Reprinted with the permission of AACE

International
1265 Suncrest Town Centre Dr.,
2010 AACE INTERNATIONAL TRANSACTIONS Morgantown, WV 26505 USA
Phone 800-858-COST/304-296-8444
RISK.01 Internet: http://www.aacei.org
E-mail: [email protected]

Alternate Methods for Integrated Cost & Schedule

Contingency Estimating
John K. Hollmann, PE CCE CEP
ABSTRACT—Schedule contingency estimating is not widely used in project planning. Even less
commonly used are methods that integrate cost and schedule contingency estimating. Only one
integrated method is well covered in the literature; i.e., range simulation applied to a cost-loaded CPM
schedule. However, this method, as typically practiced, is poorly aligned with some of AACE
International’s established principles for good contingency estimating practice. This paper presents the
most commonly used Critical Path Method (CPM) model-based methods, as well as an alternate hybrid
parametric model/expected-value method that does not rely on the CPM model and is designed to
better align with principles of good contingency estimating practice while being practical to use on all
Reprinted
projects. Pros and cons of the alternate with the are
methods permission of AACE
discussed. The paper also explains why cost and
International
schedule risk analysis must be done at the same time regardless of the contingency estimating method
1265 Suncrest Town Centre Dr.,
used. Morgantown, WV 26505 USA
Phone 800-858-COST/304-296-8444
Internet: http://www.aacei.org
E-mail: [email protected]

RISK.01.1
 2010 AACE INTERNATIONAL TRANSACTIONS
Table of Contents

Abstract …………………………………………………………………………………………………………………………………………..... 1
Table of Contents ………………………………………………………………………………………………………………………………. 2
List of Tables ………………………………………………………………………………………………………………………………………. 2
Introduction/Background………………………………………………………………………………………………………………….... 3
Assumptions Regarding This Paper’s Scope ………………………………………………………………………………………… 3
CPM-Based Activity Ranging With Cost Loading Method ………………………….………………………………………… 4
Parametric/Expected-Value Hybrid Method ………………………………………………………………………………………. 5
Method Comparison ………………………………………………………………………………………………………………………….. 6
Based on Principles of Practice ……………………………………………………………………………………………………… 6
Based on Strengths/Weaknesses ………………………………………………………………………………………………….. 8
Comparison Summary ………………………………………………………………………………………………………………………… 8
Why Cost and Schedule Risk Analysis Must Be Performed Together ………………………………………………….. 9
Conclusions …………………………………………………………………………………………………………………………………………10
References ………………………………………………………………………………………………………………………………………….10

List of Tables

Table 1—Integrated C/S Contingency Methods vs. General Principles ……………………………………………….. 7

Table 2—Integrated C/S Contingency Methods Strengths vs. Weaknesses ………………………………………… 8

RISK.01.2
 2010 AACE INTERNATIONAL TRANSACTIONS
Introduction/Background

Having consulted on capital project benchmarking and practice improvement for many years, I have
had the opportunity to observe the cost and schedule risk analysis and contingency estimating
practices used, and results obtained, by many global owner companies in the process related
industries. On the positive side, recognition of the need for better risk management processes and
methods is growing. However, the following are some cost/schedule risk analysis problems or gaps
observed:

• Few are doing schedule risk analysis or contingency estimating (further, many firms lack any
significant in-house planning and scheduling competency).
• Cost and schedule risk analysis and contingency estimating are rarely integrated.
• There is little awareness of AACE Recommended Practices or the landmark industry research behind
them.
• Project cost and schedule predictability seem to be getting worse (while the stakes are becoming
greater all the time).

Some of these gaps result from basic lacking in capital management leadership and/or cost engineering
competency (particularly scheduling) in the owner companies. However, some of the gaps result from
the industry as a whole having few effective, broadly applicable and practical tools for integrated
cost/schedule contingency estimating to chose from. As AACE International develops its Decision and
Risk Management Professional (DRMP) specialty certification, these and other gaps in the literature
and technology have become apparent to AACE International. The AACE International Technical
committees are working to close these gaps by defining recommended practices. In that light, I hope
that this paper will serve as a potential catalyst for a new AACE International Recommended Practice
(RP).

