6
Contingency

INTRODUCTION

In discussions of risk, the term “contingency” is often understood to be a number added to an estimate for project costs or durations to cover some element of risk or uncertainty. Owners establish contingency levels for each project based on acceptable risk, degree of uncertainty, and the desired confidence levels for meeting baseline requirements. When used to absorb the impacts of project uncertainty, the contingency is a form of risk mitigation, and so in evaluating potential project contingency funding, owners should apply risk assessment and probabilistic estimating techniques. However, contingency should not be a first alternative and should be used only as part of a complete risk mitigation effort.

DEFINITION

The dictionary definition of contingency is as follows:

Contingency (1): the condition that something may or may not occur: the condition of being subject to chance (2): the happening of anything by chance: fortuitousness…

a: something that is contingent: an event or condition occurring by chance and without intent, viewed as possible or eventually probable, or depending on uncertain occurrences or coincidences…b: a possible future event or condition or an unforeseen occurrence that may necessitate special measures <a reserve fund for contingencies>…c: something liable to



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 52
The Owner’s Role in Project Rick Management 6 Contingency INTRODUCTION In discussions of risk, the term “contingency” is often understood to be a number added to an estimate for project costs or durations to cover some element of risk or uncertainty. Owners establish contingency levels for each project based on acceptable risk, degree of uncertainty, and the desired confidence levels for meeting baseline requirements. When used to absorb the impacts of project uncertainty, the contingency is a form of risk mitigation, and so in evaluating potential project contingency funding, owners should apply risk assessment and probabilistic estimating techniques. However, contingency should not be a first alternative and should be used only as part of a complete risk mitigation effort. DEFINITION The dictionary definition of contingency is as follows: Contingency (1): the condition that something may or may not occur: the condition of being subject to chance (2): the happening of anything by chance: fortuitousness… a: something that is contingent: an event or condition occurring by chance and without intent, viewed as possible or eventually probable, or depending on uncertain occurrences or coincidences…b: a possible future event or condition or an unforeseen occurrence that may necessitate special measures <a reserve fund for contingencies>…c: something liable to

OCR for page 52
The Owner’s Role in Project Rick Management happen as a chance feature or accompaniment of something else. (Webster’s Third New International Dictionary) From this dictionary definition it might be supposed that the purpose of a contingency is to cover “the happening of anything by chance…uncertain occurrences or coincidences…[or] an unforeseen occurrence,” and therefore that the contingency would be expended only if the “unforeseen occurrence” actually occurred. In that case, the actual expenditure of the contingency would itself be “an event occurring by chance,” and one would therefore not expect the contingency to be used up in the normal course of activities. However, the term “contingency” is not understood in this way by many project personnel; rather, it is often taken to refer to funds that will be completely expended in the course of the project, no matter what happens. Accounting for Random Error Versus Systematic Error An example may help in illustrating the point. Suppose that someone tabulates the number of valves used on a substantial set of previous similar projects and finds that, in every case, the number of valves actually installed was always 1.17 times the number of valves counted on the engineering drawings by the quantity surveyor. Then a rule might be to purchase 17 percent more valves than the number of valves counted on the engineering drawings. This 17 percent factor is not a contingency in the sense of the dictionary definition, because there is nothing about it due to chance. Rather, it is an empirical fact that the valves are consistently undercounted, and the 17 percent factor is to offset this inherent bias or systematic error. Therefore, the additional 17 percent for the valves not counted would always be spent. (Note that the rule does not say to add 17 percent to the preliminary cost estimate, because this estimate might already have some adjustment factor in it.) Suppose now that the above example is slightly different: The investigator determines from the historical data that the factor relating actual valves used to valves counted on drawings is a random number, with average or mean 1.17 and standard deviation 0.10. While the systematic error or bias is still 17 percent, there is now a random component as well, which depends on chance events and hence is contingent. To account for both the systematic error and the random error, we have to add 17 percent to the number of valves actually counted, and then add another contingency to account for chance. When talking about chance, or contingency by the above definition, we must make probabilistic statements. If we want to be sure that we order enough valves 95 percent of the time, then in this instance we need

