APPENDIX C

Incorporating Scenario Forecasting into Monte Carlo Simulations

An extension to the Monte Carlo simulation method is to conduct a sensitivity or what-if analysis using the simulation model. In a prior project initially developed to support a North American airport’s enterprise risk management (ERM) process, members of the ACRP research team devised a system to conduct a sensitivity analysis of specific risk factor(s) in the model. The objective of the analysis was to provide insights to the airport into how specific external risks could impact airport passenger volumes and financial position. A Monte Carlo simulation forecast had been developed for this airport, including the incorporation of the airport’s own risk register of strategic risks into the forecast risk register. The scenario analysis was developed to create a series of what-if scenarios such that the airport could better understand not only the expected impacts of a risk but also how the entire forecast probability range could change should a specific risk occur.

The process devised for conducting the simulation-based scenario analysis was as follows:

• Use an existing Monte Carlo forecast model as the basis for the scenario analysis tasks.
• Identify the forecast risk factor(s) to be evaluated in the scenario.
• Set the probability of occurrence for the risk factor of interest to 100% in a given year for each scenario analysis. This will then trigger the risk factor in every scenario at a specific year to allow for a fixed comparison to the existing forecast.
• Maintain the originally forecast statistical distribution of modeled impacts for the risk factor to be analyzed. This allows the scenario analysis to retain uncertainty on how impactful a potential future risk is, even if the scenario analysis assumes that it must occur at a given time in the future.
• Run the simulation model allowing all other risks and uncertainties to be simulated as in the original forecast model. This ceteris paribus approach assumes that all other uncertain and assumed factors influencing the forecast continue to operate, with the only change to the potential future world stated in the scenario analysis being the certainty of the specific risk factor in question.
• Compare the new forecast probability range (the statistical distribution derived from the aggregation of the thousands of individual Monte Carlo simulations) and most likely forecast of the scenario analysis to the preexisting forecast as well as the recovery times and paths.
• Evaluate the impact of the specific risk over the short and long term, including the change in the probability range of outcomes. Use this information to devise mitigation and avoidance strategies.
• Repeat for additional risks to be analyzed.

An illustrative example of the process and its results are provided in Figure C-1. This illustrates how scenario analysis with simulation can be used to investigate the impact of a specific risk factor on a forecast’s outcome over time, which can be contrasted to the baseline forecast

(shown in Figure C-1). The illustrative example assumes that in forecast Year 5, one of the risk factors will occur in 100% of all simulated scenarios but allows the magnitude of the impact to vary randomly. For this example, it is assumed that the factor has a relatively low probability of risk but with a significant potential negative impact. The expected impact is a 35% reduction in traffic in the year the risk event occurs, with traffic taking 4 years to recover and a lasting negative impact on future traffic volumes (e.g., loss of a major carrier or significant shock event with long-term effects on demand/supply). From this simulation example of a high-impact, low-probability event, conclusions can be drawn, such as there is approximately a 25% probability that traffic will not recover to preshock levels by the end of the forecast or that there is a 10% to 25% probability that traffic will return to baseline, most likely levels within 1 year of the shock event.

This approach realized several benefits for the airport and the forecasting team. First, the airport gained a greater insight into the projected quantitative impacts on passenger traffic and aircraft movements of various key strategic risks in a way that the standard stochastic risk-based forecast could not easily provide. The airport ERM team and forecast users were able to see how a given risk, even with a still-uncertain and simulated range of impacts on traffic, could impact traffic both in the short and long term, holding all other forecast factors the same as in the previous forecasting efforts. Second, by evaluating how the forecast probability range shifts in response to a specific risk factor, the ERM team could better understand how a given risk would impact overall and sector-specific traffic and then devise mitigation and adaptation strategies for that risk. And as the forecast is simulation-based, those mitigation and adaptation strategies can take into account the range of potential impacts and their probabilities of a given level of traffic occurring rather than some hypothetical expected value or low/high scenario with no information on probability.

Members of the ACRP research team have conducted similar simulation-based scenario analyses for other airports as part of a forecasting process for external strategic risks, uncertainties about regulatory changes, and impacts arising from major changes to an airport’s airline