Project Risk Costs: Estimation Overruns Caused When Using Only Expected Value for Contingency Calculations

Abstract

Real project experiences show that project actual risk cost is usually higher than expected. A significant reason for that overrun could be that the overall expected risk cost (ERc) is often calculated using only the deterministic expected value method (EV), i.e., the sum of the products of probability times the cost for each risk event. This article analyzes the reliability of the so-calculated ERc value by comparing the ERc to results obtained with the Monte Carlo simulation, using thousands of random analytical models and models from 271 real projects. This article addresses the theoretical basis of the EV method and the influence of the number of risk events, their correlation, and their statistical distribution. It proves that common project risk models do not meet the premises required to use the EV method and that this method, when used alone, often results in an underestimation of the ERc; project risk cost could actually be up to three times higher. This article proposes the application of an enlargement factor to the deterministic ERc to improve its level of certainty, and provides a formula obtained with multivariate regression analysis to calculate it. Project cost overrun is a fact in project management and a risk assessment is usually conducted to prevent it. The total expected cost of the risk events identified is used as a contingency sum to cover for risks that occur, and many companies calculate it as the sum of the products of probability times the cost for each risk event. This contingency sum affects project profitability and, if it is wrongly calculated, the related decision-making process could also be wrong. This article proves that this calculation method increases the chances of an unacceptable overrun in common projects and risk scenarios because it yields just an average that will overrun 50% of the time, and real values could easily be even three times higher. This simplified method should not be used for this purpose in common projects and risk scenarios. A more sophisticated method, such as the Monte Carlo simulation, could be used instead, but many companies do not have the knowledge and skills they require. For that reason, this article proposes the application of an enlargement factor to that sum to improve its level of certainty and provides a formula to calculate it.