The paper reviews the most common integrated cost/schedule contingency estimating method (CPM-
based activity ranging with cost loading), presents an alternative new practice (parametric/expected-
value hybrid), and compares them. These two methods are then compared against AACE
International’s first principles yardstick for evaluating contingency estimating practices [1].

Assumptions Regarding This Paper’s Scope

This paper is intended for those considering which integrated cost/schedule contingency estimating
method to use on their projects. The paper further assumes that these readers already have an
intermediate level of knowledge of risk management processes, Monte-Carlo modeling, parametric
estimating, and estimating contingency based on probabilistic estimating outputs (i.e., the types of
things covered by AACE International Recommended Practices).

The scope of the methods described in the paper assumes that contingency estimating will be applied
in a risk management process. For example, such a process is described in AACE International’s Total
Cost Management (TCM) Framework Section 7.6 [2]. In the TCM process, contingency estimating is a
distinct step. Readers should note that the TCM process is the only industry risk management process
model that highlights contingency estimating. Most, as exemplified in the ISO risk management

RISK.01.3
 2010 AACE INTERNATIONAL TRANSACTIONS
process [3], do not effectively integrate risk quantification with project control planning. The TCM
process, like all risk management processes, starts with identifying risks and assessing and screening
them for the purpose of treating them (e.g., mitigate, transfer, etc). This paper does not focus on these
initial steps; it is focused on estimating contingency (i.e., time and costs allowances for uncertainty in
project plans) for the “residual” risks that remain after risk treatment. It also does not focus on the risk
workshop/analysis aspect of the contingency estimating process; workshop performance tends to be
similar for all the methods described here.

In summary, the paper is focused solely on the contingency quantification step.

CPM-Based Activity Ranging With Cost Loading Method

The purpose of the “Cost Loaded Critical Path Method (CPM)” class of methods is to leverage the
ability of modern scheduling tools to support resource-loading and Monte-Carlo modeling. The CPM-
based class of methods (which are the most commonly used although still not widely) is better
described in other papers and an AACE RP in draft at time of writing. [4,5,6]. However, in summary
(and at the risk of over simplification), the “Cost Loaded CPM” methods have the following steps:

1. Develop the “base” (contingency-free) project estimate and schedule. Optimally, these are
integrated using a common work breakdown structure (WBS). The schedule must be a high quality
CPM schedule model.
2. Resource load the schedule with all cost information. The costs include the entire project budget,
generally at a summary level (note that this is not resource-leveling; that cost/schedule integration
step should have been already done if it is a “quality” schedule). Resources that are time-
dependent (e.g., labor or equipment with a daily rate) are distinguished from those that are not
(e.g., materials).
3. Assess and enter the risk information in the schedule model (the schedule tool must support
Monte-Carlo modeling). There are several alternate methods for doing this:
a. Activity Ranging: Replace the cost-loaded activity duration with a probability distribution
(typically triangular or 3-point; low, most-likely, high) based on project team assessment of the
range (risks are not explicitly identified or linked to impacts in the model). Impacts on non-time
dependent costs are considered through traditional cost ranging.
b. Risk Driver: Starting with a list of residual risks (plus a general “background” risk), link the risk to
activity-by-activity time multipliers (factors) and also cost factors for non-time dependent costs.
Then, assess and input the probability of occurrence (e.g., 50%) of each risk, and assess and
input ranges on the activity and cost multiplier factors (again, typically triangular or 3-point;
e.g., 0.8, 1.0, 1.2).
4. Run the Monte-Carlo model and determine the cost and schedule contingency based on
management’s tolerance for risk and the model’s probabilistic output.

There are a number of potential variations from the typical approach described above. These options
include (but are not limited to):

• Risk Driver (option 3b above): The use of this method is increasing since commercial tools to
support it were introduced circa 2005.