OCR for page 52
The Owner’s Role in Project Rick Management TABLE 6-1 Contingency Factors for Systematic and Random Errors Valves counted on drawings 1,000 Valves added to correct for undercount (systematic error) 170 Valves added to reduce the probability of running out of valves to 5% or less 164 Total number of valves to be ordered 1,334 Number of valves expected to be used 1,170 Number of valves expected to be left over 164 to add 33 percent to the number counted (17 percent for the systematic bias and 1.645 times the standard deviation for the chance variation). Note that we would expect to have 16 percent of the counted valves left over, as these were for protection against running out and were expected to be used only for extraordinary circumstances. The results are shown in Table 6-1. The problem is that the term “contingency” is often used for both the amount needed to cover the systematic error and the amount needed to buffer against the risk attributable to chance. This usage can cause confusion: Some project personnel, assuming the contingency addresses systematic error or bias, expect that they are entitled to use it all up, whereas others, who believe it is allocated to cover random error or chance, expect that contingency funds will be left over and ought to be returned to the project sponsor. Not surprisingly, those who expect to expend all contingency funds tend to be project managers, and those who expect to see at least some of the contingency allowance returned tend to be owners. In practice, relatively few projects return leftover contingency funds to the sponsor. In general, project directors regard contingency funds as theirs to use; if the risks fail to materialize, the funds are expended on something else, such as project improvements. Thus there is no agreement either on (1) how large a contingency should be or (2) the basic point of whether the contingency is an unassigned cost that is intended to be spent or a reserve that is intended not to be spent. Management Reserve The term “management reserve” is used sometimes as a synonym of contingency and sometimes in distinction to it. The Electronic Industries Alliance (EIA) standard Earned Value Management Systems, EIA-748 (EIA, 1998) does not use the term contingency, but does define “budget” and “management reserve” as follows:

OCR for page 52
The Owner’s Role in Project Rick Management BUDGET AT COMPLETION: The total authorized budget for accomplishing the program scope of work. It is equal to the sum of all allocated budgets plus any undistributed budget. (Management reserve is not included.) (Emphasis added.) MANAGEMENT RESERVE: An amount of the total budget withheld for management control purposes rather than being designated for the accomplishment of a specific task or set of tasks. (EIA, 1998, p. 6) Presumably, by these definitions, contingency, if there is any, would be included in either the “allocated budgets” or the “undistributed budget,” both of which are included in the budget at completion and hence constitute something that is expected or intended to be spent. Management reserve, by contrast, is not included in the budget and therefore presumably is not expected or intended to be expended. SETTING THE CONTINGENCY Setting the contingency is a matter of some tension between the project director, who generally wants the contingency allowance set high to permit greater flexibility and protection from uncertainty, and the owner, who wants the contingency set low to maintain greater control over the project. Although the proper amount of contingency is debatable, if the contingency is set too low upper management will be in the position of micromanaging the project, and if it is set too high management may not be sufficiently involved. This point will vary with different organizations and different types of projects. However, if upper management takes the view that there are always uncertainties associated with estimating and executing projects and that competent people are hired to manage these activities, contingency can be set at a level that keeps upper management informed and involved but does not require repeated approvals for additional funding. A probabilistic view of risk can help to guide this approach. If a probability distribution that represents uncertainty in the cost or duration of an activity is assumed, contingency can then be viewed as an amount of money (or time, in the case of project schedules) added to the mean (or expected value) of the cost (or duration) of that activity. The risk of overrun is thus the probability that the actual cost or duration would be greater than the mean plus the contingency. If the project cost is considered a random variable with an associated mean value, the deterministic or single-point value that is typically quoted as the project or activity estimate may not bear any relationship to the mean or expected value of its probability distribution. The point estimate might be the mode of the distribution, or the median, but most commonly will not be based on the

