grees of risk and uncertainty and to describe them 

 clearly so that decisions can be based on the best 

 available information. The planner should also sug- 

 gest adjustments in design to reflect various atti- 

 tudes of decisionmakers toward risk and uncertain- 

 ty. If the planner can identify in qualitative terms the 

 uncertainty inherent in important design, economic, 

 and environmental variables, these judgments can 

 be transformed into or assigned subjective probabil- 

 ity distributions. A formal model characterizing the 

 relationship of these and other relevant variables 

 may be used to transform such distributions to ex- 

 hibit the uncertainty in the final outcome, which 

 again is represented by a probability distribution. 



(2) At all stages of the planning process, the 

 planning can incorporate any changes in project 

 features that, as a result of information gained at 

 that stage, could lead to a reduction in risk and un- 

 certainty at a cost consistent with improvement in 

 project performance. 



(b) Some risk and uncertainty are assumed in 

 nearly every aspect of a water resources project. 

 Some types of risk and uncertainty are dealt with in 

 terms of national planning parameters — for exam- 

 ple, ranges of population projections and other prin- 

 cipal economic and demographic variables. Other 

 types of risk and uncertainty are dealt with in terms 

 of project or regional estimates and forecasts. 

 When projects are related to other projects and 

 programs in their risk and uncertainty aspects (e.g., 

 interrelated hydrologic systems), reasonable at- 

 tempts should be made to see that the same analy- 

 ses and presumed probability distributions are used 

 for all of them. 



(c) The risk and uncertainty aspects of projects 

 are likely to be seen and analyzed differently as 

 planning proceeds from rough screening to detailed 

 project proposals. An effort should be made, there- 

 fore, to relate the techniques used in characterizing 

 and dealing with risk and uncertainty to the stage of 

 the planning process. 



(d) The resources available for analyzing aspects 

 of risk and uncertainty should be allocated to those 

 assessments that appear to be the most important 

 in their effects on project and program design. 

 Rather than assuming in advance that one or an- 

 other variable is a more important source of risk 

 and uncertainty, the planner should make a thor- 

 ough effort to determine which variables will be 

 most useful in dealing with measurement errors and 

 natural sources of risk and uncertainty. 



(e) The aspects of project evaluation that can be 

 characterized by a probability distribution based on 

 reasonably firm data, such as hydrologic risk, can 

 be treated by standard methods of risk evaluation 

 developed by Federal agencies and others. 



(f) Most risk and uncertainty aspects of projects 

 cannot be characterized by probability distributions 

 based on well established empirical data. A first 

 step in dealing with this problem is to describe why 

 the project or specific aspects of it are uncertain, 

 as well as the time periods in which different de- 

 grees of uncertainty are likely. A range of reason- 

 ably likely outcomes can then be described by 

 using sensitivity analysis — the technique of varying 

 assumptions as to alternative economic, demo- 

 graphic, environmental, and other factors, and ex- 

 amining the effects of these varying assumptions 

 on outcomes of benefits and costs. In some cases 

 and in some stages of planning, this approach, 

 when accompanied by a careful description of the 

 dimensions of uncertainty, will be sufficient. It can 

 be accompanied by descriptions of design adjust- 

 ments representing various attitudes toward uncer- 

 tainty. 



(g) It may be appropriate in some cases to char- 

 acterize the range of outcomes with a set of sub- 

 jective probability estimates, but the project report 

 should make clear that the numerical estimates are 

 subjective. Moreover, subjective probability distribu- 

 tions should be chosen and justified case by case, 

 and some description of the impact on design of 

 other subjective distributions should be given. 

 Design alternatives reflecting various attitudes 

 toward uncertainty may be suggested. 



(h) Utility functions may be used in conjunction 

 with assessments of uncertainty to explore design 

 adaptations reflecting specific preferences. Public 

 preferences, if well known, may be used to illus- 

 trate to decisionmakers what the best design would 

 be, given the uncertainties and preferences in a 

 particular case. If public preferences are not well 

 known, justification could be given for the selection 

 of various utility functions, which can be used only 

 to illustrate the effects on design of various prefer- 

 ences. 



(i) At each level of analysis, the planner should 

 take into account the differences in risk and uncer- 

 tainty among project purposes and costs, among 

 various time periods, and among different stages of 

 planning. 



(j) Adjustments to risk and uncertainty in project 

 evaluation can be characterized as general or spe- 

 cific. General adjustments include the addition of a 

 premium rate to the interest, overestimation of 

 costs, underestimation of benefits, and limitations 

 on the period of analysis. Such general adjustments 

 are usually inappropriate for public investment deci- 

 sions because they tend to obscure the different 

 degrees of uncertainty in different aspects of pro- 

 jects and programs. Specific adjustments — includ- 

 ing explicit assessments of different degrees of risk 

 and uncertainty in specific aspects of a project or 



16 



