Manage contingencies, reduce risk
the PCA technique
by Brad Eichhorn
A SUCCESSFUL PROJECT has many components. While use of a project management tool is only one critical success factor in most projects, its importance cannot be diminished. However, bridging the gap between risk reduction techniques and using the project management tool for those techniques is either never done or quickly abandoned. The Plan Contingency Allowance (PCA) technique can identify and measure both schedule risk and cost risk.
The Contingency Problem
If contingency time is to be planned, then it should be separate and distinct to allow measurement. (“If it can't be measured, it can't be managed.”) Most project managers include some contingency factor in the task's duration and work effort. For example, Exhibit 1 shows a 180-hour/12.5-day phase of a project. Each task has incorporated a 20 percent contingency factor, which produced the 216-hour/15-day project shown. Specifically, Task A is a 100-hour/2.5-week task requiring a 20 percent contingency factor. It is represented as requiring 120 hours and 3 weeks.
We have all experienced the weaknesses in this approach. For example, if Jane is assigned to Task A, she would see her assignment requiring 120 hours and 3 weeks. Unfortunately, the original estimate is still 100 hours and 2.5 weeks. Further, questions related to the cost and duration of the project's contingency cannot be answered easily.
These situations highlight the need for identifying and measuring contingency in the project management tool. The desired approach would keep each team member's tasks separate from contingency requirements and allow easy measurement of planned, earned and consumed contingency. It should be scalable, to work on small projects through megaprojects, and relatively simple to administer.
The PCA technique creates a separate contingency task for each group of related tasks, adds appropriate dependency links to the work effort, creates fictional resources and assigns those resources to the new tasks. We will look at each of these steps separately to understand how to use and implement them in any tool.
The PCA Technique
The new tasks representing the Plan Contingency Allowance are associated with a set of activities (phase) with contingency to be measured. Using a software development example, there would be five PCA tasks—one for each of the major phases (Requirements Analysis, Design, Development, Testing, and Implementation). This allows for the measurement of contingency by major phase. It also allows for each phase to have separate contingency factors. For example, a software project may use a 20 percent contingency factor. However, one particular project may be the first attempt to use a new programming language. Therefore, the Design and Development phases could have a 30 percent or 40 percent contingency factor, with the remaining phases staying at 20 percent.
The PCA technique creates a separate contingency task for each group of related tasks, adds appropriate dependency links to the work effort, creates fictional resources and assigns those resources to the new tasks.
These PCA tasks are placed in the work breakdown structure such that it follows each of the phase's tasks directly or indirectly. This is accomplished by the tool's dependency links. The phase's milestone indicating completion follows the PCA task by also using a dependency link. These links ensure that the plan's indicated completion is dependent not only on each of the detailed tasks but also on the contingency's work and duration. As one or more changes are made to tasks through the project's execution, the phase's completion milestone adjusts appropriately.
One or more new resource(s) are added to the project plan to represent the staff “assigned” to contingency tasks. The hourly rate for these new resources should represent the blended rate of the staff assigned to tasks in the corresponding phase. These fictional team members are then assigned to the PCA tasks. “Real” team members are not assigned to contingency tasks. This is necessary since an accurate resource assignment is not possible for such tasks. Further, it ensures that each team member only sees his or her assigned tasks, thereby avoiding Parkinson's Law—“work expands to fill the available time.”
Performing these steps creates the new project plan represented in Exhibit 2. Notice that it has the same total work hours and the same duration as the original project plan in Exhibit 1. This new plan has all of the benefits deemed necessary for measurable contingency. (The PCA task's hours and duration are each 20 percent of the original plan. This will often generate an overloaded PCA resource due to parallel tasks within the phase, as in this example.)
Having separate PCA tasks increases the ability to report contingency time and cost. These PCA tasks should be baselined along with the rest of the project plan. This will allow time and cost variance analysis for the project's contingency.
Now that we have these new tasks and resource assignments in the project plan, we need to administer the project plan to take advantage of this data. Here are some general guidelines that address PCA task administration.
Managing Contingency Time
Project tasks are adjusted to reflect the task status until that task is completed. The adjustable fields include the actual start date, percent complete, revised estimate to complete figures and the actual finish date. The PCA tasks are also adjusted to reflect changes in the phase's contingency measurements. However, PCA tasks are never actually “performed”—they exist to represent the remaining contingency cost in time and dollars. (The only tasks that have work performed are the original tasks.) Therefore, the PCA task's percentage complete figure will always remain zero. It will have its work effort and duration adjusted to reflect the changing project status.
On Time Scenario. There are specific situations that will be reviewed to clarify PCA task administration. The first scenario to be studied is a project that is basically on time. If every task is completed nearly on time and on budget, then at the midway point of that phase's completion, the remaining contingency would be 50 percent of the original contingency. From our original example, the PCA task would be reduced to 18 hours, 1.25 days and 0 percent complete. At 75 percent completion of the phase, the PCA task should be set to 9 hours, .6 days and still 0 percent complete. This process will move the completion milestone closer to the predicted completion date.