RISK.01.4
 2010 AACE INTERNATIONAL TRANSACTIONS
• Conditional Branching: What if a risk event happens which would cause the team to change the
schedule logic? To address this common situation, conditional branching (conditioned by
occurrence of the risk), could be incorporated in the schedule model. However, the author has
never seen or read of this being done in regular practice (i.e., “static logic”, unaffected by risk
events, is almost always assumed).
• Parametric Modeling: the typical method requires the team to directly input impact distributions
(e.g., 3-point). However, these distributions could be based on parametric models that tie project
attributes with cost and schedule impacts. This is particularly relevant for the concept of
“background” or systemic risks. Again, the author has never seen or read of this being integrated
with the Cost-Loaded CPM method.

Anyone wanting to implement any of the method variations described above should refer to the paper
references for “how-to” information. Hopefully, the brief summary above will be sufficient to support
the alternate method comparison that is a purpose of this paper.

Parametric/Expected-Value Hybrid Method

This class of methods integrates separate contingency estimating practices that have been
documented in the literature elsewhere [7]. The purpose of a hybrid method is to leverage the
strengths of each component method, including parametric estimating [8] and expected value [9]. It
does not directly use the CPM schedule; this is both a strength and a weakness as will be discussed
later. This paper is the first published description of the hybrid approach.

To understand the driver and need for this method, the user must understand the concept of risk
breakdown and the difference between “systemic” and “project-specific” risk types. Systemic risks are
those that have systematically predictable relationships to overall project cost and schedule growth
outcomes. The term systemic implies that the risk is an artifact of the project “system”, culture,
business strategy, process system complexity, technology, and so on. Systemic risks are dominant for
poorly defined projects, and their impact is not readily quantifiable by traditional risk analysis. Project-
Specific risks are those that do not have predictable, systematic relationships with outcomes; i.e., they
are specific to the project [8]. These risks are amenable to traditional risk analysis. There is no single
contingency estimating method that works well with both of these risk types.

The hybrid parametric/expected value method has the following steps:

1. Develop the “base” (contingency-free) project estimate and schedule. The estimate and schedule
can be of any level of detail and quality (the plan detail and quality attributes are in fact systemic
risks specifically addressed by the method). A CPM schedule model is not required.
2. Assess and rate/quantify the systemic risks and enter them in an empirically-based parametric cost
and schedule contingency estimating model.
a. If the estimate and schedule are AACE International Class 5 [10], the analysis is complete--
determine the cost and schedule contingency based on management’s tolerance for risk and
the parametric model’s probabilistic output. No Monte-Carlo is required.
b. If the estimate and schedule are AACE International Class 4 or better, continue to the project-
specific method in the next step

RISK.01.5
 2010 AACE INTERNATIONAL TRANSACTIONS
3. Develop a list of significant residual project-specific risks (i.e., excluding systemic risks as addressed
in your parametric model).
4. Assess and input the probability of occurrence (e.g., 50 percent) of each risk
a. This can be treated as a distribution depending on the team’s confidence in their assessment
(i.e., treat the probability as uncertain).
5. Determine and document the assumed risk response (or range of responses) most likely to be
taken if the risk occurs (e.g., let things ride, take some defined corrective action, etc.).
6. Assess and input ranges on the cost and on the schedule impacts (typically triangular or 3-point) for
the risk response(s) anticipated.
a. For schedule, identify the activity or group of activities impacted. The criticality of these
activities, and the extent that they are parallel with activities impacted by other risks must be
approximately rated (analyst’s knowledge is relied on if there is no CPM schedule).
7. Calculate the “expected value” of the cost and schedule impact of each risk (probability times
impact). If the result is unacceptable (e.g., client will not allow any schedule slip), return to step 6
and consider revised risk responses (e.g., take corrective actions to avoid any schedule impact, but
with increased cost impact).
8. To integrate the systemic and project-specific model results, enter the cost and schedule
probabilistic outputs of the Parametric model as the first risk in the expected value tool. The
parametric tool will have provided cost and schedule distributions that can be used as pre-defined
or custom distributions (with probability of 100 percent) in the expected value tool.
9. Run the Monte-Carlo model and determine the overall cost and schedule contingency based on
management’s tolerance for risk and the model’s probabilistic output.
a. Because the parametric model output is an input to the expected value model, the model
integrates all contingency risks in a single cost and schedule output.
b. Use the assumed risk responses for risk management during execution (i.e., when a risk occurs
during execution, you will have already noted your expected response).

