Study guide
5.3 Decision Trees & Expected Monetary Value (EMV)
AACE Certified Cost Technician CCT
Track your progress as you read
Finished sections get a check mark.
1 Chapter 1: AACE Certification, Ethics & Total Cost Management Framework
1.1 About the AACE CCT Exam & Certification Pathway 1.2 AACE Canons of Ethics & Professional Practice 1.3 Total Cost Management (TCM) Framework Architecture 1.4 Cost Engineering Terminology & Core Concepts 1.5 Strategic Asset Management: Asset Performance, Assessment & Configuration Control
2 Chapter 2: Elements of Cost, Labor, Materials & Indirects
2.1 Cost Classification & Cost Behavior 2.2 Labor Costs, Craft Rates & Labor Burden 2.3 Materials & Equipment Cost Fundamentals 2.4 Job-Site Indirects & Home Office Overhead Allocation 2.5 Cost Dimensions: Life Cycles, Perspective, Influence & the Legal Dimension
3 Chapter 3: Engineering Economics: Time Value of Money & Depreciation
3.1 Time Value of Money & Cash Flow Diagrams 3.2 Uniform Series & Equivalence Factors 3.3 Nominal vs. Effective Interest Rates & Inflation 3.4 Depreciation Methods & Asset Write-Down
4 Chapter 4: Capital Budgeting, NPV, IRR & Life Cycle Costing
4.1 Net Present Value (NPV) & Present Worth Analysis 4.2 Internal Rate of Return (IRR) & MARR 4.3 Payback Period & Benefit-Cost Ratio Analysis 4.4 After-Tax Cash Flow Analysis & Life Cycle Costing
5 Chapter 5: Statistics, Probability & Decision Models
5.1 Descriptive Statistics & Measures of Dispersion 5.2 Probability Concepts & Distributions 5.3 Decision Trees & Expected Monetary Value (EMV) 5.4 Linear Programming & Optimization Models 5.5 Physical Measurement, Unit Systems & Quantity Conversions
6 Chapter 6: Organization, Communications & Project Management Basics
6.1 Organizational Structures & Project Roles 6.2 Communications Management & Project Reporting 6.3 Cost Data Management & Historical Normalization
7 Chapter 7: Cost Estimating Principles & Classification Systems
7.1 Estimating Fundamentals & Basis of Estimate 7.2 AACE 18R-97 Cost Estimate Classification System 7.3 Conceptual & Parametric Estimating Methods 7.4 Detailed Estimating & Estimate Closeout
8 Chapter 8: Allowances, Escalation, Location Factors & Contingency
8.1 Allowances vs. Contingency vs. Management Reserve 8.2 Location Factors & Escalation Analysis 8.3 Contingency Determination & Risk Analysis
9 Chapter 9: Project Planning, WBS & CPM Scheduling
9.1 Work Breakdown Structure (WBS) & Scope Definition 9.2 Activity Definition, Precedence & Dependencies 9.3 Critical Path Method (CPM) Forward & Backward Pass 9.4 Float Calculations & Schedule Compression 9.5 Requirements Elicitation, Plan Implementation & Plan Validation
10 Chapter 10: PERT Estimating, Progress Measurement & Schedule Control
10.1 PERT Three-Point Estimating & Network Variance 10.2 Progress Measurement & Rules of Credit 10.3 Schedule Baseline Maintenance & Change Control
11 Chapter 11: Earned Value Management (EVM): Variances, Indices & Forecasting
11.1 EVM Foundations & Core Parameters 11.2 Cost & Schedule Variances 11.3 Performance Indices 11.4 Forecasting: EAC, ETC, VAC & TCPI
12 Chapter 12: Procurement, Contract Types & Administration
12.1 Contract Types & Risk Allocation Spectrum 12.2 Procurement Process, Solicitation & Bidding 12.3 Surety Bonds & Project Insurance 12.4 Contract Administration, Changes & Claims
13 Chapter 13: Project Risk Management: Identification, Analysis & Response
13.1 Risk Management Planning & Identification 13.2 Qualitative & Quantitative Risk Analysis 13.3 Risk Response Strategies & Risk Control
14 Chapter 14: Quality Management, Value Engineering & Resource Management
14.1 Quality Concepts & Cost of Quality (COQ) 14.2 Quality Control Tools & Continuous Improvement 14.3 Value Engineering & Value Methodology 14.4 Resource Management, Leveling & Learning Curve Theory 14.5 Environmental, Health & Safety (EHS) Cost Management
5.3 Decision Trees & Expected Monetary Value (EMV)
Key Takeaways Explain This Topic AI
Decisions under risk involve known potential outcomes with quantifiable probabilities, whereas decisions under uncertainty lack assigned probability distributions.
