Risk in Cost-Benefit Analysis

• If cost-benefit analysis is to assist in the decision-making

process the analysis must be conducted in advance of the
project being undertaken.
• This means that the value of the none of the variables
involved in the analysis can be observed, but rather has to
be predicted.
• Risk and uncertainty are always associated with predictions
about the future and should be taken into account in the
cost-benefit analysis.
• Since risk and uncertainty impose costs on decision-makers,
these costs need to be assessed and measures taken to
reduce them if possible.
• One way of reducing risk and uncertainty is to acquire 1
additional information.
 Risk in Cost-Benefit Analysis
Risk and Uncertainty
• In the context of project evaluation where the analyst
is undertaking an ex post assessment of what has
already occurred, it is clearly an unrealistic
assumption to make where the purpose of the
analysis is to undertake an appraisal of proposed
projects where one has to forecast future cost and
benefit flows.
• The future is uncertain.
• Uncertainty arises either because of factors internal
to the project or because of factors external to the
project.
2
 Risk in Cost-Benefit Analysis
• Uncertainty implies that there is more than one possible
value for any project’s annual net benefits.
• The range of possible values a variable can take may vary
considerably from one situation to another, with the two
extremes being complete certainty, where there is one
known value, to complete uncertainty where the variable
could take on any value.
• Most situations lie some-where between the two extremes.
• where it is believed that the range of possible values could
have a significant impact on the project’s profitability, a
decision about the project will involve taking risk.
3
 Risk in Cost-Benefit Analysis
• In most situations there will be some information on which
to base an assessment of the probability of possible
outcomes or values within the feasible range.
• If it is possible to estimate the probability or likelihood of
the values that an outcome could take, the degree of risk
can be quantified.
• In some situations the degree of risk can be objectively
determined, for instance, when flipping a fair coin; there is
a 50 % chance it will be leads and a 50 % change it will be
tails, about which there can be no disagreement.
• Subjective risk characterizes most situations we will
encounter in project appraisal.

4
 Risk in Cost-Benefit Analysis
• There will be some information which can be used to
assign numerical probabilities, which can in turn be used to
undertake a more rigorous analysis of possible project
outcome.
• As new information comes to hand we may revise our
probability estimates accordingly.
• The assessment of probabilities is one sense in which risk
analysis is subjective.
• It is also the case that the extent to which an individual
considers a situation to involve significant cost of risk will
vary depending on the personal circumstances and
attitude of the individual in question.
5
 Risk in Cost-Benefit Analysis

• For instance, poor and rich individuals are likely to have

different perceptions of the significance of loosing Rs.
5,000 on a wager.
• In risk analysis the following three forms of subjectivity
need to be considered;
1. Deciding what the degree of uncertainty or Probability
of an occurrence is.
2. Deciding whether or not the uncertainty constitutes a
significant risk.
3. Deciding whether the risk is acceptable or not.

6
 Risk in Cost-Benefit Analysis

• There are various ways in which the project

analyst can incorporate the analysis of risk
and uncertainty into the cost-benefit analysis.
• We consider the two most commonly used
approaches;

i) Sensitivity Analysis, and

ii) Risk Modeling.

7
 Risk in Cost-Benefit Analysis
Sensitivity Analysis
• Sensitivity Analysis describes the simple process of
establishing the extent to which the outcome of the
cost-benefit analysis is sensitive to the assumed
values of the inputs used in the analysis.
• For calculating the NPV of a project over a range, we
used discount rates.
• In this case, we should test that how sensitive the NPV
of the project is to the choice of discount rate.
• This kind of information can be very useful for the
decision-maker in various ways.
8
 Risk in Cost-Benefit Analysis

• Performing a sensitivity analysis by varying only the

discount rate constitutes one commonly used method of
taking risk into account.
• For example, we use a risk premium on the discount
rate.
• If a project is believed to be more risky than usual, the
analyst uses a higher discount rate rather than that what
would be used in the absence of risk.
• The premium representing the analyst’s perception of
the degree of riskiness of the investment.
• As a higher discount rate implies a lower NPV, ceteris
paribus.
9
 Risk in Cost-Benefit Analysis
• It will be more difficult for the project to pass the NPV
decision criterion the higher is the risk premium.
• Providing the decision-maker with NPVs for a given
project over a range of discount rates provide the
information necessary to assess the significance of
applying a risk premium and deciding whether the
project is marginal or not.
• Let us assume that the construction of a road the
engineers have indicated that the actual cost could
vary by up to 25 % above or below their “best guess”
estimate.
10
 Risk in Cost-Benefit Analysis
• In this case the analyst may wish to calculate the
present value of the road’s cost or net benefits at each
value, i.e., at 75 %, 100 % and 125 % of the best
guess vale.
• The decision-maker is then provided with a range of
possible outcomes corresponding to the range of
possible input values.
• It is assume that there is uncertainty about both the
construction costs and the future usage of the road.
• Here we used three estimates of future usage: low,
medium and high, each corresponding to a different
level of annual benefit. 11
 Risk in Cost-Benefit Analysis

• The analyst can again test the significance of

changes in this variable independently of others
by calculating the NPV for each level of road
usage, or in conjunction with others.
• For example by allowing both the capital costs
and the road usage levels to vary simultaneously
over their respective ranges.