Method Comparison

Based on Principles of Practice

AACE International RP 40R-08 (“Contingency Estimating: General Principles”) [1] provides an objective
basis or yardstick (i.e., principles) against which you can assess the suitability of any contingency
estimating method that you are considering. The principles apply to methods for estimating any kind of
risk funds or allowances including contingency, reserves, or schedule allowances. The RP also provides
a categorization framework or taxonomy for describing various methods (the methods themselves are
covered by other AACE International RPs). The following are the general principles, that any
methodology developed or selected for quantifying risk impact should address:

• Meet client objectives, expectations and requirements.

• Part of and facilitates an effective decision or risk management process (e.g., TCM).
• Fit-for-use.
• Starts with identifying the risk drivers with input from all appropriate parties.
• Methods clearly link risk drivers and cost/schedule outcomes.
• Avoids iatrogenic (self-inflicted) risks.

RISK.01.6
 2010 AACE INTERNATIONAL TRANSACTIONS
• Employs empiricism.
• Employs experience/competency. And,
• Provides probabilistic estimating results in a way the supports effective decision making and risk
management.

Methods that do not broadly respect the general principles above are not to be recommended for use.
Table 1 summarizes how the methods covered in this paper perform in consideration of the AACE
contingency estimating principles.

Integrated Cost/Schedule Contingency Estimating Methods

Cost Loaded CPM- Cost Loaded CPM- Hybrid Parametric/
First Principles
Activity Ranging Risk Driver Expected-Value
Whether a given method best meets objectives, expectations or
Meets client requirements must be determined prior to each application.
objectives and If an objective is to be applicable for all Can be used on any
requirements estimates and schedules, CPM methods are estimate or schedule
generally not applicable
Part of a risk and Process does not address risk response Considers ways that
decision (e.g., does not address changing logic in risk could impact
management response to risk; i.e. no risk to plan logic) plan logic
process
Requires a quality, fully cost-loaded CPM Can be used on any
Fit-for-use
schedule. Not applicable to Class 5. estimate or schedule
Systemic and dynamic (logic impact) risks Risk categorization is
Starts with
are not identified part of method, and
identifying risk
can deal with
drivers
alternate logic
Links risk drivers and No explicit linkage Linkages are explicit Linkages are explicit
cost/schedule except for systemic for all risk types
outcomes (background) risks
No clarity that it No clarity that it Schedule
addresses any given addresses systemic assessment is more
Avoids iatrogenic risk (understates risks (understate subjective and may
(self-inflicted) risks risks when scope risks when scope encourage
definition is poor) definition is poor) sloppiness. Requires
strong facilitation.
Generally requires the use of lessons Explicitly addressed
Employs empiricism learned, and/or validation or benchmarking in parametric model
using historical information (not explicit)
All methods require facilitator expertise Additional expertise
Employs experience
required to develop
/competency
parametric model
Provides All produce probabilistic outcomes
probabilistic
estimating results
Table 1—Integrated C/S Contingency Methods vs. General Principles
RISK.01.7
 2010 AACE INTERNATIONAL TRANSACTIONS
Based on Strengths/Weaknesses

A comparison of the methods on a strengths versus weaknesses basis is shown in table 2.