Decision trees model sequential choices using square decision nodes (managerial control), circular chance nodes (probabilistic nature), and terminal branches with financial payoffs.
Expected Monetary Value (EMV) is calculated as ∑ (Pi × Payoff_i), and decision trees are resolved via the rollback (folding back) method from right to left.
The Expected Value of Perfect Information (EVPI) represents the absolute maximum expenditure justified for perfect information: EVPI = EVwPI - Maximum EMV without Information.
Risk attitudes dictate the shape of the decision-maker's utility function: risk-neutral practitioners exhibit linear utility, risk-averse practitioners display concave utility, and risk-seeking practitioners display convex utility.
Last updated: September 2026
5.3 Decision Trees & Expected Monetary Value (EMV)
Quick Summary: Capital project managers and cost technicians regularly confront complex, multi-stage choices involving substantial capital expenditure and significant risk—such as bidding on competitive tenders, choosing between equipment leasing vs. purchasing, or selecting self-performance over subcontracting. Decision tree analysis combines visual flowcharts with Expected Monetary Value (EMV) calculations to systematically evaluate sequential choices. By mastering the rollback method, calculating the Expected Value of Perfect Information (EVPI), and understanding utility theory across differing risk attitudes, practitioners ensure decisions maximize financial outcomes while respecting enterprise risk tolerance.
1. Decision-Making Contexts: Certainty, Risk, and Uncertainty
In managerial economics and cost engineering, decision contexts are classified based on the availability and reliability of outcome data: +-----------------------------------------------------------------------------------+
- | SPECTRUM OF DECISION ENVIRONMENTS |
- +-------------------+---------------------------------------------------------------+
- | Decision Under | Complete knowledge; every alternative leads deterministically |
- | Certainty | to a single known outcome (e.g., fixed-price treasury bond). |
- +-------------------+---------------------------------------------------------------+
- | Decision Under | Multiple potential outcomes are possible for each choice, but |
- | Risk | objective or subjective probabilities CAN be assigned. |
- +-------------------+---------------------------------------------------------------+
- | Decision Under | Multiple potential outcomes exist, but probabilities CANNOT |
- | Uncertainty | be credibly assigned (novel technologies, geopolitical shifts).|
- +-------------------+---------------------------------------------------------------+
- 2. Decision Criteria Under Strict Uncertainty
When a cost engineer cannot objectively or subjectively assign numerical probabilities to competing states of nature, four classical non-probabilistic decision criteria guide strategy selection:
1. The Maximin (Wald) Criterion (Pessimistic)
Philosophy: Assumes nature will deliver the worst possible outcome regardless of choice.
Method: Identify the minimum payoff for each alternative, then select the alternative that maximizes this minimum (the "best of the worst").
Application: Highly conservative, risk-averse contractors seeking capital survival during severe recessions.
2. The Maximax Criterion (Optimistic)
Philosophy: Assumes nature will deliver the best possible outcome.
Method: Identify the maximum potential payoff for each alternative, then select the alternative that yields the highest maximum (the "best of the best").
Application: Aggressive, speculative commercial developers or venture capital contractors.
3. The Hurwicz (Criterion of Realism)
Philosophy: Strikes a weighted compromise between pure optimism and pure pessimism using an explicit coefficient of optimism ( α \alpha α , where 0 ≤ α ≤ 1 0 \le \alpha \le 1 0 ≤ α ≤ 1 ): Hurwicz Payoff = α × ( Maximum Payoff ) + ( 1 − α ) × ( Minimum Payoff ) \text{Hurwicz Payoff} = \alpha \times (\text{Maximum Payoff}) + (1 - \alpha) \times (\text{Minimum Payoff}) Hurwicz Payoff = α × ( Maximum Payoff ) + ( 1 − α ) × ( Minimum Payoff )
When α = 1.0 \alpha = 1.0 α = 1.0 , it equals Maximax; when α = 0.0 \alpha = 0.0 α = 0.0 , it equals Maximin.
4. The Minimax Regret (Savage) Criterion
Philosophy: Focuses on minimizing the post-decision regret (opportunity loss) of not having chosen the absolute best alternative for whichever state of nature actually occurs.
Method: Construct an Opportunity Loss (Regret) Matrix by subtracting each cell's payoff from the maximum payoff achievable under that specific state of nature. For each alternative, determine the maximum regret, then select the alternative that achieves the minimum of these maximum regrets .