OCR for page 52
The Owner’s Role in Project Rick Management probability distribution because it may already include an allowance for contingency inserted at some lower level. See Box 6-1 for a discussion of the fallacy of point estimates. If the cost estimates for all the work packages in a project are known to be the mean values of their individual distributions, then the mean value of the total project would be the sum of these values and the contingency could be added to it. However, individual work package estimates are seldom the mean values of their distributions. As a result, the sum of these estimates is not the mean value of the total cost or duration but a BOX 6-1 The Fallacy of Point Estimates The fallacy of point estimates has been immortalized by the legendary statistician who drowned while fording a river that was on average 3 feet deep. The fallacy of point estimates provides some insight on why projects are frequently late. Consider a project consisting of 10 parallel tasks, each of average duration 2 weeks. Many project managers are under the misconception that the average completion time of the project is therefore also 2 weeks. However, the project will finish in 2 weeks only if each of the 10 tasks finishes in average or below average time. Assuming independent symmetric distributions for each task, the chance of this is less than 1 in 1,000. The average of the maximum of 10 durations is greater than the maximum of the average of the 10 durations. This concept also provides insight on why projects are frequently over budget. Consider the example of a laboratory that must inventory cases of a perishable chemical, demand for which is uncertain but averages five cases per month. Accordingly, the lab plans to stock five cases. The cost of maintaining the inventory has two components: (1) If at the end of the month the demand has been less than the number stocked, the use-by date of the excess cases will have expired and they must be destroyed, for a loss of $50 per unit, and (2) if the demand is greater than the number stocked, the lacking units must be air freighted at an additional cost of $150 per unit. The additive cost associated with the average demand of five is zero, so most managers think this is the average cost. But the cost of five is the minimum cost. The actual cost is greater than the cost of the average demand. SOURCE: Adapted from Savage (2002).

OCR for page 52
The Owner’s Role in Project Rick Management value that includes some degree of uncertainty (i.e., some multiple of the standard deviation or variance). Therefore, summing the estimates for all work packages without the discipline of using their mean values will result in a total cost estimate that includes unknown uncertainty. For these reasons, work package estimates, even at the lowest level, often contain some contingency factor. First, there is built-in bias from the estimator, who may expect to be criticized more severely for an underestimate than for an overestimate. Second, there is a need to cover errors in estimating, which are usually errors of omission; for example, in taking quantities off drawings, a quantity estimator is more likely to undercount than to overcount. Even if a computer does the quantity estimates, some items may be missing from the drawings and others may be added later, all contributing to a bias toward undercounting. The addition of a contingency offsets this bias. Finally, the field construction operation will certainly take the position that having material left over due to overestimation is preferable to running out before the job is finished due to underestimation. For these and other reasons, estimators tend to add some contingency or safety factor to each work package estimate. For similar reasons, the next level upward typically also adds a contingency. Each person or organizational level that adds a contingency does so to provide protection from uncertainty, the consequences of which are generally considered to be more dire if the number is underestimated rather than overestimated. If each management level wishes to set the risk-adjusted cost estimate (estimate plus contingency) at some desired point on the underlying probability distribution, each manager must have some idea of the amount of contingency already incorporated in the estimates. For example, if a manager wishes to set the risk-adjusted cost estimate at approximately the 80th percentile and believes that the estimate is at the 50th percentile, then the difference in dollars (or time) between the 50th and the 80th percentiles must be added. However, if the manager believes the estimate is already at the 70th percentile, less will need to be added. This process does not require exact knowledge of what the lower level contingency or risk is, only that the manager is familiar with the basis for the estimation. Conversely, if a manager believes that the estimates have already been adjusted up to the 90th percentile, then the estimate can be reduced. This may also occur if the manager’s knowledge of the business results in a better (or different) view of the underlying uncertainties than others have. The manager may desire a risk-adjusted cost estimate at the 80th percentile and the estimates provided may have been developed at that level. However, the subjective probability distributions used by the estimators and manager may not agree as to where the 80th percentile is. Or a contractor may feel that regardless of the desirable level of contingency or safety factor, competitive conditions do not permit it—that is, although