Method Strengths Weaknesses

• Explicit cost/schedule • No explicit risk-impact linkage
integration for time dependent • No empirical basis
costs • Requires quality CPM, cost-
Cost Loaded
• Encourages use of quality loaded schedule & competent
CPM with
planning and schedule methods scheduler
Activity
• Commercial software/many • Not applicable to Class 5
Ranging
users • Static logic/risk response
assumed
• Weak for systemic risks
• Explicit risk-impact linkage (but • No empirical basis
not for systemic risks) • Requires quality CPM, cost-
• Explicit cost/schedule loaded schedule/competent
Cost Loaded integration for time dependent scheduler
CPM with costs • Not applicable to Class 5
Risk Driver • Encourages use of quality • Static logic assumed
planning and schedule methods • Weak for systemic risks
• Commercial software/some
users
• Explicit risk-impact linkage No commercial software/few
•
• Empirical basis users
• Applicable to schedules of any
• Non-CPM schedule requires
Hybrid quality more intuitive assessment
Parametric/ • Applicable to all Class of plans
• Cost/schedule integration for
Expected- • Address logic/risk response time dependent costs not
Value scenarios explicit
• Does not encourage use of
quality planning and schedule
methods
Table 2—Integrated C/S Contingency Methods Strengths vs. Weaknesses

Comparison Summary

As a consultant, I recommend that companies implement and apply the “Hybrid” approach for their
standard practice. It is most broadly consistent with first-principles (risk-impact linkage, empiricism, fit-
for-use, etc.) and it is applicable on every project without exception (it is also inexpensive). For large,
turnaround, and other projects where significant team resources can be brought to bear for upfront
planning and scheduling, I recommend that teams apply the cost-loaded CPM with risk drivers method
(and test with hybrid); however, instead of an intuitive allowance for “background” risk, I would apply
an explicit parametric analysis of “systemic” risk to derive the background distribution (i.e., make it a
hybrid method).
RISK.01.8
 2010 AACE INTERNATIONAL TRANSACTIONS
Because the cost-loaded CPM with risk drivers method requires a high level of quality in planning and
CPM scheduling, it is simply not practical for day-to-day use at most companies. Unfortunately, the
quality of project schedules is very poor at the time that contingency is estimated. A research study by
IPA, Inc. of 494 project schedules from major global companies at the time of project authorization
showed that only 13 percent had CPM schedules with resource loading [11]. While a complex schedule
model is strength for project control, it is very difficult to use it for analyzing the risk of dynamic
logic/alternate risk responses. It also does not apply to Class 5 planning. Finally, it only addresses
systemic risk to the extent that it is made a hybrid with parametric methods. Therefore, I do not
recommend cost-loaded CPM with risk drivers as a basis for standard practice.

The usage of Cost-Loaded CPM with activity ranging (not risk driven) is difficult to defend in any
respect. Along with impracticality, it violates a vital first principle of risk management which is to start
with identifying the risk and using that risk knowledge. A paper by the author in 2007 reviews why cost
or schedule “ranging” that is not linked to risk drivers is a dangerous practice [7]. I cannot think of any
reason to recommend this approach (which is unfortunately the most common integrated
cost/schedule method in use today).

Why Cost and Schedule Risk Analysis Must Be Performed Together

Cost/schedule integration in contingency estimating is important because:

• most risks impact schedule (e.g., delays).

• many costs are time dependent (e.g., idle labor during a delay). And,
• risk response, and therefore impact, depends on the project’s cost and schedule objective (e.g.,
need to assess cost/schedule trade-offs of alternate risk responses).

Few would argue the point that most risk drivers affect both cost and schedule, and because of that,
cost and schedule risk analysis should be integrated. However, in my consulting I find very few projects
that assess cost and schedule risk at the same contingency estimating workshop, at the same time, by
the same team. There are many reasons for this behavior including tool differences, staff and
consultant specialization differences, different estimate and schedule progress, limited time for
meetings, etc. However, the result of separate sessions is always poor quality contingency estimates
and risk management in general.