3. Anatomy of a Decision Tree
A decision tree is a chronological, graphical model representing sequential decisions, random chance occurrences, and financial payoffs. It consists of three standardized geometric components: DECISION NODE (Square: □) --> Management exercises direct control over alternatives.
No probabilities are assigned to outgoing branches.
CHANCE NODE (Circle: ○) --> Uncertain states of nature outside management control.
Branches represent mutually exclusive outcomes.
Probabilities on branches must sum to exactly 1.0 (ΣP = 1.0).
TERMINAL NODE (Triangle: ▷)--> Endpoint of a pathway displaying the net monetary payoff
(profit, revenue, or net incurred cost).
Structural Rules of Decision Trees
Time Flow: The tree is drawn chronologically from left to right , starting at the primary decision node and branching toward terminal outcomes.
Branch Completeness: All paths originating from a chance node must be mutually exclusive and collectively exhaustive (i.e., ∑ i = 1 n P i = 1.00 \sum_{i=1}^{n} P_i = 1.00 ∑ i = 1 n P i = 1.00 ).
Payoff Integration: Every terminal endpoint must incorporate all intermediate revenues, capital outlays, and operating losses incurred along that specific trajectory.
4. Expected Monetary Value (EMV) & The Rollback Method
Expected Monetary Value (EMV) represents the probability-weighted average payoff of an uncertain event: E M V = ∑ i = 1 n ( P i × X i ) = ( P 1 X 1 ) + ( P 2 X 2 ) + ⋯ + ( P n X n ) EMV = \sum_{i=1}^{n} (P_i \times X_i) = (P_1 X_1) + (P_2 X_2) + \dots + (P_n X_n) EM V = i = 1 ∑ n ( P i × X i ) = ( P 1 X 1 ) + ( P 2 X 2 ) + ⋯ + ( P n X n )
Where P i P_i P i is the probability of outcome i i i and X i X_i X i is the financial payoff of outcome i i i .
The Rollback (Folding Back) Method
To solve a decision tree and identify the optimal path, practitioners utilize the rollback method , working strictly backwards from right to left (from terminal nodes back to the root decision node):
At each Chance Node ( ◯ \bigcirc ◯ ): Calculate the EMV by taking the probability-weighted sum of all branches originating to the right of the node.
At each Decision Node ( □ \square □ ): Compare the calculated values of all competing alternative branches:
If maximizing profit: Select the branch with the highest EMV .
If minimizing cost: Select the branch with the lowest expected cost .
Pruning: Mark rejected, suboptimal branches with double slash marks ( / / // // ) to indicate they are pruned from consideration.
Continue Backwards: Repeat the process until the origin decision node is evaluated, revealing the optimal decision strategy.
5. Expected Value of Perfect Information (EVPI)
In project controls, managers often have opportunities to buy additional data—such as performing geotechnical test borings, running pilot plant simulations, or engaging third-party legal audits—before making a final commitment. How much should an owner pay for this advance information?
The Expected Value of Perfect Information (EVPI) establishes the absolute theoretical ceiling on what a decision maker should spend for an infallible information source that eliminates all uncertainty.
The EVPI Formulation E V P I = Expected Value with Perfect Information (EVwPI) − Maximum EMV without Information (EMV base ) EVPI = \text{Expected Value with Perfect Information (EVwPI)} - \text{Maximum EMV without Information (EMV}_{\text{base}}\text{)} E V P I = Expected Value with Perfect Information (EVwPI) − Maximum EMV without Information (EMV base )
Where:
Expected Value with Perfect Information (EVwPI): The expected payoff if the decision maker could know with 100% certainty which state of nature would occur, choosing the best alternative for that specific state every time: E V w P I = ∑ j = 1 m [ P ( State j ) × ( Best Payoff in State j ) ] EVwPI = \sum_{j=1}^{m} [ P(\text{State}_j) \times (\text{Best Payoff in } \text{State}_j) ] E V wP I = j = 1 ∑ m [ P ( State j ) × ( Best Payoff in State j )]
EMV base \text{EMV}_{\text{base}} EMV base : The maximum expected value achievable using current knowledge without any special study.
Rule of Thumb: A manager should never pay more for an engineering study or testing campaign than its calculated EVPI. If a geotechnical survey costs $80,000 but the EVPI is $45,000, the survey is economically irrational.
