Allocating, Phasing, and Converting a Risk Adjusted Cost Estimate

By selecting a contingency amount, the program estimate becomes a “risk-adjusted” estimate. Because contingency is calculated as a lump sum from the S curve, it may be necessary to allocate it throughout the WBS, phase it across future years, or convert it to budget year dollars. Allocating, phasing, and converting risk-adjusted estimates present their own sets of modeling challenges.

Allocation

Allocating contingency throughout the WBS may be necessary for a number of reasons. Analysts may want to inform the risk management process by forecasting which elements may require significant portions of the contingency. Allocation may also be necessary for organizations that prefer contingency be split between appropriations. In addition, allocation may be needed for budget submissions when the organization does not make allowance for the separate display of contingency amounts. Allocation is generally performed before the risk-adjusted estimate is phased and converted to budget dollars.

Allocating contingency is not straightforward. There are several methods for allocating contingency across WBS elements. Two common methodologies are discussed below. Analysts must ensure that contingency is assigned to WBS elements commensurate with the risk of that element, and that the sum of the risk-adjusted WBS elements equals the total risk-adjusted estimate. Each WBS element has contributed differently to the overall S curve in the Monte Carlo simulation. That is, each WBS element has its own probability distribution, with some elements more likely to underrun or overrun, and some elements positively or negatively correlated with others. Thus, contingency cannot simply be allocated equally across all cost elements, nor can it be allocated proportionally based on the contribution of each element to the total cost. Rather, contingency must be allocated according to the magnitude and variance of each WBS element’s individual probability distribution.

It is important to note that the allocation of contingency to WBS elements in this manner does not represent a commitment to fund the WBS elements. Contingency funds should be retained by management until specific needs are realized. It is likely that contingency funds will not be spent on each WBS element according to the estimated allocation. Allocation is a modeling exercise to support budget submissions, verify the prioritized risk list, and to inform the risk management process.39

Allocation Methods

Analysts that need to allocate contingency to lower-level WBS elements must ensure the allocation meets two conditions: risk-adjusted lower level elements must sum to their parents, and elements that have more risk and uncertainty should have higher amounts of contingency. The difficulty meeting the first condition arises from the fact that because each WBS element’s cost distribution is different, lower-level WBS elements at a specified percentile will not sum to the parent’s WBS percentile value. For example, the sum of level three WBS elements at the 80th percentile will be greater than their level 2 and level 1 parents’ 80th percentile value; conversely, the sum of all level three WBS elements at the 20th percentile will be less than their level 2 and level 1 parents’ 20th percentile value.

Cost analysts have proposed several methods of allocating contingency to address these conditions. Two popular methods used are allocating by standard deviation and allocating by need.

Allocation by standard deviation: Analysts may allocate contingency by adjusting the cost estimate to account for the difference in contingency between the sum of the children and the parent at a specified percentile. Analysts select the costs of children elements and the parent element at the same percentile, and calculate the difference between the sum of the children and the parent. The difference is then allocated back through the children elements proportional to the magnitude of the element’s standard deviation. Therefore, more contingency is allocated to elements that have wider cost distributions and the risk-adjusted children elements sum to the parent.

Allocation by need: Other experts have proposed that the variance of a cost element is a poor proxy for the risk associated with an element. Instead, they suggest that the analysis should take into account the probability of the element overrunning its point estimate, or its “need.” Need is represented by the difference between the cost element’s most likely estimate and the estimate at the specified percentile, and is adjusted by the correlation between lower-level elements. Contingency is allocated to lower-level elements proportional to the total need.

Other allocation methods are available in cost analysis publications, which vary in complexity and applicability. While mathematical complexity may result in more theoretically precise results, there may be a trade-off with the cost estimating team’s ability to defend the methodology to stakeholders and key decision makers.

Phasing

Contingency—either as an allocated portion of a WBS element, or a total amount that is left unallocated—may be phased, or spread, across future years of the program’s life cycle. Common methods for phasing contingency are spreading costs proportionally, through SME input, or through analogy to a similar program.

A simple approach to phasing is to spread the contingency proportionally across future years according to the phasing of the WBS element. However, proportional phasing assumes that contingency will be needed at the same rate that costs are incurred, which may not be realistic.

A common method for phasing contingency is to distribute it through future years according to the likelihood of risks being realized. For example, SMEs may assume that all risk associated with a WBS element would be realized prior to the fourth year of development, with the bulk of risk likely occurring in the third year. Cost analysts could therefore distribute half of the available contingency across years 1 and 2 of development, and assign the remaining contingency to year 3. Because this phasing method relies on the input of SMEs, it is often difficult to validate.

A third method of phasing is to spread the contingency by analogy to historical programs. For example, if the program being estimated is early in development, analysts may examine the rate of contingency expended during similar development programs. Phasing distributions modeled from historical data may serve as the basis for this phasing. As more information is known about the program, less abstract methods can be substituted for phasing. As the program moves into production, for example, the production schedule for that specific program can be the primary basis for phasing risk dollars, rather than historical programs.

Finally, cost risk software can automatically phase contingency. When using these software programs, analysts must understand the underlying assumptions used by the software and ensure that the final results are realistic and defensible.

Converting

Contingency is calculated on the total cost estimate and is therefore typically in base-year dollars. Once phased, the cost estimate must be converted into budget dollars by applying inflation indexes. Converting risk-adjusted estimates to budget year dollars is no different than converting point estimates. See step 7, chapter 10, for more information on creating budget year dollars.


  1. Experts have noted other issues with the assumption that contingency will be spent according to the forecasted allocations. For example, many risks when realized affect multiple WBS elements. In these cases, pre-allocating contingency funds to one element will underfund contingency for the other affected elements.↩︎