The reason that cost and schedule risk must be evaluated together is that cost and schedule risk are
not independent, cost and schedule impacts depend on the assumed risk response, and there is a
cost/schedule trade-off in determining which response will best meet management objectives
(hopefully, management told the team whether the objective is cost or schedule driven—most often
they have not). For example, if the risk event being evaluated is a major rainstorm with flooding, the
risk response could be to a) slow down work and let the site dry out which means a long delay but with
minimal costs (stretched out indirects), or b) initiate a massive, costly pumping and dry out recovery
effort with little schedule impact (remember, this is a “residual” risk because we presumably decided
earlier not to mitigate through design). We can see that the team is going to have to analyze the
cost/schedule tradeoffs of these and other potential responses; this simply cannot be done effectively
in separate sessions.

RISK.01.9
 2010 AACE INTERNATIONAL TRANSACTIONS
Further, defining the cost and schedule impact for any particular risk requires an iterative process
because the team and management must come to consensus on whether cost or schedule is most
important in each risk instance. The typical iterative sequence of integrated cost/schedule risk analysis
is:

1. Assume a risk response.

2. Estimate cost impact based on the assumed risk response.
3. Estimate time impact based on the assumed risk response.
4. Determine if Management finds the cost and time impacts to be acceptable (most often one or the
other is not).
5. Re-assess the response and cost/schedule impacts (repeat from step 1). And,
6. Record response assumed in planning for use in later risk management during execution.

In short, it is nearly impossible to address alternate risk responses and cost/schedule tradeoff when
cost and schedule risks are analyzed in separate sessions, particularly with separate teams. It should
also be obvious that for effective risk analysis that considers alternate risk responses, both project and
business management representation must be in the room.

Conclusions

It is hoped that the reader has a clear understanding of why I recommend that companies implement
and apply a hybrid Parametric/Expected-Value approach as their standard practice. For large,
turnaround, and other projects that can muster high quality planning and scheduling, Cost-Loaded
CPM with Risk Drivers should be considered, but only when combined in a hybrid approach using a
parametric method for systemic or background risks. Further, cost and schedule risks must be
evaluated considering alternate risk responses and cost/schedule tradeoffs and that should take place
in one integrated risk analysis workshop supporting the contingency estimating process. This is a
robust approach as most likely to contribute to overall project system success.

My apologies to those that found the method descriptions to be overly simplified; I encourage
everyone to study the References for more specific “how-to” information.

References

1. AACE International RP 40R-08: Contingency Estimating: General Principles, AACE International,

Morgantown WV, (latest revision).
2. Hollmann, John K., Editor, Total Cost Management Framework: An Integrated Approach to
Portfolio, Program and Project Management, AACE International, Morgantown WV, 2006.
3. International Standards Organization: ISO 31000:2009 Risk Management─Principles and
Guidelines, 2009 (www.iso.org).
4. Patterson, Don, “Managing Project Cost Risk”, AACE International Transactions, 2006.
5. Mulambya, Emmanuel, “The Value Added by Linking the Project Risk Register to Quantitative
Schedule Risk Analysis”, AACE International Transactions, 2007.

RISK.01.10
 2010 AACE INTERNATIONAL TRANSACTIONS
6. Hulett, David “Integrated Cost-Schedule Risk Analysis”, AACE International, Morgantown WV, (This
is a draft AACE Recommended Practice at the time of writing; see www.projectrisk.com for related
materials by Mr. Hulett).
7. Hollmann, John K., “The Monte-Carlo Challenge: A Better Approach”, AACE International
Transactions, 2007.
8. AACE International RP 42R-08: Risk Analysis and Contingency Determination Using Parametric
Estimating, AACE International, Morgantown WV, (latest revision).
9. AACE International, RP 44R-08: Risk Analysis and Contingency Determination Using Expected Value,
AACE International, Morgantown WV, (latest revision).
10. AACE International RP 18R-97: Cost Estimate Classification System – As Applied in the Engineering,
Procurement, and Construction for the Process Industries, AACE International, Morgantown WV,
(latest revision).
11. Griffith, Andrew, “Scheduling Practices and Project Success”, Cost Engineering Journal, AACE
International, Morgantown WV, 2006.

John K. Hollmann, PE CCE CEP

Validation Estimating, LLC
[email protected]

RISK.01.11