6. Risk Attitudes & Utility Theory
EMV analysis assumes that the decision maker is risk-neutral —meaning they treat every expected dollar equally, regardless of the magnitude of potential loss. In real capital projects, organizations exhibit distinct risk profiles modeled by utility functions ( U ( x ) U(x) U ( x ) ). +-----------------------------------------------------------------------------------+
- | UTILITY THEORY & RISK ATTITUDES |
- +-------------------+---------------------------------------------------------------+
- | Risk-Neutral | Utility function is a linear straight line. Decision maker |
- | | relies strictly on EMV. Indifferent between guaranteed $50k |
- | | and a 50/50 chance at $0 or $100k. |
- +-------------------+---------------------------------------------------------------+
- | Risk-Averse | Utility function is strictly concave (diminishing marginal |
- | | utility of money). Prefers a guaranteed certain return over a |
- | | risky gamble with identical EMV. Buys insurance and hedging. |
- +-------------------+---------------------------------------------------------------+
- | Risk-Seeking | Utility function is strictly convex (increasing marginal |
- | | utility). Prefers an uncertain gamble over a guaranteed cash |
- | | amount equal to the gamble's EMV. |
- +-------------------+---------------------------------------------------------------+
- The Certainty Equivalent
The Certainty Equivalent ( C E CE CE ) is the guaranteed cash amount that a decision maker would accept in lieu of undertaking an uncertain risky gamble:
For a risk-averse individual: C E < E M V CE < EMV CE < EM V . The difference ( E M V − C E EMV - CE EM V − CE ) is the risk premium the individual is willing to sacrifice to eliminate volatility.
For a risk-neutral individual: C E = E M V CE = EMV CE = EM V (risk premium is zero).
For a risk-seeking individual: C E > E M V CE > EMV CE > EM V .
7. Comprehensive Worked Mathematical Example: Self-Perform vs. Subcontract
Scenario: A general contractor is awarded a $3,500,000 highway interchange contract. The earthwork package involves substantial subsurface excavation. The cost engineer must decide between two structural execution strategies:
Alternative 1 (Self-Perform): Direct execution cost is budgeted at $ 600 , 000 \text{\textdollar}600,000 $ 600 , 000 under normal soil conditions. However, there is a 30% probability ( P = 0.30 P = 0.30 P = 0.30 ) of encountering subsurface contaminated soil, which will require specialized environmental disposal, escalating total self-performance cost to $ 1 , 100 , 000 \text{\textdollar}1,100,000 $ 1 , 100 , 000 . Under clean soil conditions ( 70% probability, P = 0.70 P = 0.70 P = 0.70 ), direct cost remains $ 600 , 000 \text{\textdollar}600,000 $ 600 , 000 .
Alternative 2 (Fixed Subcontract): A specialty earthwork subcontractor offers a binding, lump-sum turnkey contract for $ 760 , 000 \text{\textdollar}760,000 $ 760 , 000 , absorbing all environmental contamination risks.
Step 1: Draw the Decision Tree Nodes
Decision Node ( □ \square □ ): Branch A = Self-Perform; Branch B = Subcontract.
Branch A Chance Node ( ◯ \bigcirc ◯ ):
Clean Soil ( P = 0.70 P = 0.70 P = 0.70 ): Cost = $ 600 , 000 \text{\textdollar}600,000 $ 600 , 000
Contaminated Soil ( P = 0.30 P = 0.30 P = 0.30 ): Cost = $ 1 , 100 , 000 \text{\textdollar}1,100,000 $ 1 , 100 , 000
- Branch B Terminal Outcome: Guaranteed Cost = $ 760 , 000 \text{\textdollar}760,000 $ 760 , 000
- Step 2: Compute Expected Monetary Values (EMVs)
Because this is a cost minimization problem, we calculate expected costs (negative payoffs): Expected Cost (Self-Perform) = ( 0.70 × $ 600 , 000 ) + ( 0.30 × $ 1 , 100 , 000 ) = $ 420 , 000 + $ 330 , 000 = $ 750 , 000 \text{Expected Cost (Self-Perform)} = (0.70 \times \$600,000) + (0.30 \times \$1,100,000) = \$420,000 + \$330,000 = \$750,000 Expected Cost (Self-Perform) = ( 0.70 × $600 , 000 ) + ( 0.30 × $1 , 100 , 000 ) = $420 , 000 + $330 , 000 = $750 , 000 Expected Cost (Subcontract) = $ 760 , 000 \text{Expected Cost (Subcontract)} = \$760,000 Expected Cost (Subcontract) = $760 , 000
Step 3: Roll Back to the Decision Node
Comparing expected costs:
- Self-Perform Expected Cost: $ 750 , 000 \text{\textdollar}750,000 $ 750 , 000
- Subcontract Fixed Cost: $ 760 , 000 \text{\textdollar}760,000 $ 760 , 000
Decision: Under a risk-neutral EMV criterion, the contractor should Self-Perform , yielding an expected cost savings of $ 10 , 000 \text{\textdollar}10,000 $ 10 , 000 ( $ 760 , 000 − $ 750 , 000 \text{\textdollar}760,000 - \text{\textdollar}750,000 $ 760 , 000 − $ 750 , 000 ). The subcontract branch is pruned.