12
 Risk in Cost-Benefit Analysis

Construction Cost

75 % 100 % 125 %

Road $ 30
High $ 50 $ 40
Usage
Benefits
$47 $ 36 $ 25
Medium

$ 43 $ 32 $ 20
Low

Sensitivity Analysis Results: NPVs for Hypothetical Road Project

($ ‘000 at 10 % discount rate)
13
 Risk in Cost-Benefit Analysis

• If we hold the level of road usage at its ‘best guess’

or ‘medium’ level, the NPV varies from $ 25 to $ 47
thousand.
• While, if road construction costs are held at their ‘best
guess’ level (100 %), the NPV varies from $ 32 to $
40 thousand.
• The outcome is more sensitive to changes in
assumed construction cost values than to road usage
values, across their possible ranges as specified in
the earlier table.

14
 Risk in Cost-Benefit Analysis
• When we allow both input values to vary
simultaneously, then the NPV can vary from a minimum
value of $ 20 thousand (the most pessimistic scenario
when road usage is assumed to be at its lowest and
construction costs are at their highest level) to $ 50
thousand (where road usage is at its highest and
construction costs are at their lowest level).
• This information may provide the decision-maker with
sufficient information to incorporate risk in the decision-
making process, provided that it is satisfied that these
variables can be considered ‘risky’ and even in the
worst case scenario, the project’s NPV is still positive at
10 %. 15
 Risk in Cost-Benefit Analysis

• One important issue concerns possible correlation

between variations in levels of uncertain outputs and
inputs.
• Sensitivity analysis assumes that the variables in
question are all independent of one another.
• The variations in some variables will be closely
correlated with variations in others, such as, road
usage and road maintenance costs.
• Allowing these to move in opposite directions could
produce nonsensical results in the sensitivity
analysis.
16
 Risk in Cost-Benefit Analysis

• For above reason, sensitivity analysis should be used

with care and discretion.
• It can be useful first stage in determining;
1. Whether or not the project outcome is sensitive to the values of
the variables used in the analysis, and
2. To which variables the project outcome appears to be most
sensitive.
• This preliminary stage can be useful for the analyst in deciding
whether or not a more rigorous form of risk analysis is worth
undertaking, and, if so, which forecasted variables are to be
investigated further with a view to their inclusion in the more
formal risk modeling exercise.
17
 Risk in Cost-Benefit Analysis

Risk Modeling
• The preliminary steps in risk analysis allow the
analyst to determine whether or not the project
outcome under consideration can be considered risky
and, it so, whether a more detailed and rigorous form
of risk analysis can be justified.
• Assuming the decision has been made to proceed
further, the next stage is to identify and describe the
nature of the uncertainty surrounding the project
variables.
• For this, we use probability distribution.
18
 Risk in Cost-Benefit Analysis
• A probability distribution takes the description of
uncertainty one level beyond that which we used in the
sensitivity analysis.
• There we described a variable’s uncertainty purely in
terms of the range of possible values, i.e., high,
medium and low or maximum, mean and minimum.
• A probability distribution does this, but it also describes
the likelihood of occurrence of values within the given
range.
• When only a finite number of values can occur the
probability distribution is described as discrete and
when any value within the range can occur it is
continuous. 19
 Risk in Cost-Benefit Analysis

• An example of the discrete case is the probabilities of

heads or tails in the toss of a coin are as follows.

Road Construction Cost (C) Probability (P)

Low $ 50 20 %

Best Guess $ 100 60 %

High $ 125 20 %

A Discrete Probability Distribution of Road Construction Cost ($ ‘000)

20
 Risk in Cost-Benefit Analysis
• Once a probability distribution for a project input variable is
given the analyst is able to calculate the expected value of
the variable and to use this in the cash flow for the project
rather than the point estimate that we would have used if
uncertainty had been ignored.
Road Construction Probability (P) E(C)=PxC NPV E(NPV)
Cost (C)

Low $ 50 20 % $ 10 $ 86 $ 17.2

Best Guess $ 100 60 % $ 60 $ 36 $ 21.6

High $ 125 20 % $ 25 $ 11 $ 2.2

Expected value from a Discrete Probability Distribution ($ ‘000)

21
 Risk in Cost-Benefit Analysis
• The expected cost of road construction can be derived as
E(C) = $10 + $60 + $25 = $95.
• Rather than using point (best guess) estimate of $100
thousand, the analyst would use $95 thousand in the cash
flow for the project.
• It will also be possible to derive a probability distribution for
the NPV of the project.
• If we assume for simplicity that this is the only uncertain
variable, and the net benefits exclusive of road construction
costs are estimated probability of NPV = $ 86 thousand; 60
% of $36 thousand; and 20 % of $11 thousand.
E(NPV) = (17.2 + 21.6 + 2.2) thousand = $41 thousand.
22