OCR for page 52
The Owner’s Role in Project Rick Management there is a definite risk in setting contingencies too low, there is a greater risk in setting them too high if one has to bid for a fixed-price contract or receive financial authorization from the owner. In general, as estimates flow upward through different levels of organizations, higher management levels will have better knowledge of strategic business or political conditions and so may make different decisions about contingencies than their subordinates. It is likely that knowledgeable owners can make better decisions about contingencies than owners who are inexperienced or who do not make the effort to become knowledgeable about the costs and duration of the projects they typically undertake. PROJECT POLICIES AND PROCEDURES Project policies and procedures documents should address the different kinds of contingencies, the need for contingency allowances, who controls them, and what should happen to them if they are not expended. They should also include precise and consistent definitions for terminology. The term “contingency” in particular needs a consistent definition, as it means different things to different people. Equally important are discussions and examples of approaches to setting budget and schedule contingencies. Contingency is not like value engineering, change control, or other cost control methods: It is an allowance for error or a safety margin on budget overruns that does not reduce costs or risks but increases the budget. Thus, by itself, contingency is not a cost control method, as its purpose is to ensure adequate funds to pay for uncontrolled costs. The definition of contingency as a percentage of the estimated cost to complete a project, instead of a percentage of the original estimate, is an improvement, but it is a change from past practices in many cases. Project contingencies should be reported and reviewed in a consistent way that should be defined and emphasized in policy documents in order to achieve consistency across all projects. Consistency does not, however, mean the establishment of recommended or standard values or ranges for overall contingency allowances. The use of established values is a questionable practice as it encourages project managers to use these values instead of performing project-specific risk assessments. For example, a contingency percentage that might be adequate for some conventional infrastructure projects will be totally inappropriate for big science projects, waste remediation projects, and one-of-a-kind or first-of-a-kind projects, for which the technology may be new and unproven or the volume and characterization of the wastes uncertain, and which may need to retain much larger contingencies even at the final design stage.

OCR for page 52
The Owner’s Role in Project Rick Management There are at least two distinct policy issues in setting a contingency. First, one purpose of a contingency is to provide an allowance for unknowns in making estimates. Because these errors are predominantly errors of omission, some allowance must be added to cover them. Instead of counting all the valves, for example, one estimates them, and adds an allowance as a contingency. This type of contingency is bottom-up, i.e., it is estimated at the work package or activity level. Adding all these work package contingencies is not, however, supported by statistical analysis and can easily result in a very large number, which then becomes the project budget. Second, another purpose of contingency is risk mitigation, which is required not because of omissions in making estimates or any other uncertainty at the activity level but rather to allow for unknowns at the overall project level. For example, a capital acquisition project that is really a research and development project may depend on new technology, which requires more project-level contingency; or a waste remediation project may need a contingency to cover the possibility that the in situ waste may differ from the original characterization. These are not activity-level contingencies. This distinction is not merely semantic; there is an important difference in how such risks are estimated, and project policies and procedures should make this distinction clear. Moreover, there is a difference in how these different contingency factors are managed. If the project’s baseline budget includes those costs that are known and countable, with a separate allowance for the unknowns in estimating these costs, then one expects that over the life of the project all or most of this allowance will be transferred to the budget, as these actual quantities and costs are identified. But a project contingency or management reserve may cover risk factors that would have a very high impact if they occurred but that are highly unlikely to occur. Thus, for example, if the contingency allowance for a possible flood is not used because no flood occurs, then this contingency allowance should not be automatically transferred to the base budget to cover overruns in other areas. Who owns the contingency and what should happen to it if it is not expended are important issues that should be addressed in the policies and procedures documents. It can be argued that management reserves for high-impact, low-probability events should be held at the program level, not at the project level. At a minimum, these program-level contingencies alert management that there are large risks inherent in projects. Contingencies are known as risk funds in some organizations and are separated from all other budget funds unless needed for a specific project event. If a contract is for a fixed price, the contractor owns the contingency within the bid price and is entitled to it if there is any left at the end. But this is not the case with cost-plus contracts. Even with a fixed-price con-

OCR for page 52
The Owner’s Role in Project Rick Management tract, the owner needs to hold some contingency to cover potential change orders. Contingency policies need to distinguish between fixed-price, cost-plus, cost plus incentive fee, and other common types of contracts in the discussion of risk and contingency and should state whether contingency is controlled at the project director level or at the owner’s program level.