Step 4: Break-Even Sensitivity Analysis
At what probability of contaminated soil ( p p p ) does subcontracting become the preferred economic choice? Expected Cost (Self-Perform) = ( 1 − p ) ( $ 600 , 000 ) + p ( $ 1 , 100 , 000 ) = 600 , 000 + 500 , 000 p \text{Expected Cost (Self-Perform)} = (1 - p)(\$600,000) + p(\$1,100,000) = 600,000 + 500,000p Expected Cost (Self-Perform) = ( 1 − p ) ( $600 , 000 ) + p ( $1 , 100 , 000 ) = 600 , 000 + 500 , 000 p
Set this equal to the fixed subcontract price of $ 760 , 000 \text{\textdollar}760,000 $ 760 , 000 : 600 , 000 + 500 , 000 p = 760 , 000 ⟹ 500 , 000 p = 160 , 000 ⟹ p = 160 , 000 500 , 000 = 0.32 ( 32.0 % 600,000 + 500,000p = 760,000 \implies 500,000p = 160,000 \implies p = \frac{160,000}{500,000} = 0.32 \quad (32.0\% 600 , 000 + 500 , 000 p = 760 , 000 ⟹ 500 , 000 p = 160 , 000 ⟹ p = 500 , 000 160 , 000 = 0.32 ( 32.0%
If the probability of contamination is less than 32% , Self-Perform is superior.
If the probability of contamination is greater than 32% , Subcontracting is superior.
Step 5: Calculate Expected Value of Perfect Information (EVPI)
Suppose an environmental testing agency can conduct advanced sonic core testing that reveals the exact soil condition with 100% certainty before the contract is awarded. What is the EVPI?
If Clean Soil occurs ( P = 0.70 P = 0.70 P = 0.70 ): Management chooses Self-Perform at a cost of $ 600 , 000 \text{\textdollar}600,000 $ 600 , 000 .
If Contaminated Soil occurs ( P = 0.30 P = 0.30 P = 0.30 ): Management chooses Subcontracting at a cost of $ 760 , 000 \text{\textdollar}760,000 $ 760 , 000 (avoiding the $ 1 , 100 , 000 \text{\textdollar}1,100,000 $ 1 , 100 , 000 self-perform blowout).
Expected Cost with Perfect Information (EVwPI): E V w P I = ( 0.70 × $ 600 , 000 ) + ( 0.30 × $ 760 , 000 ) = $ 420 , 000 + $ 228 , 000 = $ 648 , 000 EVwPI = (0.70 \times \$600,000) + (0.30 \times \$760,000) = \$420,000 + \$228,000 = \$648,000 E V wP I = ( 0.70 × $600 , 000 ) + ( 0.30 × $760 , 000 ) = $420 , 000 + $228 , 000 = $648 , 000
EVPI Calculation: E V P I = Base Expected Cost without Info − E V w P I = $ 750 , 000 − $ 648 , 000 = $ 102 , 000 EVPI = \text{Base Expected Cost without Info} - EVwPI = \$750,000 - \$648,000 = \$102,000 E V P I = Base Expected Cost without Info − E V wP I = $750 , 000 − $648 , 000 = $102 , 000
Strategic Implication: The contractor should be willing to pay up to $102,000 for an infallible geotechnical testing program to guide this execution decision.
8. Exam Traps & Decision Modeling Pitfalls
Warning
The Cost Minimization vs. Profit Maximization Inversion: Always check whether tree payoffs represent profits or costs! For profit trees, you select the highest EMV at a decision node. For cost trees, you select the lowest EMV. Candidates frequently solve the math correctly and then pick the highest number out of habit, recommending the most expensive execution plan!
Caution
Node Symbol Confusion: Proctors love testing basic symbology. Squares represent Decisions (points of active management choice where branches have no probabilities). Circles represent Chance/Event Nodes (points of nature where branches MUST carry probabilities summing to 1.0). Never put a probability on a branch leaving a square node!
Tip
Risk-Averse Bias in Real Practice: While textbook CCT questions assume risk-neutrality (picking the highest/lowest EMV), in actual construction, a contractor with limited working capital may gladly pay the $10,000 expected premium to subcontract the work, completely neutralizing the existential threat of an $1,100,000 overrun. In essay or scenario questions, mention that risk-averse leadership will often sacrifice EMV to cap downside loss.