Ahead of Schedule Scenario. A project performing ahead of schedule has PCA tasks adjusted similarly to the project that is basically on time. However, the original estimates for the project were apparently conservative. If a phase of a project is 15 percent ahead of schedule, then there may be justification to reduce the contingency factor for the remaining time by up to that same 15 percent. Using our example, the contingency factor would become 17 percent (15 percent below the original 20 percent). At the midway point of this phase, the PCA would be only 15.3 hours (17 percent of 90) with a duration of 1.1 days. This process will move the completion milestone even closer to the predicted completion date.
Behind Schedule Scenario. The project falling behind schedule needs to use a different process for adjusting Plan Contingency Allowance. The original project plan documented a particular completion date. Since we are assuming there has been no Project Change Request to account for a delayed completion, we need to “consume” the planned contingency to keep the project on schedule. If the team members are now giving an estimate to complete date of 3 days and 10 hours greater than originally planned for Task A, the PCA task should be reduced by 3 days and 10 hours. (If the task requiring greater time does not lie on the phase's critical path, such as Tasks B and C, then the PCA task's duration does not have to be reduced, only the PCA task work hours.)
The result is a phase with a completion date that still matches the original plan, task work effort that indicates an unfavorable variance of 10 hours, and reduced contingency available for other tasks in that phase. Note that a project phase running significantly late will consume all of the available contingency time. However, remember that contingency exists to cushion the effect of errors and problems. This method of tracking Plan Contingency Allowance will quickly highlight this situation and allow an early resolution to the problem.
When the phase's tasks are completed in each scenario, the PCA task is updated so that its work hours equal zero, the duration equals zero and the percent complete is 100 percent. This technique will facilitate cost calculations that reflect contingency variances and place the phase's completion milestone on the actual completion date.
Managing Cost Risk
The discussion so far has related to managing schedule risk. The measurement of cost risk follows in a similar manner. Earned value analysis (EVA) will include the time and cost of contingency throughout the project even as the contingency is adjusted—whether favorable or unfavorable. Periodic adjustments to PCA tasks will improve EVA's forecasting accuracy.
Exhibit 1. These 180 hours of project activities have 36 hours of embedded contingency time. However, the contingency time cannot be properly managed since it is not visible and may not be assigned to the proper resource.
Exhibit 2. The revised project plan clearly identifies contingency time and duration and provides for separate resource assignment. Further, it allows for accurate variance calculations of the project tasks and the opportunity to baseline plan contingency allowance time and duration for variance calculations.
The dollar cost of contingency can be measured in real terms and as a variance from the baselined project. These variances are also expressed in dollars since each PCA resource was assigned a blended rate of the corresponding staff.
“Earned contingency” is observed when the actual work hours for the phase are less than the baselined work hours. Therefore it is important to set the PCA task's actual work hours to zero when the phase completes. (Some professionals will argue that the “earned contingency” only occurs when the total actual work hours are less than the baseline plan, excluding the PCA work hours. These can be lively debates. However, regardless of your particular definition, this process will provide the data for whatever calculation is deemed appropriate.)
Similarly, contingency is “consumed” when the actual work hours for the phase are greater than the baselined work hours. Information on contingency variances, earned or consumed, is valuable for the next project planning effort since it provides feedback on the estimation process.
There are a few miscellaneous topics to be considered. It is possible to micromanage the PCA tasks by recalculating their work effort and duration with every update to the project plan. The PCA technique is a risk identification and measurement tool and should not require significant effort to administer. Biweekly PCA calculations should be sufficient for project accuracy. Additional recalculations would be warranted only if there are significant changes to the project plan. Also, the data should represent the risk level. In other words, focus on the order of magnitude of the PCA work and duration, not the fourth significant digit of the calculations.
Special circumstances may benefit from having multiple PCA resources. One large program required three PCA resources—one for each of the two major subcontractors and one for the internal staff. The PCA work effort was allocated to each of the PCA resources proportionate to that vendor's activity in each phase. This helped manage costs and helped each subcontractor anticipate resource requirements that included their planned contingency. Another example of needing multiple PCA resources is if there is a need to track contingency by resource skill set. Be wary of the trap exposed in the previous paragraph—that of making the precision gained too expensive to implement and administer by creating too many PCA resources.
Date-constrained projects (such as those that must meet regulatory dates or Year 2000 projects) have a unique characteristic. There may be recognition of a certain level of risk but insufficient calendar time to be feasible. This can be demonstrated by PCA tasks with significant work effort given the task's duration. The PCA technique allows the project manager a visible way to demonstrate the schedule risk.
Some organizations may manage two completion dates on projects. The “internal date” is calculated by setting the PCA durations to zero days, whereas the “external date” is calculated using the agreed-upon PCA durations.
Finally, large projects may use Monte Carlo analysis to identify more probable completion dates and costs. The PCA work effort and duration variables can be studied separately to determine their impact to the project's risk.
ONCE UNDERSTOOD, the PCA technique for risk identification and measurement will take a small percentage of project management time to implement and administer, but the benefits significantly outweigh the costs. ■
Bradford R. Eichhorn, senior manager with Claremont Technology Group, Inc., has 21 years in information technology experience, both as a consultant and in diverse industries.
PM Network • October 1997