 23 USC 150(b).

The ultimate objective is to provide the decision-maker with tools that add value to the decision-making process and improve the robustness of the infrastructure network as a whole. In that sense, novel approaches for the evaluation of risk will be sought to capture the stochastic nature of interdependent infrastructure. A graph theory approach to evaluate criticality of network node failure as shown by Buldyrev and colleagues (2010) may prove interesting for the evaluation of consequences, and thus the real option value for the infrastructure, simulated by network programming methods.

The application and evaluation of a large sample of data and data simulations is computationally challenging. Furthermore, decision-making tools are urged to be simple and understandable. As big data may improve predictability and performance of models, strong emphasis must be laid on the usability of such models. In this project, it is suggested that particular focus will be on addressing these challenges with the outlook of combining big data and the model’s user interface design.

References:
Buldyrev, S. V., R. Parshani, G. Paul, H. E. Stanley and S. Havlin (2010) Catastrophic cascade of failures in interdependent networks, Nature, 464, 1025-1028.
de Neufville, R. and S. Scholtes (2011) Flexibility in Engineering Design, Engineering Systems, MIT Press, ISBN 978-0262297332.
Savage, S. (2012) The Flaw of Averages: Why we underestimate Risk in the face of Uncertainty, Wiley, ISBN 978-1118073759.
Prof. Dr. Rade Hajdin, July 2019

Research Plan:

TBD

Duration: n/a months
Project Budget: \$0,000

Topics: Analytical Tools and Models
Subjects: Risk Management